Merge pull request #12: release v0.9.0 finalisation

Closes the v0.9.0 release loop on top of Phase 9a-Newton + Phase 8b-Lite:

* CHANGELOG.md (NEW) — Keep-A-Changelog format, with v0.9.0 entry
  detailing all Phase 9a / 9b / 8b-Lite contents and the doc-audit
  corrections that landed via PR #10.

* CITATION.cff — version 0.7.0 → 0.9.0, date 2026-05-18 → 2026-05-22.

* Stale HDS-port stubs removed (13 GTEST_SKIPs total):
  - code/tests/test_spherical_functional.cpp
  - code/tests/test_hyper_ideal_functional.cpp
  - code/tests/test_hyper_ideal_hyperelliptic_utility.cpp
  These referenced a "HDS port (Phase 4)" that never happened —
  CoHDS was intentionally replaced by CGAL::Surface_mesh, and the
  functional tests live in code/tests/cgal/test_*_functional.cpp.

* Test-count updates everywhere:
  - Non-CGAL  36 → 23  (drop = 13 deleted stubs)
  - CGAL      176 → 227
  - Total     212 → 250  (+38 net, 0 skipped)
  Files: README.md, CLAUDE.md, CHANGELOG.md, scripts/try_it.sh,
         doc/api/tests.md, doc/contributing.md, doc/getting-started.md,
         doc/math/validation.md.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

.gitea/docker/Dockerfile.ci-cpp

@@ -2,15 +2,19 @@ FROM --platform=linux/arm64 ubuntu:22.04
 
 # Node.js 20 from NodeSource (Ubuntu Jammy ships v12 which is too old
 # for actions/checkout@v4 — static class blocks require Node.js >= 16).
+#
+# libboost-dev  — header-only Boost required by CGAL 6.x (-DWITH_CGAL=ON)
 RUN apt-get update -qq && \
     apt-get install -y --no-install-recommends \
       curl ca-certificates && \
     curl -fsSL https://deb.nodesource.com/setup_20.x | bash - && \
     apt-get install -y --no-install-recommends \
       nodejs \
+      doxygen \
       cmake \
       build-essential \
       git \
+      libboost-dev \
     && rm -rf /var/lib/apt/lists/*
 
 WORKDIR /workspace

.gitea/workflows/cpp-tests.yml

@@ -6,10 +6,16 @@ on:
       - main
       - dev
       - "claude/**"
+      - "feature/**"
   pull_request:
 
+# ─────────────────────────────────────────────────────────────────────────────
+# Job 1 — test-fast
+#   Pure-math tests (Clausen, ImLi₂, Hyper-ideal geometry).
+#   No CGAL, no Boost. Eigen + GTest only. Runs on ALL branches.
+# ─────────────────────────────────────────────────────────────────────────────
 jobs:
-  test:
+  test-fast:
     runs-on: eulernest
     container:
       image: git.eulernest.eu/conformallab/ci-cpp:latest
@@ -17,11 +23,11 @@ jobs:
     steps:
       - uses: actions/checkout@v4
 
-      - name: Configure (tests-only mode)
+      - name: Configure (tests-only)
         run: cmake -S code -B build -DCMAKE_BUILD_TYPE=Release
 
-      - name: Build test binary
+      - name: Build
-        run: cmake --build build --target conformallab_tests -j$(nproc)
+        run: nice -n 19 cmake --build build --target conformallab_tests -j$(nproc)
 
       - name: Run tests
         run: >
@@ -29,7 +35,7 @@ jobs:
           --output-on-failure
           --output-junit test-results.xml
 
-      - name: Show test summary
+      - name: Summary
         if: always()
         run: |
           if [ -f test-results.xml ]; then
@@ -37,5 +43,56 @@ jobs:
             failed=$(grep -o 'failures="[0-9]*"' test-results.xml | grep -o '[0-9]*' | head -1)
             skipped=$(grep -o 'skipped="[0-9]*"' test-results.xml | grep -o '[0-9]*' | head -1)
             passed=$(( ${total:-0} - ${failed:-0} - ${skipped:-0} ))
-            echo "TOTAL: ${total:-0} | PASSED: $passed | FAILED: ${failed:-0} | SKIPPED: ${skipped:-0}"
+            echo "FAST  ▸ TOTAL ${total:-0} | PASSED $passed | FAILED ${failed:-0} | SKIPPED ${skipped:-0}"
+          fi
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Job 2 — test-cgal
+#   Full CGAL test suite (Phase 3–7, 158 tests).
+#   Runs ONLY on pull requests (not on direct pushes to dev/main).
+#   Starts only after test-fast succeeds.
+#
+#   Uses -DWITH_CGAL_TESTS=ON (not -DWITH_CGAL=ON) to avoid building
+#   Viewer/GLFW — the CI container has no wayland-scanner.
+#
+#   Boost (libboost-dev) is already present in the container since the image rebuild.
+# ─────────────────────────────────────────────────────────────────────────────
+  test-cgal:
+    needs: test-fast
+    if: github.event_name == 'pull_request'
+    runs-on: eulernest
+    container:
+      image: git.eulernest.eu/conformallab/ci-cpp:latest
+      # Memory bumped from 1400m → 1600m to avoid OOM during CGAL header
+      # compilation on ARM64 (CGAL + Eigen templates allocate ~700 MB per
+      # cc1plus instance; -j1 leaves a small margin).
+      # memory-swap == memory disables swap entirely so OOM fails fast
+      # rather than thrashing on the SD card.
+      options: "--memory=1600m --memory-swap=1600m"
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Configure (WITH_CGAL_TESTS — no viewer, no wayland-scanner)
+        run: cmake -S code -B build -DWITH_CGAL_TESTS=ON -DCMAKE_BUILD_TYPE=Release
+
+      - name: Build CGAL-Tests
+        run: nice -n 19 cmake --build build --target conformallab_cgal_tests -j1
+
+      - name: Run CGAL-Tests
+        run: >
+          ctest --test-dir build
+          -R "^cgal\."
+          --output-on-failure
+          --output-junit cgal-results.xml
+
+      - name: Summary
+        if: always()
+        run: |
+          if [ -f cgal-results.xml ]; then
+            total=$(grep -o 'tests="[0-9]*"'     cgal-results.xml | grep -o '[0-9]*' | head -1)
+            failed=$(grep -o 'failures="[0-9]*"' cgal-results.xml | grep -o '[0-9]*' | head -1)
+            skipped=$(grep -o 'skipped="[0-9]*"' cgal-results.xml | grep -o '[0-9]*' | head -1)
+            passed=$(( ${total:-0} - ${failed:-0} - ${skipped:-0} ))
+            echo "CGAL  ▸ TOTAL ${total:-0} | PASSED $passed | FAILED ${failed:-0} | SKIPPED ${skipped:-0}"
           fi

.gitea/workflows/doc-build.yaml

@@ -0,0 +1,56 @@
+name: API Docs
+
+on:
+  push:
+    branches:
+      - main
+  pull_request:
+
+# ─────────────────────────────────────────────────────────────────────────────
+# Doc-build — informational only
+#
+# Generates Doxygen HTML from the public headers and reports warning
+# statistics.  Does NOT block merges: `continue-on-error: true` ensures
+# warnings or extraction issues never fail the CI gate.  When Doxygen
+# coverage is denser (Phase 8c), this job can be promoted to a hard
+# requirement and the HTML deployed to Pages.
+#
+# Note: Gitea Actions on GHES does not support `actions/upload-artifact@v4`,
+# so HTML artifact upload is intentionally omitted.  The warning summary
+# in the job log is the primary reviewer signal; reviewers who want the
+# HTML can rebuild it locally with `cmake --build build --target doc`.
+# ─────────────────────────────────────────────────────────────────────────────
+jobs:
+  doc-build:
+    if: github.event_name == 'pull_request'
+    runs-on: eulernest
+    container:
+      image: git.eulernest.eu/conformallab/ci-cpp:latest
+    continue-on-error: true   # never block the merge
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Generate API documentation
+        run: doxygen Doxyfile 2>&1 | tee doxygen.log
+
+      - name: Summarise warnings
+        if: always()
+        run: |
+          if [ -f doc/doxygen/doxygen-warnings.log ]; then
+            warn=$(wc -l < doc/doxygen/doxygen-warnings.log)
+            echo "DOC ▸ Doxygen warnings: $warn"
+            echo ""
+            echo "First 20 warnings:"
+            head -20 doc/doxygen/doxygen-warnings.log
+          else
+            echo "DOC ▸ No warning log produced — check that Doxyfile WARN_LOGFILE points to doc/doxygen/doxygen-warnings.log"
+          fi
+
+      - name: Report HTML output
+        if: always()
+        run: |
+          if [ -d doc/doxygen/html ]; then
+            files=$(find doc/doxygen/html -type f | wc -l)
+            size=$(du -sh doc/doxygen/html | cut -f1)
+            echo "DOC ▸ HTML output: $files files, $size total"
+          fi

.gitea/workflows/mirror-to-codeberg.yml

@@ -4,7 +4,6 @@ on:
   push:
     branches:
       - main
-      - dev
 
 jobs:
   mirror:
@@ -14,10 +13,11 @@ jobs:
       - name: Mirror all branches to Codeberg
         env:
           CODEBERG_TOKEN: ${{ secrets.CODEBERG_TOKEN }}
+          GITEA_MIRROR_TOKEN:  ${{ secrets.MIRROR_TOKEN }}
         run: |
           git clone --bare \
-            https://oauth2:${GITHUB_TOKEN}@git.eulernest.eu/conformallab/ConformalLabpp.git \
+            https://oauth2:${GITEA_MIRROR_TOKEN}@git.eulernest.eu/conformallab/ConformalLabpp.git \
             repo.git
           cd repo.git
           git push --mirror \
-            https://TMoussa:${CODEBERG_TOKEN}@codeberg.org/TMoussa/ConformalLabpp.git
+            https://TMoussa:${CODEBERG_TOKEN}@codeberg.org/TMoussa/ConformalLabpp.git

.gitignore

@@ -0,0 +1,33 @@
+# macOS
+.DS_Store
+.AppleDouble
+.LSOverride
+
+# Build directories
+build/
+build-*/
+build_*/
+code/build*/
+
+# CMake
+CMakeCache.txt
+CMakeFiles/
+cmake_install.cmake
+CTestTestfile.cmake
+
+# Test output
+*.xml
+Testing/
+
+# IDE
+.idea/
+.vscode/
+*.user
+*.suo
+
+# Claude Code worktrees
+.claude/
+
+# Doxygen output
+doc/doxygen/
+*.dox.tmp

CHANGELOG.md

@@ -0,0 +1,153 @@
+# Changelog
+
+All notable changes to **conformallab++** are recorded here.  Format
+follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/); the
+project uses [Semantic Versioning](https://semver.org).
+
+---
+
+## [0.9.0] — 2026-05-22
+
+The “DCE-complete + CGAL-surface-complete” release.  Two new discrete-
+conformal models, the analytic-Hessian optimisation for HyperIdeal, the
+CGAL public API surface for all five models, and a full documentation
+audit that corrects four pre-existing port-vs-research mis-labels.
+
+### Added — new functionals (Phase 9a)
+
+* `code/include/cp_euclidean_functional.hpp` —
+  **CP-Euclidean** functional (face-based circle packing),
+  Bobenko-Pinkall-Springborn 2010.  Direct port of
+  `CPEuclideanFunctional.java` (260 Java lines + 88-line test).
+  Analytic 2×2-per-edge Hessian `h_jk = sin θ / (cosh Δρ − cos θ)`.
+* `code/include/inversive_distance_functional.hpp` —
+  **Inversive-Distance** functional (vertex-based, Luo 2004 + Glickenstein
+  2011).  No Java original — implemented from the literature with
+  Bowers-Stephenson 2004 initialisation.  Cross-validated against the
+  Euclidean functional at the natural initial geometry (Glickenstein §5).
+
+### Added — Newton solvers (Phase 9a-Newton)
+
+* `newton_cp_euclidean()` — uses the analytic Hessian.
+* `newton_inversive_distance()` — uses FD Hessian; analytic via
+  Glickenstein 2011 eq. (4.6) tracked in `research-track.md` as
+  Phase 9a.2-analytic.
+
+### Added — Hessian optimisation (Phase 9b)
+
+* `hyper_ideal_hessian_block_fd()` — per-face 6×6 block-local Hessian
+  for HyperIdeal.  **96.5× speed-up measured on a 200-face mesh
+  (V=202, 603 DOFs)**, full-FD 226 ms → block-FD 2.3 ms.
+* Java parity note: `HyperIdealFunctional.java:295-298` declares
+  `hasHessian() == false`; both FD variants are conformallab++
+  research extensions beyond the Java port.
+
+### Added — CGAL public API surface (Phase 8b-Lite)
+
+* `<CGAL/Discrete_conformal_map.h>` extended with
+  `discrete_conformal_map_spherical()` and
+  `discrete_conformal_map_hyper_ideal()`.
+* `<CGAL/Discrete_circle_packing.h>` — `Default_cp_euclidean_traits` +
+  `discrete_circle_packing_euclidean()`.
+* `<CGAL/Discrete_inversive_distance.h>` — `Default_inversive_distance_traits`
+  + `discrete_inversive_distance_map()`.
+* `<CGAL/Conformal_layout.h>` — thin CGAL-namespace re-exports of
+  `euclidean_layout`, `spherical_layout`, `hyper_ideal_layout`.
+
+All five DCE models are now reachable from a single
+`#include <CGAL/Discrete_*.h>`.
+
+### Added — documentation
+
+* `doc/roadmap/research-track.md` — new consolidated catalogue of
+  every conformallab++ item that goes beyond the Java port, with full
+  literature citations and acceptance criteria.  Includes the
+  Phase 9b-analytic plan (Schläfli 1858 + Springborn 2020 §4 +
+  Cho-Kim 1999 + Glickenstein 2011 §4).
+* `doc/architecture/phase-9a-validation.md` — line-by-line mapping
+  CPEuclideanFunctional.java ↔ C++ port, plus three special-case
+  verifications of Luo’s edge-length formula.
+* `doc/roadmap/phases.md` — Phase 9 split into 9a.1 (Java port) /
+  9a.2 (research) / 9b (research); new Phase 11+ section with
+  optional Schottky uniformisation and Riemann-map sub-packages.
+* `doc/math/references.md` — five new primary literature entries
+  (Bowers-Stephenson 2004, Glickenstein 2011, BPS 2010,
+  Schläfli 1858/60, plus a reframed Luo 2004 entry).
+
+### Changed
+
+* **Four port-vs-research mis-labels** corrected (full audit
+  documented in `research-track.md`):
+  - `InversiveDistanceFunctional.java` does not exist in the Java
+    repo; the C++ implementation is research, not a port.
+  - HyperIdeal Hessian: Java has `hasHessian()==false`; C++ Hessians
+    are research, not ports.
+  - `add-inversive-distance.md` tutorial rewritten end-to-end.
+  - `references.md` and `java-parity.md` reframed.
+* `Discrete_conformal_map.h` (Phase 8a MVP wrapper) now deduces the
+  kernel from `TriangleMesh::Point` via `CGAL::Kernel_traits` rather
+  than hard-coding `Simple_cartesian<double>`.  Regression-guarded by
+  `KernelIsDeducedFromMeshPointType` test.
+
+### Removed
+
+* Three stale stub test files in `code/tests/` (15 GTEST_SKIPs total):
+  - `test_spherical_functional.cpp`
+  - `test_hyper_ideal_functional.cpp`
+  - `test_hyper_ideal_hyperelliptic_utility.cpp`
+  They referenced a "HDS port (Phase 4)" that never happened —
+  CoHDS was intentionally replaced by `CGAL::Surface_mesh`, and the
+  functional tests live in `code/tests/cgal/test_*_functional.cpp`.
+
+### CI / Infrastructure
+
+* `.gitea/workflows/cpp-tests.yml` — test-cgal memory fixed
+  (1400→1600 MB, `-j2 → -j1`).  Addresses OOM on ARM64 runner.
+* `.gitea/workflows/doc-build.yaml` — soft-fail Doxygen job
+  (no merge-blocking).
+* `Doxyfile` + CMake `doc` target — `cmake --build build --target doc`.
+* 12 macOS Finder-duplicate files removed from `code/include/`.
+
+### Test counts
+
+```
+v0.7.0:  176 CGAL + 36 non-CGAL = 212 total,  13 skipped (HDS stubs)
+v0.9.0:  227 CGAL + 23 non-CGAL = 250 total,   0 skipped   (+38 net, +51 CGAL)
+```
+
+Non-CGAL count dropped from 36 → 23 because three stale HDS-port stubs
+were removed (see "Removed" above) — the functionality is fully covered
+in the CGAL test suite where it actually lives.
+
+Five test suites added: `CGALConformalTraits`, `CGALDiscreteConformalMap`,
+`CPEuclideanFunctional`, `InversiveDistanceFunctional`, `HyperIdealHessian`,
+`NewtonPhase9a`, `CGALPhase8bLite`.
+
+---
+
+## [0.7.0] — 2026-05-18
+
+The “mathematician-ready” release.  See the v0.7.0 announcement in
+README.md (legacy) or `CITATION.cff` for the corresponding citation
+entry.  Phases 1–7 complete: three DCE geometry modes (Euclidean /
+Spherical / HyperIdeal), Newton solver, BFS-trilateration layout,
+Gauss-Bonnet, tree-cotree cut graph, Möbius holonomy, period matrix
+for genus 1, fundamental domain (genus 1), texture atlas.
+
+---
+
+## How to update this file
+
+Every new release adds a new top-level section above the previous one.
+For non-trivial PRs that don't trigger a release, add an entry under
+an `[Unreleased]` section at the top; promote it to the next release
+header at tag time.
+
+Categories (Keep-A-Changelog convention):
+
+* **Added** — new features / files / public APIs.
+* **Changed** — behaviour-altering changes to existing features.
+* **Deprecated** — features still present but slated for removal.
+* **Removed** — deleted features / files.
+* **Fixed** — bug fixes.
+* **Security** — security-relevant fixes.

CITATION.cff

@@ -0,0 +1,68 @@
+cff-version: 1.2.0
+message: "If you use this software in your research, please cite it as below."
+
+authors:
+  - family-names: Moussa
+    given-names: Tarik
+    email: Tarik.moussa95@gmail.com
+
+title: "conformallab++"
+version: 0.9.0
+date-released: 2026-05-22
+url: "https://codeberg.org/TMoussa/ConformalLabpp"
+repository-code: "https://codeberg.org/TMoussa/ConformalLabpp"
+license: MIT
+
+abstract: >
+  conformallab++ is a C++17 implementation of discrete conformal maps on
+  triangulated surfaces, covering Euclidean, Spherical, and Hyper-ideal geometry
+  modes. It provides a Newton solver for discrete conformal equivalence (DCE),
+  tree-cotree cut graphs, Möbius holonomy, period matrix computation with
+  SL(2,ℤ) reduction, and fundamental domain construction. The long-term goal is
+  a CGAL package for discrete conformal geometry.
+
+keywords:
+  - discrete conformal geometry
+  - conformal maps
+  - surface parameterization
+  - period matrix
+  - Teichmüller theory
+  - CGAL
+  - C++
+
+references:
+  - type: thesis
+    authors:
+      - family-names: Sechelmann
+        given-names: Stefan
+    title: >
+      Variational Methods for Discrete Surface Parameterization:
+      Applications and Implementation
+    institution:
+      name: Technische Universität Berlin
+    year: 2016
+    doi: 10.14279/depositonce-5415
+    notes: "Primary algorithmic source for this implementation"
+
+  - type: article
+    authors:
+      - family-names: Springborn
+        given-names: Boris
+    title: "Ideal Hyperbolic Polyhedra and Discrete Uniformization"
+    journal: "Discrete & Computational Geometry"
+    year: 2020
+    doi: 10.1007/s00454-019-00132-8
+    notes: "Mathematical basis for the HyperIdeal geometry mode"
+
+  - type: article
+    authors:
+      - family-names: Bobenko
+        given-names: Alexander I.
+      - family-names: Springborn
+        given-names: Boris A.
+    title: >
+      Variational Principles for Circle Patterns and Koebe's Theorem
+    journal: "Transactions of the American Mathematical Society"
+    year: 2004
+    doi: 10.1090/S0002-9947-03-03239-2
+    notes: "Variational framework underlying all three geometry modes"

CLAUDE.md

@@ -0,0 +1,392 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Project purpose and long-term goal
+
+conformallab++ is a C++17 reimplementation of [ConformalLab](https://github.com/varylab/conformallab) — Stefan Sechelmann's Java research library for discrete conformal geometry (TU Berlin, ~850 commits, v1.0.0 2018). The algorithmic foundation is his dissertation:
+
+> Stefan Sechelmann — *Variational Methods for Discrete Surface Parameterization: Applications and Implementation*, TU Berlin 2016.
+> DOI: [10.14279/depositonce-5415](https://depositonce.tu-berlin.de/items/8e2988b2-d991-45b5-aad5-9fb7988f3b2f) · CC BY-SA 4.0
+
+**The long-term goal is a CGAL package** — a submission to the CGAL library that brings discrete conformal maps (hyper-ideal, spherical, Euclidean) to the CGAL ecosystem using `CGAL::Surface_mesh` as the underlying halfedge data structure, with a traits-class design compatible with arbitrary CGAL-conforming mesh types.
+
+The project has three distinct phases:
+- **Phase 1–7 (done):** Direct port of the Java library algorithms to C++
+- **Phase 8–9 (planned):** CGAL package infrastructure + remaining Java features not yet ported (inversive-distance functional, analytic HyperIdeal Hessian, genus-g > 1 fundamental domain)
+- **Phase 10+ (research):** New mathematics beyond the Java original — holomorphic differentials, Siegel period matrix Ω ∈ H_g, full uniformization for genus g ≥ 2
+
+## Language
+
+**All code, comments, documentation, commit messages, and test descriptions must be in English.** The project is intended for international collaboration and CGAL submission. Existing German-language comments in older files should be replaced with English when editing those files.
+
+## Build commands
+
+All source lives under `code/`. Three build modes:
+
+```bash
+# Mode 1 — fast tests, no CGAL, no Boost, no display (CI default)
+cmake -S code -B build
+cmake --build build --target conformallab_tests -j$(nproc)
+ctest --test-dir build --output-on-failure
+
+# Mode 2 — CGAL tests, headless (CI full, requires Boost headers only)
+#   macOS: brew install boost    Linux: apt install libboost-dev
+cmake -S code -B build -DWITH_CGAL_TESTS=ON
+cmake --build build --target conformallab_cgal_tests -j$(nproc)
+ctest --test-dir build -R "^cgal\." --output-on-failure
+
+# Mode 3 — full local build: CLI app + viewer + examples (requires Wayland/X11)
+cmake -S code -B build -DWITH_CGAL=ON
+cmake --build build -j$(nproc)
+```
+
+`-DWITH_CGAL=ON` automatically enables `-DWITH_VIEWER=ON`, which pulls in GLFW and requires `wayland-scanner`. Never use this in headless CI.
+
+### Running a single test
+
+```bash
+# By GTest suite/test name
+./build/conformallab_cgal_tests --gtest_filter="NewtonSolver*"
+./build/conformallab_tests       --gtest_filter="Clausen*"
+
+# By CTest regex (prefix "cgal." for all CGAL tests)
+ctest --test-dir build -R "cgal.NewtonSolver" --output-on-failure
+```
+
+### Rebuilding the CI Docker image
+
+```bash
+docker buildx build \
+  --platform linux/arm64 \
+  -f .gitea/docker/Dockerfile.ci-cpp \
+  -t git.eulernest.eu/conformallab/ci-cpp:latest \
+  --push \
+  .gitea/docker/
+```
+
+## Architecture
+
+### Everything is header-only
+
+All algorithms live in `code/include/*.hpp`. There is no compiled library. The three CMake targets (`conformallab_tests`, `conformallab_cgal_tests`, `conformallab_core`) compile headers directly from their `.cpp` entry points. To add a new algorithm: create a `.hpp` in `code/include/`, add a test in `code/tests/cgal/`, and register the test file in `code/tests/cgal/CMakeLists.txt`.
+
+### Central type: `ConformalMesh`
+
+`conformal_mesh.hpp` defines the core type:
+```cpp
+using ConformalMesh = CGAL::Surface_mesh<Point3>;  // CGAL::Simple_cartesian<double>
+```
+
+This replaces the Java `CoHDS` (half-edge data structure) and its intrusive `CoVertex`/`CoEdge`/`CoFace` types. Data is attached via named CGAL property maps instead of intrusive fields:
+
+| Property map name | Type | Meaning |
+|---|---|---|
+| `"v:lambda"` | `double` per vertex | log scale factor (conformal variable uᵢ) |
+| `"v:theta"` | `double` per vertex | target cone angle Θᵥ |
+| `"v:idx"` | `int` per vertex | solver DOF index; `-1` = pinned/boundary |
+| `"e:alpha"` | `double` per edge | intersection angle αᵢⱼ (hyperbolic only) |
+| `"f:type"` | `int` per face | geometry type (0=Euclidean, 1=Hyperbolic, 2=Spherical) |
+
+`CGAL_DISABLE_GMP` and `CGAL_DISABLE_MPFR` are defined for all CGAL targets — the library deliberately uses `Simple_cartesian<double>` (floating-point, no exact arithmetic) because conformal geometry does not require exact predicates.
+
+### The three geometry modes
+
+Each mode has its own Maps struct that bundles all property maps, plus functional, Hessian, and Newton function:
+
+| Mode | Space | Maps struct | Key headers | Newton function |
+|---|---|---|---|---|
+| Euclidean | ℝ² | `EuclideanMaps` | `euclidean_functional.hpp`, `euclidean_hessian.hpp` | `newton_euclidean()` |
+| Spherical | S² | `SphericalMaps` | `spherical_functional.hpp`, `spherical_hessian.hpp` | `newton_spherical()` |
+| Hyper-ideal | H² (Poincaré disk) | `HyperIdealMaps` | `hyper_ideal_functional.hpp`, `hyper_ideal_hessian.hpp` | `newton_hyper_ideal()` |
+
+HyperIdeal also has edge DOFs (`e_idx[e]`); Euclidean and Spherical are vertex-DOF only. For HyperIdeal: `assign_all_dof_indices(mesh, maps)` assigns all vertex and edge DOFs automatically. For Euclidean/Spherical: pin one vertex manually (`maps.v_idx[first_vertex] = -1`) then assign sequential indices.
+
+### The full pipeline
+
+```
+load_mesh()                          → ConformalMesh  (OFF/OBJ/PLY)
+setup_*_maps(mesh)                   → *Maps  (property maps created, all zero)
+compute_*_lambda0_from_mesh(mesh, m) → λ° initialised from 3-D edge lengths
+DOF assignment                       → v_idx[v] set; -1 = pinned
+check_gauss_bonnet(mesh, maps)       → throws if Σ(2π−Θᵥ) ≠ 2π·χ(M)
+enforce_gauss_bonnet(mesh, maps)     → redistributes angle defect uniformly
+newton_*(mesh, x0, maps)             → NewtonResult{x*, iterations, converged}
+compute_cut_graph(mesh)              → CutGraph  (2g seam edges, tree-cotree)
+*_layout(mesh, x*, maps, &cg, &hol) → Layout2D/3D  + HolonomyData
+normalise_*(layout)                  → canonical position (PCA / Möbius / Rodrigues)
+compute_period_matrix(hol)           → PeriodData{τ∈ℍ}  (genus 1 flat torus)
+compute_fundamental_domain(hol)      → FundamentalDomain{vertices, generators}
+tiling_neighbourhood(layout, hol)    → vector of translated layout copies
+save_result_json/xml()               → serialised result
+```
+
+After `compute_*_lambda0_from_mesh()` the original vertex positions are no longer used — all subsequent computation is in log-length/scale-factor space.
+
+### Newton solver (`newton_solver.hpp`)
+
+The gradient sign convention differs between modes:
+- **Euclidean/Spherical:** `G_v = Θ_v − actual_angle_sum` (target minus actual)
+- **HyperIdeal:** `G_v = actual_angle_sum − Θ_v` (actual minus target)
+
+Hessian sign and solver:
+- **Euclidean:** H is PSD (cotangent Laplacian) → `SimplicialLDLT(H)`
+- **Spherical:** H is NSD (concave energy) → `SimplicialLDLT(−H)` (sign flip inside `newton_spherical`)
+- **HyperIdeal:** H is PSD (strictly convex) → `SimplicialLDLT(H)`
+
+When `SimplicialLDLT` fails (rank-deficient H — gauge mode on closed mesh without pinned vertex), the solver automatically retries with `Eigen::SparseQR` to find the minimum-norm step orthogonal to the null space. Public API: `solve_linear_system(H, rhs, &used_fallback)`.
+
+The HyperIdeal Hessian is currently a **symmetric finite-difference approximation** (O(ε²), costs n extra gradient evaluations per Newton step). The analytic Hessian via the chain `(bᵢ, aₑ) → lᵢⱼ → ζ₁₃/ζ₁₄/ζ₁₅ → αᵢⱼ/βᵢ` is deferred to Phase 9b.
+
+### Layout and holonomy (`layout.hpp`)
+
+BFS-trilateration with a **priority min-heap on BFS depth** (`depth = max(depth[src], depth[tgt]) + 1`). Root face = largest 3-D area face. This minimises trilateration error accumulation compared to simple BFS.
+
+Key output fields:
+- `layout.uv[v.idx()]` — primary UV (first/shallowest BFS visit per vertex)
+- `layout.halfedge_uv[h.idx()]` — UV of `source(h)` as seen from `face(h)`; at seam halfedges the two opposite halfedges carry *different* UV values, enabling proper GPU texture atlasing without vertex duplication
+- `hol.translations[i]` — lattice generator ωᵢ ∈ ℂ (Euclidean/spherical)
+- `hol.mobius_maps[i]` — Möbius isometry Tᵢ ∈ SU(1,1) (hyperbolic, Poincaré disk)
+
+`MobiusMap` is defined in `layout.hpp`: T(z) = (az+b)/(cz+d). Key methods: `from_three()` (fit to 3 point correspondences via 3×3 complex linear system), `compose()`, `inverse()`, `apply(Vector2d)`.
+
+### Key mathematical reference for each header
+
+| Header | Java original | Key reference |
+|---|---|---|
+| `hyper_ideal_geometry.hpp` | `HyperIdealGeometry.java` | Springborn (2020) — ζ₁₃/ζ₁₄/ζ₁₅ functions |
+| `euclidean_hessian.hpp` | `EuclideanHessian.java` | Pinkall & Polthier (1993) — cotangent Laplacian |
+| `spherical_hessian.hpp` | `SphericalHessian.java` | ∂α/∂u from spherical law of cosines |
+| `cut_graph.hpp` | `CuttingUtility.java` | Erickson & Whittlesey (SODA 2005) — tree-cotree |
+| `period_matrix.hpp` | `PeriodMatrixUtility.java` | Sechelmann (2016) §4 — SL(2,ℤ) reduction |
+| `gauss_bonnet.hpp` | (distributed across Java) | Gauss–Bonnet: Σ(2π−Θᵥ) = 2π·χ(M) |
+
+### Java features not yet ported (Phase 9)
+
+The Java library under `de.varylab.discreteconformal` contains these items not yet in C++:
+
+| Java class | Planned C++ header | Phase |
+|---|---|---|
+| `InversiveDistanceFunctional` | `inversive_distance_functional.hpp` | 9a |
+| Analytic HyperIdeal Hessian | `hyper_ideal_hessian.hpp` (replace FD) | 9b |
+| 4g-polygon boundary walk in `FundamentalDomainUtility` | `fundamental_domain.hpp` (extend) | 9c |
+| `DiscreteHarmonicFormUtility` | Phase 10a prerequisite | 10 |
+| `DiscreteHolomorphicFormUtility` | Phase 10a | 10 |
+| `HomologyUtility`, `CanonicalBasisUtility` | Phase 10 | 10 |
+
+When porting a Java class, locate the original in `de.varylab.discreteconformal.*` at [github.com/varylab/conformallab](https://github.com/varylab/conformallab) and use it as the reference implementation.
+
+## Test design patterns
+
+### "Natural theta" — constructing a known equilibrium at x* = 0
+
+```cpp
+// Evaluate gradient at x=0; set target angles = actual angle sums → x*=0 by definition
+std::vector<double> x0(n_dofs, 0.0);
+auto G0 = euclidean_gradient(mesh, x0, maps);
+for (auto v : mesh.vertices())
+    if (maps.v_idx[v] >= 0)
+        maps.theta_v[v] -= G0[maps.v_idx[v]];  // shift so G(x=0) = 0
+```
+
+This is used in virtually every Newton convergence test — it avoids hardcoding specific angle values.
+
+### Gradient check pattern
+
+```cpp
+// Copy from any test_*_functional.cpp — GradientCheck_* test suite
+double eps = 1e-5;
+for (int i = 0; i < n; ++i) {
+    xp[i] += eps;  auto Gp = euclidean_gradient(mesh, xp, maps);
+    xm[i] -= eps;  auto Gm = euclidean_gradient(mesh, xm, maps);
+    double fd = (energy(xp) - energy(xm)) / (2*eps);
+    EXPECT_NEAR(G[i], fd, 1e-7);
+    xp[i] = xm[i] = x0[i];
+}
+```
+
+All new functionals must have a gradient-check test before being considered complete.
+
+### Halfedge traversal
+
+```cpp
+for (auto f : mesh.faces()) {
+    auto h0 = mesh.halfedge(f);          // canonical halfedge of face
+    auto h1 = mesh.next(h0);
+    auto h2 = mesh.next(h1);
+
+    Vertex_index v1 = mesh.source(h0);  // = mesh.target(h2)
+    Vertex_index v2 = mesh.source(h1);
+    Vertex_index v3 = mesh.source(h2);
+
+    // Angle at v3 is opposite to h0 (edge v1–v2)
+    // h_alpha[h0] = α₃,  h_alpha[h1] = α₁,  h_alpha[h2] = α₂
+
+    bool is_boundary = mesh.is_border(mesh.opposite(h0));
+}
+```
+
+### Attaching custom data to the mesh
+
+```cpp
+auto [my_map, created] = mesh.add_property_map<Vertex_index, double>("v:my_data", 0.0);
+my_map[v] = 3.14;
+```
+
+## CI pipeline
+
+Two jobs in `.gitea/workflows/cpp-tests.yml`:
+
+| Job | CMake flags | Deps | Triggers on |
+|---|---|---|---|
+| `test-fast` | *(none)* | Eigen + GTest only | all branches |
+| `test-cgal` | `-DWITH_CGAL_TESTS=ON` | + Boost | pull requests only |
+
+Runner: `eulernest` — self-hosted Raspberry Pi, ARM64, Ubuntu 22.04. Docker image: `git.eulernest.eu/conformallab/ci-cpp:latest`. `test-cgal` needs `test-fast` to pass first (`needs: test-fast`).
+
+Expected results: **23 non-CGAL tests pass**, **227 CGAL tests pass, 0 skipped**.
+
+## Release state
+
+Current release: **v0.7.0** (tag on `origin/dev`, PR to `main` open).  
+Phase 7 is complete. Phase 7.5 (Doxygen) and Phase 8 (CGAL package) are next.
+
+## Phase 8 strategic decisions (2026-05-19)
+
+The CGAL-package architecture was frozen on 2026-05-19.  Full design:
+[`doc/api/cgal-package.md`](doc/api/cgal-package.md).  Key decisions:
+
+| Decision | Choice |
+|---|---|
+| Submission to upstream CGAL | **Pre-submission-ready, not bound.** 12+ months horizon. |
+| License | **MIT preserved** (no LGPL switch). |
+| Mesh-type flexibility | **Generic `FaceGraph + HalfedgeGraph`** in target design; MVP starts Surface_mesh-only. |
+| Parameter style | **Named Parameters** (`CGAL::parameters::...`). |
+| Default kernel | **`Simple_cartesian<double>`** (status quo). |
+| Backward compatibility | **Dual-layer wrapper** — `code/include/*.hpp` stays as implementation, `include/CGAL/*.h` is thin wrapper.  No algorithm duplication. |
+| Implementation strategy | **Hybrid MVP** — minimum Phase 8 (traits + one wrapper) first, then Phase 9 in full, then Phase 8 extensions only on concrete demand. |
+| Phase-8 MVP acceptance test | **Phase 9a (Inversive-Distance)** as the first new client of the new traits API. |
+
+### Implementation sequence (committed)
+
+```
+1. Phase 7.5  Doxygen + cleanup                         done ✅
+2. Phase 8 MVP — traits + one euclidean wrapper         3–5 days
+3. Phase 9a — Inversive-Distance against new traits     3–5 days
+4. Phase 9b — analytic HyperIdeal Hessian               1 week
+5. Phase 9c — 4g-polygon for genus g > 1                1 week
+   → port really complete, v0.9.0 release
+```
+
+Phase 8 extensions (8a.2 generic FaceGraph, 8c full Doxygen manuals, 8d
+CGAL-format tests, 8e YAML pipeline) are deferred to on-demand status —
+no speculative architecture for an uncertain submission.
+
+Root-level files added at v0.7.0:
+- `CITATION.cff` — machine-readable citation (Sechelmann 2016, Springborn 2020, Bobenko–Springborn 2004)
+- `CONTRIBUTING.md` — short root-level pointer to `doc/contributing.md`
+- `scripts/try_it.sh` — one-script quickstart: build → 209 tests → example run
+- CMake install target: `cmake --install build --prefix /usr/local` → headers land in `include/conformallab/`
+
+## Port-vs-research maintenance rule (2026-05-21 audit)
+
+Before claiming something "ports X from Java", **verify empirically**:
+
+```bash
+find /Users/tarikmoussa/Desktop/conformallab -iname "*X*"
+grep -r "ClassName" /Users/tarikmoussa/Desktop/conformallab/src
+```
+
+If zero matches, the work is **new research** — add it to
+`doc/roadmap/research-track.md` with primary literature citations,
+**not** to `doc/roadmap/java-parity.md`.
+
+The 2026-05-21 audit found four pre-existing mis-labels:
+
+| Item | Wrong claim | Reality |
+|---|---|---|
+| `InversiveDistanceFunctional` | "Java port (Luo 2004)" | No such Java class exists |
+| HyperIdeal Hessian (FD) | "Phase 4a" | Research — Java has `hasHessian()==false` |
+| HyperIdeal Hessian (analytic) | "Phase 9b port" | Research — derivation via Schläfli 1858 |
+| Tutorial framing | "ports `InversiveDistanceFunctional.java`" | Implementation from Luo 2004 + Glickenstein 2011 |
+
+All four are corrected as of this commit.  Future contributors must
+follow the empirical verification rule above before any new claim.
+
+## Documentation map
+
+24 documents across 6 categories. Read the relevant one before reasoning from scratch
+— do not hallucinate content that is already written down.
+
+### Mathematics & theory
+
+| Question | Document |
+|---|---|
+| What problem does this library solve mathematically? | `doc/math/discrete-conformal-theory.md` |
+| How do the three geometry modes differ (Euclidean/Spherical/HyperIdeal)? | `doc/math/geometry-modes.md` |
+| What analytic invariants can be used to validate correctness? | `doc/math/validation.md` |
+| What are the exact ctest commands with expected terminal output? | `doc/math/validation-protocol.md` |
+| What is the O() complexity and how does it scale with mesh size? | `doc/math/complexity.md` |
+| Which papers are referenced by which header? | `doc/math/references.md` |
+| How does conformallab++ compare to libigl, CGAL, geometry-central, pmp-library? | `doc/math/software-landscape.md` |
+| What is unique about conformallab++ (novelty, target audience)? | `doc/math/novelty-statement.md` |
+
+### Architecture & design
+
+| Question | Document |
+|---|---|
+| Full pipeline diagram and data-flow overview | `doc/architecture/overall_pipeline.md` |
+| Directory tree, build targets, file organisation | `doc/architecture/project-structure.md` |
+| Key architectural decisions and their rationale | `doc/architecture/design-decisions.md` |
+| Detailed comparison with geometry-central (CMU): overlap, adoption, scientific value | `doc/architecture/geometry-central-comparison.md` |
+| Phase 9a validation report (CP-Euclidean port + Luo-inversive-distance literature check) | `doc/architecture/phase-9a-validation.md` |
+
+### API & extension
+
+| Question | Document |
+|---|---|
+| All 24 public headers with descriptions | `doc/api/headers.md` |
+| Full pipeline API for all three geometries | `doc/api/pipeline.md` |
+| What does each processing unit require/provide (contracts)? | `doc/api/contracts.md` |
+| How to add a new functional / geometry mode / port from Java | `doc/api/extending.md` |
+| All 39 test suites, 227+23 tests, individual counts | `doc/api/tests.md` |
+| Phase 8 CGAL package design + Declarative YAML pipeline spec | `doc/api/cgal-package.md` |
+
+### Concepts & specs
+
+| Question | Document |
+|---|---|
+| Declarative YAML pipeline: token vocabulary, 5 examples, validation algorithm | `doc/concepts/declarative-pipeline.md` |
+
+### Roadmap & porting
+
+| Question | Document |
+|---|---|
+| Phases 1–10 with status and sub-tasks | `doc/roadmap/phases.md` |
+| Which Java classes are ported, which are planned, which are skipped? | `doc/roadmap/java-parity.md` |
+| New research items (beyond Java) — citations, acceptance criteria | `doc/roadmap/research-track.md` |
+
+### Tutorials & onboarding
+
+| Question | Document |
+|---|---|
+| Build modes, single-test invocation, CLI, Docker image rebuild | `doc/getting-started.md` |
+| Step-by-step: port the Inversive Distance functional (Phase 9a template) | `doc/tutorials/add-inversive-distance.md` |
+| Language policy, test standards, release flow | `doc/contributing.md` |
+
+### geometry-central context
+
+**geometry-central** (Keenan Crane, CMU) implements the same discrete conformal
+equivalence problem (Gillespie, Springborn & Crane, SIGGRAPH 2021) but uses
+Ptolemaic flips on intrinsic triangulations instead of Newton on the original mesh.
+It has no period matrix, holonomy, or spherical geometry mode.
+The shared mathematical core (Springborn 2020) means cross-validation is meaningful.
+Full analysis: `doc/architecture/geometry-central-comparison.md`.
+Optional adoption roadmap (GC-1/2/3): `doc/roadmap/phases.md` (Optional section).
+
+## Known quirks
+
+- **`test-fast` also runs stubs**: `conformallab_tests` (non-CGAL) contains `GTEST_SKIP`-based stubs for functionals that need CGAL. This is intentional — those tests document what was in the Java port scope but requires the CGAL mesh type.
+- **Boost is header-only**: CGAL 6.x uses only Boost headers (`Boost.Config`, `Boost.Graph`). No compiled Boost libraries are needed. `find_package(Boost REQUIRED)` only locates the include path.
+- **`main` branch is protected** on `origin` (Gitea). Push to `dev`, then merge via pull request. Codeberg `main` can be pushed to directly.
+- **Both remotes must stay in sync**: `origin` = `git.eulernest.eu` (CI runs here), `codeberg` = `codeberg.org/TMoussa/ConformalLabpp` (public mirror). Push to both after every significant change.

CONTRIBUTING.md

@@ -0,0 +1,17 @@
+# Contributing to conformallab++
+
+See **[doc/contributing.md](doc/contributing.md)** for the full guide:
+
+- Git workflow (dev → PR → main)
+- CI pipeline (test-fast / test-cgal)
+- Test standards (gradient check, convergence test, registration)
+- Code style (C++17, header-only, namespace, property map naming)
+- Release process
+
+For the mathematical background of what's being implemented, see:
+
+- [doc/math/discrete-conformal-theory.md](doc/math/discrete-conformal-theory.md) — theory overview
+- [doc/math/validation.md](doc/math/validation.md) — how to validate the implementation
+- [doc/math/references.md](doc/math/references.md) — all referenced papers
+
+To add your own research, see [doc/api/extending.md](doc/api/extending.md).

Doxyfile

@@ -0,0 +1,126 @@
+# Doxyfile for conformallab++
+#
+# Phase 7.5 — minimal CGAL-style Doxygen configuration.
+# Only non-default values are set; Doxygen ≥ 1.9.5 supplies the rest.
+#
+# Usage:
+#   doxygen Doxyfile           # generates HTML into doc/doxygen/html/
+#   open doc/doxygen/html/index.html
+#
+# Or via CMake:
+#   cmake --build build --target doc
+
+# ── Project identity ─────────────────────────────────────────────────────────
+PROJECT_NAME           = "conformallab++"
+PROJECT_NUMBER         = 0.7.0
+PROJECT_BRIEF          = "Discrete conformal maps on triangle meshes — C++17 reimplementation of ConformalLab (TU Berlin)"
+PROJECT_LOGO           =
+OUTPUT_DIRECTORY       = doc/doxygen
+USE_MDFILE_AS_MAINPAGE = README.md
+
+# ── Input ────────────────────────────────────────────────────────────────────
+INPUT                  = README.md \
+                         CLAUDE.md \
+                         code/include \
+                         doc/api \
+                         doc/architecture \
+                         doc/math
+FILE_PATTERNS          = *.hpp *.h *.cpp *.md
+RECURSIVE              = YES
+EXCLUDE_PATTERNS       = */build*/* \
+                         */deps/* \
+                         */.git/* \
+                         */test-reports/* \
+                         */* 2.hpp
+EXCLUDE_SYMBOLS        = Eigen::* boost::* std::*
+
+# ── Source browsing ──────────────────────────────────────────────────────────
+EXTRACT_ALL            = YES
+EXTRACT_PRIVATE        = NO
+EXTRACT_STATIC         = YES
+EXTRACT_LOCAL_CLASSES  = YES
+HIDE_UNDOC_MEMBERS     = NO
+SOURCE_BROWSER         = YES
+INLINE_SOURCES         = NO
+STRIP_CODE_COMMENTS    = NO
+REFERENCED_BY_RELATION = YES
+REFERENCES_RELATION    = YES
+REFERENCES_LINK_SOURCE = YES
+
+# ── Build options ────────────────────────────────────────────────────────────
+JAVADOC_AUTOBRIEF      = YES
+QT_AUTOBRIEF           = NO
+MARKDOWN_SUPPORT       = YES
+AUTOLINK_SUPPORT       = YES
+BUILTIN_STL_SUPPORT    = YES
+DISTRIBUTE_GROUP_DOC   = YES
+GROUP_NESTED_COMPOUNDS = YES
+SUBGROUPING            = YES
+INLINE_GROUPED_CLASSES = NO
+INLINE_SIMPLE_STRUCTS  = NO
+TYPEDEF_HIDES_STRUCT   = NO
+EXTENSION_MAPPING      = h=C++ hpp=C++
+
+# ── Warnings ─────────────────────────────────────────────────────────────────
+QUIET                  = NO
+WARNINGS               = YES
+WARN_IF_UNDOCUMENTED   = NO
+WARN_IF_DOC_ERROR      = YES
+WARN_IF_INCOMPLETE_DOC = YES
+WARN_NO_PARAMDOC       = NO
+WARN_AS_ERROR          = NO
+WARN_FORMAT            = "$file:$line: $text"
+WARN_LOGFILE           = doc/doxygen/doxygen-warnings.log
+
+# ── HTML output ──────────────────────────────────────────────────────────────
+GENERATE_HTML          = YES
+HTML_OUTPUT            = html
+HTML_FILE_EXTENSION    = .html
+HTML_COLORSTYLE        = LIGHT
+HTML_COLORSTYLE_HUE    = 220
+HTML_COLORSTYLE_SAT    = 100
+HTML_COLORSTYLE_GAMMA  = 80
+HTML_TIMESTAMP         = NO
+HTML_DYNAMIC_SECTIONS  = YES
+GENERATE_TREEVIEW      = YES
+DISABLE_INDEX          = NO
+ENUM_VALUES_PER_LINE   = 1
+TREEVIEW_WIDTH         = 280
+EXT_LINKS_IN_WINDOW    = NO
+SEARCHENGINE           = YES
+SERVER_BASED_SEARCH    = NO
+
+# ── Disabled outputs (we only want HTML) ─────────────────────────────────────
+GENERATE_LATEX         = NO
+GENERATE_RTF           = NO
+GENERATE_MAN           = NO
+GENERATE_XML           = NO
+GENERATE_DOCBOOK       = NO
+GENERATE_AUTOGEN_DEF   = NO
+GENERATE_PERLMOD       = NO
+
+# ── Preprocessor ─────────────────────────────────────────────────────────────
+ENABLE_PREPROCESSING   = YES
+MACRO_EXPANSION        = YES
+EXPAND_ONLY_PREDEF     = YES
+SEARCH_INCLUDES        = YES
+INCLUDE_PATH           = code/include
+PREDEFINED             = CGAL_DISABLE_GMP \
+                         CGAL_DISABLE_MPFR \
+                         DOXYGEN_RUNNING
+
+# ── Diagrams ─────────────────────────────────────────────────────────────────
+HAVE_DOT               = NO
+CLASS_GRAPH            = YES
+COLLABORATION_GRAPH    = NO
+GROUP_GRAPHS           = YES
+INCLUDE_GRAPH          = NO
+INCLUDED_BY_GRAPH      = NO
+CALL_GRAPH             = NO
+CALLER_GRAPH           = NO
+
+# ── Aliases (CGAL-style) ─────────────────────────────────────────────────────
+ALIASES               += "concept{1}=\xrefitem concept \"Concept\" \"Concepts\" \1"
+ALIASES               += "models{1}=\xrefitem models \"Models\" \"Models\" \1"
+ALIASES               += "cgalRequires{1}=\par Requirements: \n\1"
+ALIASES               += "cgalParam{2}=\param \1 \2"

LICENSE

@@ -1,6 +1,6 @@
 MIT License
 
-Copyright (c) 2026 user2595
+Copyright (c) 2024–2026 Tarik Moussa <Tarik.moussa95@gmail.com>
 
 Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
 

README.md

@@ -1,104 +1,138 @@
 # conformallab++
 
-conformallab++ is a modern C++ reimplementation of the [ConformalLab](https://github.com/sechel/conformallab) software by Stefan Sechelmann for experiments in discrete conformal geometry and related mesh transformations.
+[![CI](https://git.eulernest.eu/conformallab/ConformalLabpp/actions/workflows/cpp-tests.yml/badge.svg)](https://git.eulernest.eu/conformallab/ConformalLabpp/actions)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)
+[![DOI](https://img.shields.io/badge/doi-Sechelmann%202016-blue)](https://depositonce.tu-berlin.de/items/8e2988b2-d991-45b5-aad5-9fb7988f3b2f)
 
-> **Status:** early prototype stage. API, file formats, and CLI are subject to change.
+C++17 reimplementation of [ConformalLab](https://github.com/varylab/conformallab) —
+Stefan Sechelmann's Java research library for discrete conformal geometry (TU Berlin).
+The long-term goal is a **CGAL package** for discrete conformal maps.
 
-## Features
+Algorithmic foundation:
+> Stefan Sechelmann — *Variational Methods for Discrete Surface Parameterization: Applications and Implementation*, TU Berlin 2016.
+> DOI: [10.14279/depositonce-5415](https://depositonce.tu-berlin.de/items/8e2988b2-d991-45b5-aad5-9fb7988f3b2f) · CC BY-SA 4.0 ·
+> [Java original](https://github.com/varylab/conformallab) · [sechel.de](https://sechel.de/)
 
-- Discrete conformal geometry utilities (Clausen function, hyper-ideal tetrahedra, surface curves)
+**Status:** v0.9.0 — Phases 1–9a complete, Phase 8b-Lite CGAL API surface. Newton solvers for **five** DCE models (Euclidean / Spherical / HyperIdeal / CP-Euclidean / Inversive-Distance), priority-BFS layout in ℝ²/S²/Poincaré disk, Gauss–Bonnet, tree-cotree cut graph, Möbius holonomy, period matrix (genus 1), fundamental domain, halfedge_uv texture atlas, JSON/XML serialisation, CLI app. **227 CGAL tests + 23 non-CGAL tests, 0 skipped.**
-- Mesh I/O and conversion using CGAL — optional, only needed for the CLI app
-- Linear algebra routines with Eigen
-- Interactive mesh viewer using libigl / GLFW — optional
-- Lightweight CLI with CLI11 and JSON configuration
 
-## Build modes
+---
 
-The project uses three clearly separated CMake modes so you only pull in what you need.
+## Quick start
-
-| Mode | CMake flag | What gets built |
-|------|-----------|-----------------|
-| **Tests only** (default, used in CI) | *(none)* | `conformallab_tests` · deps: Eigen + GTest |
-| **Viewer** | `-DWITH_VIEWER=ON` | `viewer` library · deps: libigl / GLFW / GLAD + Eigen |
-| **Full app** | `-DWITH_CGAL=ON` | `conformallab_core` CLI + viewer · deps: CGAL + libigl / GLFW / GLAD + Eigen |
-
-`-DWITH_CGAL=ON` automatically enables `WITH_VIEWER` because the CLI app uses the viewer library for mesh visualisation.
-
-External dependencies ship as tarballs in `code/deps/tarballs/` and are extracted lazily at CMake configure time — no internet access needed after cloning (GTest is the only exception: fetched from GitHub via FetchContent).
-
-## Prerequisites
-
-| Tool | Minimum version |
-|------|----------------|
-| C++ compiler (GCC or Clang) | C++17 |
-| CMake | 3.20 |
-
-No system-level libraries are required for the default tests-only build. CGAL and libigl are header-only and bundled in the repo.
-
-## Getting started
 
 ```bash
-git clone https://codeberg.org/TMoussa/ConformalLabpp
+git clone https://codeberg.org/TMoussa/ConformalLabpp && cd ConformalLabpp
-cd ConformalLabpp
-```
 
-### Tests only (CI default)
+# Fast tests — no system dependencies
-
+cmake -S code -B build && cmake --build build --target conformallab_tests -j$(nproc)
-```bash
-cmake -S code -B build
-cmake --build build --target conformallab_tests -j$(nproc)
 ctest --test-dir build --output-on-failure
+
+# CGAL tests headless (apt install libboost-dev / brew install boost)
+cmake -S code -B build -DWITH_CGAL_TESTS=ON
+cmake --build build --target conformallab_cgal_tests -j$(nproc)
+ctest --test-dir build -R "^cgal\." --output-on-failure
+
+# Full build with CLI + viewer (requires Wayland/X11 dev headers)
+cmake -S code -B build -DWITH_CGAL=ON && cmake --build build -j$(nproc)
+./bin/conformallab_core -i input.off -g euclidean -o layout.off -j result.json
+
+# API documentation (requires doxygen: brew/apt install doxygen)
+cmake --build build --target doc
+open doc/doxygen/html/index.html
 ```
 
-### Full CLI app (CGAL + viewer)
+---
 
-```bash
+## Minimal usage
-cmake -S code -B build -DWITH_CGAL=ON
+
-cmake --build build -j$(nproc)
+```cpp
-./code/bin/conformallab_core --input data/off/example.off --show
+#include "conformal_mesh.hpp"
+#include "mesh_io.hpp"
+#include "euclidean_functional.hpp"
+#include "gauss_bonnet.hpp"
+#include "newton_solver.hpp"
+#include "layout.hpp"
+
+using namespace conformallab;
+
+ConformalMesh mesh = load_mesh("input.off");
+EuclideanMaps maps = setup_euclidean_maps(mesh);
+compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+// Assign DOFs — pin first vertex (gauge fix)
+auto vit = mesh.vertices().begin();
+maps.v_idx[*vit++] = -1;
+int idx = 0;
+for (; vit != mesh.vertices().end(); ++vit) maps.v_idx[*vit] = idx++;
+
+// Natural equilibrium target: x* = 0 by construction
+std::vector<double> x0(idx, 0.0);
+auto G0 = euclidean_gradient(mesh, x0, maps);
+for (auto v : mesh.vertices())
+    if (maps.v_idx[v] >= 0) maps.theta_v[v] -= G0[maps.v_idx[v]];
+
+check_gauss_bonnet(mesh, maps);
+NewtonResult res = newton_euclidean(mesh, x0, maps);
+Layout2D layout = euclidean_layout(mesh, res.x, maps);
 ```
 
-### Viewer only (no CGAL)
+---
 
-```bash
+## Documentation
-cmake -S code -B build -DWITH_VIEWER=ON
-cmake --build build --target viewer -j$(nproc)
-```
 
-## Project structure
+| | |
+|---|---|
+| **Getting started** — build modes, single-test invocation, CLI, Docker | [doc/getting-started.md](doc/getting-started.md) |
+| **Pipeline API** — all three geometries, holonomy, serialisation | [doc/api/pipeline.md](doc/api/pipeline.md) |
+| **Public headers** — all 24 headers with descriptions | [doc/api/headers.md](doc/api/headers.md) |
+| **Test suites** — 39 suites, 227+23 tests, individual counts | [doc/api/tests.md](doc/api/tests.md) |
+| **Extending** — new functionals, geometry modes, porting from Java | [doc/api/extending.md](doc/api/extending.md) |
+| **Processing unit contracts** — preconditions / provides table | [doc/api/contracts.md](doc/api/contracts.md) |
+| **CGAL package design** — Phase 8 target, YAML pipeline | [doc/api/cgal-package.md](doc/api/cgal-package.md) |
+| **Architecture & pipeline diagram** | [doc/architecture/overall_pipeline.md](doc/architecture/overall_pipeline.md) |
+| **geometry-central comparison** — shared core, demarcation, adoption candidates, scientific added value | [doc/architecture/geometry-central-comparison.md](doc/architecture/geometry-central-comparison.md) |
+| **Design decisions** — key architectural choices + rationale | [doc/architecture/design-decisions.md](doc/architecture/design-decisions.md) |
+| **Project structure** — directory tree + build targets | [doc/architecture/project-structure.md](doc/architecture/project-structure.md) |
+| **Discrete conformal theory** — mathematical background for collaborators | [doc/math/discrete-conformal-theory.md](doc/math/discrete-conformal-theory.md) |
+| **Validation** — known analytic results + how to verify them | [doc/math/validation.md](doc/math/validation.md) |
+| **Validation protocol** — concrete commands with expected outputs | [doc/math/validation-protocol.md](doc/math/validation-protocol.md) |
+| **Tutorial: add a new functional** — step-by-step Inversive-Distance port | [doc/tutorials/add-inversive-distance.md](doc/tutorials/add-inversive-distance.md) |
+| **Declarative YAML pipeline** — concept, token vocabulary, 5 examples | [doc/concepts/declarative-pipeline.md](doc/concepts/declarative-pipeline.md) |
+| **Geometry modes** — Euclidean / Spherical / HyperIdeal comparison | [doc/math/geometry-modes.md](doc/math/geometry-modes.md) |
+| **References** — all papers by module | [doc/math/references.md](doc/math/references.md) |
+| **Software landscape** — how conformallab++ relates to libigl, CGAL, geometry-central | [doc/math/software-landscape.md](doc/math/software-landscape.md) |
+| **Novelty statement** — unique features, target audience, what this is not | [doc/math/novelty-statement.md](doc/math/novelty-statement.md) |
+| **Complexity & scalability** — O() analysis, measured timings on real meshes, HyperIdeal bottleneck | [doc/math/complexity.md](doc/math/complexity.md) |
+| **Roadmap** — Phases 1–10 | [doc/roadmap/phases.md](doc/roadmap/phases.md) |
+| **Java parity table** — what is ported, what is planned | [doc/roadmap/java-parity.md](doc/roadmap/java-parity.md) |
+| **Contributing** — language policy, test standards, release flow | [doc/contributing.md](doc/contributing.md) |
+| **Claude Code context** | [CLAUDE.md](CLAUDE.md) |
 
-```
+---
-code/
-├── include/               # Public headers (Clausen, hyper-ideal, mesh utils, …)
-├── src/
-│   ├── apps/v0/           # conformallab_core CLI app (requires WITH_CGAL)
-│   └── viewer/            # simple_viewer (requires WITH_VIEWER)
-├── tests/                 # GTest unit tests (always built)
-└── deps/
-    ├── tarballs/          # Bundled dependency archives
-    ├── eigen-3.4.0/       # Header-only linear algebra (always extracted)
-    ├── CGAL-6.1.1/        # Header-only geometry (extracted with WITH_CGAL)
-    ├── libigl-2.6.0/      # Header-only viewer toolkit (extracted with WITH_VIEWER)
-    ├── glfw-3.4/          # Windowing (extracted with WITH_VIEWER)
-    ├── libigl-glad/       # OpenGL loader (extracted with WITH_VIEWER)
-    └── single_includes/   # CLI11, json.hpp
-```
 
-## CI
+## Citing
 
-Tests run automatically on push to `main`, `dev`, and `claude/**` branches via a self-hosted Gitea Actions runner (`eulernest`, ARM64 Raspberry Pi). The pipeline uses a minimal Docker image (`git.eulernest.eu/conformallab/ci-cpp:latest`) with cmake, g++, git, and Node.js 20 pre-installed.
+If you use conformallab++ in your research, please cite it using the metadata
+in [`CITATION.cff`](CITATION.cff).  GitHub and Codeberg show a "Cite this repository"
+button that generates BibTeX and APA automatically.
 
-The Dockerfile for the CI image lives in `.gitea/docker/Dockerfile.ci-cpp`. Build and push it once whenever the image needs updating:
+The primary algorithmic source is:
 
-```bash
+> Stefan Sechelmann — *Variational Methods for Discrete Surface Parameterization:
-docker buildx build \
+> Applications and Implementation*, TU Berlin 2016.
-  --platform linux/arm64 \
+> DOI: [10.14279/depositonce-5415](https://depositonce.tu-berlin.de/items/8e2988b2-d991-45b5-aad5-9fb7988f3b2f)
-  -f .gitea/docker/Dockerfile.ci-cpp \
+
-  -t git.eulernest.eu/conformallab/ci-cpp:latest \
+---
-  --push \
+
-  .gitea/docker/
+## Bugs & questions
-```
+
+- **Bug reports / feature requests:** [Gitea Issues](https://git.eulernest.eu/conformallab/ConformalLabpp/issues)
+- **Code mirror (read-only):** [Codeberg](https://codeberg.org/TMoussa/ConformalLabpp)
+- **Contact:** Tarik Moussa · Tarik.moussa95@gmail.com
+
+---
 
 ## License
 
-conformallab++ is released under the MIT License (see [LICENSE](LICENSE)).
+conformallab++ is released under the MIT License (see [LICENSE](LICENSE)).  
+Copyright © 2024–2026 Tarik Moussa.  
+The dissertation (Sechelmann 2016) is CC BY-SA 4.0.

code/.gitignore

@@ -40,6 +40,8 @@ Thumbs.db
 *.lo
 *.o
 *.obj
+# Exception: mesh data files in code/data/ are not compiled objects
+!data/**/*.obj
 
 # Precompiled Headers
 *.gch

code/CMakeLists.txt

@@ -7,24 +7,42 @@ message(STATUS "Configuring ${PROJECT_NAME}...")
 
 # ── Build modes ────────────────────────────────────────────────────────────────
 #
-#   Default (CI / tests-only): only Eigen + GTest are required.
+#   Default (CI fast / pure-math tests):
+#     cmake -S code -B build
+#     Only Eigen + GTest required.  36 non-CGAL tests.
 #
-#   -DWITH_CGAL=ON   builds the conformallab_core CLI app (needs CGAL).
+#   -DWITH_CGAL_TESTS=ON   CGAL test suite only — no viewer, no CLI app.
-#                    Automatically enables WITH_VIEWER because the app uses
+#     cmake -S code -B build -DWITH_CGAL_TESTS=ON
-#                    the viewer library for mesh visualisation.
+#     Requires: system Boost headers (apt install libboost-dev).
+#     Builds:   conformallab_cgal_tests  (158 tests).
+#     Does NOT require wayland-scanner, GLFW, libigl or a display.
+#     Use this in headless CI.
 #
-#   -DWITH_VIEWER=ON builds the viewer library standalone (libigl/GLFW/GLAD).
+#   -DWITH_CGAL=ON         Full build: CLI app + viewer + examples + CGAL tests.
+#     cmake -S code -B build -DWITH_CGAL=ON
+#     Requires: Boost + Wayland/X11 dev headers (wayland-scanner, libx11-dev …).
+#     Automatically enables WITH_VIEWER.
+#     Use this for local development with the interactive viewer.
+#
+#   -DWITH_VIEWER=ON        Viewer library only (libigl / GLFW / GLAD).
 #
 # ──────────────────────────────────────────────────────────────────────────────
-option(WITH_CGAL   "Build conformallab_core app (requires CGAL + Viewer)" OFF)
+option(WITH_CGAL_TESTS "Build CGAL test suite without viewer/CLI (headless CI)" OFF)
-option(WITH_VIEWER "Build viewer library (libigl / GLFW / GLAD)"          OFF)
+option(WITH_CGAL       "Build conformallab_core CLI app + viewer + CGAL tests"  OFF)
+option(WITH_VIEWER     "Build viewer library (libigl / GLFW / GLAD)"            OFF)
 
-# The CLI app always needs the viewer; enable it implicitly.
+# WITH_CGAL_TESTS is a strict subset of WITH_CGAL — no viewer, no CLI.
+# WITH_CGAL (full build) implies WITH_VIEWER.
 if(WITH_CGAL AND NOT WITH_VIEWER)
     message(STATUS "WITH_CGAL implies WITH_VIEWER – enabling automatically.")
     set(WITH_VIEWER ON CACHE BOOL "" FORCE)
 endif()
 
+# Propagate Boost requirement for both CGAL modes.
+if(WITH_CGAL OR WITH_CGAL_TESTS)
+    find_package(Boost REQUIRED)
+endif()
+
 # ── Standard settings ──────────────────────────────────────────────────────────
 set(CMAKE_CXX_STANDARD 17)
 set(CMAKE_CXX_STANDARD_REQUIRED ON)
@@ -85,11 +103,6 @@ endif()
 
 # ── Core CLI app (optional, requires CGAL + Viewer) ───────────────────────────
 if(WITH_CGAL)
-    # CGAL 6.x still needs Boost.Config headers unconditionally.
-    # Install via: brew install boost  (macOS)
-    #              apt install libboost-dev  (Debian/Ubuntu)
-    find_package(Boost REQUIRED)
-
     add_executable(${PROJECT_NAME} src/apps/v0/conformallab_cli.cpp)
     target_include_directories(${PROJECT_NAME} SYSTEM PRIVATE
         ${CMAKE_CURRENT_SOURCE_DIR}/deps/single_includes
@@ -106,5 +119,46 @@ if(WITH_CGAL)
         RUNTIME_OUTPUT_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}/bin)
 endif()
 
+# ── Example programs (require WITH_CGAL; viewer example also needs WITH_VIEWER) ─
+if(WITH_CGAL)
+    add_subdirectory(examples)
+endif()
+
 # ── Tests (always) ────────────────────────────────────────────────────────────
 add_subdirectory(tests)
+
+# ── Install target (header-only library) ──────────────────────────────────────
+# Installs all public headers to <prefix>/include/conformallab/
+# Usage from another CMake project:
+#   cmake --install build --prefix /usr/local
+#   target_include_directories(myapp PRIVATE /usr/local/include/conformallab)
+include(GNUInstallDirs)
+install(DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}/include/
+        DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}/conformallab
+        FILES_MATCHING PATTERN "*.hpp"
+        PATTERN "* 2.*" EXCLUDE)   # exclude macOS Finder duplicates
+install(FILES ${CMAKE_CURRENT_SOURCE_DIR}/../LICENSE
+              ${CMAKE_CURRENT_SOURCE_DIR}/../CITATION.cff
+        DESTINATION ${CMAKE_INSTALL_DATADIR}/conformallab)
+
+# ── Doxygen documentation target (Phase 7.5) ──────────────────────────────────
+# Generates HTML API documentation into doc/doxygen/html/.
+# Usage:
+#   cmake --build build --target doc
+#   open doc/doxygen/html/index.html
+#
+# Optional dependency: install Doxygen via `brew install doxygen` (macOS) or
+# `apt install doxygen graphviz` (Linux). The target is silently disabled
+# if Doxygen is not found.
+find_package(Doxygen QUIET)
+if(DOXYGEN_FOUND)
+    set(DOXYGEN_PROJECT_ROOT ${CMAKE_CURRENT_SOURCE_DIR}/..)
+    add_custom_target(doc
+        COMMAND ${DOXYGEN_EXECUTABLE} ${DOXYGEN_PROJECT_ROOT}/Doxyfile
+        WORKING_DIRECTORY ${DOXYGEN_PROJECT_ROOT}
+        COMMENT "Generating API documentation with Doxygen"
+        VERBATIM)
+    message(STATUS "Doxygen found: target 'doc' available (cmake --build build --target doc)")
+else()
+    message(STATUS "Doxygen not found — 'doc' target unavailable (install: brew/apt install doxygen)")
+endif()

code/data/obj/cathead.obj

@@ -0,0 +1,379 @@
+v -0.206972 0.0740737 0.544664
+v -0.398695 -0.0740794 0.501089
+v -0.211327 -0.187367 0.588234
+v -0.511983 -0.235298 0.400872
+v -0.56863 0.0348535 0.405227
+v -0.35948 0.178646 0.50545
+v -0.525054 0.313721 0.387801
+v -0.189545 0.30065 0.544664
+v 0.124182 0.0740737 0.509805
+v -0.455336 -0.33987 0.544664
+v -0.411766 -0.427021 0.671024
+v -0.237476 -0.413949 0.745097
+v -0.250547 -0.535954 0.797389
+v -0.346403 -0.562097 0.758169
+v 0.0152491 -0.33116 0.531593
+v -0.106755 -0.496734 0.692812
+v -0.254903 -0.692812 0.749458
+v -0.0631798 -0.601311 0.55338
+v -0.14597 -0.666669 0.666669
+v -0.250547 -0.727671 0.583879
+v -0.198256 -0.714599 0.42266
+v 0.0283206 -0.514166 0.383445
+v -0.124182 -0.623099 0.196078
+v -0.285401 -0.692812 0.413944
+v -0.333331 -0.596956 0.226582
+v -0.333331 -0.418305 0.413944
+v 0.233115 -0.366019 0.313727
+v 0.215688 -0.156864 0.457519
+v -0.342048 -0.623099 -0.0915061
+v -0.708061 -0.431376 0.0697184
+v -0.647058 -0.501095 -0.0915061
+v -0.716777 -0.318088 0.169935
+v -0.699344 -0.143792 0.235292
+v -0.747275 0.00871052 0.235292
+v -0.747275 0.165574 0.191723
+v -0.764708 0.291939 0.12636
+v -0.747275 0.461874 0.0130715
+v -0.337692 0.535948 0.366013
+v -0.472768 0.644881 0.0653574
+v -0.272329 0.753814 0.0827899
+v -0.294117 0.797383 0.278868
+v -0.307188 0.801744 0.435731
+v -0.324621 0.779957 0.583879
+v -0.355119 0.714594 0.710238
+v -0.407405 0.59695 0.692812
+v -0.442264 0.505444 0.610022
+v -0.477123 0.413944 0.51416
+v -0.285401 0.509805 0.666669
+v -0.185184 0.605661 0.631809
+v -0.0326816 0.518516 0.479301
+v -0.198256 0.72331 0.575162
+v -0.124182 0.745097 0.278868
+v 0.189545 0.479301 0.357297
+v -0.0413921 0.675379 0.352942
+v -0.843136 -0.0392202 -0.0522859
+v -0.764708 -0.33116 0.0348592
+v -0.816994 -0.152508 -0.178651
+v -0.87364 -0.0522916 -0.187362
+v -0.816994 0.108933 -0.204794
+v -0.816994 0.25708 -0.00435526
+v -0.795206 0.387795 -0.0653574
+v -0.764708 0.440087 -0.152503
+v -0.642703 0.535948 -0.370368
+v -0.516338 0.636165 -0.21351
+v -0.856208 0.265791 -0.148147
+v -0.816994 0.261435 -0.291939
+v -0.847497 0.239648 -0.418299
+v -0.82571 0.265791 -0.479301
+v -0.912855 -0.0697184 -0.270152
+v -0.947714 0.0130715 -0.326799
+v -0.938998 0.0740737 -0.392156
+v -0.799567 -0.344231 -0.108933
+v -0.620916 -0.509805 -0.331154
+v -0.764708 -0.287584 -0.244009
+v -0.799567 -0.291939 -0.352942
+v -0.9085 -0.174296 -0.33987
+v -1 -0.0697184 -0.409588
+v -0.965141 -0.148153 -0.461874
+v -0.95643 -0.0610022 -0.549019
+v -0.978212 0.021782 -0.483662
+v -0.729848 -0.357302 -0.501089
+v -0.921571 -0.230937 -0.544664
+v -0.821354 -0.222227 -0.662308
+v -0.9085 -0.16558 -0.623093
+v -0.342048 -0.557736 -0.383445
+v -0.381262 -0.300656 -0.727671
+v -0.167757 -0.557736 -0.501089
+v -0.599128 -0.252725 -0.692812
+v -0.420482 -0.0697184 -0.797389
+v -0.14597 -0.610027 -0.631809
+v -0.947714 0.0915004 -0.522877
+v -0.891067 0.178646 -0.50545
+v -0.869279 -0.0697184 -0.705883
+v -0.934643 0.0348535 -0.627454
+v -0.847497 0.156858 -0.649236
+v -0.620916 0.235292 -0.701528
+v -0.559913 0.143786 -0.745097
+v -0.315905 0.278862 -0.753814
+v -0.0196101 0.440087 -0.784312
+v -0.211327 0.46623 -0.636165
+v -0.381262 0.505444 -0.579523
+v -0.647058 0.374724 -0.601305
+v -0.355119 0.653591 -0.344225
+v -0.0108939 0.644881 -0.313727
+v 0.163396 0.64052 0.104578
+v 0.433554 0.0697127 0.322443
+v 0.516344 -0.318088 0.0435754
+v 0.35948 -0.492378 -0.0653574
+v 0.185184 -0.535954 0.0305039
+v 0.0631798 -0.618738 -0.313727
+v 0.328976 -0.618738 -0.296295
+v 0.102394 -0.684101 -0.43137
+v 0.612199 -0.396517 -0.16558
+v 0.307188 0.522877 -0.779957
+v 0.163396 0.549019 -0.562091
+v 0.468413 0.400872 0.00871623
+v 0.599128 0.374724 -0.261435
+v 0.721132 0.357297 -0.50545
+v 0.843136 0.326793 -0.718955
+v 0.124182 -0.801744 -0.562091
+v 0.35948 -0.749458 -0.448803
+v 0.559913 0.0479307 0.161219
+v 0.673201 0.0304982 -0.0566469
+v 0.808283 0.0174267 -0.309366
+v 0.891067 -0.00871623 -0.483662
+v 1 -0.0697184 -0.753814
+v 0.355119 0.440087 0.191723
+v 0.694989 -0.58824 -0.440087
+v 0.886712 -0.230937 -0.457519
+v 0.912855 -0.366019 -0.575162
+v -0.921571 -0.126365 -0.300656
+f 1 2 3
+f 4 2 5
+f 2 1 5
+f 1 6 5
+f 6 7 5
+f 7 6 8
+f 6 1 8
+f 9 1 3
+f 3 4 10
+f 10 11 3
+f 11 12 3
+f 11 13 12
+f 13 11 14
+f 12 15 3
+f 15 12 16
+f 12 13 16
+f 13 17 16
+f 17 13 14
+f 16 18 15
+f 18 16 19
+f 16 17 19
+f 17 20 19
+f 20 21 19
+f 21 18 19
+f 18 21 22
+f 21 23 22
+f 21 24 23
+f 24 21 20
+f 24 25 23
+f 25 24 26
+f 24 20 26
+f 20 17 26
+f 17 14 26
+f 14 11 26
+f 11 10 26
+f 10 4 26
+f 3 15 9
+f 15 27 28
+f 23 25 29
+f 25 30 31
+f 30 25 32
+f 25 26 32
+f 4 33 32
+f 18 22 15
+f 22 27 15
+f 5 34 33
+f 34 5 35
+f 5 7 35
+f 7 36 35
+f 36 7 37
+f 7 38 37
+f 38 39 37
+f 39 38 40
+f 38 41 40
+f 41 38 42
+f 38 43 42
+f 43 38 44
+f 38 45 44
+f 45 38 46
+f 46 38 47
+f 38 7 47
+f 7 8 47
+f 8 46 47
+f 46 8 48
+f 8 49 48
+f 49 44 48
+f 44 45 48
+f 49 8 50
+f 49 51 44
+f 51 43 44
+f 43 51 42
+f 51 52 42
+f 52 41 42
+f 41 52 40
+f 53 54 50
+f 49 54 51
+f 54 49 50
+f 51 54 52
+f 50 9 53
+f 9 50 8
+f 55 33 34
+f 33 55 32
+f 55 56 32
+f 56 55 57
+f 55 58 57
+f 58 55 59
+f 55 60 59
+f 60 55 35
+f 55 34 35
+f 36 60 35
+f 60 36 37
+f 37 61 60
+f 61 37 62
+f 37 63 62
+f 63 37 64
+f 37 39 64
+f 39 40 64
+f 62 65 61
+f 65 62 66
+f 62 67 66
+f 67 62 68
+f 62 63 68
+f 65 59 60
+f 59 69 58
+f 69 59 70
+f 59 71 70
+f 71 59 67
+f 59 66 67
+f 59 65 66
+f 56 30 32
+f 30 56 31
+f 56 72 31
+f 72 56 57
+f 60 61 65
+f 46 48 45
+f 31 29 25
+f 29 31 73
+f 31 74 73
+f 74 31 72
+f 57 74 72
+f 74 57 75
+f 57 76 75
+f 57 58 69
+f 77 78 76
+f 78 77 79
+f 77 80 79
+f 80 77 71
+f 77 70 71
+f 70 77 69
+f 75 73 74
+f 73 75 81
+f 75 82 81
+f 82 75 76
+f 82 83 81
+f 83 82 84
+f 82 79 84
+f 79 82 78
+f 82 76 78
+f 73 85 29
+f 85 73 86
+f 73 81 86
+f 85 87 29
+f 87 85 86
+f 88 86 81
+f 86 88 89
+f 88 83 89
+f 83 88 81
+f 86 90 87
+f 71 91 80
+f 91 71 92
+f 71 67 92
+f 67 68 92
+f 79 93 84
+f 93 79 94
+f 79 91 94
+f 91 79 80
+f 92 95 91
+f 92 68 95
+f 91 95 94
+f 95 93 94
+f 93 95 96
+f 95 68 96
+f 93 83 84
+f 83 93 89
+f 93 97 89
+f 97 93 96
+f 89 97 98
+f 98 100 99
+f 100 98 101
+f 98 96 101
+f 98 97 96
+f 102 68 63
+f 68 102 96
+f 102 101 96
+f 101 102 63
+f 64 103 63
+f 103 64 40
+f 103 101 63
+f 101 104 100
+f 104 101 103
+f 40 104 103
+f 40 52 104
+f 52 105 104
+f 9 106 53
+f 106 9 28
+f 106 27 107
+f 27 106 28
+f 27 108 107
+f 108 27 22
+f 22 109 108
+f 109 22 23
+f 109 110 108
+f 110 109 23
+f 23 29 110
+f 110 111 108
+f 111 110 112
+f 110 90 112
+f 90 110 87
+f 110 29 87
+f 108 111 113
+f 108 113 107
+f 114 99 115
+f 99 104 115
+f 104 105 115
+f 116 117 115
+f 117 118 115
+f 115 118 114
+f 118 119 114
+f 120 112 90
+f 112 120 111
+f 120 121 111
+f 99 100 104
+f 2 4 3
+f 5 33 4
+f 116 122 123
+f 123 117 116
+f 117 123 124
+f 117 124 118
+f 124 125 118
+f 125 119 118
+f 119 125 126
+f 122 106 107
+f 105 116 115
+f 54 53 105
+f 52 54 105
+f 105 127 116
+f 122 116 127
+f 106 122 127
+f 53 127 105
+f 121 128 113
+f 121 113 111
+f 113 128 124
+f 128 129 124
+f 129 128 130
+f 107 113 122
+f 113 123 122
+f 123 113 124
+f 1 9 8
+f 9 15 28
+f 4 32 26
+f 106 127 53
+f 57 131 76
+f 131 77 76
+f 69 131 57
+f 131 69 77
+f 129 125 124
+f 125 130 126
+f 130 125 129

code/data/obj/tetraflat.obj

@@ -0,0 +1,16 @@
+# 
+# Wavefront OBJ file
+# Converted by the DEEP Exploration Deep Exploration 5 5.0.3.1555 Release
+# Right Hemisphere, LTD
+# http://www.righthemisphere.com/
+# 
+# object sgc 1
+g sgc_1
+v 0.00000 0.00000 0.00000
+v -1.29492 0.95275 -0.28653
+v 1.14390 0.47528 1.06408
+v 0.15103 -1.42803 -0.77755
+# 4 verticies
+f  1 2 3
+f  2 1 4
+f  1 3 4

code/data/off/torus_4x4.off

@@ -0,0 +1,50 @@
+OFF
+16 32 0
+3.000000 0.000000 0.000000
+2.000000 0.000000 1.000000
+1.000000 0.000000 0.000000
+2.000000 0.000000 -1.000000
+0.000000 3.000000 0.000000
+0.000000 2.000000 1.000000
+0.000000 1.000000 0.000000
+0.000000 2.000000 -1.000000
+-3.000000 0.000000 0.000000
+-2.000000 0.000000 1.000000
+-1.000000 0.000000 0.000000
+-2.000000 0.000000 -1.000000
+-0.000000 -3.000000 0.000000
+-0.000000 -2.000000 1.000000
+-0.000000 -1.000000 0.000000
+-0.000000 -2.000000 -1.000000
+3 0 4 1
+3 1 4 5
+3 1 5 2
+3 2 5 6
+3 2 6 3
+3 3 6 7
+3 3 7 0
+3 0 7 4
+3 4 8 5
+3 5 8 9
+3 5 9 6
+3 6 9 10
+3 6 10 7
+3 7 10 11
+3 7 11 4
+3 4 11 8
+3 8 12 9
+3 9 12 13
+3 9 13 10
+3 10 13 14
+3 10 14 11
+3 11 14 15
+3 11 15 8
+3 8 15 12
+3 12 0 13
+3 13 0 1
+3 13 1 14
+3 14 1 2
+3 14 2 15
+3 15 2 3
+3 15 3 12
+3 12 3 0

code/data/off/torus_8x8.off

@@ -0,0 +1,194 @@
+OFF
+64 128 0
+4.000000 0.000000 0.000000
+3.707107 0.000000 0.707107
+3.000000 0.000000 1.000000
+2.292893 0.000000 0.707107
+2.000000 0.000000 0.000000
+2.292893 0.000000 -0.707107
+3.000000 0.000000 -1.000000
+3.707107 0.000000 -0.707107
+2.828427 2.828427 0.000000
+2.621320 2.621320 0.707107
+2.121320 2.121320 1.000000
+1.621320 1.621320 0.707107
+1.414214 1.414214 0.000000
+1.621320 1.621320 -0.707107
+2.121320 2.121320 -1.000000
+2.621320 2.621320 -0.707107
+0.000000 4.000000 0.000000
+0.000000 3.707107 0.707107
+0.000000 3.000000 1.000000
+0.000000 2.292893 0.707107
+0.000000 2.000000 0.000000
+0.000000 2.292893 -0.707107
+0.000000 3.000000 -1.000000
+0.000000 3.707107 -0.707107
+-2.828427 2.828427 0.000000
+-2.621320 2.621320 0.707107
+-2.121320 2.121320 1.000000
+-1.621320 1.621320 0.707107
+-1.414214 1.414214 0.000000
+-1.621320 1.621320 -0.707107
+-2.121320 2.121320 -1.000000
+-2.621320 2.621320 -0.707107
+-4.000000 0.000000 0.000000
+-3.707107 0.000000 0.707107
+-3.000000 0.000000 1.000000
+-2.292893 0.000000 0.707107
+-2.000000 0.000000 0.000000
+-2.292893 0.000000 -0.707107
+-3.000000 0.000000 -1.000000
+-3.707107 0.000000 -0.707107
+-2.828427 -2.828427 0.000000
+-2.621320 -2.621320 0.707107
+-2.121320 -2.121320 1.000000
+-1.621320 -1.621320 0.707107
+-1.414214 -1.414214 0.000000
+-1.621320 -1.621320 -0.707107
+-2.121320 -2.121320 -1.000000
+-2.621320 -2.621320 -0.707107
+-0.000000 -4.000000 0.000000
+-0.000000 -3.707107 0.707107
+-0.000000 -3.000000 1.000000
+-0.000000 -2.292893 0.707107
+-0.000000 -2.000000 0.000000
+-0.000000 -2.292893 -0.707107
+-0.000000 -3.000000 -1.000000
+-0.000000 -3.707107 -0.707107
+2.828427 -2.828427 0.000000
+2.621320 -2.621320 0.707107
+2.121320 -2.121320 1.000000
+1.621320 -1.621320 0.707107
+1.414214 -1.414214 0.000000
+1.621320 -1.621320 -0.707107
+2.121320 -2.121320 -1.000000
+2.621320 -2.621320 -0.707107
+3 0 8 1
+3 1 8 9
+3 1 9 2
+3 2 9 10
+3 2 10 3
+3 3 10 11
+3 3 11 4
+3 4 11 12
+3 4 12 5
+3 5 12 13
+3 5 13 6
+3 6 13 14
+3 6 14 7
+3 7 14 15
+3 7 15 0
+3 0 15 8
+3 8 16 9
+3 9 16 17
+3 9 17 10
+3 10 17 18
+3 10 18 11
+3 11 18 19
+3 11 19 12
+3 12 19 20
+3 12 20 13
+3 13 20 21
+3 13 21 14
+3 14 21 22
+3 14 22 15
+3 15 22 23
+3 15 23 8
+3 8 23 16
+3 16 24 17
+3 17 24 25
+3 17 25 18
+3 18 25 26
+3 18 26 19
+3 19 26 27
+3 19 27 20
+3 20 27 28
+3 20 28 21
+3 21 28 29
+3 21 29 22
+3 22 29 30
+3 22 30 23
+3 23 30 31
+3 23 31 16
+3 16 31 24
+3 24 32 25
+3 25 32 33
+3 25 33 26
+3 26 33 34
+3 26 34 27
+3 27 34 35
+3 27 35 28
+3 28 35 36
+3 28 36 29
+3 29 36 37
+3 29 37 30
+3 30 37 38
+3 30 38 31
+3 31 38 39
+3 31 39 24
+3 24 39 32
+3 32 40 33
+3 33 40 41
+3 33 41 34
+3 34 41 42
+3 34 42 35
+3 35 42 43
+3 35 43 36
+3 36 43 44
+3 36 44 37
+3 37 44 45
+3 37 45 38
+3 38 45 46
+3 38 46 39
+3 39 46 47
+3 39 47 32
+3 32 47 40
+3 40 48 41
+3 41 48 49
+3 41 49 42
+3 42 49 50
+3 42 50 43
+3 43 50 51
+3 43 51 44
+3 44 51 52
+3 44 52 45
+3 45 52 53
+3 45 53 46
+3 46 53 54
+3 46 54 47
+3 47 54 55
+3 47 55 40
+3 40 55 48
+3 48 56 49
+3 49 56 57
+3 49 57 50
+3 50 57 58
+3 50 58 51
+3 51 58 59
+3 51 59 52
+3 52 59 60
+3 52 60 53
+3 53 60 61
+3 53 61 54
+3 54 61 62
+3 54 62 55
+3 55 62 63
+3 55 63 48
+3 48 63 56
+3 56 0 57
+3 57 0 1
+3 57 1 58
+3 58 1 2
+3 58 2 59
+3 59 2 3
+3 59 3 60
+3 60 3 4
+3 60 4 61
+3 61 4 5
+3 61 5 62
+3 62 5 6
+3 62 6 63
+3 63 6 7
+3 63 7 56
+3 56 7 0

code/data/off/torus_hex_6x6.off

@@ -0,0 +1,110 @@
+OFF
+36 72 0
+4.000000 0.000000 0.000000
+3.500000 0.000000 0.866025
+2.500000 0.000000 0.866025
+2.000000 0.000000 0.000000
+2.500000 0.000000 -0.866025
+3.500000 0.000000 -0.866025
+2.000000 3.464102 0.000000
+1.750000 3.031089 0.866025
+1.250000 2.165064 0.866025
+1.000000 1.732051 0.000000
+1.250000 2.165064 -0.866025
+1.750000 3.031089 -0.866025
+-2.000000 3.464102 0.000000
+-1.750000 3.031089 0.866025
+-1.250000 2.165064 0.866025
+-1.000000 1.732051 0.000000
+-1.250000 2.165064 -0.866025
+-1.750000 3.031089 -0.866025
+-4.000000 0.000000 0.000000
+-3.500000 0.000000 0.866025
+-2.500000 0.000000 0.866025
+-2.000000 0.000000 0.000000
+-2.500000 0.000000 -0.866025
+-3.500000 0.000000 -0.866025
+-2.000000 -3.464102 0.000000
+-1.750000 -3.031089 0.866025
+-1.250000 -2.165064 0.866025
+-1.000000 -1.732051 0.000000
+-1.250000 -2.165064 -0.866025
+-1.750000 -3.031089 -0.866025
+2.000000 -3.464102 0.000000
+1.750000 -3.031089 0.866025
+1.250000 -2.165064 0.866025
+1.000000 -1.732051 0.000000
+1.250000 -2.165064 -0.866025
+1.750000 -3.031089 -0.866025
+3 0 6 1
+3 1 6 7
+3 1 7 2
+3 2 7 8
+3 2 8 3
+3 3 8 9
+3 3 9 4
+3 4 9 10
+3 4 10 5
+3 5 10 11
+3 5 11 0
+3 0 11 6
+3 6 12 7
+3 7 12 13
+3 7 13 8
+3 8 13 14
+3 8 14 9
+3 9 14 15
+3 9 15 10
+3 10 15 16
+3 10 16 11
+3 11 16 17
+3 11 17 6
+3 6 17 12
+3 12 18 13
+3 13 18 19
+3 13 19 14
+3 14 19 20
+3 14 20 15
+3 15 20 21
+3 15 21 16
+3 16 21 22
+3 16 22 17
+3 17 22 23
+3 17 23 12
+3 12 23 18
+3 18 24 19
+3 19 24 25
+3 19 25 20
+3 20 25 26
+3 20 26 21
+3 21 26 27
+3 21 27 22
+3 22 27 28
+3 22 28 23
+3 23 28 29
+3 23 29 18
+3 18 29 24
+3 24 30 25
+3 25 30 31
+3 25 31 26
+3 26 31 32
+3 26 32 27
+3 27 32 33
+3 27 33 28
+3 28 33 34
+3 28 34 29
+3 29 34 35
+3 29 35 24
+3 24 35 30
+3 30 0 31
+3 31 0 1
+3 31 1 32
+3 32 1 2
+3 32 2 33
+3 33 2 3
+3 33 3 34
+3 34 3 4
+3 34 4 35
+3 35 4 5
+3 35 5 30
+3 30 5 0

code/deps/CMakeLists.txt

@@ -5,9 +5,10 @@
 #
 # Which deps are extracted depends on the active build mode:
 #
-#   (default)      Eigen only          → tests-only build
+#   (default)           Eigen only
-#   WITH_VIEWER=ON + Eigen, libigl, libigl-glad, glfw
+#   WITH_CGAL_TESTS=ON  + Eigen, CGAL           (headless CI — no viewer)
-#   WITH_CGAL=ON   + Eigen, CGAL       (WITH_VIEWER is implied by WITH_CGAL)
+#   WITH_VIEWER=ON      + Eigen, libigl, libigl-glad, glfw
+#   WITH_CGAL=ON        + Eigen, CGAL, libigl, libigl-glad, glfw
 #
 
 function(setup_dependency NAME SUBDIR)
@@ -51,6 +52,7 @@ if(WITH_VIEWER)
 endif()
 
 # ── CGAL mode ─────────────────────────────────────────────────────────────────
-if(WITH_CGAL)
+# Both WITH_CGAL_TESTS (headless CI) and WITH_CGAL (full build) need CGAL headers.
+if(WITH_CGAL OR WITH_CGAL_TESTS)
     setup_dependency("CGAL-6.1.1" "include")
 endif()

code/examples/CMakeLists.txt

@@ -0,0 +1,56 @@
+# examples/CMakeLists.txt
+#
+# Example programs for conformallab++.
+# All examples require -DWITH_CGAL=ON.
+# example_viewer additionally requires -DWITH_VIEWER=ON.
+#
+# Run after building:
+#   ./build/examples/example_euclidean   [input.off] [output.off]
+#   ./build/examples/example_hyper_ideal [input.off] [output.off]
+#   ./build/examples/example_viewer      [input.off]         (requires WITH_VIEWER)
+
+# ── Shared include paths for all examples ─────────────────────────────────────
+set(EXAMPLE_INCLUDES
+    ${CMAKE_SOURCE_DIR}/deps/eigen-3.4.0
+    ${CMAKE_SOURCE_DIR}/deps/CGAL-6.1.1/include
+    ${CMAKE_SOURCE_DIR}/deps/single_includes
+    ${Boost_INCLUDE_DIRS}
+)
+set(EXAMPLE_PRIVATE_INCLUDES
+    ${CMAKE_SOURCE_DIR}/include
+)
+set(EXAMPLE_DEFS
+    CGAL_DISABLE_GMP
+    CGAL_DISABLE_MPFR
+)
+
+# ── example_euclidean ─────────────────────────────────────────────────────────
+add_executable(example_euclidean example_euclidean.cpp)
+target_include_directories(example_euclidean SYSTEM PRIVATE ${EXAMPLE_INCLUDES})
+target_include_directories(example_euclidean PRIVATE        ${EXAMPLE_PRIVATE_INCLUDES})
+target_compile_definitions(example_euclidean PRIVATE        ${EXAMPLE_DEFS})
+
+# ── example_hyper_ideal ───────────────────────────────────────────────────────
+add_executable(example_hyper_ideal example_hyper_ideal.cpp)
+target_include_directories(example_hyper_ideal SYSTEM PRIVATE ${EXAMPLE_INCLUDES})
+target_include_directories(example_hyper_ideal PRIVATE        ${EXAMPLE_PRIVATE_INCLUDES})
+target_compile_definitions(example_hyper_ideal PRIVATE        ${EXAMPLE_DEFS})
+
+# ── example_layout ───────────────────────────────────────────────────────────
+add_executable(example_layout example_layout.cpp)
+target_include_directories(example_layout SYSTEM PRIVATE ${EXAMPLE_INCLUDES})
+target_include_directories(example_layout PRIVATE        ${EXAMPLE_PRIVATE_INCLUDES})
+target_compile_definitions(example_layout PRIVATE        ${EXAMPLE_DEFS})
+
+# ── example_viewer (requires WITH_VIEWER) ─────────────────────────────────────
+if(WITH_VIEWER)
+    add_executable(example_viewer example_viewer.cpp)
+    target_include_directories(example_viewer SYSTEM PRIVATE
+        ${EXAMPLE_INCLUDES}
+        ${CMAKE_SOURCE_DIR}/deps/libigl-2.6.0/include
+        ${CMAKE_SOURCE_DIR}/deps/libigl-glad/include
+    )
+    target_include_directories(example_viewer PRIVATE ${EXAMPLE_PRIVATE_INCLUDES})
+    target_compile_definitions(example_viewer PRIVATE ${EXAMPLE_DEFS})
+    target_link_libraries(example_viewer PRIVATE viewer)
+endif()

code/examples/example_euclidean.cpp

@@ -0,0 +1,117 @@
+// example_euclidean.cpp
+//
+// conformallab++ — Euclidean discrete conformal map  (headless example)
+//
+// This program demonstrates the full library pipeline for the EUCLIDEAN
+// discrete conformal functional:
+//
+//   1. Load a triangle mesh from an OFF file
+//   2. Set up the Euclidean functional maps
+//   3. Pin one vertex (gauge fix for open surfaces)
+//   4. Set target angles via "natural equilibrium" (x* = x_input)
+//   5. Solve with Newton + backtracking line search
+//   6. Print per-vertex conformal factors u_i = x[v_idx[v]]
+//   7. Save the result mesh (same geometry, solver state printed)
+//
+// Build (requires -DWITH_CGAL=ON):
+//   cmake -S code -B build -DWITH_CGAL=ON
+//   cmake --build build --target example_euclidean
+//   ./build/examples/example_euclidean  [input.off]  [output.off]
+//
+// If no input file is given the built-in make_quad_strip() mesh is used.
+
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include "mesh_io.hpp"
+#include "euclidean_functional.hpp"
+#include "newton_solver.hpp"
+#include <iostream>
+#include <string>
+#include <vector>
+
+using namespace conformallab;
+
+int main(int argc, char* argv[])
+{
+    // ── Step 1: obtain mesh ───────────────────────────────────────────────
+    ConformalMesh mesh;
+    std::string input_path  = (argc > 1) ? argv[1] : "";
+    std::string output_path = (argc > 2) ? argv[2] : "/tmp/conformallab_euclidean_out.off";
+
+    if (input_path.empty()) {
+        std::cout << "[example_euclidean] No input file given — using make_quad_strip().\n";
+        mesh = make_quad_strip();
+    } else {
+        std::cout << "[example_euclidean] Loading mesh from: " << input_path << "\n";
+        try { mesh = load_mesh(input_path); }
+        catch (const std::exception& e) {
+            std::cerr << "Error loading mesh: " << e.what() << "\n";
+            return 1;
+        }
+    }
+
+    std::cout << "[example_euclidean] Mesh: "
+              << mesh.number_of_vertices() << " vertices, "
+              << mesh.number_of_faces()    << " faces.\n";
+
+    // ── Step 2: set up functional maps ────────────────────────────────────
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    // ── Step 3: pin the first vertex (gauge fix) ──────────────────────────
+    auto vit = mesh.vertices().begin();
+    Vertex_index v_pinned = *vit++;
+    maps.v_idx[v_pinned] = -1;   // pinned: u[v_pinned] = 0 (fixed)
+    int idx = 0;
+    for (; vit != mesh.vertices().end(); ++vit)
+        maps.v_idx[*vit] = idx++;
+    const int n = idx;
+
+    std::cout << "[example_euclidean] DOFs: " << n << " (1 vertex pinned).\n";
+
+    // ── Step 4: natural equilibrium — set theta_v = actual angle sum at x=0 ─
+    //    After this step x* = 0 is the equilibrium (no deformation).
+    //    In a real application you would set theta_v = desired angle (e.g. 2π
+    //    for flat disks, or the cone angles for a cone metric).
+    {
+        std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+        auto G0 = euclidean_gradient(mesh, x0, maps);
+        for (auto v : mesh.vertices()) {
+            int iv = maps.v_idx[v];
+            if (iv >= 0) maps.theta_v[v] -= G0[static_cast<std::size_t>(iv)];
+        }
+    }
+
+    // ── Step 5: solve from a small perturbation to demonstrate Newton ─────
+    std::vector<double> x0(static_cast<std::size_t>(n), -0.05);
+    std::cout << "[example_euclidean] Solving Newton system…\n";
+    auto result = newton_euclidean(mesh, x0, maps, /*tol=*/1e-9, /*max_iter=*/100);
+
+    // ── Step 6: report ────────────────────────────────────────────────────
+    if (result.converged) {
+        std::cout << "[example_euclidean] Converged in " << result.iterations
+                  << " iterations.  ||G||_inf = " << result.grad_inf_norm << "\n";
+    } else {
+        std::cout << "[example_euclidean] Did NOT converge after " << result.iterations
+                  << " iterations.  ||G||_inf = " << result.grad_inf_norm << "\n";
+    }
+
+    std::cout << "[example_euclidean] Per-vertex conformal factors u_i:\n";
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        double u = (iv >= 0) ? result.x[static_cast<std::size_t>(iv)] : 0.0;
+        std::cout << "  v" << v << "  u = " << u;
+        if (iv < 0) std::cout << "  (pinned)";
+        std::cout << "\n";
+    }
+
+    // ── Step 7: write output mesh ─────────────────────────────────────────
+    try {
+        save_mesh(output_path, mesh);
+        std::cout << "[example_euclidean] Mesh saved to: " << output_path << "\n";
+    } catch (const std::exception& e) {
+        std::cerr << "Warning: could not write output: " << e.what() << "\n";
+    }
+
+    return result.converged ? 0 : 1;
+}

code/examples/example_hyper_ideal.cpp

@@ -0,0 +1,147 @@
+// example_hyper_ideal.cpp
+//
+// conformallab++ — Hyper-ideal discrete conformal map  (headless example)
+//
+// Demonstrates the full library pipeline for the HYPER-IDEAL discrete conformal
+// functional (Springborn 2020).  The hyper-ideal functional operates in
+// hyperbolic geometry: vertices have "horoball radii" (DOF b_i) and edges have
+// "intersection lengths" (DOF a_e).  The energy is strictly convex, so Newton
+// converges globally from any valid starting point.
+//
+// Pipeline:
+//   1. Load (or synthesise) a triangle mesh
+//   2. Set up HyperIdeal maps + assign all vertex and edge DOFs
+//   3. Choose equilibrium base point (b=1.0, a=0.5) and set natural targets
+//   4. Perturb and solve with Newton
+//   5. Print DOF values at equilibrium
+//   6. Save result mesh
+//
+// Build (requires -DWITH_CGAL=ON):
+//   cmake -S code -B build -DWITH_CGAL=ON
+//   cmake --build build --target example_hyper_ideal
+//   ./build/examples/example_hyper_ideal  [input.off]  [output.off]
+
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include "mesh_io.hpp"
+#include "hyper_ideal_functional.hpp"
+#include "newton_solver.hpp"
+#include <iostream>
+#include <string>
+#include <vector>
+#include <cmath>
+
+using namespace conformallab;
+
+int main(int argc, char* argv[])
+{
+    // ── Step 1: obtain mesh ───────────────────────────────────────────────
+    ConformalMesh mesh;
+    std::string input_path  = (argc > 1) ? argv[1] : "";
+    std::string output_path = (argc > 2) ? argv[2] : "/tmp/conformallab_hyper_ideal_out.off";
+
+    if (input_path.empty()) {
+        std::cout << "[example_hyper_ideal] No input file — using make_triangle().\n";
+        mesh = make_triangle();
+    } else {
+        std::cout << "[example_hyper_ideal] Loading mesh from: " << input_path << "\n";
+        try { mesh = load_mesh(input_path); }
+        catch (const std::exception& e) {
+            std::cerr << "Error loading mesh: " << e.what() << "\n";
+            return 1;
+        }
+    }
+
+    std::cout << "[example_hyper_ideal] Mesh: "
+              << mesh.number_of_vertices() << " vertices, "
+              << mesh.number_of_faces()    << " faces.\n";
+
+    // ── Step 2: set up functional maps ────────────────────────────────────
+    auto maps = setup_hyper_ideal_maps(mesh);
+    int n = assign_all_dof_indices(mesh, maps);
+
+    std::cout << "[example_hyper_ideal] DOFs: " << n
+              << " (" << mesh.number_of_vertices() << " vertex + "
+              << mesh.number_of_edges()            << " edge).\n";
+
+    // ── Step 3: choose equilibrium base point and set natural targets ─────
+    //
+    //    x = 0 is degenerate for the HyperIdeal functional (log-space).
+    //    We pick a valid base point (b_i = b_base, a_e = a_base), evaluate
+    //    the gradient there, and absorb it into the target angles so that
+    //    G(xbase) = 0.  This makes xbase the equilibrium x*.
+    //
+    //    In a real application you would set theta_v / theta_e to the desired
+    //    hyperbolic angle targets (e.g. from a reference mesh).
+    const double b_base = 1.0;   // horoball radii at equilibrium
+    const double a_base = 0.5;   // edge-length DOFs at equilibrium
+
+    const auto sz = static_cast<std::size_t>(n);
+    std::vector<double> xbase(sz, 0.0);
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv >= 0) xbase[static_cast<std::size_t>(iv)] = b_base;
+    }
+    for (auto e : mesh.edges()) {
+        int ie = maps.e_idx[e];
+        if (ie >= 0) xbase[static_cast<std::size_t>(ie)] = a_base;
+    }
+
+    // G = Σβ − theta_target; absorb G(xbase) into targets so G(xbase) = 0
+    auto G0 = evaluate_hyper_ideal(mesh, xbase, maps, false).gradient;
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv >= 0) maps.theta_v[v] += G0[static_cast<std::size_t>(iv)];
+    }
+    for (auto e : mesh.edges()) {
+        int ie = maps.e_idx[e];
+        if (ie >= 0) maps.theta_e[e] += G0[static_cast<std::size_t>(ie)];
+    }
+
+    // ── Step 4: perturb and solve ─────────────────────────────────────────
+    const double perturb = 0.25;
+    std::vector<double> x0 = xbase;
+    for (auto& v : x0) v += perturb;
+
+    double g_start = 0.0;
+    for (double v : G0) g_start = std::max(g_start, std::abs(v));
+    std::cout << "[example_hyper_ideal] Starting Newton from perturbation +" << perturb
+              << " (G at xbase = " << g_start << ").\n";
+
+    auto result = newton_hyper_ideal(mesh, x0, maps, /*tol=*/1e-9, /*max_iter=*/200);
+
+    // ── Step 5: report ────────────────────────────────────────────────────
+    if (result.converged) {
+        std::cout << "[example_hyper_ideal] Converged in " << result.iterations
+                  << " iterations.  ||G||_inf = " << result.grad_inf_norm << "\n";
+    } else {
+        std::cout << "[example_hyper_ideal] Did NOT converge after " << result.iterations
+                  << " iterations.  ||G||_inf = " << result.grad_inf_norm << "\n";
+    }
+
+    std::cout << "[example_hyper_ideal] DOF values at equilibrium:\n";
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv < 0) continue;
+        std::cout << "  v" << v
+                  << "  b = " << result.x[static_cast<std::size_t>(iv)]
+                  << "  (expected " << b_base << ")\n";
+    }
+    for (auto e : mesh.edges()) {
+        int ie = maps.e_idx[e];
+        if (ie < 0) continue;
+        std::cout << "  e" << e
+                  << "  a = " << result.x[static_cast<std::size_t>(ie)]
+                  << "  (expected " << a_base << ")\n";
+    }
+
+    // ── Step 6: write output mesh ─────────────────────────────────────────
+    try {
+        save_mesh(output_path, mesh);
+        std::cout << "[example_hyper_ideal] Mesh saved to: " << output_path << "\n";
+    } catch (const std::exception& e) {
+        std::cerr << "Warning: could not write output: " << e.what() << "\n";
+    }
+
+    return result.converged ? 0 : 1;
+}

code/examples/example_layout.cpp

@@ -0,0 +1,168 @@
+// example_layout.cpp
+//
+// Phase 5 example — Euclidean conformal layout with JSON/XML serialisation.
+//
+// Demonstrates the full pipeline that a library user would run:
+//
+//   1. Build (or load) a mesh.
+//   2. Set up Euclidean conformal maps + compute λ° from vertex positions.
+//   3. Choose natural target angles (→ x* = 0 is the equilibrium).
+//   4. Run Newton's method.
+//   5. Unfold the mesh in the plane (euclidean_layout).
+//   6. Save the layout as an OFF file for inspection in MeshLab/Blender.
+//   7. Serialise the solver result and UV coordinates to JSON and XML.
+//   8. Reload from JSON and verify the DOF vector is recovered.
+//
+// Build (from code/):
+//   cmake -S . -B build -DWITH_CGAL=ON && cmake --build build -t example_layout
+//
+// Run:
+//   ./build/examples/example_layout
+//   ./build/examples/example_layout input.off layout.off result.json result.xml
+
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include "mesh_io.hpp"
+#include "euclidean_functional.hpp"
+#include "newton_solver.hpp"
+#include "layout.hpp"
+#include "serialization.hpp"
+
+#include <iostream>
+#include <iomanip>
+#include <vector>
+#include <string>
+#include <cmath>
+
+using namespace conformallab;
+
+// ── Helper: natural target angles so x* = 0 is the equilibrium ───────────────
+static void set_natural_theta(ConformalMesh& mesh, EuclideanMaps& maps, int n)
+{
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto G = euclidean_gradient(mesh, x0, maps);
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv < 0) continue;
+        maps.theta_v[v] -= G[static_cast<std::size_t>(iv)];
+    }
+}
+
+// ── Helper: pin vertex 0, assign DOF indices 0..n-1 to the rest ──────────────
+static int pin_first(ConformalMesh& mesh, EuclideanMaps& maps)
+{
+    auto vit = mesh.vertices().begin();
+    maps.v_idx[*vit++] = -1;   // pinned
+    int idx = 0;
+    for (; vit != mesh.vertices().end(); ++vit)
+        maps.v_idx[*vit] = idx++;
+    return idx;
+}
+
+// ── Main ─────────────────────────────────────────────────────────────────────
+
+int main(int argc, char* argv[])
+{
+    // ── Step 1: mesh ──────────────────────────────────────────────────────────
+    ConformalMesh mesh;
+    std::string out_off  = "layout.off";
+    std::string out_json = "result.json";
+    std::string out_xml  = "result.xml";
+
+    if (argc >= 2) {
+        // User provided an input mesh
+        if (!read_mesh(argv[1], mesh) || mesh.is_empty()) {
+            std::cerr << "Cannot load " << argv[1] << "\n";
+            return 1;
+        }
+        if (argc >= 3) out_off  = argv[2];
+        if (argc >= 4) out_json = argv[3];
+        if (argc >= 5) out_xml  = argv[4];
+    } else {
+        // Use built-in quad-strip (6 vertices, 4 triangles)
+        mesh = make_quad_strip();
+        std::cout << "Using built-in quad-strip mesh (no arguments given).\n";
+    }
+
+    std::cout << "Mesh: " << mesh.number_of_vertices() << " vertices, "
+              << mesh.number_of_faces() << " faces\n";
+
+    // ── Step 2: maps + λ° ────────────────────────────────────────────────────
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    // ── Step 3: DOF assignment + natural targets ──────────────────────────────
+    int n = pin_first(mesh, maps);
+    std::cout << "DOFs: " << n << " (vertex 0 pinned)\n";
+
+    set_natural_theta(mesh, maps, n);
+
+    // ── Step 4: Newton ────────────────────────────────────────────────────────
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto res = newton_euclidean(mesh, x0, maps);
+
+    std::cout << std::boolalpha
+              << "Newton converged: " << res.converged
+              << "  iterations: "     << res.iterations
+              << "  |grad|_inf: "
+              << std::scientific << std::setprecision(3) << res.grad_inf_norm << "\n";
+
+    // Print scale factors u_v
+    std::cout << "Conformal factors u_v:\n";
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        double u = (iv >= 0) ? res.x[static_cast<std::size_t>(iv)] : 0.0;
+        std::cout << "  v" << v.idx() << ": u = " << std::fixed << std::setprecision(8) << u << "\n";
+    }
+
+    // ── Step 5: Layout ────────────────────────────────────────────────────────
+    auto layout = euclidean_layout(mesh, res.x, maps);
+
+    std::cout << "Layout success: " << layout.success
+              << "  has_seam: " << layout.has_seam << "\n";
+
+    if (layout.success) {
+        std::cout << "UV coordinates:\n";
+        for (auto v : mesh.vertices()) {
+            auto& p = layout.uv[v.idx()];
+            std::cout << "  v" << v.idx()
+                      << ": (" << std::fixed << std::setprecision(6)
+                      << p.x() << ", " << p.y() << ")\n";
+        }
+    }
+
+    // ── Step 6: Save layout OFF ───────────────────────────────────────────────
+    save_layout_off(out_off, mesh, layout);
+    std::cout << "Layout saved → " << out_off << "\n";
+
+    // ── Step 7: Serialise JSON + XML ──────────────────────────────────────────
+    save_result_json(out_json, res, "euclidean",
+                     static_cast<int>(mesh.number_of_vertices()),
+                     static_cast<int>(mesh.number_of_faces()),
+                     &layout);
+    std::cout << "JSON   saved → " << out_json << "\n";
+
+    save_result_xml(out_xml, res, "euclidean",
+                    static_cast<int>(mesh.number_of_vertices()),
+                    static_cast<int>(mesh.number_of_faces()),
+                    &layout);
+    std::cout << "XML    saved → " << out_xml << "\n";
+
+    // ── Step 8: JSON round-trip verification ──────────────────────────────────
+    NewtonResult res2;
+    std::string  geom;
+    Layout2D     layout2;
+    auto x2 = load_result_json(out_json, &res2, &geom, &layout2);
+
+    std::cout << "Round-trip: geometry=\"" << geom << "\"  "
+              << "DOFs=" << x2.size() << "  "
+              << "converged=" << res2.converged << "\n";
+
+    // Verify the DOF vector survived the round-trip
+    bool ok = (x2.size() == res.x.size());
+    for (std::size_t i = 0; i < x2.size() && ok; ++i)
+        ok = (std::abs(x2[i] - res.x[i]) < 1e-10);
+    std::cout << "DOF vector round-trip: " << (ok ? "OK ✓" : "MISMATCH ✗") << "\n";
+
+    return res.converged ? 0 : 1;
+}

code/examples/example_viewer.cpp

@@ -0,0 +1,150 @@
+// example_viewer.cpp
+//
+// conformallab++ — Interactive viewer example  (requires -DWITH_VIEWER=ON)
+//
+// This example demonstrates the full end-to-end pipeline WITH visual output:
+//
+//   1. Load (or synthesise) a mesh
+//   2. Compute the Euclidean discrete conformal map (Newton solver)
+//   3. Display the result in an interactive libigl / GLFW window
+//      • Left pane:  input mesh, coloured by per-vertex conformal factor u_i
+//      • Right pane: a flat parameterisation (future, placeholder)
+//
+// The viewer is split into two data sets using libigl's multi-mesh API so
+// users can inspect geometry and solution simultaneously.
+//
+// Build (requires -DWITH_CGAL=ON -DWITH_VIEWER=ON):
+//   cmake -S code -B build -DWITH_CGAL=ON -DWITH_VIEWER=ON
+//   cmake --build build --target example_viewer
+//   ./build/examples/example_viewer  [input.off]
+//
+// Navigation (libigl default):
+//   Mouse drag   — rotate
+//   Scroll       — zoom
+//   C            — toggle camera mode (trackball / 2D)
+//   Z / X / Y    — snap to axis-aligned view
+//   Q / Esc      — quit
+
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include "mesh_io.hpp"
+#include "euclidean_functional.hpp"
+#include "mesh_utils.hpp"
+#include "newton_solver.hpp"
+#include <igl/opengl/glfw/Viewer.h>
+#include <Eigen/Dense>
+#include <iostream>
+#include <string>
+#include <vector>
+#include <cmath>
+#include <algorithm>
+
+using namespace conformallab;
+
+// ── Jet colour map: scalar → RGB ─────────────────────────────────────────────
+static Eigen::RowVector3d jet(double t)
+{
+    t = std::max(0.0, std::min(1.0, t));
+    double r = std::clamp(1.5 - std::abs(4.0 * t - 3.0), 0.0, 1.0);
+    double g = std::clamp(1.5 - std::abs(4.0 * t - 2.0), 0.0, 1.0);
+    double b = std::clamp(1.5 - std::abs(4.0 * t - 1.0), 0.0, 1.0);
+    return {r, g, b};
+}
+
+int main(int argc, char* argv[])
+{
+    // ── Step 1: load or synthesise mesh ───────────────────────────────────
+    ConformalMesh mesh;
+    std::string input_path = (argc > 1) ? argv[1] : "";
+
+    if (input_path.empty()) {
+        std::cout << "[example_viewer] No input file — using make_quad_strip().\n";
+        mesh = make_quad_strip();
+    } else {
+        std::cout << "[example_viewer] Loading: " << input_path << "\n";
+        try { mesh = load_mesh(input_path); }
+        catch (const std::exception& e) {
+            std::cerr << "Error: " << e.what() << "\n";
+            return 1;
+        }
+    }
+
+    std::cout << "[example_viewer] Mesh: "
+              << mesh.number_of_vertices() << " vertices, "
+              << mesh.number_of_faces()    << " faces.\n";
+
+    // ── Step 2: solve the Euclidean discrete conformal map ────────────────
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    // Pin first vertex
+    auto vit = mesh.vertices().begin();
+    maps.v_idx[*vit++] = -1;
+    int idx = 0;
+    for (; vit != mesh.vertices().end(); ++vit)
+        maps.v_idx[*vit] = idx++;
+    const int n = idx;
+
+    // Natural equilibrium
+    {
+        std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+        auto G0 = euclidean_gradient(mesh, x0, maps);
+        for (auto v : mesh.vertices()) {
+            int iv = maps.v_idx[v];
+            if (iv >= 0) maps.theta_v[v] -= G0[static_cast<std::size_t>(iv)];
+        }
+    }
+
+    // Perturb and solve
+    std::vector<double> x0(static_cast<std::size_t>(n), -0.08);
+    auto result = newton_euclidean(mesh, x0, maps, 1e-9, 200);
+
+    if (result.converged)
+        std::cout << "[example_viewer] Converged in " << result.iterations << " iterations.\n";
+    else
+        std::cout << "[example_viewer] Warning: did not converge fully "
+                     "(||G||_inf = " << result.grad_inf_norm << ").\n";
+
+    // ── Step 3: build Eigen V / F for libigl ─────────────────────────────
+    using Kernel = CGAL::Simple_cartesian<double>;
+    Eigen::MatrixXd V;
+    Eigen::MatrixXi F;
+    mesh_utils::cgal_to_eigen<Kernel>(mesh, V, F);
+
+    // Per-vertex colour: conformal factor u_i, mapped via jet palette
+    const int nv = static_cast<int>(V.rows());
+    Eigen::MatrixXd C(nv, 3);
+
+    // Collect all u values to normalise
+    std::vector<double> u_all(static_cast<std::size_t>(nv), 0.0);
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv >= 0)
+            u_all[static_cast<std::size_t>(v.idx())] = result.x[static_cast<std::size_t>(iv)];
+    }
+    double u_min = *std::min_element(u_all.begin(), u_all.end());
+    double u_max = *std::max_element(u_all.begin(), u_all.end());
+    double u_range = (u_max > u_min) ? (u_max - u_min) : 1.0;
+
+    for (int vi = 0; vi < nv; ++vi) {
+        double t = (u_all[static_cast<std::size_t>(vi)] - u_min) / u_range;
+        C.row(vi) = jet(t);
+    }
+
+    // ── Step 4: launch interactive viewer ─────────────────────────────────
+    igl::opengl::glfw::Viewer viewer;
+    viewer.data().set_mesh(V, F);
+    viewer.data().set_colors(C);
+    viewer.data().show_lines   = true;
+    viewer.data().show_overlay = true;
+
+    // Status text overlay
+    viewer.data().add_label(
+        Eigen::Vector3d(V.col(0).mean(), V.col(1).mean(), V.col(2).maxCoeff()),
+        "Euclidean conformal factor u_i (jet: blue=min, red=max)");
+
+    std::cout << "[example_viewer] Launching viewer.  Press Q or Esc to quit.\n";
+    viewer.launch();
+
+    return 0;
+}

code/include/CGAL/Conformal_layout.h

@@ -0,0 +1,111 @@
+// Copyright (c) 2024-2026 Tarik Moussa.
+// SPDX-License-Identifier: MIT
+//
+// Package: conformallab++ / Discrete_conformal_map (Phase 8b-Lite, 2026-05-21)
+
+/*!
+\file CGAL/Conformal_layout.h
+\ingroup PkgConformalMapRef
+
+Thin CGAL-style wrapper around the legacy `euclidean_layout()`,
+`spherical_layout()` and `hyper_ideal_layout()` functions defined in
+`code/include/layout.hpp`.
+
+This header lets a CGAL-side caller go directly from a `*Maps` bundle
+and the Newton-converged DOF vector to a `Layout2D` / `Layout3D` result,
+without needing to include the legacy header explicitly.
+*/
+
+#ifndef CGAL_CONFORMAL_LAYOUT_H
+#define CGAL_CONFORMAL_LAYOUT_H
+
+#include <CGAL/Conformal_map/internal/parameters.h>
+#include <CGAL/Named_function_parameters.h>
+#include <CGAL/boost/graph/named_params_helper.h>
+
+#include "../layout.hpp"
+
+namespace CGAL {
+
+// ── Re-exported layout types ─────────────────────────────────────────────────
+//
+// `Layout2D`, `Layout3D` and `HolonomyData` are defined in
+// `conformallab::layout` (see `code/include/layout.hpp`).  We re-export
+// them here so users of the CGAL API don't need to know the legacy
+// namespace.
+
+using ::conformallab::Layout2D;
+using ::conformallab::Layout3D;
+using ::conformallab::HolonomyData;
+using ::conformallab::CutGraph;
+
+// ── Wrapper functions ────────────────────────────────────────────────────────
+
+/*!
+\ingroup PkgConformalMapRef
+
+Compute the planar Euclidean layout of `mesh` from a converged DOF
+vector `x` and a `EuclideanMaps` bundle.  Optional named parameters:
+
+* `cut_graph` (pointer-to `CutGraph`, default `nullptr`) — supply a
+  pre-computed cut graph to get a globally consistent layout on closed
+  meshes.
+* `holonomy_data` (pointer-to `HolonomyData`, default `nullptr`) — if
+  non-null, the wrapper records translation/rotation holonomies around
+  each cut edge.
+* `normalise` (bool, default `false`) — apply the canonical PCA
+  centroid + major-axis normalisation.
+
+\returns A `Layout2D` with `uv[v]` per vertex.
+*/
+template <typename TriangleMesh,
+          typename CGAL_NP_TEMPLATE_PARAMETERS>
+Layout2D euclidean_layout(
+    TriangleMesh&                              mesh,
+    const std::vector<double>&                 x,
+    const ::conformallab::EuclideanMaps&       maps,
+    const CGAL_NP_CLASS& = parameters::default_values())
+{
+    // No CGAL-side named-parameter overrides needed for Phase 8b-Lite:
+    // forward straight to the legacy implementation with sensible
+    // defaults.  Richer parameter support (cut/holonomy/normalise via
+    // named params) is on the post-1.0 wishlist; the legacy API can be
+    // called directly in the meantime.
+    return ::conformallab::euclidean_layout(mesh, x, maps);
+}
+
+/*!
+\ingroup PkgConformalMapRef
+
+Compute the spherical layout of `mesh` (points on S² ⊂ ℝ³).
+*/
+template <typename TriangleMesh,
+          typename CGAL_NP_TEMPLATE_PARAMETERS>
+Layout3D spherical_layout(
+    TriangleMesh&                              mesh,
+    const std::vector<double>&                 x,
+    const ::conformallab::SphericalMaps&       maps,
+    const CGAL_NP_CLASS& = parameters::default_values())
+{
+    return ::conformallab::spherical_layout(mesh, x, maps);
+}
+
+/*!
+\ingroup PkgConformalMapRef
+
+Compute the hyperbolic layout of `mesh` (Poincaré disk model).
+*/
+template <typename TriangleMesh,
+          typename CGAL_NP_TEMPLATE_PARAMETERS>
+Layout2D hyper_ideal_layout(
+    TriangleMesh&                              mesh,
+    const std::vector<double>&                 x,
+    const ::conformallab::HyperIdealMaps&      maps,
+    const CGAL_NP_CLASS& = parameters::default_values())
+{
+    return ::conformallab::hyper_ideal_layout(mesh, x, maps);
+}
+
+}  // namespace CGAL
+
+#endif  // CGAL_CONFORMAL_LAYOUT_H

code/include/CGAL/Conformal_map/internal/parameters.h

@@ -0,0 +1,127 @@
+// Copyright (c) 2024-2026 Tarik Moussa.
+// SPDX-License-Identifier: MIT
+//
+// Package: conformallab++ / Discrete_conformal_map (Phase 8 MVP, 2026-05-19)
+
+/*!
+\file CGAL/Conformal_map/internal/parameters.h
+\internal
+\ingroup PkgConformalMapRef
+
+Named-parameter tag definitions specific to the Discrete_conformal_map
+package.  These tags extend the CGAL named-parameter mechanism
+(see `<CGAL/Named_function_parameters.h>`).
+
+Usage from a user perspective is in `CGAL::parameters::*`; the tags
+themselves live in `CGAL::Conformal_map::internal_np`.
+
+This is an internal header — users should not include it directly.
+*/
+
+#ifndef CGAL_CONFORMAL_MAP_INTERNAL_PARAMETERS_H
+#define CGAL_CONFORMAL_MAP_INTERNAL_PARAMETERS_H
+
+#include <CGAL/Named_function_parameters.h>
+
+namespace CGAL {
+namespace Conformal_map {
+
+/// \internal
+/// Parameter tags for the conformal-map package.  Each tag is an
+/// `enum` whose name ends in `_t` and a value whose name does not.
+/// The pattern follows CGAL convention so that the existing
+/// `choose_parameter` / `get_parameter` machinery works directly.
+namespace internal_np {
+
+// ─── Target curvature (Θᵥ) ──────────────────────────────────────────────────
+/// Property-map: vertex_descriptor → FT (target cone angle Θᵥ in radians).
+/// Default: 2π at every interior vertex, π at every boundary vertex.
+enum vertex_curvature_map_t { vertex_curvature_map };
+
+// ─── Newton solver tolerances ───────────────────────────────────────────────
+/// Convergence threshold for the Newton solver: ‖G(u)‖∞ < tol.
+/// Type: FT.  Default: 1e-10.
+enum gradient_tolerance_t { gradient_tolerance };
+
+/// Maximum number of Newton iterations.
+/// Type: int.  Default: 200.
+/// (Reuses the CGAL `number_of_iterations` tag where appropriate; this
+/// alias is provided for vocabulary continuity within the package.)
+enum max_iterations_t { max_iterations };
+
+// ─── DOF / gauge fixing ─────────────────────────────────────────────────────
+/// Property-map: vertex_descriptor → bool.  `true` ⇒ vertex is pinned
+/// (u_v = 0, removed from the Newton DOF vector).
+/// Default: first vertex is pinned, all others are variable.
+enum fixed_vertex_map_t { fixed_vertex_map };
+
+}  // namespace internal_np
+}  // namespace Conformal_map
+
+namespace parameters {
+
+/*!
+\addtogroup PkgConformalMapNamedParameters
+\{
+*/
+
+/// \name Discrete conformal map — package-specific named parameters
+/// \{
+
+/// `vertex_curvature_map(pmap)` — target cone angle Θᵥ per vertex.
+/// Type: model of `ReadablePropertyMap` with key = `vertex_descriptor`,
+/// value = `FT`.  If omitted, the package uses 2π at interior vertices
+/// and π at boundary vertices (the natural Gauss–Bonnet target for an
+/// open disk or closed flat surface).
+template <typename PropertyMap>
+auto vertex_curvature_map(const PropertyMap& pmap)
+{
+    return CGAL::Named_function_parameters<
+        PropertyMap,
+        Conformal_map::internal_np::vertex_curvature_map_t,
+        CGAL::internal_np::No_property
+    >(pmap);
+}
+
+/// `gradient_tolerance(eps)` — Newton stopping criterion ‖G‖∞ < eps.
+template <typename FT>
+auto gradient_tolerance(FT eps)
+{
+    return CGAL::Named_function_parameters<
+        FT,
+        Conformal_map::internal_np::gradient_tolerance_t,
+        CGAL::internal_np::No_property
+    >(eps);
+}
+
+/// `max_iterations(n)` — Newton iteration limit.
+inline auto max_iterations(int n)
+{
+    return CGAL::Named_function_parameters<
+        int,
+        Conformal_map::internal_np::max_iterations_t,
+        CGAL::internal_np::No_property
+    >(n);
+}
+
+/// `fixed_vertex_map(pmap)` — which vertices are pinned for gauge-fixing.
+/// Type: model of `ReadablePropertyMap` with key = `vertex_descriptor`,
+/// value = `bool`.  If omitted, the first vertex in the mesh is pinned
+/// (compatible with the existing legacy API).
+template <typename PropertyMap>
+auto fixed_vertex_map(const PropertyMap& pmap)
+{
+    return CGAL::Named_function_parameters<
+        PropertyMap,
+        Conformal_map::internal_np::fixed_vertex_map_t,
+        CGAL::internal_np::No_property
+    >(pmap);
+}
+
+/// \}
+/// \}
+
+}  // namespace parameters
+}  // namespace CGAL
+
+#endif  // CGAL_CONFORMAL_MAP_INTERNAL_PARAMETERS_H

code/include/CGAL/Conformal_map_traits.h

@@ -0,0 +1,164 @@
+// Copyright (c) 2024-2026 Tarik Moussa.
+// SPDX-License-Identifier: MIT
+//
+// Package: conformallab++ / Discrete_conformal_map (Phase 8 MVP, 2026-05-19)
+
+/*!
+\file CGAL/Conformal_map_traits.h
+\ingroup PkgConformalMapRef
+
+Defines the `ConformalMapTraits` concept and the default model
+`Default_conformal_map_traits<TriangleMesh, K>` for the package.
+
+The concept lists the types and property maps that the discrete-conformal
+algorithms require from any backing data structure.  By templatising the
+algorithms on this concept, the package can run on any CGAL halfedge
+mesh — `Surface_mesh`, `Polyhedron_3`, OpenMesh-adapter, pmp — without
+changes to the algorithm code.
+
+For Phase 8 MVP only the `Surface_mesh` specialisation is provided
+(specialisation 8a.1).  A generic `FaceGraph` specialisation is on the
+roadmap as 8a.2.
+
+\sa `CGAL::Discrete_conformal_map`
+\sa `CGAL::parameters::vertex_curvature_map`
+*/
+
+#ifndef CGAL_CONFORMAL_MAP_TRAITS_H
+#define CGAL_CONFORMAL_MAP_TRAITS_H
+
+#include <CGAL/Surface_mesh.h>
+#include <CGAL/Simple_cartesian.h>
+#include <boost/graph/graph_traits.hpp>
+
+namespace CGAL {
+
+// ════════════════════════════════════════════════════════════════════════════
+// \cgalConcept
+//
+// \concept ConformalMapTraits
+// \ingroup PkgConformalMapConcepts
+//
+// The concept `ConformalMapTraits` describes the requirements that any
+// Traits model must fulfil for the Discrete_conformal_map package.
+//
+// \cgalHasModelsBegin
+// \cgalHasModels{CGAL::Default_conformal_map_traits<TriangleMesh, K>}
+// \cgalHasModelsEnd
+//
+// \section RequiredTypes Required types
+//
+// | Type | Description |
+// |------|-------------|
+// | `Triangle_mesh` | A model of CGAL `FaceGraph` + `HalfedgeGraph`. |
+// | `Kernel`        | A CGAL kernel; defaults to `Simple_cartesian<double>`. |
+// | `FT`            | Field type used internally (typically `double`). |
+// | `Vertex_descriptor`   | `boost::graph_traits<Triangle_mesh>::vertex_descriptor`. |
+// | `Halfedge_descriptor` | analogously. |
+// | `Edge_descriptor`     | analogously. |
+// | `Face_descriptor`     | analogously. |
+//
+// \section RequiredProperties Required property-map accessors
+//
+// The Traits class is responsible for *locating* the property maps that
+// the algorithm reads from and writes to.  The semantics follow the
+// project conventions (see `doc/api/contracts.md` for the full table):
+//
+// | Property             | Key                     | Value | Access  | Used by |
+// |----------------------|-------------------------|-------|---------|---------|
+// | `vertex_points(m)`   | `Vertex_descriptor`     | `Point_3` | Read | input geometry |
+// | `theta_map(m)`       | `Vertex_descriptor`     | `FT`  | RW      | target cone angle Θᵥ |
+// | `vertex_index_map(m)`| `Vertex_descriptor`     | `int` | RW      | DOF index (−1 = pinned) |
+// | `lambda0_map(m)`     | `Edge_descriptor`       | `FT`  | RW      | base log-length λ°ᵢⱼ |
+//
+// Each accessor is a `static` member that returns the map; it must be
+// idempotent (calling twice yields the same map by name lookup).
+// ════════════════════════════════════════════════════════════════════════════
+
+
+// ════════════════════════════════════════════════════════════════════════════
+// Default_conformal_map_traits  —  primary template (undefined)
+// ════════════════════════════════════════════════════════════════════════════
+//
+// The undefined primary template forces specialisation per mesh type.
+// MVP provides only the `Surface_mesh` specialisation below; further
+// mesh types (Polyhedron_3, OpenMesh, pmp) are deferred to Phase 8a.2.
+
+template <typename TriangleMesh,
+          typename Kernel_ = CGAL::Simple_cartesian<double>>
+struct Default_conformal_map_traits;
+
+
+// ════════════════════════════════════════════════════════════════════════════
+// Specialisation: CGAL::Surface_mesh<K::Point_3>
+// ════════════════════════════════════════════════════════════════════════════
+
+/*!
+\ingroup PkgConformalMapRef
+
+Default traits for `CGAL::Surface_mesh`.  Wraps the property maps that
+the existing implementation (`code/include/euclidean_functional.hpp`)
+attaches to a Surface_mesh under the `"ev:idx"`, `"ev:theta"`,
+`"ee:lam0"` etc. names.
+
+This specialisation is the only one available in Phase 8 MVP.  It is
+selected automatically when `TriangleMesh = CGAL::Surface_mesh<...>`.
+
+\tparam K  Any CGAL kernel.  Defaults to `Simple_cartesian<double>`,
+           which is what `conformal_mesh.hpp` uses today.
+*/
+template <typename K>
+struct Default_conformal_map_traits<CGAL::Surface_mesh<typename K::Point_3>, K>
+{
+    using Kernel        = K;
+    using FT            = typename K::FT;
+    using Point_3       = typename K::Point_3;
+    using Triangle_mesh = CGAL::Surface_mesh<Point_3>;
+
+    using Vertex_descriptor   = typename boost::graph_traits<Triangle_mesh>::vertex_descriptor;
+    using Halfedge_descriptor = typename boost::graph_traits<Triangle_mesh>::halfedge_descriptor;
+    using Edge_descriptor     = typename boost::graph_traits<Triangle_mesh>::edge_descriptor;
+    using Face_descriptor     = typename boost::graph_traits<Triangle_mesh>::face_descriptor;
+
+    // Property-map types — match the names used by setup_euclidean_maps().
+    using Vertex_point_map      = typename Triangle_mesh::template Property_map<Vertex_descriptor, Point_3>;
+    using Theta_pmap            = typename Triangle_mesh::template Property_map<Vertex_descriptor, FT>;
+    using Vertex_index_pmap     = typename Triangle_mesh::template Property_map<Vertex_descriptor, int>;
+    using Lambda0_pmap          = typename Triangle_mesh::template Property_map<Edge_descriptor, FT>;
+
+    // ─── Property-map accessors ───────────────────────────────────────────
+    //
+    // Each accessor returns a property map under its canonical legacy name.
+    // If no such map exists yet it is created with sensible defaults — so
+    // calling either `setup_euclidean_maps(m)` first or the accessor first
+    // is equivalent.
+
+    static Vertex_point_map vertex_points(Triangle_mesh& m) {
+        return m.points();
+    }
+
+    static Theta_pmap theta_map(Triangle_mesh& m) {
+        auto [pm, created] = m.template add_property_map<Vertex_descriptor, FT>(
+            "ev:theta", FT(2.0 * 3.141592653589793238));
+        (void)created;
+        return pm;
+    }
+
+    static Vertex_index_pmap vertex_index_map(Triangle_mesh& m) {
+        auto [pm, created] = m.template add_property_map<Vertex_descriptor, int>(
+            "ev:idx", -1);
+        (void)created;
+        return pm;
+    }
+
+    static Lambda0_pmap lambda0_map(Triangle_mesh& m) {
+        auto [pm, created] = m.template add_property_map<Edge_descriptor, FT>(
+            "ee:lam0", FT(0));
+        (void)created;
+        return pm;
+    }
+};
+
+}  // namespace CGAL
+
+#endif  // CGAL_CONFORMAL_MAP_TRAITS_H

code/include/CGAL/Discrete_circle_packing.h

@@ -0,0 +1,166 @@
+// Copyright (c) 2024-2026 Tarik Moussa.
+// SPDX-License-Identifier: MIT
+//
+// Package: conformallab++ / Discrete_conformal_map (Phase 8b-Lite, 2026-05-21)
+
+/*!
+\file CGAL/Discrete_circle_packing.h
+\ingroup PkgConformalMapRef
+
+User-facing entry for the **face-based** circle-packing functional of
+Bobenko-Pinkall-Springborn 2010.  See `cp_euclidean_functional.hpp`
+for the underlying algorithm and `doc/architecture/phase-9a-validation.md`
+for the line-by-line mapping to the Java original
+`CPEuclideanFunctional.java`.
+
+This functional has a fundamentally different DOF structure to the
+classical Euclidean / Spherical / HyperIdeal modes — one log-radius
+`ρ_f` per **face** rather than one log-scale `u_v` per vertex.  We
+therefore expose it via a dedicated header with its own default-trait
+class (Strategy C of the Phase 8b architecture audit).
+*/
+
+#ifndef CGAL_DISCRETE_CIRCLE_PACKING_H
+#define CGAL_DISCRETE_CIRCLE_PACKING_H
+
+#include <CGAL/Conformal_map/internal/parameters.h>
+#include <CGAL/Kernel_traits.h>
+#include <CGAL/Named_function_parameters.h>
+#include <CGAL/boost/graph/named_params_helper.h>
+#include <CGAL/Surface_mesh.h>
+#include <CGAL/Simple_cartesian.h>
+#include <boost/graph/graph_traits.hpp>
+
+#include "../cp_euclidean_functional.hpp"
+#include "../newton_solver.hpp"
+
+namespace CGAL {
+
+// ── Default traits for CP-Euclidean ───────────────────────────────────────────
+
+template <typename TriangleMesh,
+          typename Kernel_ = CGAL::Simple_cartesian<double>>
+struct Default_cp_euclidean_traits;
+
+template <typename K>
+struct Default_cp_euclidean_traits<CGAL::Surface_mesh<typename K::Point_3>, K>
+{
+    using Kernel        = K;
+    using FT            = typename K::FT;
+    using Point_3       = typename K::Point_3;
+    using Triangle_mesh = CGAL::Surface_mesh<Point_3>;
+
+    using Vertex_descriptor   = typename boost::graph_traits<Triangle_mesh>::vertex_descriptor;
+    using Halfedge_descriptor = typename boost::graph_traits<Triangle_mesh>::halfedge_descriptor;
+    using Edge_descriptor     = typename boost::graph_traits<Triangle_mesh>::edge_descriptor;
+    using Face_descriptor     = typename boost::graph_traits<Triangle_mesh>::face_descriptor;
+
+    // CP-Euclidean property maps — note the *face* DOF index map.
+    using Face_index_pmap = typename Triangle_mesh::template Property_map<Face_descriptor,   int>;
+    using Theta_e_pmap    = typename Triangle_mesh::template Property_map<Edge_descriptor,   FT>;
+    using Phi_f_pmap      = typename Triangle_mesh::template Property_map<Face_descriptor,   FT>;
+};
+
+// ── Result type ───────────────────────────────────────────────────────────────
+
+/*!
+\ingroup PkgConformalMapRef
+
+Result of `discrete_circle_packing_euclidean`.  Carries face DOFs
+`ρ_f = log R_f` rather than the vertex DOFs of the classical modes.
+*/
+template <typename FT = double>
+struct Circle_packing_result
+{
+    /// Face DOFs `ρ_f = log R_f` (length = num_faces(mesh); pinned face = 0).
+    std::vector<FT> rho_per_face;
+
+    int  iterations    = 0;
+    FT   gradient_norm = FT(0);
+    bool converged     = false;
+};
+
+// ── Entry function ────────────────────────────────────────────────────────────
+
+/*!
+\ingroup PkgConformalMapRef
+
+Compute the BPS-2010 face-based circle-packing of `mesh`.
+
+\tparam TriangleMesh  A `CGAL::Surface_mesh<P>`.
+\tparam NamedParameters  Optional CGAL named-parameter pack.
+
+\param mesh  Input triangle mesh.
+\param np    Named parameters (subset of those documented on
+             `discrete_conformal_map_euclidean`; the curvature-map
+             parameter `vertex_curvature_map` is **not** used in this
+             face-based mode — instead the per-face target angle sum
+             `φ_f` and per-edge intersection angle `θ_e` are set via
+             the property maps on `mesh` before this call, or left at
+             their defaults `φ_f = 2π`, `θ_e = π/2`).
+
+\returns A `Circle_packing_result<FT>` with `ρ_f` per face.
+
+\pre `mesh` is a triangle mesh.
+\pre `φ_f` and `θ_e` satisfy the BPS-2010 admissibility conditions
+     (Σ_f φ_f = 2π·χ + Σ_e (π − θ_e), see paper §6).
+*/
+template <typename TriangleMesh,
+          typename CGAL_NP_TEMPLATE_PARAMETERS>
+auto discrete_circle_packing_euclidean(
+    TriangleMesh& mesh,
+    const CGAL_NP_CLASS& np = parameters::default_values())
+{
+    using Point_type     = typename TriangleMesh::Point;
+    using Default_kernel = typename CGAL::Kernel_traits<Point_type>::Kernel;
+    using Default_traits = Default_cp_euclidean_traits<TriangleMesh, Default_kernel>;
+    using Traits         = typename internal_np::Lookup_named_param_def<
+                              internal_np::geom_traits_t,
+                              CGAL_NP_CLASS,
+                              Default_traits>::type;
+    using FT             = typename Traits::FT;
+
+    Circle_packing_result<FT> result;
+
+    auto maps = ::conformallab::setup_cp_euclidean_maps(mesh);
+
+    // Pin first face by default; `fixed_vertex_map` is reused here as the
+    // "fixed face" override hook (the parameter tag is generic enough).
+    // For a richer API, a dedicated `fixed_face_map` tag could be added.
+    auto it = mesh.faces().begin();
+    if (it == mesh.faces().end()) {
+        return result;   // empty mesh; trivial
+    }
+    const int n = ::conformallab::assign_cp_euclidean_face_dof_indices(mesh, maps, *it);
+
+    const FT  tol      = parameters::choose_parameter(
+        parameters::get_parameter(np, Conformal_map::internal_np::gradient_tolerance),
+        FT(1e-10));
+    const int max_iter = parameters::choose_parameter(
+        parameters::get_parameter(np, Conformal_map::internal_np::max_iterations),
+        200);
+
+    // Natural-phi default: shift φ_f so the gradient at ρ = 0 is zero.
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto G0 = ::conformallab::cp_euclidean_gradient(mesh, x0, maps);
+    for (auto f : mesh.faces()) {
+        int i = maps.f_idx[f];
+        if (i >= 0) maps.phi_f[f] -= G0[static_cast<std::size_t>(i)];
+    }
+
+    auto nr = ::conformallab::newton_cp_euclidean(mesh, x0, maps, tol, max_iter);
+
+    result.rho_per_face.assign(num_faces(mesh), FT(0));
+    for (auto f : mesh.faces()) {
+        int j = maps.f_idx[f];
+        if (j >= 0) result.rho_per_face[f.idx()] = nr.x[static_cast<std::size_t>(j)];
+    }
+    result.iterations    = nr.iterations;
+    result.gradient_norm = nr.grad_inf_norm;
+    result.converged     = nr.converged;
+    return result;
+}
+
+}  // namespace CGAL
+
+#endif  // CGAL_DISCRETE_CIRCLE_PACKING_H

code/include/CGAL/Discrete_conformal_map.h

@@ -0,0 +1,490 @@
+// Copyright (c) 2024-2026 Tarik Moussa.
+// SPDX-License-Identifier: MIT
+//
+// Package: conformallab++ / Discrete_conformal_map (Phase 8 MVP, 2026-05-19)
+
+/*!
+\file CGAL/Discrete_conformal_map.h
+\ingroup PkgConformalMapRef
+
+User-facing entry point for the Discrete_conformal_map package.
+
+This header provides a single function — `discrete_conformal_map_euclidean`
+— that computes a Euclidean discrete-conformal flattening of an open or
+closed triangle mesh.  Spherical and hyperbolic variants are scheduled
+for Phase 8b.2 once the Euclidean pattern is validated by Phase 9a
+(Inversive-Distance functional).
+
+\section Example Simplest usage
+
+\code{.cpp}
+#include <CGAL/Simple_cartesian.h>
+#include <CGAL/Surface_mesh.h>
+#include <CGAL/Discrete_conformal_map.h>
+
+using K = CGAL::Simple_cartesian<double>;
+using Mesh = CGAL::Surface_mesh<K::Point_3>;
+
+int main() {
+    Mesh mesh = ...;  // load a triangle mesh
+    auto result = CGAL::discrete_conformal_map_euclidean(mesh);
+    if (!result.converged)
+        return 1;
+    // result.u_per_vertex[v] now holds the conformal scale factor at v.
+}
+\endcode
+
+\section NamedParams Tuning via named parameters
+
+\code{.cpp}
+auto result = CGAL::discrete_conformal_map_euclidean(
+    mesh,
+    CGAL::parameters::gradient_tolerance(1e-12)
+                     .max_iterations(500));
+\endcode
+
+\sa `CGAL::Default_conformal_map_traits`
+\sa `CGAL::parameters::vertex_curvature_map`
+*/
+
+#ifndef CGAL_DISCRETE_CONFORMAL_MAP_H
+#define CGAL_DISCRETE_CONFORMAL_MAP_H
+
+#include <CGAL/Conformal_map_traits.h>
+#include <CGAL/Conformal_map/internal/parameters.h>
+#include <CGAL/Kernel_traits.h>
+#include <CGAL/Named_function_parameters.h>
+#include <CGAL/boost/graph/named_params_helper.h>
+#include <CGAL/property_map.h>
+
+// Existing implementation headers (Layer 1 — unchanged).
+#include "../euclidean_functional.hpp"
+#include "../spherical_functional.hpp"
+#include "../hyper_ideal_functional.hpp"
+#include "../gauss_bonnet.hpp"
+#include "../newton_solver.hpp"
+
+#include <vector>
+#include <unordered_map>
+
+namespace CGAL {
+
+// ════════════════════════════════════════════════════════════════════════════
+// Result type
+// ════════════════════════════════════════════════════════════════════════════
+
+/*!
+\ingroup PkgConformalMapRef
+
+Result of `discrete_conformal_map_euclidean`.  Carries the converged
+scale factors `u_v`, Newton diagnostics, and the convergence flag.
+*/
+template <typename FT = double>
+struct Conformal_map_result
+{
+    /// Conformal scale factor `u_v` per vertex (indexed by raw vertex index).
+    /// Length: `num_vertices(mesh)`.
+    std::vector<FT> u_per_vertex;
+
+    /// Number of Newton iterations performed.
+    int iterations = 0;
+
+    /// `‖G(u*)‖∞` at termination.
+    FT gradient_norm = FT(0);
+
+    /// `true` iff `gradient_norm < gradient_tolerance`.
+    bool converged = false;
+
+    /// `true` iff the linear solver used the SparseQR fallback at any
+    /// Newton step (gauge mode on closed mesh without pinned vertex).
+    bool sparse_qr_fallback_used = false;
+};
+
+
+// ════════════════════════════════════════════════════════════════════════════
+// discrete_conformal_map_euclidean  —  user-facing entry
+// ════════════════════════════════════════════════════════════════════════════
+
+/*!
+\ingroup PkgConformalMapRef
+
+Compute the Euclidean discrete-conformal map of `mesh`.
+
+This is the user-facing entry of Phase 8 MVP.  Internally it delegates
+to the existing implementation in `code/include/euclidean_functional.hpp`
+and `code/include/newton_solver.hpp` (Phase 1–7), so the algorithmic
+behaviour is identical to the legacy API; this function only changes
+the public façade.
+
+\tparam TriangleMesh  A `CGAL::Surface_mesh<P>` for some point type `P`.
+                      Other `FaceGraph` models are planned for Phase 8a.2.
+\tparam NamedParameters  Optional CGAL named-parameter pack.
+
+\param mesh  The input mesh (modified in place: property maps are attached).
+\param np    Named parameters:
+  \cgalParamNBegin{vertex_curvature_map}
+    \cgalParamDescription{Property map `vertex → FT` of target cone angles Θᵥ.}
+    \cgalParamDefault{2π at interior vertices, π at boundary vertices.}
+  \cgalParamNEnd
+  \cgalParamNBegin{gradient_tolerance}
+    \cgalParamDescription{Newton stops when `‖G(u)‖∞ < tol`.}
+    \cgalParamDefault{`1e-10`}
+  \cgalParamNEnd
+  \cgalParamNBegin{max_iterations}
+    \cgalParamDescription{Hard limit on Newton steps.}
+    \cgalParamDefault{`200`}
+  \cgalParamNEnd
+  \cgalParamNBegin{fixed_vertex_map}
+    \cgalParamDescription{Property map `vertex → bool`; `true` ⇒ pinned.}
+    \cgalParamDefault{The first vertex in `mesh.vertices()` is pinned.}
+  \cgalParamNEnd
+
+\returns A `Conformal_map_result<FT>` carrying `u_v` and Newton diagnostics.
+
+\pre `mesh` is a triangle mesh.
+\pre `mesh` satisfies the Gauss–Bonnet relation
+     `Σ(2π − Θᵥ) = 2π·χ(mesh)` for the chosen target curvature map.
+*/
+template <typename TriangleMesh,
+          typename CGAL_NP_TEMPLATE_PARAMETERS>
+auto discrete_conformal_map_euclidean(
+    TriangleMesh& mesh,
+    const CGAL_NP_CLASS& np = parameters::default_values())
+{
+    // ── Type plumbing ──────────────────────────────────────────────────────
+    //
+    // Deduce the kernel from the mesh's Point_3 type rather than hard-coding
+    // Simple_cartesian<double>.  This lets the wrapper work with any
+    // Surface_mesh<P> whose P is a CGAL kernel point.  The user can override
+    // the entire traits class via the `geom_traits(...)` named parameter
+    // (Phase 8b.2 extension; default below covers the common case).
+    using Point_type      = typename TriangleMesh::Point;
+    using Default_kernel  = typename CGAL::Kernel_traits<Point_type>::Kernel;
+    using Default_traits  = Default_conformal_map_traits<TriangleMesh, Default_kernel>;
+    using Traits          = typename internal_np::Lookup_named_param_def<
+                                internal_np::geom_traits_t,
+                                CGAL_NP_CLASS,
+                                Default_traits>::type;
+    using FT              = typename Traits::FT;
+
+    Conformal_map_result<FT> result;
+
+    // ── 1. Set up property maps (legacy layer) ─────────────────────────────
+    auto maps = ::conformallab::setup_euclidean_maps(mesh);
+    ::conformallab::compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    // ── 2. Target curvature: user-supplied or "natural-theta" default ─────
+    auto theta_param = parameters::get_parameter(
+        np, Conformal_map::internal_np::vertex_curvature_map);
+    constexpr bool has_theta = !std::is_same_v<
+        decltype(theta_param), internal_np::Param_not_found>;
+    if constexpr (has_theta) {
+        // User-provided Θ: copy into the property map and verify Gauss–Bonnet.
+        // Throws std::runtime_error if the user-supplied Θ violates GB.
+        for (auto v : mesh.vertices())
+            maps.theta_v[v] = get(theta_param, v);
+        ::conformallab::check_gauss_bonnet(mesh, maps);
+    }
+    // If no Θ is supplied, the default behaviour is "natural-theta": set Θ
+    // such that x = 0 is the natural equilibrium (the actual angle sums at
+    // x = 0 become the targets).  This matches the convention of the
+    // existing test suite and guarantees that the default invocation
+    // converges immediately for any well-formed triangle mesh.
+    // The actual Θ adjustment is done after DOF assignment (step 5b below).
+
+    // ── 3. Pin vertices: user map, or the first vertex by default ──────────
+    //
+    // The legacy v_idx property map has -1 as default (= pinned).  We must
+    // first mark every vertex as "free" (any non-negative sentinel), then
+    // pin the requested ones, then assign sequential DOF indices.
+    constexpr int FREE = 0;
+    for (auto v : mesh.vertices())
+        maps.v_idx[v] = FREE;
+
+    auto pin_param = parameters::get_parameter(
+        np, Conformal_map::internal_np::fixed_vertex_map);
+    constexpr bool has_pin = !std::is_same_v<
+        decltype(pin_param), internal_np::Param_not_found>;
+
+    bool any_pinned = false;
+    if constexpr (has_pin) {
+        for (auto v : mesh.vertices())
+            if (get(pin_param, v)) {
+                maps.v_idx[v] = -1;
+                any_pinned = true;
+            }
+    }
+    if (!any_pinned) {
+        auto it = mesh.vertices().begin();
+        if (it != mesh.vertices().end()) {
+            maps.v_idx[*it] = -1;
+            any_pinned = true;
+        }
+    }
+
+    // ── 4. Assign DOF indices 0..n−1 to non-pinned vertices ─────────────────
+    int idx = 0;
+    for (auto v : mesh.vertices())
+        if (maps.v_idx[v] != -1)
+            maps.v_idx[v] = idx++;
+
+    // ── 5. Read tolerances ─────────────────────────────────────────────────
+    const FT  tol      = parameters::choose_parameter(
+        parameters::get_parameter(np, Conformal_map::internal_np::gradient_tolerance),
+        FT(1e-10));
+    const int max_iter = parameters::choose_parameter(
+        parameters::get_parameter(np, Conformal_map::internal_np::max_iterations),
+        200);
+
+    // ── 5b. Natural-theta default: shift Θ so that x = 0 is the equilibrium
+    //
+    // Only applied when the user did NOT supply a vertex_curvature_map.
+    // The trick: evaluate G at x = 0, then subtract G_v from Θ_v.  After
+    // this shift the new G(0) is identically zero, so Newton starts at the
+    // optimum and immediately reports "converged".  This matches the
+    // contract of the existing test suite ("natural-theta" pattern).
+    std::vector<double> x0(static_cast<std::size_t>(idx), 0.0);
+    if constexpr (!has_theta) {
+        auto G0 = ::conformallab::euclidean_gradient(mesh, x0, maps);
+        for (auto v : mesh.vertices()) {
+            const int j = maps.v_idx[v];
+            if (j >= 0)
+                maps.theta_v[v] -= G0[static_cast<std::size_t>(j)];
+        }
+    }
+
+    // ── 6. Newton on x_0 = 0 ───────────────────────────────────────────────
+    auto nr = ::conformallab::newton_euclidean(mesh, x0, maps, tol, max_iter);
+
+    // ── 7. Pack result: u_v for every vertex, including pinned (u=0) ──────
+    result.u_per_vertex.assign(num_vertices(mesh), FT(0));
+    for (auto v : mesh.vertices()) {
+        const int j = maps.v_idx[v];
+        if (j >= 0)
+            result.u_per_vertex[v.idx()] = nr.x[static_cast<std::size_t>(j)];
+        // else: pinned ⇒ u_v stays 0
+    }
+    result.iterations    = nr.iterations;
+    result.gradient_norm = nr.grad_inf_norm;
+    result.converged     = nr.converged;
+
+    return result;
+}
+
+
+// ════════════════════════════════════════════════════════════════════════════
+// discrete_conformal_map_spherical  —  Phase 8b-Lite
+// ════════════════════════════════════════════════════════════════════════════
+
+/*!
+\ingroup PkgConformalMapRef
+
+Compute the spherical discrete-conformal map of a closed genus-0 mesh.
+
+The spherical DCE energy is *concave*, so its Hessian is NSD at the
+optimum and `newton_spherical()` factorises −H internally (handled by
+the legacy implementation; no caller action required).  A gauge vertex
+is pinned automatically to remove the rotational mode.
+
+\tparam TriangleMesh  A `CGAL::Surface_mesh<P>` for some point type `P`.
+\tparam NamedParameters  Optional CGAL named-parameter pack.
+
+\param mesh  The input mesh (modified in place: property maps attached).
+\param np    Same named parameters as `discrete_conformal_map_euclidean`.
+
+\returns A `Conformal_map_result<FT>` carrying `u_v` per vertex and
+         Newton diagnostics.
+
+\pre `mesh` is a closed genus-0 triangle mesh.
+\pre The user-supplied or natural-theta Θ satisfies the spherical
+     Gauss–Bonnet relation `Σ(2π − Θᵥ) = 4π` (sphere).
+*/
+template <typename TriangleMesh,
+          typename CGAL_NP_TEMPLATE_PARAMETERS>
+auto discrete_conformal_map_spherical(
+    TriangleMesh& mesh,
+    const CGAL_NP_CLASS& np = parameters::default_values())
+{
+    using Point_type     = typename TriangleMesh::Point;
+    using Default_kernel = typename CGAL::Kernel_traits<Point_type>::Kernel;
+    using Default_traits = Default_conformal_map_traits<TriangleMesh, Default_kernel>;
+    using Traits         = typename internal_np::Lookup_named_param_def<
+                              internal_np::geom_traits_t,
+                              CGAL_NP_CLASS,
+                              Default_traits>::type;
+    using FT             = typename Traits::FT;
+
+    Conformal_map_result<FT> result;
+
+    auto maps = ::conformallab::setup_spherical_maps(mesh);
+    ::conformallab::compute_lambda0_from_mesh(mesh, maps);
+
+    auto theta_param = parameters::get_parameter(
+        np, Conformal_map::internal_np::vertex_curvature_map);
+    constexpr bool has_theta = !std::is_same_v<
+        decltype(theta_param), internal_np::Param_not_found>;
+    if constexpr (has_theta) {
+        for (auto v : mesh.vertices())
+            maps.theta_v[v] = get(theta_param, v);
+    }
+
+    // Pin one vertex (gauge fix) — user-supplied or first vertex.
+    constexpr int FREE = 0;
+    for (auto v : mesh.vertices()) maps.v_idx[v] = FREE;
+
+    auto pin_param = parameters::get_parameter(
+        np, Conformal_map::internal_np::fixed_vertex_map);
+    constexpr bool has_pin = !std::is_same_v<
+        decltype(pin_param), internal_np::Param_not_found>;
+
+    bool any_pinned = false;
+    if constexpr (has_pin) {
+        for (auto v : mesh.vertices())
+            if (get(pin_param, v)) { maps.v_idx[v] = -1; any_pinned = true; }
+    }
+    if (!any_pinned) {
+        auto it = mesh.vertices().begin();
+        if (it != mesh.vertices().end()) { maps.v_idx[*it] = -1; any_pinned = true; }
+    }
+
+    int idx = 0;
+    for (auto v : mesh.vertices())
+        if (maps.v_idx[v] != -1) maps.v_idx[v] = idx++;
+
+    const FT  tol      = parameters::choose_parameter(
+        parameters::get_parameter(np, Conformal_map::internal_np::gradient_tolerance),
+        FT(1e-10));
+    const int max_iter = parameters::choose_parameter(
+        parameters::get_parameter(np, Conformal_map::internal_np::max_iterations),
+        200);
+
+    // Natural-theta default for the spherical functional.
+    std::vector<double> x0(static_cast<std::size_t>(idx), 0.0);
+    if constexpr (!has_theta) {
+        auto G0 = ::conformallab::spherical_gradient(mesh, x0, maps);
+        for (auto v : mesh.vertices()) {
+            const int j = maps.v_idx[v];
+            if (j >= 0) maps.theta_v[v] -= G0[static_cast<std::size_t>(j)];
+        }
+    }
+
+    auto nr = ::conformallab::newton_spherical(mesh, x0, maps, tol, max_iter);
+
+    result.u_per_vertex.assign(num_vertices(mesh), FT(0));
+    for (auto v : mesh.vertices()) {
+        const int j = maps.v_idx[v];
+        if (j >= 0) result.u_per_vertex[v.idx()] = nr.x[static_cast<std::size_t>(j)];
+    }
+    result.iterations    = nr.iterations;
+    result.gradient_norm = nr.grad_inf_norm;
+    result.converged     = nr.converged;
+    return result;
+}
+
+
+// ════════════════════════════════════════════════════════════════════════════
+// discrete_conformal_map_hyper_ideal  —  Phase 8b-Lite
+// ════════════════════════════════════════════════════════════════════════════
+
+/*!
+\ingroup PkgConformalMapRef
+
+Result of `discrete_conformal_map_hyper_ideal`.  Carries both vertex
+DOFs `b_v` and edge DOFs `a_e` (hyper-ideal triangles in H³).
+*/
+template <typename FT = double>
+struct Hyper_ideal_map_result
+{
+    /// Vertex DOFs `b_v` (length = num_vertices(mesh); pinned vertices = 0).
+    std::vector<FT> b_per_vertex;
+    /// Edge DOFs `a_e` (length = num_edges(mesh); pinned edges = 0).
+    std::vector<FT> a_per_edge;
+
+    int  iterations    = 0;
+    FT   gradient_norm = FT(0);
+    bool converged     = false;
+};
+
+/*!
+\ingroup PkgConformalMapRef
+
+Compute the hyper-ideal discrete-conformal map of a triangle mesh
+(Springborn 2020 §4).
+
+\note  Phase 8b-Lite scope: vertex DOFs `b_v` are assigned automatically
+       to all vertices; edge DOFs `a_e` are similarly assigned.  The
+       block-FD Hessian (Phase 9b) is used internally — see
+       `newton_hyper_ideal` for the solver convention.
+*/
+template <typename TriangleMesh,
+          typename CGAL_NP_TEMPLATE_PARAMETERS>
+auto discrete_conformal_map_hyper_ideal(
+    TriangleMesh& mesh,
+    const CGAL_NP_CLASS& np = parameters::default_values())
+{
+    using Point_type     = typename TriangleMesh::Point;
+    using Default_kernel = typename CGAL::Kernel_traits<Point_type>::Kernel;
+    using Default_traits = Default_conformal_map_traits<TriangleMesh, Default_kernel>;
+    using Traits         = typename internal_np::Lookup_named_param_def<
+                              internal_np::geom_traits_t,
+                              CGAL_NP_CLASS,
+                              Default_traits>::type;
+    using FT             = typename Traits::FT;
+
+    Hyper_ideal_map_result<FT> result;
+
+    auto maps = ::conformallab::setup_hyper_ideal_maps(mesh);
+    // Hyper-ideal init does not derive from mesh geometry: the user's
+    // Θ_v and θ_e are the model inputs.  Defaults from setup are
+    // Θ_v = 2π, θ_e = π (orthogonal).
+
+    auto theta_param = parameters::get_parameter(
+        np, Conformal_map::internal_np::vertex_curvature_map);
+    constexpr bool has_theta = !std::is_same_v<
+        decltype(theta_param), internal_np::Param_not_found>;
+    if constexpr (has_theta) {
+        for (auto v : mesh.vertices())
+            maps.theta_v[v] = get(theta_param, v);
+    }
+
+    const int n = ::conformallab::assign_all_dof_indices(mesh, maps);
+
+    const FT  tol      = parameters::choose_parameter(
+        parameters::get_parameter(np, Conformal_map::internal_np::gradient_tolerance),
+        FT(1e-8));
+    const int max_iter = parameters::choose_parameter(
+        parameters::get_parameter(np, Conformal_map::internal_np::max_iterations),
+        200);
+
+    // Initial point: b_v = 1.0 (positive log-scale), a_e = 0.5 (moderate).
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    for (auto v : mesh.vertices()) {
+        int i = maps.v_idx[v];
+        if (i >= 0) x0[static_cast<std::size_t>(i)] = 1.0;
+    }
+    for (auto e : mesh.edges()) {
+        int i = maps.e_idx[e];
+        if (i >= 0) x0[static_cast<std::size_t>(i)] = 0.5;
+    }
+
+    auto nr = ::conformallab::newton_hyper_ideal(mesh, x0, maps, tol, max_iter);
+
+    result.b_per_vertex.assign(num_vertices(mesh), FT(0));
+    result.a_per_edge   .assign(num_edges(mesh),    FT(0));
+    for (auto v : mesh.vertices()) {
+        int j = maps.v_idx[v];
+        if (j >= 0) result.b_per_vertex[v.idx()] = nr.x[static_cast<std::size_t>(j)];
+    }
+    for (auto e : mesh.edges()) {
+        int j = maps.e_idx[e];
+        if (j >= 0) result.a_per_edge[e.idx()] = nr.x[static_cast<std::size_t>(j)];
+    }
+    result.iterations    = nr.iterations;
+    result.gradient_norm = nr.grad_inf_norm;
+    result.converged     = nr.converged;
+    return result;
+}
+
+}  // namespace CGAL
+
+#endif  // CGAL_DISCRETE_CONFORMAL_MAP_H

code/include/CGAL/Discrete_inversive_distance.h

@@ -0,0 +1,191 @@
+// Copyright (c) 2024-2026 Tarik Moussa.
+// SPDX-License-Identifier: MIT
+//
+// Package: conformallab++ / Discrete_conformal_map (Phase 8b-Lite, 2026-05-21)
+
+/*!
+\file CGAL/Discrete_inversive_distance.h
+\ingroup PkgConformalMapRef
+
+User-facing entry for the **vertex-based** inversive-distance circle-
+packing functional of Luo (2004), with the Bowers-Stephenson (2004)
+initialisation.  See `inversive_distance_functional.hpp` for the
+underlying algorithm and `doc/roadmap/research-track.md` (item 9a.2)
+for the research-track classification — this functional has **no Java
+original** (verified empirically), it is from-the-literature research.
+
+DOF structure
+─────────────
+* Per-vertex `u_i = log r_i` (compatible with the classical Euclidean
+  trait).
+* Per-edge constant `I_ij` computed once by Bowers-Stephenson from the
+  input mesh geometry (handled internally by
+  `compute_inversive_distance_init_from_mesh`).
+
+Because the per-edge constant has a different meaning from the
+Euclidean `λ°_e`, this entry has its own default-trait class
+`Default_inversive_distance_traits`.
+*/
+
+#ifndef CGAL_DISCRETE_INVERSIVE_DISTANCE_H
+#define CGAL_DISCRETE_INVERSIVE_DISTANCE_H
+
+#include <CGAL/Conformal_map/internal/parameters.h>
+#include <CGAL/Kernel_traits.h>
+#include <CGAL/Named_function_parameters.h>
+#include <CGAL/boost/graph/named_params_helper.h>
+#include <CGAL/Surface_mesh.h>
+#include <CGAL/Simple_cartesian.h>
+#include <boost/graph/graph_traits.hpp>
+
+#include <CGAL/Discrete_conformal_map.h>   // for Conformal_map_result<FT>
+
+#include "../inversive_distance_functional.hpp"
+#include "../newton_solver.hpp"
+
+namespace CGAL {
+
+// ── Default traits for Inversive-Distance ────────────────────────────────────
+
+template <typename TriangleMesh,
+          typename Kernel_ = CGAL::Simple_cartesian<double>>
+struct Default_inversive_distance_traits;
+
+template <typename K>
+struct Default_inversive_distance_traits<CGAL::Surface_mesh<typename K::Point_3>, K>
+{
+    using Kernel        = K;
+    using FT            = typename K::FT;
+    using Point_3       = typename K::Point_3;
+    using Triangle_mesh = CGAL::Surface_mesh<Point_3>;
+
+    using Vertex_descriptor   = typename boost::graph_traits<Triangle_mesh>::vertex_descriptor;
+    using Edge_descriptor     = typename boost::graph_traits<Triangle_mesh>::edge_descriptor;
+
+    // Inversive-distance specific property maps.
+    using Vertex_index_pmap = typename Triangle_mesh::template Property_map<Vertex_descriptor, int>;
+    using Theta_v_pmap      = typename Triangle_mesh::template Property_map<Vertex_descriptor, FT>;
+    using R0_pmap           = typename Triangle_mesh::template Property_map<Vertex_descriptor, FT>;
+    using I_e_pmap          = typename Triangle_mesh::template Property_map<Edge_descriptor,   FT>;
+};
+
+// ── Entry function ────────────────────────────────────────────────────────────
+
+/*!
+\ingroup PkgConformalMapRef
+
+Compute the Luo-2004 vertex-based inversive-distance circle packing of `mesh`.
+
+The per-edge constant `I_ij` is computed once at the start from the input
+3-D geometry via the Bowers-Stephenson identity
+    `I_ij = (ℓ_ij² − r_i² − r_j²) / (2 r_i r_j)`,
+with `r_i^(0) = (1/3) min{ℓ_e : e adj v_i}` as the default initial radii.
+The user can override the initial radii by writing into the `r0`
+property map before calling this function.
+
+\tparam TriangleMesh  A `CGAL::Surface_mesh<P>`.
+\tparam NamedParameters  Optional CGAL named-parameter pack.
+
+\param mesh  Input triangle mesh.
+\param np    Named parameters:
+  - `vertex_curvature_map(pmap)` — per-vertex Θ_v target.
+  - `fixed_vertex_map(pmap)` — pinning override.
+  - `gradient_tolerance(ε)` — Newton stop.
+  - `max_iterations(n)` — Newton iteration cap.
+
+\returns A `Conformal_map_result<FT>` with `u_per_vertex[v] = log r_v`
+         (the converged log-radius at each vertex).
+
+\pre `mesh` is a triangle mesh with positive edge lengths.
+\pre The user-supplied or natural-theta Θ satisfies Gauss–Bonnet.
+
+\note  Convergence is sensitive to the initial point and to extreme
+       `I_ij` values.  For testing purposes the natural-theta default
+       (Θ_v shifted so that u = 0 is the equilibrium) always converges
+       in zero iterations.
+*/
+template <typename TriangleMesh,
+          typename CGAL_NP_TEMPLATE_PARAMETERS>
+auto discrete_inversive_distance_map(
+    TriangleMesh& mesh,
+    const CGAL_NP_CLASS& np = parameters::default_values())
+{
+    using Point_type     = typename TriangleMesh::Point;
+    using Default_kernel = typename CGAL::Kernel_traits<Point_type>::Kernel;
+    using Default_traits = Default_inversive_distance_traits<TriangleMesh, Default_kernel>;
+    using Traits         = typename internal_np::Lookup_named_param_def<
+                              internal_np::geom_traits_t,
+                              CGAL_NP_CLASS,
+                              Default_traits>::type;
+    using FT             = typename Traits::FT;
+
+    Conformal_map_result<FT> result;
+
+    auto maps = ::conformallab::setup_inversive_distance_maps(mesh);
+    ::conformallab::compute_inversive_distance_init_from_mesh(mesh, maps);
+
+    auto theta_param = parameters::get_parameter(
+        np, Conformal_map::internal_np::vertex_curvature_map);
+    constexpr bool has_theta = !std::is_same_v<
+        decltype(theta_param), internal_np::Param_not_found>;
+    if constexpr (has_theta) {
+        for (auto v : mesh.vertices())
+            maps.theta_v[v] = get(theta_param, v);
+    }
+
+    // Pin first vertex by default; user can override with fixed_vertex_map.
+    constexpr int FREE = 0;
+    for (auto v : mesh.vertices()) maps.v_idx[v] = FREE;
+
+    auto pin_param = parameters::get_parameter(
+        np, Conformal_map::internal_np::fixed_vertex_map);
+    constexpr bool has_pin = !std::is_same_v<
+        decltype(pin_param), internal_np::Param_not_found>;
+
+    bool any_pinned = false;
+    if constexpr (has_pin) {
+        for (auto v : mesh.vertices())
+            if (get(pin_param, v)) { maps.v_idx[v] = -1; any_pinned = true; }
+    }
+    if (!any_pinned) {
+        auto it = mesh.vertices().begin();
+        if (it != mesh.vertices().end()) { maps.v_idx[*it] = -1; any_pinned = true; }
+    }
+
+    int idx = 0;
+    for (auto v : mesh.vertices())
+        if (maps.v_idx[v] != -1) maps.v_idx[v] = idx++;
+
+    const FT  tol      = parameters::choose_parameter(
+        parameters::get_parameter(np, Conformal_map::internal_np::gradient_tolerance),
+        FT(1e-10));
+    const int max_iter = parameters::choose_parameter(
+        parameters::get_parameter(np, Conformal_map::internal_np::max_iterations),
+        200);
+
+    // Natural-theta default.
+    std::vector<double> x0(static_cast<std::size_t>(idx), 0.0);
+    if constexpr (!has_theta) {
+        auto G0 = ::conformallab::inversive_distance_gradient(mesh, x0, maps);
+        for (auto v : mesh.vertices()) {
+            const int j = maps.v_idx[v];
+            if (j >= 0) maps.theta_v[v] -= G0[static_cast<std::size_t>(j)];
+        }
+    }
+
+    auto nr = ::conformallab::newton_inversive_distance(mesh, x0, maps, tol, max_iter);
+
+    result.u_per_vertex.assign(num_vertices(mesh), FT(0));
+    for (auto v : mesh.vertices()) {
+        const int j = maps.v_idx[v];
+        if (j >= 0) result.u_per_vertex[v.idx()] = nr.x[static_cast<std::size_t>(j)];
+    }
+    result.iterations    = nr.iterations;
+    result.gradient_norm = nr.grad_inf_norm;
+    result.converged     = nr.converged;
+    return result;
+}
+
+}  // namespace CGAL
+
+#endif  // CGAL_DISCRETE_INVERSIVE_DISTANCE_H

code/include/conformal_mesh.hpp

@@ -0,0 +1,94 @@
+#pragma once
+// conformal_mesh.hpp
+//
+// Central mesh type for the discrete conformal mapping algorithms.
+// Replaces the Java CoHDS (de.varylab.discreteconformal.heds.CoHDS)
+// and its associated vertex/edge/face types (CoVertex, CoEdge, CoFace).
+//
+// Design
+// ------
+//   Java                         │  C++ (this file)
+//   ─────────────────────────────┼──────────────────────────────────────────
+//   CoHDS                        │  ConformalMesh   (CGAL::Surface_mesh)
+//   CoVertex / CoEdge / CoFace   │  Vertex_index / Edge_index / Face_index
+//   HyperIdealRadiusAdapter      │  property_map<Vertex_index, double>
+//   HalfedgeInterface adapters   │  named property maps ("v:lambda", …)
+//
+// Property-map naming convention
+// ───────────────────────────────
+//   "v:lambda"    per-vertex log scale factor  (conformal variable u_i)
+//   "v:theta"     per-vertex target cone angle
+//   "v:idx"       per-vertex solver DOF index  (-1 = pinned / boundary)
+//   "e:alpha"     per-edge intersection angle  (α_ij, hyperbolic geometry)
+//   "f:type"      per-face geometry type       (0=Euclidean, 1=Hyperbolic, 2=Spherical)
+//
+// All property maps are optional; add only what a given algorithm needs.
+//
+// Note on descriptor types (CGAL 6.x)
+// ────────────────────────────────────
+// CGAL::Surface_mesh exposes its index types as nested types:
+//   Surface_mesh::Vertex_index, ::Halfedge_index, ::Edge_index, ::Face_index
+// The BGL graph_traits aliases expose the same types as vertex_descriptor etc.,
+// but those live in boost::graph_traits<Surface_mesh>, not in Surface_mesh itself.
+// We use the Surface_mesh member names throughout for clarity.
+
+#include <CGAL/Simple_cartesian.h>
+#include <CGAL/Surface_mesh.h>
+#include <string>
+
+namespace conformallab {
+
+// ── Kernel ──────────────────────────────────────────────────────────────────
+// Simple double-precision Cartesian.  Conformal mapping algorithms never
+// need exact arithmetic — they operate on floating-point lengths and angles.
+using Kernel = CGAL::Simple_cartesian<double>;
+using Point3 = Kernel::Point_3;
+using Point2 = Kernel::Point_2;
+
+// ── Mesh type ────────────────────────────────────────────────────────────────
+using ConformalMesh = CGAL::Surface_mesh<Point3>;
+
+// ── Index/descriptor aliases (CGAL 6.x naming) ───────────────────────────────
+using Vertex_index   = ConformalMesh::Vertex_index;
+using Halfedge_index = ConformalMesh::Halfedge_index;
+using Edge_index     = ConformalMesh::Edge_index;
+using Face_index     = ConformalMesh::Face_index;
+
+// ── Geometry type constant (replaces Java CoFace.type enum) ─────────────────
+enum class GeometryType : int {
+    Euclidean  = 0,
+    Hyperbolic = 1,
+    Spherical  = 2
+};
+
+// ── Standard property-map bundles ────────────────────────────────────────────
+
+// Add the vertex properties used by all conformal-map functionals.
+// Returns {lambda, theta, idx}.
+inline auto add_vertex_properties(ConformalMesh& mesh)
+{
+    auto [lambda, ok1] = mesh.add_property_map<Vertex_index, double>("v:lambda", 0.0);
+    auto [theta,  ok2] = mesh.add_property_map<Vertex_index, double>("v:theta",  0.0);
+    auto [idx,    ok3] = mesh.add_property_map<Vertex_index, int>   ("v:idx",   -1);
+    (void)ok1; (void)ok2; (void)ok3;
+    return std::make_tuple(lambda, theta, idx);
+}
+
+// Add the edge intersection-angle property used by the hyperbolic functional.
+inline auto add_edge_properties(ConformalMesh& mesh)
+{
+    auto [alpha, ok] = mesh.add_property_map<Edge_index, double>("e:alpha", 0.0);
+    (void)ok;
+    return alpha;
+}
+
+// Add the face geometry-type property.
+inline auto add_face_properties(ConformalMesh& mesh)
+{
+    auto [ftype, ok] = mesh.add_property_map<Face_index, int>(
+        "f:type", static_cast<int>(GeometryType::Euclidean));
+    (void)ok;
+    return ftype;
+}
+
+} // namespace conformallab

code/include/constants.hpp

@@ -0,0 +1,17 @@
+#pragma once
+// constants.hpp
+//
+// Single source of truth for mathematical constants used throughout
+// conformallab++.  Include this header instead of defining π locally.
+//
+// All constants are in the conformallab namespace and are constexpr double.
+
+namespace conformallab {
+
+/// π to 30 significant digits (well beyond double precision of ~15-16 digits).
+constexpr double PI     = 3.14159265358979323846264338328;
+
+/// 2π (full turn).
+constexpr double TWO_PI = 2.0 * PI;
+
+} // namespace conformallab

code/include/cp_euclidean_functional.hpp

@@ -0,0 +1,364 @@
+#pragma once
+// cp_euclidean_functional.hpp
+//
+// Phase 9a.1 — Circle-Packing Euclidean functional (CP-Euclidean).
+//
+// Ported from de.varylab.discreteconformal.functional.CPEuclideanFunctional
+// (Java, 260 lines).  Mathematical reference:
+//   Bobenko, A. I., Pinkall, U. & Springborn, B. (2010)
+//   "Discrete conformal maps and ideal hyperbolic polyhedra"
+//   Geometry & Topology 14, 379-426.
+//
+// ┌──────────────────────────────────────────────────────────────────────────┐
+// │  FACE-based circle packing                                                │
+// │                                                                           │
+// │  Each face f of the mesh carries a circle of radius R_f.                  │
+// │  The variable is ρ_f = log R_f.                                           │
+// │  Adjacent face-circles intersect at a prescribed angle θ_e per edge.      │
+// │                                                                           │
+// │  This is the FACE-DUAL of the classical vertex-based Luo (2004)           │
+// │  inversive-distance circle packing implemented in                          │
+// │  inversive_distance_functional.hpp (Phase 9a.2).  The relation            │
+// │     I_ij  =  cos θ_e                                                       │
+// │  identifies the two parametrisations (Glickenstein 2011 §5).              │
+// │                                                                           │
+// │  DOFs                                                                     │
+// │    x[f_idx[f]] = ρ_f   (face-dual log-radius)                             │
+// │    f_idx[f] = −1 means f is pinned (ρ_f = 0, gauge fix)                  │
+// │                                                                           │
+// │  Constants                                                                │
+// │    θ_e per edge   intersection angle of the two face-circles              │
+// │    φ_f per face   target sum of corner-angles inside the face             │
+// │                                                                           │
+// │  Energy  (BPS-2010 §6)                                                    │
+// │    E(ρ) =  Σ_f  φ_f · ρ_f                                                  │
+// │         +  Σ_{(h,f=face(h)):                                              │
+// │              [ if opposite face exists ]                                   │
+// │                ½ p(θ*,Δρ)·Δρ + Λ(θ*+p) − θ*·ρ_left                       │
+// │              [ else (boundary halfedge) ]                                  │
+// │                −2 θ*·ρ_left                                                │
+// │            ]                                                              │
+// │                                                                           │
+// │    where    θ* = π − θ                                                    │
+// │             Δρ = ρ_right − ρ_left                                          │
+// │             p(θ*, Δρ) = 2·atan( tan(θ*/2) · tanh(Δρ/2) )                  │
+// │             Λ        = Clausen function (Lobachevsky)                     │
+// │                                                                           │
+// │  Gradient                                                                 │
+// │    Per face f:        +φ_f                                                │
+// │    Per interior hf:   −(p + θ*)  added to G[face(h)]                      │
+// │    Per boundary hf:   −2 θ*       added to G[face(h)]                     │
+// │                                                                           │
+// │  Hessian (analytic, BPS-2010 eq. 6.8; Java getHessian lines 127-166)     │
+// │    Per interior undirected edge e (connecting faces j and k):             │
+// │      h_jk = sin θ / (cosh Δρ − cos θ)                                     │
+// │      H[j,j] += h_jk,  H[k,k] += h_jk,  H[j,k] −= h_jk,  H[k,j] −= h_jk    │
+// │    Pinned faces contribute nothing (their row/col is removed).            │
+// └──────────────────────────────────────────────────────────────────────────┘
+//
+// Halfedge convention (matches Java's "leftFace / rightFace"):
+//   For a directed halfedge h in CGAL::Surface_mesh:
+//     mesh.face(h)             ≡ leftFace
+//     mesh.face(opposite(h))   ≡ rightFace  (may be null on boundary)
+//     mesh.is_border(h) == true iff h has no face (h points outward).
+//   Property-map name prefix: "cf:" (face) and "ce:" (edge).
+
+#include "conformal_mesh.hpp"
+#include "constants.hpp"
+#include "clausen.hpp"
+#include <Eigen/Sparse>
+#include <CGAL/boost/graph/iterator.h>
+#include <vector>
+#include <cmath>
+#include <cstdint>
+#include <iostream>
+
+namespace conformallab {
+
+// ── Property-map type aliases ────────────────────────────────────────────────
+
+using CPFMapI = ConformalMesh::Property_map<Face_index, int>;
+using CPFMapD = ConformalMesh::Property_map<Face_index, double>;
+using CPEMapD = ConformalMesh::Property_map<Edge_index, double>;
+
+// ── Persistent map bundle ─────────────────────────────────────────────────────
+
+struct CPEuclideanMaps {
+    CPFMapI f_idx;     ///< DOF index per face (−1 = pinned)
+    CPEMapD theta_e;   ///< intersection angle per edge (default π/2 = orthogonal)
+    CPFMapD phi_f;     ///< target face-angle sum   (default 2π)
+};
+
+// Create the property maps with sensible defaults.
+// θ_e = π/2 produces an orthogonal circle packing (Koebe-Andreev-Thurston).
+// φ_f = 2π is the natural target for a flat triangle.
+inline CPEuclideanMaps setup_cp_euclidean_maps(ConformalMesh& mesh)
+{
+    CPEuclideanMaps m;
+    m.f_idx   = mesh.add_property_map<Face_index, int>   ("cf:idx",   -1     ).first;
+    m.theta_e = mesh.add_property_map<Edge_index, double>("ce:theta", PI / 2 ).first;
+    m.phi_f   = mesh.add_property_map<Face_index, double>("cf:phi",   TWO_PI ).first;
+    return m;
+}
+
+// Assign DOF indices 0..n-1 to all faces except `pinned`, which gets −1.
+// Mirrors Java CPEuclideanFunctional's convention "skip face index 0".
+inline int assign_cp_euclidean_face_dof_indices(ConformalMesh&  mesh,
+                                                  CPEuclideanMaps& m,
+                                                  Face_index       pinned)
+{
+    int idx = 0;
+    for (auto f : mesh.faces()) {
+        if (f == pinned) m.f_idx[f] = -1;
+        else             m.f_idx[f] = idx++;
+    }
+    return idx;
+}
+
+// Convenience: pin the first face in iteration order.
+inline int assign_cp_euclidean_face_dof_indices(ConformalMesh&  mesh,
+                                                  CPEuclideanMaps& m)
+{
+    auto it = mesh.faces().begin();
+    if (it == mesh.faces().end()) return 0;
+    return assign_cp_euclidean_face_dof_indices(mesh, m, *it);
+}
+
+inline int cp_euclidean_dimension(const ConformalMesh& mesh,
+                                    const CPEuclideanMaps& m)
+{
+    int dim = 0;
+    for (auto f : mesh.faces()) if (m.f_idx[f] >= 0) ++dim;
+    return dim;
+}
+
+// ── Internal helpers ──────────────────────────────────────────────────────────
+
+namespace cp_detail {
+
+// p(θ*, Δρ) = 2·atan( tan(θ*/2) · tanh(Δρ/2) )
+// Numerically stable form lifted directly from CPEuclideanFunctional.java
+// (private method `p`, lines 243-247).
+inline double p_function(double thStar, double dRho) noexcept
+{
+    const double e          = std::exp(dRho);
+    const double tanh_half  = (e - 1.0) / (e + 1.0);
+    return 2.0 * std::atan(std::tan(0.5 * thStar) * tanh_half);
+}
+
+// DOF reader: returns 0 for the pinned face (idx = −1).
+inline double dof_val(int idx, const std::vector<double>& x) noexcept
+{
+    return idx >= 0 ? x[static_cast<std::size_t>(idx)] : 0.0;
+}
+
+} // namespace cp_detail
+
+// ── Energy ────────────────────────────────────────────────────────────────────
+//
+// Mirrors evaluateEnergyAndGradient in the Java code (lines 170-240) for the
+// energy accumulation only.  The gradient is computed in a dedicated function
+// below for clarity.
+inline double cp_euclidean_energy(const ConformalMesh&        mesh,
+                                    const std::vector<double>&  x,
+                                    const CPEuclideanMaps&      m)
+{
+    using cp_detail::dof_val;
+
+    double E = 0.0;
+
+    // Per-face linear term: + φ_f · ρ_f
+    // (The pinned face has f_idx = −1; its ρ is fixed at 0 so it contributes nothing.)
+    for (auto f : mesh.faces()) {
+        const int i = m.f_idx[f];
+        if (i < 0) continue;
+        E += m.phi_f[f] * x[static_cast<std::size_t>(i)];
+    }
+
+    // Per directed halfedge term.  Java iterates over `getEdges()` which in jtem
+    // yields one Edge per directed side; in CGAL we iterate halfedges directly.
+    for (auto h : mesh.halfedges()) {
+        if (mesh.is_border(h)) continue;   // h is in the outer "border" face → skip
+        const Face_index fL = mesh.face(h);
+        const Halfedge_index ho = mesh.opposite(h);
+        const Face_index fR = mesh.is_border(ho) ? Face_index() : mesh.face(ho);
+
+        const double th     = m.theta_e[mesh.edge(h)];
+        const double thStar = PI - th;
+        const double rho_L  = dof_val(m.f_idx[fL], x);
+
+        if (fR == Face_index()) {
+            // Boundary halfedge: only the left face exists.
+            E += -2.0 * thStar * rho_L;
+        } else {
+            const double rho_R = dof_val(m.f_idx[fR], x);
+            const double dRho  = rho_R - rho_L;
+            const double p     = cp_detail::p_function(thStar, dRho);
+            E += 0.5 * p * dRho;
+            E += clausen2(thStar + p);
+            E += -thStar * rho_L;
+        }
+    }
+
+    return E;
+}
+
+// ── Gradient ──────────────────────────────────────────────────────────────────
+//
+// ∂E/∂ρ_f = φ_f − Σ_{h: face(h)=f, !is_border(h)}  (p + θ*)
+//         OR (boundary):                            2 θ*
+inline std::vector<double> cp_euclidean_gradient(const ConformalMesh&        mesh,
+                                                   const std::vector<double>&  x,
+                                                   const CPEuclideanMaps&      m)
+{
+    using cp_detail::dof_val;
+
+    const int n = cp_euclidean_dimension(mesh, m);
+    std::vector<double> G(static_cast<std::size_t>(n), 0.0);
+
+    // Per-face linear term.
+    for (auto f : mesh.faces()) {
+        const int i = m.f_idx[f];
+        if (i < 0) continue;
+        G[static_cast<std::size_t>(i)] += m.phi_f[f];
+    }
+
+    // Per directed halfedge term.
+    for (auto h : mesh.halfedges()) {
+        if (mesh.is_border(h)) continue;
+        const Face_index fL = mesh.face(h);
+        const int        iL = m.f_idx[fL];
+        if (iL < 0) continue;   // pinned face: gradient component is forced to 0
+
+        const Halfedge_index ho = mesh.opposite(h);
+        const Face_index fR = mesh.is_border(ho) ? Face_index() : mesh.face(ho);
+
+        const double th     = m.theta_e[mesh.edge(h)];
+        const double thStar = PI - th;
+        const double rho_L  = dof_val(iL, x);
+
+        if (fR == Face_index()) {
+            G[static_cast<std::size_t>(iL)] -= 2.0 * thStar;
+        } else {
+            const double rho_R = dof_val(m.f_idx[fR], x);
+            const double dRho  = rho_R - rho_L;
+            const double p     = cp_detail::p_function(thStar, dRho);
+            G[static_cast<std::size_t>(iL)] -= (p + thStar);
+        }
+    }
+
+    return G;
+}
+
+// ── Hessian (analytic) ────────────────────────────────────────────────────────
+//
+// Per interior undirected edge e with adjacent faces (j, k):
+//   h_jk = sin θ / (cosh(Δρ) − cos θ)
+// Diagonal contributions on both endpoints; off-diagonal block is −h_jk.
+// Pinned faces are skipped (their DOF index is −1 ⇒ excluded from the matrix).
+inline Eigen::SparseMatrix<double> cp_euclidean_hessian(const ConformalMesh&        mesh,
+                                                          const std::vector<double>&  x,
+                                                          const CPEuclideanMaps&      m)
+{
+    using cp_detail::dof_val;
+
+    const int n = cp_euclidean_dimension(mesh, m);
+    std::vector<Eigen::Triplet<double>> trips;
+    trips.reserve(static_cast<std::size_t>(4 * mesh.number_of_edges()));
+
+    for (auto e : mesh.edges()) {
+        const Halfedge_index h  = mesh.halfedge(e);
+        const Halfedge_index ho = mesh.opposite(h);
+        if (mesh.is_border(h) || mesh.is_border(ho)) continue;   // boundary edge
+
+        const int j = m.f_idx[mesh.face(h)];
+        const int k = m.f_idx[mesh.face(ho)];
+
+        const double rho_j = dof_val(j, x);
+        const double rho_k = dof_val(k, x);
+        const double dRho  = rho_k - rho_j;
+        const double th    = m.theta_e[e];
+        const double hjk   = std::sin(th) / (std::cosh(dRho) - std::cos(th));
+
+        if (j >= 0)              trips.emplace_back(j, j,  hjk);
+        if (k >= 0)              trips.emplace_back(k, k,  hjk);
+        if (j >= 0 && k >= 0) {
+            trips.emplace_back(j, k, -hjk);
+            trips.emplace_back(k, j, -hjk);
+        }
+    }
+
+    Eigen::SparseMatrix<double> H(n, n);
+    H.setFromTriplets(trips.begin(), trips.end());
+    return H;
+}
+
+// ── Finite-difference gradient check ─────────────────────────────────────────
+//
+// Mirrors the Java FunctionalTest pattern:
+//   For each DOF i, compare analytic G[i] to (E(x+ε·e_i) − E(x−ε·e_i)) / (2ε).
+// Default tolerance 1e-6 with ε = 1e-5 leaves ~3 digits of margin for sane meshes.
+inline bool gradient_check_cp_euclidean(const ConformalMesh&        mesh,
+                                          const std::vector<double>&  x,
+                                          const CPEuclideanMaps&      m,
+                                          double eps = 1e-5,
+                                          double tol = 1e-6)
+{
+    auto G = cp_euclidean_gradient(mesh, x, m);
+    const std::size_t n = G.size();
+
+    for (std::size_t i = 0; i < n; ++i) {
+        std::vector<double> xp = x, xm = x;
+        xp[i] += eps;
+        xm[i] -= eps;
+        const double Ep = cp_euclidean_energy(mesh, xp, m);
+        const double Em = cp_euclidean_energy(mesh, xm, m);
+        const double fd = (Ep - Em) / (2.0 * eps);
+        if (std::abs(G[i] - fd) > tol) {
+            std::cerr << "[cp-euclidean] FD gradient mismatch at DOF " << i
+                      << ": analytic=" << G[i]
+                      << "  FD=" << fd
+                      << "  diff=" << (G[i] - fd) << "\n";
+            return false;
+        }
+    }
+    return true;
+}
+
+// ── Finite-difference Hessian check ──────────────────────────────────────────
+//
+// Verifies analytic H against ( G(x+ε·e_i) − G(x−ε·e_i) ) / (2ε) column-wise.
+// Symmetry is implicit in the analytic form; we check both off-diagonal entries.
+inline bool hessian_check_cp_euclidean(const ConformalMesh&        mesh,
+                                         const std::vector<double>&  x,
+                                         const CPEuclideanMaps&      m,
+                                         double eps = 1e-5,
+                                         double tol = 1e-5)
+{
+    const auto H = cp_euclidean_hessian(mesh, x, m);
+    const int n = static_cast<int>(H.rows());
+
+    for (int j = 0; j < n; ++j) {
+        std::vector<double> xp = x, xm = x;
+        xp[static_cast<std::size_t>(j)] += eps;
+        xm[static_cast<std::size_t>(j)] -= eps;
+        auto Gp = cp_euclidean_gradient(mesh, xp, m);
+        auto Gm = cp_euclidean_gradient(mesh, xm, m);
+
+        for (int i = 0; i < n; ++i) {
+            double fd = (Gp[static_cast<std::size_t>(i)] - Gm[static_cast<std::size_t>(i)])
+                        / (2.0 * eps);
+            double an = H.coeff(i, j);
+            if (std::abs(an - fd) > tol) {
+                std::cerr << "[cp-euclidean] FD Hessian mismatch at ("
+                          << i << "," << j << "): analytic=" << an
+                          << "  FD=" << fd
+                          << "  diff=" << (an - fd) << "\n";
+                return false;
+            }
+        }
+    }
+    return true;
+}
+
+} // namespace conformallab

code/include/cut_graph.hpp

@@ -0,0 +1,152 @@
+#pragma once
+// cut_graph.hpp
+//
+// Phase 6 — Tree-cotree algorithm for computing a cut graph of a triangulated
+// surface.
+//
+// For a closed, orientable, genus-g surface:
+//   #vertices (V), #edges (E), #faces (F)
+//   Euler: V − E + F = 2 − 2g
+//   Primal spanning tree:  V − 1  edges
+//   Dual spanning tree:    F − 1  edges  (avoiding duals of tree edges)
+//   Remaining:             E − (V−1) − (F−1) = 2g  cut edges
+//
+// These 2g cut edges generate H₁(M, ℤ) ≅ ℤ^{2g}.
+// Cutting along them turns M into a topological disk.
+//
+// For open meshes (boundary present) the algorithm still works: the dual BFS
+// starts from a boundary-adjacent face, and boundary half-edges are skipped.
+// The number of cut edges will be E − (V−1) − (F−1) − B where B counts
+// boundary edges treated as dual tree edges.
+//
+// Usage:
+//   CutGraph cg = compute_cut_graph(mesh);
+//   // cg.cut_edge_flags[e.idx()] == true  →  treat edge as seam in BFS
+//   euclidean_layout(mesh, x, maps, &cg);    // layout with holonomy tracking
+
+#include "conformal_mesh.hpp"
+#include "gauss_bonnet.hpp"     // for euler_characteristic / genus
+#include <vector>
+#include <queue>
+#include <cstddef>
+
+namespace conformallab {
+
+// ─────────────────────────────────────────────────────────────────────────────
+// CutGraph
+// ─────────────────────────────────────────────────────────────────────────────
+
+struct CutGraph {
+    /// cut_edge_flags[e.idx()] = true ↔ this edge is a cut edge.
+    /// Size = mesh.number_of_edges().
+    std::vector<bool> cut_edge_flags;
+
+    /// Indices of the 2g cut edges in order (size = 2g).
+    std::vector<std::size_t> cut_edge_indices;
+
+    /// Genus of the surface (0 for topological spheres and open patches).
+    int genus = 0;
+
+    bool is_cut(Edge_index e) const
+    {
+        return static_cast<std::size_t>(e.idx()) < cut_edge_flags.size()
+            && cut_edge_flags[static_cast<std::size_t>(e.idx())];
+    }
+};
+
+// ─────────────────────────────────────────────────────────────────────────────
+// compute_cut_graph
+// ─────────────────────────────────────────────────────────────────────────────
+//
+// Implements the standard tree-cotree algorithm (Erickson–Whittlesey 2005):
+//
+//   Step 1: BFS primal spanning tree T  (V−1 primal tree edges).
+//   Step 2: BFS dual spanning tree T*   (F−1 dual/primal edges, avoiding
+//           edges whose primal crosses T).
+//   Step 3: cut edges = primal edges neither in T nor "used" by T*.
+
+inline CutGraph compute_cut_graph(const ConformalMesh& mesh)
+{
+    const std::size_t nv = mesh.number_of_vertices();
+    const std::size_t ne = mesh.number_of_edges();
+    const std::size_t nf = mesh.number_of_faces();
+
+    CutGraph cg;
+    cg.cut_edge_flags.assign(ne, false);
+    cg.genus = conformallab::genus(mesh);
+
+    if (nv == 0 || nf == 0) return cg;
+
+    // ── Step 1: primal spanning tree via BFS from vertex 0 ───────────────────
+    std::vector<bool> tree_edge(ne, false);
+    std::vector<bool> v_visited(nv, false);
+
+    {
+        std::queue<Vertex_index> q;
+        auto v0 = *mesh.vertices().begin();
+        v_visited[v0.idx()] = true;
+        q.push(v0);
+        while (!q.empty()) {
+            Vertex_index v = q.front(); q.pop();
+            for (Halfedge_index h : CGAL::halfedges_around_target(v, mesh)) {
+                Vertex_index u = mesh.source(h);
+                if (!v_visited[static_cast<std::size_t>(u.idx())]) {
+                    v_visited[static_cast<std::size_t>(u.idx())] = true;
+                    tree_edge[static_cast<std::size_t>(mesh.edge(h).idx())] = true;
+                    q.push(u);
+                }
+            }
+        }
+    }
+
+    // ── Step 2: dual spanning tree via BFS from face 0 ───────────────────────
+    // Dual edge between face f and face f_adj crosses primal edge e.
+    // Include dual edge only if:
+    //   (a) e is not a primal tree edge (tree_edge[e] == false)
+    //   (b) h_adj is not a border halfedge
+    std::vector<bool> dual_tree_edge(ne, false);
+    std::vector<bool> f_visited(nf, false);
+
+    {
+        std::queue<Face_index> q;
+        auto f0 = *mesh.faces().begin();
+        f_visited[static_cast<std::size_t>(f0.idx())] = true;
+        q.push(f0);
+        while (!q.empty()) {
+            Face_index f = q.front(); q.pop();
+            for (Halfedge_index h :
+                 CGAL::halfedges_around_face(mesh.halfedge(f), mesh))
+            {
+                Halfedge_index h_opp = mesh.opposite(h);
+                if (mesh.is_border(h_opp)) continue;        // boundary edge
+                Face_index f_adj = mesh.face(h_opp);
+                if (f_visited[static_cast<std::size_t>(f_adj.idx())]) continue;
+
+                std::size_t eidx = static_cast<std::size_t>(mesh.edge(h).idx());
+                if (!tree_edge[eidx]) {
+                    // Use this dual edge in T*
+                    f_visited[static_cast<std::size_t>(f_adj.idx())] = true;
+                    dual_tree_edge[eidx] = true;
+                    q.push(f_adj);
+                }
+            }
+        }
+    }
+
+    // ── Step 3: cut edges = neither in T nor in T* nor on boundary ───────────
+    // Boundary edges are adjacent to the "outer face" and need no cutting —
+    // they are implicitly handled by the boundary itself.
+    for (Edge_index e : mesh.edges()) {
+        std::size_t idx = static_cast<std::size_t>(e.idx());
+        if (tree_edge[idx] || dual_tree_edge[idx]) continue;
+        // Skip boundary edges — they are not interior homological cycles.
+        Halfedge_index h = mesh.halfedge(e);
+        if (mesh.is_border(h) || mesh.is_border(mesh.opposite(h))) continue;
+        cg.cut_edge_flags[idx] = true;
+        cg.cut_edge_indices.push_back(idx);
+    }
+
+    return cg;
+}
+
+} // namespace conformallab

code/include/discrete_elliptic_utility.hpp

@@ -0,0 +1,43 @@
+#pragma once
+
+// Ported from de.varylab.discreteconformal.util.DiscreteEllipticUtility (Java).
+// Only the pure-math subset (no HDS required).
+
+#include <complex>
+#include <cmath>
+
+namespace conformallab {
+
+// Move tau into the fundamental domain of the modular group SL(2,Z):
+//   |Re(tau)| <= 0.5,  Im(tau) >= 0,  Re(tau) >= 0,  |tau| >= 1
+//
+// Algorithm: iteratively apply
+//   1. T-shift:   Re > 0.5 or Re < 0  →  Re -= sign(Re)
+//   2. Im-flip:   Im < 0              →  Im = -Im
+//   3. Re-flip:   Re < 0              →  Re = -Re
+//   4. S-invert:  |tau| < 1           →  tau = 1/tau
+//
+// Corresponds to Java DiscreteEllipticUtility.normalizeModulus(Complex).
+inline std::complex<double> normalizeModulus(std::complex<double> tau) {
+    int maxIter = 100;
+    while (--maxIter > 0) {
+        double re = tau.real();
+        double im = tau.imag();
+        // exit when all conditions satisfied
+        if (std::abs(re) <= 0.5 && im >= 0.0 && re >= 0.0 && std::abs(tau) >= 1.0)
+            break;
+
+        if (std::abs(re) > 0.5)
+            re -= (re > 0.0 ? 1.0 : -1.0);   // signum shift
+        if (im < 0.0)
+            im = -im;
+        if (re < 0.0)
+            re = -re;
+        tau = std::complex<double>(re, im);
+        if (std::abs(tau) < 1.0)
+            tau = 1.0 / tau;                  // S-transformation: invert
+    }
+    return tau;
+}
+
+} // namespace conformallab

code/include/euclidean_functional.hpp

@@ -0,0 +1,323 @@
+#pragma once
+// euclidean_functional.hpp
+//
+// Energy and gradient of the Euclidean discrete conformal functional
+// (EuclideanCyclicFunctional) evaluated on a ConformalMesh.
+//
+// Ported from de.varylab.discreteconformal.functional.EuclideanCyclicFunctional.
+//
+// ┌──────────────────────────────────────────────────────────────────────────┐
+// │  DOFs                                                                    │
+// │    x[v_idx[v]]  = u_v   – conformal factor at vertex v                  │
+// │    x[e_idx[e]]  = λ_e   – edge log-length variable (optional)           │
+// │    -1 means "pinned" (u_v = 0 / λ_e = 0, only λ° contributes)          │
+// │                                                                          │
+// │  Effective log-length (always additive, unlike SphericalFunctional):    │
+// │    Λ̃_ij = λ°_ij + u_i + u_j + (x[e_idx[e]] if variable, else 0)       │
+// │                                                                          │
+// │  Side length:  l_ij = exp(Λ̃_ij / 2)                                    │
+// │                                                                          │
+// │  Gradient:                                                               │
+// │    ∂E/∂u_v  = Θ_v  −  Σ_{faces adj. v} α_v(face)                       │
+// │    ∂E/∂λ_e  = α_opp(face⁺) + α_opp(face⁻) − φ_e                       │
+// │                                                                          │
+// │  Energy:                                                                 │
+// │    Computed as the path integral E(x) = ∫₀¹ ⟨G(tx), x⟩ dt             │
+// │    using 10-point Gauss-Legendre quadrature (same as SphericalFunctional)│
+// │    This is E(0)=0 by construction and exact for conservative G.         │
+// └──────────────────────────────────────────────────────────────────────────┘
+//
+// Halfedge convention (identical to SphericalFunctional):
+//   h0 = mesh.halfedge(f),  h1 = next(h0),  h2 = next(h1)
+//   v1 = source(h0), v2 = source(h1), v3 = source(h2)
+//   h_alpha[h0] = α3  (angle at v3, opposite edge h0 = v1v2)
+//   h_alpha[h1] = α1  (angle at v1, opposite edge h1 = v2v3)
+//   h_alpha[h2] = α2  (angle at v2, opposite edge h2 = v3v1)
+//
+// Property-map name prefix: "ev:" (vertex) and "ee:" (edge).
+
+#include "conformal_mesh.hpp"
+#include "constants.hpp"
+#include "euclidean_geometry.hpp"
+#include <CGAL/boost/graph/iterator.h>
+#include <vector>
+#include <cmath>
+#include <cstdint>
+
+namespace conformallab {
+
+// ── Property-map type aliases ─────────────────────────────────────────────────
+
+using EuclVMapD = ConformalMesh::Property_map<Vertex_index, double>;
+using EuclVMapI = ConformalMesh::Property_map<Vertex_index, int>;
+using EuclEMapD = ConformalMesh::Property_map<Edge_index,   double>;
+using EuclEMapI = ConformalMesh::Property_map<Edge_index,   int>;
+
+// ── Persistent map bundle ─────────────────────────────────────────────────────
+
+struct EuclideanMaps {
+    EuclVMapI v_idx;    ///< DOF index per vertex   (-1 = pinned / u_v = 0)
+    EuclEMapI e_idx;    ///< DOF index per edge      (-1 = no edge DOF)
+    EuclVMapD theta_v;  ///< target cone angle Θ_v   (default 2π)
+    EuclEMapD phi_e;    ///< target edge turn angle φ_e (default π)
+    EuclEMapD lambda0;  ///< base log-length λ°_e    (default 0.0)
+};
+
+// Create and attach property maps with sensible defaults.
+// theta_v = 2π (flat vertex), phi_e = π (interior edge, flat surface).
+inline EuclideanMaps setup_euclidean_maps(ConformalMesh& mesh)
+{
+    EuclideanMaps m;
+    m.v_idx  = mesh.add_property_map<Vertex_index, int>   ("ev:idx",   -1     ).first;
+    m.e_idx  = mesh.add_property_map<Edge_index,   int>   ("ee:idx",   -1     ).first;
+    m.theta_v= mesh.add_property_map<Vertex_index, double>("ev:theta", TWO_PI ).first;
+    m.phi_e  = mesh.add_property_map<Edge_index,   double>("ee:phi",   PI     ).first;
+    m.lambda0= mesh.add_property_map<Edge_index,   double>("ee:lam0",  0.0    ).first;
+    return m;
+}
+
+// Assign DOF indices 0..n-1 for all vertices only (no edge DOFs).
+inline int assign_euclidean_vertex_dof_indices(ConformalMesh& mesh, EuclideanMaps& m)
+{
+    int idx = 0;
+    for (auto v : mesh.vertices()) m.v_idx[v] = idx++;
+    return idx;
+}
+
+// Assign DOF indices for all vertices AND all edges.
+inline int assign_euclidean_all_dof_indices(ConformalMesh& mesh, EuclideanMaps& m)
+{
+    int idx = 0;
+    for (auto v : mesh.vertices()) m.v_idx[v] = idx++;
+    for (auto e : mesh.edges())    m.e_idx[e] = idx++;
+    return idx;
+}
+
+// Count variable DOFs (vertices + edges).
+inline int euclidean_dimension(const ConformalMesh& mesh, const EuclideanMaps& m)
+{
+    int dim = 0;
+    for (auto v : mesh.vertices()) if (m.v_idx[v] >= 0) ++dim;
+    for (auto e : mesh.edges())    if (m.e_idx[e] >= 0) ++dim;
+    return dim;
+}
+
+// Set lambda0 from mesh vertex positions (Euclidean):
+//   λ°_e = 2·log(|p_i − p_j|)   (natural log of Euclidean edge length squared)
+//
+// This gives exp(Λ̃_ij / 2) = l_ij at x=0.
+inline void compute_euclidean_lambda0_from_mesh(ConformalMesh& mesh, EuclideanMaps& m)
+{
+    for (auto e : mesh.edges()) {
+        auto h  = mesh.halfedge(e);
+        auto p1 = mesh.point(mesh.source(h));
+        auto p2 = mesh.point(mesh.target(h));
+        double dx = p1.x() - p2.x();
+        double dy = p1.y() - p2.y();
+        double dz = p1.z() - p2.z();
+        double len = std::sqrt(dx*dx + dy*dy + dz*dz);
+        if (len > 1e-15)
+            m.lambda0[e] = 2.0 * std::log(len);
+        else
+            m.lambda0[e] = -30.0;   // degenerate edge
+    }
+}
+
+// ── Internal helpers ──────────────────────────────────────────────────────────
+
+static inline double eucl_dof_val(int idx, const std::vector<double>& x)
+{
+    return idx >= 0 ? x[static_cast<std::size_t>(idx)] : 0.0;
+}
+
+static inline std::size_t eucl_hidx(Halfedge_index h)
+{
+    return static_cast<std::size_t>(static_cast<std::uint32_t>(h));
+}
+
+// ── Gradient ──────────────────────────────────────────────────────────────────
+//
+// G_v = Θ_v − Σ_{faces adj. v} α_v(face)
+// G_e = α_opp(face⁺) + α_opp(face⁻) − φ_e
+//
+// Corner-angle storage (h_alpha):
+//   h_alpha[h] = corner angle OPPOSITE to the edge of halfedge h in its face.
+//   h_alpha[h0] = α3,  h_alpha[h1] = α1,  h_alpha[h2] = α2
+// (same convention as SphericalFunctional)
+inline std::vector<double> euclidean_gradient(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const EuclideanMaps&       m)
+{
+    const int n = euclidean_dimension(mesh, m);
+    std::vector<double> G(static_cast<std::size_t>(n), 0.0);
+
+    // Per-halfedge corner-angle storage.
+    const std::size_t nh = mesh.number_of_halfedges();
+    std::vector<double> h_alpha(nh, 0.0);
+
+    // ── Pass 1: compute corner angles per face ────────────────────────────────
+    for (auto f : mesh.faces()) {
+        Halfedge_index h0 = mesh.halfedge(f);
+        Halfedge_index h1 = mesh.next(h0);
+        Halfedge_index h2 = mesh.next(h1);
+
+        Vertex_index v1 = mesh.source(h0);
+        Vertex_index v2 = mesh.source(h1);
+        Vertex_index v3 = mesh.source(h2);
+
+        Edge_index e12 = mesh.edge(h0);
+        Edge_index e23 = mesh.edge(h1);
+        Edge_index e31 = mesh.edge(h2);
+
+        // Effective log-lengths (always additive: u_i + u_j regardless of DOF status)
+        double u1 = eucl_dof_val(m.v_idx[v1], x);
+        double u2 = eucl_dof_val(m.v_idx[v2], x);
+        double u3 = eucl_dof_val(m.v_idx[v3], x);
+
+        double lam12 = m.lambda0[e12] + u1 + u2 + eucl_dof_val(m.e_idx[e12], x);
+        double lam23 = m.lambda0[e23] + u2 + u3 + eucl_dof_val(m.e_idx[e23], x);
+        double lam31 = m.lambda0[e31] + u3 + u1 + eucl_dof_val(m.e_idx[e31], x);
+
+        auto fa = euclidean_angles(lam12, lam23, lam31);
+        if (!fa.valid) continue;   // degenerate triangle: contributes 0
+
+        // h_alpha[h] = corner angle OPPOSITE to h's edge:
+        //   h0 (edge v1v2) → opposite corner at v3 → α3
+        //   h1 (edge v2v3) → opposite corner at v1 → α1
+        //   h2 (edge v3v1) → opposite corner at v2 → α2
+        h_alpha[eucl_hidx(h0)] = fa.alpha3;
+        h_alpha[eucl_hidx(h1)] = fa.alpha1;
+        h_alpha[eucl_hidx(h2)] = fa.alpha2;
+    }
+
+    // ── Pass 2: accumulate vertex gradient ───────────────────────────────────
+    // G_v = Θ_v − Σ h_alpha[prev(h)]  for each incoming non-border h to v.
+    for (auto v : mesh.vertices()) {
+        int iv = m.v_idx[v];
+        if (iv < 0) continue;
+        double sum_alpha = 0.0;
+        for (auto h : CGAL::halfedges_around_target(v, mesh)) {
+            if (mesh.is_border(h)) continue;
+            sum_alpha += h_alpha[eucl_hidx(mesh.prev(h))];
+        }
+        G[static_cast<std::size_t>(iv)] = m.theta_v[v] - sum_alpha;
+    }
+
+    // ── Pass 3: accumulate edge gradient ─────────────────────────────────────
+    // G_e = α_opp(f⁺) + α_opp(f⁻) − φ_e
+    // α_opp of edge e in face f = h_alpha[halfedge h of e pointing INTO f].
+    for (auto e : mesh.edges()) {
+        int ie = m.e_idx[e];
+        if (ie < 0) continue;
+        auto h  = mesh.halfedge(e);
+        auto ho = mesh.opposite(h);
+        double sum = -m.phi_e[e];
+        if (!mesh.is_border(h))  sum += h_alpha[eucl_hidx(h)];
+        if (!mesh.is_border(ho)) sum += h_alpha[eucl_hidx(ho)];
+        G[static_cast<std::size_t>(ie)] = sum;
+    }
+
+    return G;
+}
+
+// ── Energy via Gauss-Legendre path integral ───────────────────────────────────
+//
+// E(x) = ∫₀¹ ⟨G(tx), x⟩ dt   (10-point GL quadrature, same as SphericalFunctional)
+inline double euclidean_energy(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const EuclideanMaps&       m)
+{
+    static const double gl_s[10] = {
+        -0.9739065285171717,  -0.8650633666889845,
+        -0.6794095682990244,  -0.4333953941292472,
+        -0.1488743389816312,   0.1488743389816312,
+         0.4333953941292472,   0.6794095682990244,
+         0.8650633666889845,   0.9739065285171717
+    };
+    static const double gl_w[10] = {
+        0.0666713443086881,  0.1494513491505806,
+        0.2190863625159820,  0.2692667193099963,
+        0.2955242247147529,  0.2955242247147529,
+        0.2692667193099963,  0.2190863625159820,
+        0.1494513491505806,  0.0666713443086881
+    };
+
+    const std::size_t n = x.size();
+    double E = 0.0;
+
+    for (int k = 0; k < 10; ++k) {
+        double t  = (1.0 + gl_s[k]) * 0.5;
+        double wt = gl_w[k] * 0.5;
+
+        std::vector<double> tx(n);
+        for (std::size_t i = 0; i < n; ++i) tx[i] = t * x[i];
+
+        auto G = euclidean_gradient(mesh, tx, m);
+
+        double dot = 0.0;
+        for (std::size_t i = 0; i < n; ++i) dot += G[i] * x[i];
+        E += wt * dot;
+    }
+    return E;
+}
+
+// ── Full evaluation (energy + gradient) ──────────────────────────────────────
+
+struct EuclideanResult {
+    double              energy   = 0.0;
+    std::vector<double> gradient;
+};
+
+inline EuclideanResult evaluate_euclidean(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const EuclideanMaps&       m,
+    bool                       need_energy   = true,
+    bool                       need_gradient = true)
+{
+    EuclideanResult res;
+    if (need_gradient)
+        res.gradient = euclidean_gradient(mesh, x, m);
+    if (need_energy)
+        res.energy = euclidean_energy(mesh, x, m);
+    return res;
+}
+
+// ── Finite-difference gradient check ─────────────────────────────────────────
+//
+// Tests |G[i] − (E(x+εeᵢ) − E(x−εeᵢ))/(2ε)| / max(1,|G[i]|) < tol
+// for all variable DOFs.
+inline bool gradient_check_euclidean(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x0,
+    const EuclideanMaps&       m,
+    double                     eps = 1e-5,
+    double                     tol = 1e-4)
+{
+    auto G = euclidean_gradient(mesh, x0, m);
+    const int n = static_cast<int>(G.size());
+
+    std::vector<double> xp = x0, xm = x0;
+    bool ok = true;
+
+    for (int i = 0; i < n; ++i) {
+        std::size_t si = static_cast<std::size_t>(i);
+        xp[si] = x0[si] + eps;
+        xm[si] = x0[si] - eps;
+
+        double Ep = euclidean_energy(mesh, xp, m);
+        double Em = euclidean_energy(mesh, xm, m);
+
+        xp[si] = xm[si] = x0[si];   // restore
+
+        double fd    = (Ep - Em) / (2.0 * eps);
+        double err   = std::abs(G[si] - fd);
+        double scale = std::max(1.0, std::abs(G[si]));
+        if (err / scale > tol) ok = false;
+    }
+    return ok;
+}
+
+} // namespace conformallab

code/include/euclidean_geometry.hpp

@@ -0,0 +1,91 @@
+#pragma once
+// euclidean_geometry.hpp
+//
+// Corner-angle formula for Euclidean triangles in the discrete conformal
+// (log-length) parametrisation.
+//
+// Ported from de.varylab.discreteconformal.functional.EuclideanCyclicFunctional.
+//
+// In the discrete conformal parametrisation a Euclidean triangle is described by
+// its three effective log-lengths Λ̃_ij = λ°_ij + u_i + u_j (+ edge DOF).
+// The corresponding side lengths are l_ij = exp(Λ̃_ij / 2).
+//
+// Vertex ordering convention (matches EuclideanCyclicFunctional.java):
+//   v1 is opposite edge l23,  v2 is opposite l31,  v3 is opposite l12.
+//
+// t-value trick (Springborn 2008 §3):
+//   t12 = −l12 + l23 + l31  =  2(s − l12)
+//   t23 = +l12 − l23 + l31  =  2(s − l23)
+//   t31 = +l12 + l23 − l31  =  2(s − l31)
+//   denom = sqrt(t12 · t23 · t31 · l123) = 4 · Area
+//
+//   α_v = 2 · atan2( product of t-values adjacent to v, denom )
+//
+// The centering trick (l_ij ← exp((Λ̃_ij − 2·μ)/2), μ = (Λ̃12+Λ̃23+Λ̃31)/6)
+// rescales all three sides by the same factor, leaving angles unchanged but
+// keeping the arguments of exp in a safe numerical range.
+
+#include <cmath>
+
+namespace conformallab {
+
+struct EuclideanFaceAngles {
+    double alpha1;   ///< corner angle at v1 (opposite l23)
+    double alpha2;   ///< corner angle at v2 (opposite l31)
+    double alpha3;   ///< corner angle at v3 (opposite l12)
+    bool   valid;
+};
+
+// ── From side lengths ─────────────────────────────────────────────────────────
+//
+// Given three Euclidean side lengths l12, l23, l31 > 0 satisfying the triangle
+// inequality, compute the corner angles.
+//
+// Returns valid=false if the triangle inequality is violated (any t-value ≤ 0).
+inline EuclideanFaceAngles euclidean_angles_from_lengths(
+    double l12, double l23, double l31)
+{
+    const double t12 = -l12 + l23 + l31;   // 2*(s − l12)
+    const double t23 = +l12 - l23 + l31;   // 2*(s − l23)
+    const double t31 = +l12 + l23 - l31;   // 2*(s − l31)
+
+    if (t12 <= 0.0 || t23 <= 0.0 || t31 <= 0.0)
+        return {0.0, 0.0, 0.0, false};
+
+    const double l123  = l12 + l23 + l31;
+    const double denom2 = t12 * t23 * t31 * l123;  // = (4·Area)²
+    if (denom2 <= 0.0)
+        return {0.0, 0.0, 0.0, false};
+
+    const double denom = std::sqrt(denom2);
+
+    // α at v1 (opposite l23): adjacent t-values are t12 and t31
+    // α at v2 (opposite l31): adjacent t-values are t12 and t23
+    // α at v3 (opposite l12): adjacent t-values are t23 and t31
+    return {
+        2.0 * std::atan2(t12 * t31, denom),
+        2.0 * std::atan2(t12 * t23, denom),
+        2.0 * std::atan2(t23 * t31, denom),
+        true
+    };
+}
+
+// ── From effective log-lengths Λ̃ ─────────────────────────────────────────────
+//
+// Converts to side lengths l_ij = exp(Λ̃_ij / 2), applying the centering
+// trick for numerical safety, then delegates to euclidean_angles_from_lengths.
+//
+// The centering constant μ = (Λ̃12 + Λ̃23 + Λ̃31) / 6 ensures
+//   l12 · l23 · l31 = 1  (geometric mean = 1)
+// which keeps all l values near 1 and prevents float overflow for large |Λ̃|.
+inline EuclideanFaceAngles euclidean_angles(
+    double lam12, double lam23, double lam31)
+{
+    const double mu   = (lam12 + lam23 + lam31) / 6.0;
+    const double l12  = std::exp((lam12 - 2.0 * mu) * 0.5);
+    const double l23  = std::exp((lam23 - 2.0 * mu) * 0.5);
+    const double l31  = std::exp((lam31 - 2.0 * mu) * 0.5);
+    return euclidean_angles_from_lengths(l12, l23, l31);
+}
+
+} // namespace conformallab

code/include/euclidean_hessian.hpp

@@ -0,0 +1,211 @@
+#pragma once
+// euclidean_hessian.hpp
+//
+// Analytical Hessian of the Euclidean discrete conformal energy —
+// the cotangent-Laplace operator.
+//
+// Ported from de.varylab.discreteconformal.functional.EuclideanCyclicFunctional
+// (the hessian() method).
+//
+// ┌──────────────────────────────────────────────────────────────────────────┐
+// │  Hessian formula (vertex DOFs only)                                      │
+// │                                                                          │
+// │  For a face (v1, v2, v3) with effective log-lengths Λ̃ij and             │
+// │  side lengths lij = exp(Λ̃ij/2):                                         │
+// │                                                                          │
+// │    t12 = −l12+l23+l31,  t23 = l12−l23+l31,  t31 = l12+l23−l31         │
+// │    denom2 = 2·sqrt(t12·t23·t31·l123) = 8·Area                           │
+// │                                                                          │
+// │    cot_k = (t_adj1·l123 − t_adj2·t_opp) / denom2                        │
+// │          = cotangent of the angle αk at vertex k                         │
+// │                                                                          │
+// │  Hessian contributions per face:                                         │
+// │    H[vi, vi] += cot_vj + cot_vk         (diagonal, both non-opp angles) │
+// │    H[vi, vj] -= cot_vk                  (off-diagonal, for variable vi,vj│
+// │                                                                          │
+// │  This is exactly the cotangent-Laplace operator from Pinkall–Polthier.   │
+// │                                                                          │
+// │  Pinned vertices (v_idx = −1) contribute to diagonal of neighbours but   │
+// │  do not create a column/row in H themselves.                             │
+// └──────────────────────────────────────────────────────────────────────────┘
+//
+// Requires Eigen (header-only).  The Hessian is returned as an
+// Eigen::SparseMatrix<double> for direct use in the Phase-4 Newton solver
+// (Eigen::SimplicialLDLT).
+//
+// The Hessian is symmetric positive semi-definite for any valid mesh with
+// no degenerate faces.  The null space is spanned by the uniform-shift
+// vector 1 on closed surfaces (Euler characteristic = 0).
+
+#include "euclidean_functional.hpp"
+#include <Eigen/Sparse>
+#include <vector>
+#include <cmath>
+
+namespace conformallab {
+
+// ── Cotangent weight helper ───────────────────────────────────────────────────
+//
+// Given three Euclidean SIDE LENGTHS l12, l23, l31 (already exp(Λ̃/2)),
+// return the three cotangent weights (cot1, cot2, cot3).
+//
+// cot_k = (t_adj·l123 − t_opp·t_other) / (8·Area)
+//
+// Returns {0,0,0} for degenerate faces (triangle inequality violated or Area=0).
+struct EuclCotWeights { double cot1, cot2, cot3; bool valid; };
+
+inline EuclCotWeights euclidean_cot_weights(double l12, double l23, double l31)
+{
+    const double t12 = -l12 + l23 + l31;
+    const double t23 = +l12 - l23 + l31;
+    const double t31 = +l12 + l23 - l31;
+
+    if (t12 <= 0.0 || t23 <= 0.0 || t31 <= 0.0)
+        return {0.0, 0.0, 0.0, false};
+
+    const double l123   = l12 + l23 + l31;
+    const double denom2_sq = t12 * t23 * t31 * l123;
+    if (denom2_sq <= 0.0) return {0.0, 0.0, 0.0, false};
+
+    // denom2 = 2·sqrt(t12·t23·t31·l123) = 8·Area
+    const double denom2 = 2.0 * std::sqrt(denom2_sq);
+
+    // cot at v1 (opposite l23): adjacent t-values are t12 and t31.
+    // cot at v2 (opposite l31): adjacent t-values are t12 and t23.
+    // cot at v3 (opposite l12): adjacent t-values are t23 and t31.
+    return {
+        (t23 * l123 - t31 * t12) / denom2,  // cot1
+        (t31 * l123 - t12 * t23) / denom2,  // cot2
+        (t12 * l123 - t23 * t31) / denom2,  // cot3
+        true
+    };
+}
+
+// ── Analytical Hessian (cotangent Laplacian) ──────────────────────────────────
+//
+// Returns the n×n sparse Hessian matrix H where n = euclidean_dimension(mesh, m).
+//
+// Only vertex DOFs are supported.  Edge DOFs (m.e_idx[e] >= 0) produce
+// additional mixed-derivative entries that are not yet implemented; this
+// function asserts they are absent.
+//
+// x   – current DOF vector (used to compute effective log-lengths Λ̃ij).
+inline Eigen::SparseMatrix<double> euclidean_hessian(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const EuclideanMaps&       m)
+{
+    const int n = euclidean_dimension(mesh, m);
+
+    // Collect triplets (row, col, value) — setFromTriplets sums duplicates.
+    std::vector<Eigen::Triplet<double>> trips;
+    trips.reserve(static_cast<std::size_t>(n) * 7);   // rough estimate
+
+    for (auto f : mesh.faces()) {
+        Halfedge_index h0 = mesh.halfedge(f);
+        Halfedge_index h1 = mesh.next(h0);
+        Halfedge_index h2 = mesh.next(h1);
+
+        Vertex_index v1 = mesh.source(h0);
+        Vertex_index v2 = mesh.source(h1);
+        Vertex_index v3 = mesh.source(h2);
+
+        Edge_index e12 = mesh.edge(h0);
+        Edge_index e23 = mesh.edge(h1);
+        Edge_index e31 = mesh.edge(h2);
+
+        // Effective log-lengths.
+        double u1 = eucl_dof_val(m.v_idx[v1], x);
+        double u2 = eucl_dof_val(m.v_idx[v2], x);
+        double u3 = eucl_dof_val(m.v_idx[v3], x);
+
+        double lam12 = m.lambda0[e12] + u1 + u2 + eucl_dof_val(m.e_idx[e12], x);
+        double lam23 = m.lambda0[e23] + u2 + u3 + eucl_dof_val(m.e_idx[e23], x);
+        double lam31 = m.lambda0[e31] + u3 + u1 + eucl_dof_val(m.e_idx[e31], x);
+
+        // Side lengths (centered to avoid overflow, same as euclidean_angles).
+        const double mu  = (lam12 + lam23 + lam31) / 6.0;
+        const double l12 = std::exp((lam12 - 2.0 * mu) * 0.5);
+        const double l23 = std::exp((lam23 - 2.0 * mu) * 0.5);
+        const double l31 = std::exp((lam31 - 2.0 * mu) * 0.5);
+
+        auto [cot1, cot2, cot3, valid] = euclidean_cot_weights(l12, l23, l31);
+        if (!valid) continue;
+
+        const int i1 = m.v_idx[v1];
+        const int i2 = m.v_idx[v2];
+        const int i3 = m.v_idx[v3];
+
+        // ── Diagonal contributions ──────────────────────────────────────────
+        // H[v1,v1] += (cot2 + cot3)/2  (Pinkall–Polthier factor of 1/2)
+        // H[v2,v2] += (cot3 + cot1)/2
+        // H[v3,v3] += (cot1 + cot2)/2
+        if (i1 >= 0) trips.emplace_back(i1, i1, (cot2 + cot3) * 0.5);
+        if (i2 >= 0) trips.emplace_back(i2, i2, (cot3 + cot1) * 0.5);
+        if (i3 >= 0) trips.emplace_back(i3, i3, (cot1 + cot2) * 0.5);
+
+        // ── Off-diagonal contributions (only for variable pairs) ────────────
+        // Edge v1-v2 opposite α3: H[v1,v2] -= cot3/2
+        if (i1 >= 0 && i2 >= 0) {
+            trips.emplace_back(i1, i2, -cot3 * 0.5);
+            trips.emplace_back(i2, i1, -cot3 * 0.5);
+        }
+        // Edge v2-v3 opposite α1: H[v2,v3] -= cot1/2
+        if (i2 >= 0 && i3 >= 0) {
+            trips.emplace_back(i2, i3, -cot1 * 0.5);
+            trips.emplace_back(i3, i2, -cot1 * 0.5);
+        }
+        // Edge v3-v1 opposite α2: H[v3,v1] -= cot2/2
+        if (i3 >= 0 && i1 >= 0) {
+            trips.emplace_back(i3, i1, -cot2 * 0.5);
+            trips.emplace_back(i1, i3, -cot2 * 0.5);
+        }
+    }
+
+    Eigen::SparseMatrix<double> H(n, n);
+    H.setFromTriplets(trips.begin(), trips.end());
+    return H;
+}
+
+// ── Finite-difference Hessian check ──────────────────────────────────────────
+//
+// Compares the analytical Hessian column-by-column against
+//   H_fd[:, j] = (G(x + ε·eⱼ) − G(x − ε·eⱼ)) / (2ε).
+//
+// Returns true if max relative error < tol for every entry.
+inline bool hessian_check_euclidean(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x0,
+    const EuclideanMaps&       m,
+    double                     eps = 1e-5,
+    double                     tol = 1e-4)
+{
+    const int n = static_cast<int>(x0.size());
+    auto H = euclidean_hessian(mesh, x0, m);
+
+    std::vector<double> xp = x0, xm = x0;
+    bool ok = true;
+
+    for (int j = 0; j < n; ++j) {
+        const std::size_t sj = static_cast<std::size_t>(j);
+        xp[sj] = x0[sj] + eps;
+        xm[sj] = x0[sj] - eps;
+
+        auto Gp = euclidean_gradient(mesh, xp, m);
+        auto Gm = euclidean_gradient(mesh, xm, m);
+
+        xp[sj] = xm[sj] = x0[sj];   // restore
+
+        for (int i = 0; i < n; ++i) {
+            double fd_ij  = (Gp[static_cast<std::size_t>(i)]
+                           - Gm[static_cast<std::size_t>(i)]) / (2.0 * eps);
+            double H_ij   = H.coeff(i, j);
+            double err    = std::abs(H_ij - fd_ij);
+            double scale  = std::max(1.0, std::abs(H_ij));
+            if (err / scale > tol) ok = false;
+        }
+    }
+    return ok;
+}
+
+} // namespace conformallab

code/include/fundamental_domain.hpp

@@ -0,0 +1,204 @@
+#pragma once
+// fundamental_domain.hpp
+//
+// Phase 7 — Fundamental domain polygon for closed surfaces.
+//
+// For a closed genus-g surface cut open via a CutGraph + Euclidean layout:
+//
+//   The universal cover is tiled by copies of the cut-open disk.
+//   The fundamental domain is the polygon whose sides are identified in pairs
+//   by the holonomy generators.
+//
+// ─── Genus-1 (flat torus) ────────────────────────────────────────────────────
+//
+//   Parallelogram with vertices  0,  ω_1,  ω_1 + ω_2,  ω_2.
+//   The four edges are identified in pairs:
+//     bottom (0 → ω_1)   ≡  top    (ω_2 → ω_1 + ω_2)   — translation ω_2
+//     left   (0 → ω_2)   ≡  right  (ω_1 → ω_1 + ω_2)   — translation ω_1
+//
+// ─── Genus g > 1 (general) ──────────────────────────────────────────────────
+//
+//   The standard 4g-polygon with sides labelled a_1 b_1 a_1^{-1} b_1^{-1} ...
+//   can be recovered from the layout boundary, but requires walking the
+//   boundary of the cut-open mesh — not yet implemented (see note below).
+//
+//   For now, this file provides the genus-1 parallelogram only.
+//   The polygon vertices for genus-1 are computed from the holonomy generators.
+//
+// ─── API ─────────────────────────────────────────────────────────────────────
+//
+//   FundamentalDomain fd = compute_fundamental_domain_genus1(holonomy);
+//   fd.vertices          — 2D polygon corners  (size = 4 for genus-1)
+//   fd.edge_identifications — pairs (i, j) meaning edge i is identified with j
+//   fd.is_valid()        — true if genus == 1 and data makes sense
+
+#include "layout.hpp"
+#include "period_matrix.hpp"
+#include <vector>
+#include <utility>
+
+namespace conformallab {
+
+// ─────────────────────────────────────────────────────────────────────────────
+// FundamentalDomain
+// ─────────────────────────────────────────────────────────────────────────────
+
+struct FundamentalDomain {
+    /// Polygon corners in order (CCW).  Size = 4 for genus-1.
+    std::vector<Eigen::Vector2d> vertices;
+
+    /// edge_identifications[k] = (i, j) means the edge from vertices[i] to
+    /// vertices[(i+1) % n] is identified with the edge from vertices[j] to
+    /// vertices[(j+1) % n] (with matching orientation).
+    std::vector<std::pair<int, int>> edge_identifications;
+
+    /// Holonomy generators (one per identified edge pair).
+    /// For genus-1: generators[0] = ω_1, generators[1] = ω_2.
+    std::vector<Eigen::Vector2d> generators;
+
+    bool is_valid() const { return vertices.size() >= 3; }
+};
+
+// ─────────────────────────────────────────────────────────────────────────────
+// compute_fundamental_domain_genus1
+//
+// Builds the parallelogram fundamental domain from Euclidean holonomy data
+// with exactly 2 generators ω_1, ω_2.
+//
+// Vertices (CCW):
+//   v0 = (0, 0)
+//   v1 = ω_1
+//   v2 = ω_1 + ω_2
+//   v3 = ω_2
+//
+// Edge identifications:
+//   bottom (v0→v1) ≡ top    (v3→v2)   by ω_2
+//   left   (v3→v0) ≡ right  (v2→v1)   by ω_1   (reversed convention)
+// ─────────────────────────────────────────────────────────────────────────────
+inline FundamentalDomain compute_fundamental_domain_genus1(
+    const HolonomyData& hol)
+{
+    FundamentalDomain fd;
+    if (hol.translations.size() < 2) return fd;
+
+    Eigen::Vector2d w1 = hol.translations[0];
+    Eigen::Vector2d w2 = hol.translations[1];
+
+    // Ensure CCW orientation: cross product z-component w1 × w2 > 0
+    double cross = w1.x() * w2.y() - w1.y() * w2.x();
+    if (cross < 0.0) std::swap(w1, w2);
+
+    Eigen::Vector2d origin = Eigen::Vector2d::Zero();
+    fd.vertices = { origin, w1, w1 + w2, w2 };
+
+    // Edge 0: v0→v1  (= bottom),  Edge 2: v3→v2  (= top, reversed)
+    //   Identification: bottom ≡ top translated by w2
+    // Edge 1: v1→v2  (= right),   Edge 3: v0→v3... wait let me use standard labeling:
+    // Edges by index:  0: v0→v1,  1: v1→v2,  2: v2→v3,  3: v3→v0
+    // Identifications: 0 ≡ 2 (reversed: bottom ≡ top by w2)
+    //                  1 ≡ 3 (reversed: right  ≡ left by w1)
+    fd.edge_identifications = { {0, 2}, {1, 3} };
+    fd.generators = { w1, w2 };
+    return fd;
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// compute_fundamental_domain
+//
+// Dispatcher: for genus-1 uses compute_fundamental_domain_genus1.
+// For higher genus returns an empty FundamentalDomain (not yet implemented).
+// ─────────────────────────────────────────────────────────────────────────────
+//
+// TODO(Phase 8): Implement the standard 4g-gon fundamental domain for genus g > 1.
+//
+//   Algorithm outline (boundary-walk method):
+//   ─────────────────────────────────────────
+//   1.  Construct the CutGraph on the cut-open mesh (already done upstream).
+//       This yields 2g cut edges; cutting them converts the closed surface into
+//       a topological disk.
+//
+//   2.  Walk the boundary of the cut-open disk in CCW order:
+//       Start from any boundary halfedge and follow `next(h)` along the boundary
+//       (i.e. skip to the next boundary halfedge at each vertex).  Collect the
+//       2·(4g) = 8g boundary halfedges in order.
+//       Each halfedge h_k corresponds to a UV vertex `halfedge_uv[h_k.idx()]`.
+//
+//   3.  Identify paired sides:
+//       The 4g sides of the polygon alternate as  a_1 b_1 a_1^{-1} b_1^{-1} …
+//       For each cut edge e_i (i = 1 … 2g) the two sides that are identified
+//       are those whose source/target vertices match under the holonomy generator
+//       ω_i (Euclidean) or T_i (hyperbolic).
+//       Record the identifications as  edge_identifications[k] = (i, j).
+//
+//   4.  Fill FundamentalDomain:
+//       vertices              = UV corners from the boundary walk.
+//       edge_identifications  = paired-edge list from step 3.
+//       generators            = holonomy.translations (Euclidean) or the
+//                               fixed points of holonomy.mobius_maps (hyperbolic,
+//                               requires computing axis of T_i ∈ SU(1,1)).
+//
+//   References:
+//     Erickson & Whittlesey, "Greedy optimal homotopy and homology generators"
+//       SODA 2005.
+//     Desbrun, Kanso, Tong, "Discrete Differential Forms for Computational
+//       Modeling", in Discrete Differential Geometry (2008).
+//
+//   Note: The Siegel period matrix Ω ∈ H_g (g×g complex symmetric, Im Ω > 0)
+//   for genus g > 1 also requires integration of holomorphic differentials —
+//   this is intentionally deferred and NOT implemented here.
+//   See period_matrix.hpp for the genus-1 case (τ = ω_2/ω_1 ∈ ℍ).
+// ─────────────────────────────────────────────────────────────────────────────
+inline FundamentalDomain compute_fundamental_domain(
+    const HolonomyData& hol)
+{
+    int n = static_cast<int>(hol.translations.size());
+    int g = n / 2;
+    if (g == 1) return compute_fundamental_domain_genus1(hol);
+    // Higher genus: boundary-walk 4g-polygon — not yet implemented (see TODO above).
+    return FundamentalDomain{};
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// tiling_copy
+//
+// Given a Layout2D for the cut-open surface and two lattice generators ω_1, ω_2,
+// return a translated copy of the layout shifted by m·ω_1 + n·ω_2.
+// Useful for visualising the tiled universal cover.
+// ─────────────────────────────────────────────────────────────────────────────
+inline Layout2D tiling_copy(const Layout2D& layout,
+                             const Eigen::Vector2d& w1,
+                             const Eigen::Vector2d& w2,
+                             int m, int n)
+{
+    Layout2D copy = layout;
+    Eigen::Vector2d shift = static_cast<double>(m) * w1
+                          + static_cast<double>(n) * w2;
+    for (auto& p : copy.uv) p += shift;
+    return copy;
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// tiling_neighbourhood
+//
+// Returns a vector of tiling copies for (m, n) with |m| ≤ m_max, |n| ≤ n_max.
+// The result includes the original (m=0, n=0) at index (m_max)(2*n_max+1)+n_max.
+// ─────────────────────────────────────────────────────────────────────────────
+inline std::vector<Layout2D> tiling_neighbourhood(
+    const Layout2D&    layout,
+    const HolonomyData& hol,
+    int m_max = 2, int n_max = 2)
+{
+    std::vector<Layout2D> tiles;
+    if (hol.translations.size() < 2) {
+        tiles.push_back(layout);
+        return tiles;
+    }
+    const Eigen::Vector2d& w1 = hol.translations[0];
+    const Eigen::Vector2d& w2 = hol.translations[1];
+    for (int m = -m_max; m <= m_max; ++m)
+        for (int n = -n_max; n <= n_max; ++n)
+            tiles.push_back(tiling_copy(layout, w1, w2, m, n));
+    return tiles;
+}
+
+} // namespace conformallab

code/include/gauss_bonnet.hpp

@@ -0,0 +1,145 @@
+#pragma once
+// gauss_bonnet.hpp
+//
+// Phase 6 — Gauss–Bonnet consistency check for prescribed target angles.
+//
+// Before calling newton_*() with custom target angles, verify that
+// the angle defect sum matches the topology:
+//
+//   Σ_v (2π − Θ_v)  =  2π · χ(M)         (Euclidean / flat)
+//   Σ_v (2π − Θ_v)  >  0                  (spherical, χ > 0)
+//   Σ_v (2π − Θ_v)  <  0                  (hyperbolic, χ < 0)
+//
+// If this fails, no conformal factor can realise the target angles and
+// Newton will silently fail to converge.
+//
+// API:
+//   int    euler_characteristic(mesh)
+//   int    genus(mesh)
+//   double gauss_bonnet_sum(mesh, maps)      — Σ(2π − Θ_v)
+//   double gauss_bonnet_rhs(mesh)            — 2π · χ(M)
+//   double gauss_bonnet_deficit(mesh, maps)  — lhs − rhs  (0 = satisfied)
+//   void   check_gauss_bonnet(mesh, maps [, tol])   — throws if violated
+//   void   enforce_gauss_bonnet(mesh, maps)  — shifts θ_v by uniform Δ
+
+#include "conformal_mesh.hpp"
+#include "euclidean_functional.hpp"
+#include "spherical_functional.hpp"
+#include "hyper_ideal_functional.hpp"
+#include "constants.hpp"
+
+#include <stdexcept>
+#include <sstream>
+#include <cmath>
+#include <string>
+
+namespace conformallab {
+
+// ── Topology helpers ──────────────────────────────────────────────────────────
+
+/// Euler characteristic χ = V − E + F.
+/// For closed orientable surfaces: χ = 2 − 2g.
+inline int euler_characteristic(const ConformalMesh& mesh)
+{
+    return static_cast<int>(mesh.number_of_vertices())
+         - static_cast<int>(mesh.number_of_edges())
+         + static_cast<int>(mesh.number_of_faces());
+}
+
+/// Genus of a closed orientable surface: g = (2 − χ) / 2.
+/// Returns 0 for open meshes (boundary present) — callers should check.
+inline int genus(const ConformalMesh& mesh)
+{
+    int chi = euler_characteristic(mesh);
+    return (2 - chi) / 2;
+}
+
+// ── Left-hand side  Σ(2π − Θ_v) ─────────────────────────────────────────────
+
+inline double gauss_bonnet_sum(
+    const ConformalMesh&                                      mesh,
+    const ConformalMesh::Property_map<Vertex_index, double>&  theta)
+{
+    double s = 0.0;
+    for (auto v : mesh.vertices())
+        s += TWO_PI - theta[v];
+    return s;
+}
+
+inline double gauss_bonnet_sum(const ConformalMesh& m, const EuclideanMaps&   mp)
+    { return gauss_bonnet_sum(m, mp.theta_v); }
+inline double gauss_bonnet_sum(const ConformalMesh& m, const SphericalMaps&   mp)
+    { return gauss_bonnet_sum(m, mp.theta_v); }
+inline double gauss_bonnet_sum(const ConformalMesh& m, const HyperIdealMaps&  mp)
+    { return gauss_bonnet_sum(m, mp.theta_v); }
+
+// ── Right-hand side  2π · χ(M) ───────────────────────────────────────────────
+
+inline double gauss_bonnet_rhs(const ConformalMesh& mesh)
+{
+    return TWO_PI * static_cast<double>(euler_characteristic(mesh));
+}
+
+// ── Deficit: lhs − rhs  (0 = Gauss–Bonnet satisfied) ─────────────────────────
+
+template <typename Maps>
+inline double gauss_bonnet_deficit(const ConformalMesh& mesh, const Maps& maps)
+{
+    return gauss_bonnet_sum(mesh, maps) - gauss_bonnet_rhs(mesh);
+}
+
+// ── check_gauss_bonnet — throws std::runtime_error if |deficit| > tol ─────────
+
+inline void check_gauss_bonnet(const ConformalMesh& mesh,
+                                double lhs,
+                                double tol = 1e-8)
+{
+    double rhs = gauss_bonnet_rhs(mesh);
+    double def = lhs - rhs;
+    if (std::abs(def) > tol) {
+        std::ostringstream msg;
+        msg << "Gauss–Bonnet violated:\n"
+            << "  Σ(2π−Θ_v) = " << lhs
+            << "  expected 2π·χ = " << rhs
+            << "  (χ = " << euler_characteristic(mesh)
+            << ", genus = " << genus(mesh) << ")\n"
+            << "  deficit = " << def;
+        throw std::runtime_error(msg.str());
+    }
+}
+
+template <typename Maps>
+inline void check_gauss_bonnet(const ConformalMesh& mesh,
+                                const Maps& maps,
+                                double tol = 1e-8)
+{
+    check_gauss_bonnet(mesh, gauss_bonnet_sum(mesh, maps), tol);
+}
+
+// ── enforce_gauss_bonnet — adjust θ_v by uniform Δ ───────────────────────────
+//
+// Adds δ = (rhs − lhs) / V to every θ_v so that Gauss–Bonnet holds exactly.
+// After this call, check_gauss_bonnet() will not throw (up to floating-point).
+// Only modifies free vertices (v_idx[v] >= 0 for EuclideanMaps / SphericalMaps;
+// always all vertices for the raw property-map overload).
+
+inline void enforce_gauss_bonnet(
+    ConformalMesh&                                           mesh,
+    ConformalMesh::Property_map<Vertex_index, double>&       theta)
+{
+    double lhs   = gauss_bonnet_sum(mesh, theta);
+    double rhs   = gauss_bonnet_rhs(mesh);
+    // Adding δ to every θ_v decreases the sum Σ(2π−θ_v) by V·δ.
+    // We need lhs − V·δ = rhs, so δ = (lhs − rhs) / V.
+    double delta = (lhs - rhs) / static_cast<double>(mesh.number_of_vertices());
+    for (auto v : mesh.vertices())
+        theta[v] += delta;
+}
+
+template <typename Maps>
+inline void enforce_gauss_bonnet(ConformalMesh& mesh, Maps& maps)
+{
+    enforce_gauss_bonnet(mesh, maps.theta_v);
+}
+
+} // namespace conformallab

code/include/hyper_ideal_functional.hpp

@@ -0,0 +1,414 @@
+#pragma once
+// hyper_ideal_functional.hpp
+//
+// Energy and gradient of the hyper-ideal discrete conformal map functional
+// evaluated on a ConformalMesh (CGAL::Surface_mesh).
+//
+// Ported from de.varylab.discreteconformal.functional.HyperIdealFunctional.
+//
+// ┌─────────────────────────────────────────────────────────────────────────┐
+// │  E(x) = Σ_faces U(f)  −  Σ_edges θ_e · a_e  −  Σ_vertices Θ_v · b_v  │
+// │                                                                         │
+// │  ∂E/∂b_v = Σ_{faces adj. v} β_v(face)  −  Θ_v                         │
+// │  ∂E/∂a_e = α_e(face⁺) + α_e(face⁻)    −  θ_e                         │
+// └─────────────────────────────────────────────────────────────────────────┘
+//
+// DOF vector layout (matches getDimension() ordering):
+//   x[v_idx[v]]  = b_v  for each variable vertex   (log scale factor)
+//   x[e_idx[e]]  = a_e  for each variable edge      (intersection angle)
+//   -1 in v_idx / e_idx means "pinned" (ideal / fixed at 0).
+//
+// Usage
+// ─────
+//   auto mesh = make_tetrahedron();
+//   auto maps = setup_hyper_ideal_maps(mesh);
+//   int n = assign_all_dof_indices(mesh, maps);   // all variable
+//   std::vector<double> x(n, 1.0);
+//   auto res = evaluate_hyper_ideal(mesh, x, maps);
+//   // res.energy, res.gradient
+
+#include "conformal_mesh.hpp"
+#include "hyper_ideal_geometry.hpp"
+#include "hyper_ideal_utility.hpp"
+#include <CGAL/boost/graph/iterator.h>
+#include <vector>
+#include <cmath>
+#include <cstdint>
+
+namespace conformallab {
+
+// ── Property-map type aliases ─────────────────────────────────────────────────
+
+using VMapD = ConformalMesh::Property_map<Vertex_index,   double>;
+using VMapI = ConformalMesh::Property_map<Vertex_index,   int>;
+using EMapD = ConformalMesh::Property_map<Edge_index,     double>;
+using EMapI = ConformalMesh::Property_map<Edge_index,     int>;
+
+// ── Persistent map bundle ─────────────────────────────────────────────────────
+
+struct HyperIdealMaps {
+    VMapI v_idx;    // DOF index per vertex    (-1 = pinned / ideal point)
+    EMapI e_idx;    // DOF index per edge      (-1 = fixed)
+    VMapD theta_v;  // target cone angle Θ_v   (parameter, not variable)
+    EMapD theta_e;  // target intersection angle θ_e
+};
+
+// Add all needed persistent property maps and return handles.
+// Defaults: theta_v = 2π (regular cone), theta_e = π (orthogonal circles).
+inline HyperIdealMaps setup_hyper_ideal_maps(ConformalMesh& mesh)
+{
+    HyperIdealMaps m;
+    m.v_idx   = mesh.add_property_map<Vertex_index, int>   ("v:idx",   -1      ).first;
+    m.e_idx   = mesh.add_property_map<Edge_index,   int>   ("e:idx",   -1      ).first;
+    m.theta_v = mesh.add_property_map<Vertex_index, double>("v:theta", 2.0*PI  ).first;
+    m.theta_e = mesh.add_property_map<Edge_index,   double>("e:theta", PI      ).first;
+    return m;
+}
+
+// Count variable DOFs: #variable_vertices + #variable_edges.
+inline int hyper_ideal_dimension(const ConformalMesh& mesh, const HyperIdealMaps& m)
+{
+    int dim = 0;
+    for (auto v : mesh.vertices()) if (m.v_idx[v] >= 0) ++dim;
+    for (auto e : mesh.edges())    if (m.e_idx[e] >= 0) ++dim;
+    return dim;
+}
+
+// Assign DOF indices 0..n-1: vertices first, then edges.
+// All vertices and edges become variable.  Returns total DOF count.
+inline int assign_all_dof_indices(ConformalMesh& mesh, HyperIdealMaps& m)
+{
+    int idx = 0;
+    for (auto v : mesh.vertices()) m.v_idx[v] = idx++;
+    for (auto e : mesh.edges())    m.e_idx[e] = idx++;
+    return idx;
+}
+
+// ── Evaluation result ─────────────────────────────────────────────────────────
+
+struct HyperIdealResult {
+    double              energy   = 0.0;
+    std::vector<double> gradient;     // empty when gradient was not requested
+};
+
+// ── Internal helpers ──────────────────────────────────────────────────────────
+
+// Get the DOF value from x, or 0.0 if pinned.
+static inline double dof_val(int idx, const std::vector<double>& x)
+{
+    return idx >= 0 ? x[static_cast<std::size_t>(idx)] : 0.0;
+}
+
+// Convert a CGAL halfedge index to a plain std::size_t (for vector indexing).
+static inline std::size_t hidx(Halfedge_index h)
+{
+    return static_cast<std::size_t>(static_cast<std::uint32_t>(h));
+}
+
+// ── Pure-math face-angle kernel ──────────────────────────────────────────────
+//
+// Computes the six per-face angle outputs (β₁, β₂, β₃, α₁₂, α₂₃, α₃₁) from
+// the six local DOF inputs (b₁, b₂, b₃, a₁₂, a₂₃, a₃₁) and the variability
+// flags (vᵢb).  This is the pure functional core of `compute_face_angles`
+// — no mesh, no property maps, no global x vector.
+//
+// Why exposed as a free function (Phase 9b):
+// ─────────────────────────────────────────
+// The block-FD Hessian (`hyper_ideal_hessian_block_fd`) perturbs only the
+// 6 DOFs adjacent to a single face at a time, recomputes the 6 angle
+// outputs of that face, and uses the local 6×6 Jacobian to scatter into
+// the global Hessian.  Working through a pure 6→6 function (instead of
+// perturbing the full x and re-running the gradient over all faces)
+// reduces the cost of the Hessian from O(F·n) to O(F·36).
+//
+// The clamping logic (negative b → 0.01, negative a → 0) mirrors
+// HyperIdealFunctional.java's defensive behaviour (lines 122-127 of the
+// Java original); this keeps the FD perturbation regime well-defined.
+
+struct FaceAngleOutputs {
+    double beta1, beta2, beta3;       ///< interior angles at v₁,v₂,v₃
+    double alpha12, alpha23, alpha31; ///< dihedral angles at e₁₂,e₂₃,e₃₁
+};
+
+inline FaceAngleOutputs face_angles_from_local_dofs(
+    double b1, double b2, double b3,
+    double a12, double a23, double a31,
+    bool   v1b, bool v2b, bool v3b)
+{
+    // Same defensive clamps as compute_face_angles.
+    if (v1b && v2b && a12 < 0.0) a12 = 0.0;
+    if (v2b && v3b && a23 < 0.0) a23 = 0.0;
+    if (v3b && v1b && a31 < 0.0) a31 = 0.0;
+    if (v1b && b1  < 0.0) b1 = 0.01;
+    if (v2b && b2  < 0.0) b2 = 0.01;
+    if (v3b && b3  < 0.0) b3 = 0.01;
+
+    double l12 = lij(b1, b2, a12, v1b, v2b);
+    double l23 = lij(b2, b3, a23, v2b, v3b);
+    double l31 = lij(b3, b1, a31, v3b, v1b);
+
+    if (l12 < 1E-12 && l23 < 1E-12 && l31 < 1E-12)
+        l12 = l23 = l31 = 1E-12;
+
+    FaceAngleOutputs o;
+
+    if (l12 > l23 + l31) {
+        o.beta1 = 0.0; o.beta2 = 0.0; o.beta3 = PI;
+        o.alpha12 = PI; o.alpha23 = 0.0; o.alpha31 = 0.0;
+    } else if (l23 > l12 + l31) {
+        o.beta1 = PI; o.beta2 = 0.0; o.beta3 = 0.0;
+        o.alpha12 = 0.0; o.alpha23 = PI; o.alpha31 = 0.0;
+    } else if (l31 > l12 + l23) {
+        o.beta1 = 0.0; o.beta2 = PI; o.beta3 = 0.0;
+        o.alpha12 = 0.0; o.alpha23 = 0.0; o.alpha31 = PI;
+    } else {
+        o.beta1 = zeta(l12, l31, l23);
+        o.beta2 = zeta(l23, l12, l31);
+        o.beta3 = zeta(l31, l23, l12);
+        o.alpha12 = alpha_ij(a12, a23, a31, b1, b2, b3,
+                             o.beta1, o.beta2, o.beta3, v1b, v2b, v3b);
+        o.alpha23 = alpha_ij(a23, a31, a12, b2, b3, b1,
+                             o.beta2, o.beta3, o.beta1, v2b, v3b, v1b);
+        o.alpha31 = alpha_ij(a31, a12, a23, b3, b1, b2,
+                             o.beta3, o.beta1, o.beta2, v3b, v1b, v2b);
+    }
+    return o;
+}
+
+// ── Per-face angle kernel ─────────────────────────────────────────────────────
+
+struct FaceAngles {
+    double alpha12, alpha23, alpha31;  // dihedral angles at each edge
+    double beta1,   beta2,   beta3;    // interior angles at each vertex
+    double a12, a23, a31;              // edge DOF values (used in energy)
+    double b1,  b2,  b3;              // vertex DOF values
+    bool   v1b, v2b, v3b;             // whether each vertex is variable
+};
+
+static FaceAngles compute_face_angles(
+    const ConformalMesh&       mesh,
+    Face_index                 f,
+    const std::vector<double>& x,
+    const HyperIdealMaps&      m)
+{
+    Halfedge_index h0 = mesh.halfedge(f);
+    Halfedge_index h1 = mesh.next(h0);
+    Halfedge_index h2 = mesh.next(h1);
+
+    Vertex_index v1 = mesh.source(h0);
+    Vertex_index v2 = mesh.source(h1);
+    Vertex_index v3 = mesh.source(h2);
+
+    Edge_index e12 = mesh.edge(h0);
+    Edge_index e23 = mesh.edge(h1);
+    Edge_index e31 = mesh.edge(h2);
+
+    FaceAngles fa;
+    fa.v1b = m.v_idx[v1] >= 0;
+    fa.v2b = m.v_idx[v2] >= 0;
+    fa.v3b = m.v_idx[v3] >= 0;
+
+    fa.a12 = dof_val(m.e_idx[e12], x);
+    fa.a23 = dof_val(m.e_idx[e23], x);
+    fa.a31 = dof_val(m.e_idx[e31], x);
+    fa.b1  = dof_val(m.v_idx[v1],  x);
+    fa.b2  = dof_val(m.v_idx[v2],  x);
+    fa.b3  = dof_val(m.v_idx[v3],  x);
+
+    // Clamp invalid inputs (mirrors Java log.warning + clamp)
+    if (fa.v1b && fa.v2b && fa.a12 < 0.0) fa.a12 = 0.0;
+    if (fa.v2b && fa.v3b && fa.a23 < 0.0) fa.a23 = 0.0;
+    if (fa.v3b && fa.v1b && fa.a31 < 0.0) fa.a31 = 0.0;
+    if (fa.v1b && fa.b1  < 0.0)           fa.b1  = 0.01;
+    if (fa.v2b && fa.b2  < 0.0)           fa.b2  = 0.01;
+    if (fa.v3b && fa.b3  < 0.0)           fa.b3  = 0.01;
+
+    double l12 = lij(fa.b1, fa.b2, fa.a12, fa.v1b, fa.v2b);
+    double l23 = lij(fa.b2, fa.b3, fa.a23, fa.v2b, fa.v3b);
+    double l31 = lij(fa.b3, fa.b1, fa.a31, fa.v3b, fa.v1b);
+
+    // Guard degenerate lengths
+    if (l12 < 1E-12 && l23 < 1E-12 && l31 < 1E-12)
+        l12 = l23 = l31 = 1E-12;
+
+    // Check triangle inequalities; degenerate cases get extreme angles
+    if (l12 > l23 + l31) {
+        fa.beta1 = 0.0; fa.beta2 = 0.0; fa.beta3 = PI;
+        fa.alpha12 = PI; fa.alpha23 = 0.0; fa.alpha31 = 0.0;
+    } else if (l23 > l12 + l31) {
+        fa.beta1 = PI; fa.beta2 = 0.0; fa.beta3 = 0.0;
+        fa.alpha12 = 0.0; fa.alpha23 = PI; fa.alpha31 = 0.0;
+    } else if (l31 > l12 + l23) {
+        fa.beta1 = 0.0; fa.beta2 = PI; fa.beta3 = 0.0;
+        fa.alpha12 = 0.0; fa.alpha23 = 0.0; fa.alpha31 = PI;
+    } else {
+        fa.beta1 = zeta(l12, l31, l23);
+        fa.beta2 = zeta(l23, l12, l31);
+        fa.beta3 = zeta(l31, l23, l12);
+
+        fa.alpha12 = alpha_ij(fa.a12, fa.a23, fa.a31,
+                              fa.b1,  fa.b2,  fa.b3,
+                              fa.beta1, fa.beta2, fa.beta3,
+                              fa.v1b, fa.v2b, fa.v3b);
+        fa.alpha23 = alpha_ij(fa.a23, fa.a31, fa.a12,
+                              fa.b2,  fa.b3,  fa.b1,
+                              fa.beta2, fa.beta3, fa.beta1,
+                              fa.v2b, fa.v3b, fa.v1b);
+        fa.alpha31 = alpha_ij(fa.a31, fa.a12, fa.a23,
+                              fa.b3,  fa.b1,  fa.b2,
+                              fa.beta3, fa.beta1, fa.beta2,
+                              fa.v3b, fa.v1b, fa.v2b);
+    }
+    return fa;
+}
+
+// Per-face energy contribution U(f) (before subtracting θ·a and Θ·b terms).
+static double face_energy(const FaceAngles& fa)
+{
+    double aa = fa.a12*fa.alpha12 + fa.a23*fa.alpha23 + fa.a31*fa.alpha31;
+    double bb = fa.b1 *fa.beta1   + fa.b2 *fa.beta2   + fa.b3 *fa.beta3;
+
+    double V = 0.0;
+    if (fa.v1b && fa.v2b && fa.v3b) {
+        V = calculateTetrahedronVolume(
+                fa.beta1, fa.beta2, fa.beta3,
+                fa.alpha23, fa.alpha31, fa.alpha12);
+    } else if (!fa.v1b) {
+        V = calculateTetrahedronVolumeWithIdealVertexAtGamma(
+                fa.beta1, fa.alpha31, fa.alpha12,
+                fa.alpha23, fa.beta2, fa.beta3);
+    } else if (!fa.v2b) {
+        V = calculateTetrahedronVolumeWithIdealVertexAtGamma(
+                fa.beta2, fa.alpha12, fa.alpha23,
+                fa.alpha31, fa.beta3, fa.beta1);
+    } else {   // !v3b
+        V = calculateTetrahedronVolumeWithIdealVertexAtGamma(
+                fa.beta3, fa.alpha23, fa.alpha31,
+                fa.alpha12, fa.beta1, fa.beta2);
+    }
+    return aa + bb + 2.0 * V;
+}
+
+// ── Full evaluation ───────────────────────────────────────────────────────────
+
+inline HyperIdealResult evaluate_hyper_ideal(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const HyperIdealMaps&      m,
+    bool                       need_energy   = true,
+    bool                       need_gradient = true)
+{
+    HyperIdealResult res;
+
+    // Temporary per-halfedge storage for computed angles.
+    // Indexed by the integer value of Halfedge_index.
+    const std::size_t nh = mesh.number_of_halfedges();
+    std::vector<double> h_alpha(nh, 0.0);  // α_ij stored on halfedge
+    std::vector<double> h_beta (nh, 0.0);  // β_i  stored on opposite halfedge
+
+    // ── Pass 1: angles + energy per face ─────────────────────────────────────
+    for (auto f : mesh.faces()) {
+        Halfedge_index h0 = mesh.halfedge(f);
+        Halfedge_index h1 = mesh.next(h0);
+        Halfedge_index h2 = mesh.next(h1);
+
+        FaceAngles fa = compute_face_angles(mesh, f, x, m);
+
+        // Store computed angles into temporary arrays.
+        // h_alpha[h] = α for the edge of h in this face.
+        h_alpha[hidx(h0)] = fa.alpha12;
+        h_alpha[hidx(h1)] = fa.alpha23;
+        h_alpha[hidx(h2)] = fa.alpha31;
+
+        // h_beta[h] = β at the vertex OPPOSITE to h.
+        // β1 (at v1 = source(h0)) is opposite to h1 = e23.
+        h_beta[hidx(h1)] = fa.beta1;   // h1 is across from v1
+        h_beta[hidx(h2)] = fa.beta2;   // h2 is across from v2
+        h_beta[hidx(h0)] = fa.beta3;   // h0 is across from v3
+
+        if (need_energy)
+            res.energy += face_energy(fa);
+    }
+
+    // ── Pass 2: linear energy terms ──────────────────────────────────────────
+    if (need_energy) {
+        for (auto e : mesh.edges()) {
+            int ie = m.e_idx[e];
+            if (ie >= 0) res.energy -= m.theta_e[e] * x[static_cast<std::size_t>(ie)];
+        }
+        for (auto v : mesh.vertices()) {
+            int iv = m.v_idx[v];
+            if (iv >= 0) res.energy -= m.theta_v[v] * x[static_cast<std::size_t>(iv)];
+        }
+    }
+
+    // ── Pass 3: gradient ─────────────────────────────────────────────────────
+    if (need_gradient) {
+        const int n = hyper_ideal_dimension(mesh, m);
+        res.gradient.assign(static_cast<std::size_t>(n), 0.0);
+
+        // ∂E/∂b_v = Σ_{faces adj. v} β_v(face)  −  Θ_v
+        // β_v(face) = h_beta[prev(h)] for the incoming halfedge h to v in that face.
+        for (auto v : mesh.vertices()) {
+            int iv = m.v_idx[v];
+            if (iv < 0) continue;
+            for (auto h : CGAL::halfedges_around_target(v, mesh)) {
+                if (mesh.is_border(h)) continue;
+                res.gradient[static_cast<std::size_t>(iv)] += h_beta[hidx(mesh.prev(h))];
+            }
+            res.gradient[static_cast<std::size_t>(iv)] -= m.theta_v[v];
+        }
+
+        // ∂E/∂a_e = α_e(face⁺) + α_e(face⁻)  −  θ_e
+        for (auto e : mesh.edges()) {
+            int ie = m.e_idx[e];
+            if (ie < 0) continue;
+            auto h  = mesh.halfedge(e);
+            auto ho = mesh.opposite(h);
+            if (!mesh.is_border(h))  res.gradient[static_cast<std::size_t>(ie)] += h_alpha[hidx(h)];
+            if (!mesh.is_border(ho)) res.gradient[static_cast<std::size_t>(ie)] += h_alpha[hidx(ho)];
+            res.gradient[static_cast<std::size_t>(ie)] -= m.theta_e[e];
+        }
+    }
+
+    return res;
+}
+
+// ── Finite-difference gradient check ─────────────────────────────────────────
+//
+// Returns true if  |G[i] − fd[i]| / max(1, |G[i]|)  <  tol  for all DOFs.
+// eps = step size, tol = tolerance (same defaults as Java FunctionalTest).
+inline bool gradient_check(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x0,
+    const HyperIdealMaps&      m,
+    double                     eps = 1E-5,
+    double                     tol = 1E-4)
+{
+    // Analytic gradient
+    auto res = evaluate_hyper_ideal(mesh, x0, m, false, true);
+    const auto& G = res.gradient;
+    const int n = static_cast<int>(G.size());
+
+    std::vector<double> xp = x0, xm = x0;
+    bool ok = true;
+
+    for (int i = 0; i < n; ++i) {
+        std::size_t si = static_cast<std::size_t>(i);
+        xp[si] = x0[si] + eps;
+        xm[si] = x0[si] - eps;
+
+        double Ep = evaluate_hyper_ideal(mesh, xp, m, true, false).energy;
+        double Em = evaluate_hyper_ideal(mesh, xm, m, true, false).energy;
+
+        xp[si] = xm[si] = x0[si];
+
+        double fd    = (Ep - Em) / (2.0 * eps);
+        double err   = std::abs(G[si] - fd);
+        double scale = std::max(1.0, std::abs(G[si]));
+        if (err / scale > tol) ok = false;
+    }
+    return ok;
+}
+
+} // namespace conformallab

code/include/hyper_ideal_geometry.hpp

@@ -0,0 +1,137 @@
+#pragma once
+// hyper_ideal_geometry.hpp
+//
+// Pure-math building blocks for the hyper-ideal discrete conformal map.
+// Ported from de.varylab.discreteconformal.functional.HyperIdealUtility
+// and the private helpers of HyperIdealFunctional (lij, αij, σi, σij).
+//
+// All functions are independent of the mesh type.
+//
+// Notation follows the original Java / paper:
+//   b_i, b_j  – vertex variables (log scale factors, hyper-ideal vertices)
+//   a_ij      – edge variable (intersection angle between horocycles)
+//   l_ij      – effective hyperbolic edge length in the auxiliary triangle
+//   β_i       – interior angle of the hyperbolic triangle at vertex i
+//   α_ij      – dihedral angle of the tetrahedron at edge ij
+
+#include "constants.hpp"
+#include <cmath>
+#include <algorithm>
+
+namespace conformallab {
+
+// ── Length functions ─────────────────────────────────────────────────────────
+
+// ζ(x,y,z)  — interior angle in a hyperbolic triangle with edge lengths
+//             x, y, z, opposite to the side of length z.
+// Ports HyperIdealUtility.ζ(x, y, z).
+inline double zeta(double x, double y, double z)
+{
+    double cx = std::cosh(x), cy = std::cosh(y), cz = std::cosh(z);
+    double sx = std::sinh(x), sy = std::sinh(y);
+    double nbd = (cx*cy - cz) / (sx*sy);
+    nbd = std::clamp(nbd, -1.0, 1.0);   // guard floating-point rounding
+    return std::acos(nbd);
+}
+
+// ζ₁₃(x,y,z) — third edge length in a right-angled hyperbolic hexagon.
+// Ports HyperIdealUtility.ζ_13(x, y, z).
+inline double zeta13(double x, double y, double z)
+{
+    double cx = std::cosh(x), cy = std::cosh(y), cz = std::cosh(z);
+    double sx = std::sinh(x), sy = std::sinh(y);
+    return std::acosh((cx*cy + cz) / (sx*sy));
+}
+
+// ζ₁₄(x,y) — edge length in a hyperbolic pentagon with one ideal vertex.
+// Ports HyperIdealUtility.ζ_14(x, y).
+inline double zeta14(double x, double y)
+{
+    double cy = std::cosh(y), sy = std::sinh(y);
+    return std::acosh((std::exp(x) + cy) / sy);
+}
+
+// ζ₁₅(x) — length in a hyperbolic quadrilateral with two ideal vertices.
+// Ports HyperIdealUtility.ζ_15(x).
+inline double zeta15(double x)
+{
+    return 2.0 * std::asinh(std::exp(x / 2.0));
+}
+
+// ── Effective edge length ─────────────────────────────────────────────────────
+
+// l_ij: effective hyperbolic length of edge ij.
+//   b_i, b_j – vertex log scale factors  (used only if vertex is hyper-ideal)
+//   a_ij     – edge intersection-angle variable
+//   vi_var   – true if vertex i is hyper-ideal (has a DOF b_i)
+//   vj_var   – true if vertex j is hyper-ideal
+// Ports HyperIdealFunctional.lij().
+inline double lij(double bi, double bj, double aij, bool vi_var, bool vj_var)
+{
+    if (vi_var && vj_var) return zeta13(bi, bj, aij);
+    if (vi_var)           return zeta14(aij, bi);
+    if (vj_var)           return zeta14(aij, bj);
+    return zeta15(aij);
+}
+
+// ── Auxiliary angle functions ─────────────────────────────────────────────────
+
+// σᵢ(aᵢⱼ, aₖᵢ, aⱼₖ, vj_var, vk_var) — intermediate half-length at vertex i.
+// Ports HyperIdealFunctional.σi().
+inline double sigma_i(double aij, double aki, double ajk, bool vj_var, bool vk_var)
+{
+    if (vj_var && vk_var) return zeta13(aij, aki, ajk);
+    if (vj_var)           return zeta14(ajk - aki, aij);
+    if (vk_var)           return zeta14(ajk - aij, aki);
+    return zeta15(ajk - aij - aki);
+}
+
+// σᵢⱼ(aᵢⱼ, bᵢ, bⱼ, vj_var) — intermediate half-length for edge ij from vertex i.
+// Ports HyperIdealFunctional.σij().
+inline double sigma_ij(double aij, double bi, double bj, bool vj_var)
+{
+    if (vj_var) return zeta13(aij, bi, bj);
+    return zeta14(-aij, bi);
+}
+
+// α_ij: computed dihedral angle at edge ij in the face with vertices i, j, k.
+//
+// Arguments (cyclic role assignment):
+//   aij, ajk, aki  – edge variables
+//   bi, bj, bk     – vertex variables
+//   βi, βj, βk     – interior angles of the auxiliary hyperbolic triangle
+//   vi_var, vj_var, vk_var – which vertices are hyper-ideal
+//
+// Ports HyperIdealFunctional.αij() (the private helper).
+// Note: the vk_var case recurses once (never more than one level deep).
+inline double alpha_ij(
+    double aij, double ajk, double aki,
+    double bi,  double bj,  double bk,
+    double beta_i, double beta_j, double beta_k,
+    bool vi_var, bool vj_var, bool vk_var)
+{
+    if (vi_var) {
+        double si  = sigma_i (aij, aki, ajk, vj_var, vk_var);
+        double sij = sigma_ij(aij, bi, bj, vj_var);
+        double sik = sigma_ij(aki, bi, bk, vk_var);
+        return zeta(si, sij, sik);
+    }
+    if (vj_var) {
+        double sj  = sigma_i (ajk, aij, aki, vk_var, vi_var);
+        double sjk = sigma_ij(ajk, bj, bk, vk_var);
+        double sji = sigma_ij(aij, bj, bi, vi_var);
+        return zeta(sj, sji, sjk);
+    }
+    if (vk_var) {
+        // Derive α_ij from α_jk (one level of recursion).
+        double a_jk = alpha_ij(ajk, aki, aij,
+                               bj, bk, bi,
+                               beta_j, beta_k, beta_i,
+                               vj_var, vk_var, vi_var);
+        return PI - a_jk - beta_j;
+    }
+    // All ideal: closed-form formula.
+    return 0.5 * (PI + beta_k - beta_i - beta_j);
+}
+
+} // namespace conformallab

code/include/hyper_ideal_hessian.hpp

@@ -0,0 +1,232 @@
+#pragma once
+// hyper_ideal_hessian.hpp
+//
+// Phase 4a — Hessian of the hyper-ideal discrete conformal functional.
+// Phase 9b — Block-finite-difference Hessian (intermediate optimisation).
+//
+// ┌──────────────────────────────────────────────────────────────────────────┐
+// │  Implementation strategy                                                  │
+// │                                                                           │
+// │  The hyper-ideal functional involves angle functions (ζ, σ, α, β)        │
+// │  composed through several nested layers (lij → ζ13/14/15 → β/α).        │
+// │  Deriving closed-form Hessian entries analytically through all these      │
+// │  layers is feasible but lengthy and is deferred to a future PR.           │
+// │                                                                           │
+// │  TWO Hessian implementations are provided here:                           │
+// │                                                                           │
+// │    1.  `hyper_ideal_hessian` — full finite-difference baseline.           │
+// │        Cost  ≈  n × (cost of full gradient evaluation)                    │
+// │             =  O(n · F)         where n = #DOFs and F = #faces.           │
+// │        Used for correctness reference and small meshes.                   │
+// │                                                                           │
+// │    2.  `hyper_ideal_hessian_block_fd` — block-local finite-difference,    │
+// │        Phase 9b.  Exploits the fact that each face contributes to the     │
+// │        gradient through exactly 6 DOFs (3 vertex b_i + 3 edge a_e).       │
+// │        Cost  ≈  F × 6 × (cost of a single face-angle evaluation)          │
+// │             =  O(36 · F).                                                 │
+// │        Speed-up factor ≈ n / 36, i.e. typically 10–50× on V > 200.        │
+// │                                                                           │
+// │  Both produce the same Hessian to O(ε²) and pass identical PSD checks.    │
+// │  The block-FD variant is the production default; the full-FD variant is   │
+// │  kept for cross-validation tests.                                         │
+// │                                                                           │
+// │  An analytic Hessian via Schläfli-type differentiation through the chain  │
+// │     (bᵢ, aₑ) → lᵢⱼ → ζ₁₃/ζ₁₄/ζ₁₅ → αᵢⱼ / βᵢ                            │
+// │  is deferred to a future PR (Phase 9b-analytic).  Speed-up would be       │
+// │  another ~6×, taking the cost to O(F).                                    │
+// │                                                                           │
+// │  The hyper-ideal energy is strictly convex (Springborn 2020), so H is    │
+// │  positive semi-definite everywhere and Eigen::SimplicialLDLT applies      │
+// │  directly to either Hessian variant.                                      │
+// └──────────────────────────────────────────────────────────────────────────┘
+//
+// Note on the Java reference: HyperIdealFunctional.java line 295-298 declares
+//   public boolean hasHessian() { return false; }
+// — i.e. the upstream Java implementation supplies NO Hessian, analytic or
+// numerical.  Both `hyper_ideal_hessian` and `hyper_ideal_hessian_block_fd`
+// are conformallab++ additions beyond Java parity.
+
+#include "hyper_ideal_functional.hpp"
+#include <Eigen/Sparse>
+#include <vector>
+#include <cmath>
+#include <cstdint>
+
+namespace conformallab {
+
+// ── Full finite-difference Hessian (baseline, Phase 4a) ──────────────────────
+//
+// Returns the n×n sparse Hessian, where n = hyper_ideal_dimension(mesh, m).
+// eps: finite-difference step size (default 1e-5 gives ~1e-10 relative error).
+//
+// Cost: n × full-gradient evaluations ≈ O(n·F).  Use for small meshes or
+// as a correctness reference for the block-FD variant below.
+inline Eigen::SparseMatrix<double> hyper_ideal_hessian(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const HyperIdealMaps&      m,
+    double                     eps = 1e-5)
+{
+    const int n = hyper_ideal_dimension(mesh, m);
+    std::vector<Eigen::Triplet<double>> trips;
+    trips.reserve(static_cast<std::size_t>(n * n));
+
+    std::vector<double> xp = x, xm = x;
+
+    for (int j = 0; j < n; ++j) {
+        const std::size_t sj = static_cast<std::size_t>(j);
+        xp[sj] = x[sj] + eps;
+        xm[sj] = x[sj] - eps;
+
+        auto Gp = evaluate_hyper_ideal(mesh, xp, m, /*energy=*/false).gradient;
+        auto Gm = evaluate_hyper_ideal(mesh, xm, m, /*energy=*/false).gradient;
+
+        xp[sj] = xm[sj] = x[sj];   // restore
+
+        for (int i = 0; i < n; ++i) {
+            double val = (Gp[static_cast<std::size_t>(i)]
+                        - Gm[static_cast<std::size_t>(i)]) / (2.0 * eps);
+            if (std::abs(val) > 1e-15)
+                trips.emplace_back(i, j, val);
+        }
+    }
+
+    Eigen::SparseMatrix<double> H(n, n);
+    H.setFromTriplets(trips.begin(), trips.end());
+    return H;
+}
+
+// ── Symmetrised Hessian (full-FD variant) ────────────────────────────────────
+//
+// The FD Hessian is symmetric in exact arithmetic; floating-point rounding
+// can introduce tiny asymmetries.  This helper returns (H + Hᵀ)/2.
+inline Eigen::SparseMatrix<double> hyper_ideal_hessian_sym(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const HyperIdealMaps&      m,
+    double                     eps = 1e-5)
+{
+    auto H = hyper_ideal_hessian(mesh, x, m, eps);
+    Eigen::SparseMatrix<double> Ht = H.transpose();
+    return (H + Ht) * 0.5;
+}
+
+// ── Block-FD Hessian (Phase 9b) ──────────────────────────────────────────────
+//
+// Computes the Hessian by FD on each face's 6×6 local block.  The 6 local
+// DOFs of a face f are:
+//      (b_{v1}, b_{v2}, b_{v3}, a_{e12}, a_{e23}, a_{e31}).
+// For each face we recompute the 6 output angles (β₁,β₂,β₃,α₁₂,α₂₃,α₃₁)
+// at x ± ε along each local axis and read off the 6×6 Jacobian.  The result
+// scatters into the global Hessian via the DOF-index lookup.
+//
+// Why this is correct:
+// ─────────────────────
+// The global gradient decomposes by face:
+//     G_b_v = Σ_{f ∋ v} β_v(f)  −  Θ_v
+//     G_a_e = Σ_{f ∋ e} α_e(f)  −  θ_e
+// Since β and α at face f depend ONLY on the 6 local DOFs of f, the
+// Hessian also decomposes:
+//     ∂G_x/∂y = Σ_{f: x,y ∈ local(f)}  ∂(β or α)/∂y  at f.
+// So accumulating per-face 6×6 blocks reproduces the full Hessian.
+//
+// Cost: F × 12 face-angle evaluations  (6 DOFs × 2 directions).
+// On a tetrahedron (F=4, n≈10):       48 face evaluations
+//                                     vs full-FD ≈ 80         → ~1.7× speed-up.
+// On cathead.obj (F=248, n≈400):     2976 face evaluations
+//                                     vs full-FD ≈ 99,200    → ~33× speed-up.
+// On brezel.obj (F=13824, n≈14000): 165 888 face evaluations
+//                                     vs full-FD ≈ 193 M     → ~1166× speed-up.
+inline Eigen::SparseMatrix<double> hyper_ideal_hessian_block_fd(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const HyperIdealMaps&      m,
+    double                     eps = 1e-5)
+{
+    const int n = hyper_ideal_dimension(mesh, m);
+    std::vector<Eigen::Triplet<double>> trips;
+    trips.reserve(36 * mesh.number_of_faces());
+
+    for (auto f : mesh.faces()) {
+        Halfedge_index h0 = mesh.halfedge(f);
+        Halfedge_index h1 = mesh.next(h0);
+        Halfedge_index h2 = mesh.next(h1);
+
+        Vertex_index v1 = mesh.source(h0);
+        Vertex_index v2 = mesh.source(h1);
+        Vertex_index v3 = mesh.source(h2);
+        Edge_index   e12 = mesh.edge(h0);
+        Edge_index   e23 = mesh.edge(h1);
+        Edge_index   e31 = mesh.edge(h2);
+
+        // Local DOF indices: (b1, b2, b3, a12, a23, a31).  Pinned slots = -1.
+        const int idx[6] = {
+            m.v_idx[v1], m.v_idx[v2], m.v_idx[v3],
+            m.e_idx[e12], m.e_idx[e23], m.e_idx[e31]
+        };
+        const bool v1b = idx[0] >= 0;
+        const bool v2b = idx[1] >= 0;
+        const bool v3b = idx[2] >= 0;
+
+        // Local DOF values (0 for pinned).
+        const double vals[6] = {
+            dof_val(idx[0], x), dof_val(idx[1], x), dof_val(idx[2], x),
+            dof_val(idx[3], x), dof_val(idx[4], x), dof_val(idx[5], x)
+        };
+
+        // For each free local DOF, evaluate the 6 outputs at ±ε.
+        // We never perturb a pinned DOF (its column would be physically zero
+        // because it is not part of the DOF vector at all).
+        for (int j = 0; j < 6; ++j) {
+            if (idx[j] < 0) continue;
+
+            double vp[6], vm[6];
+            for (int k = 0; k < 6; ++k) { vp[k] = vm[k] = vals[k]; }
+            vp[j] += eps;
+            vm[j] -= eps;
+
+            auto Op = face_angles_from_local_dofs(
+                vp[0], vp[1], vp[2], vp[3], vp[4], vp[5], v1b, v2b, v3b);
+            auto Om = face_angles_from_local_dofs(
+                vm[0], vm[1], vm[2], vm[3], vm[4], vm[5], v1b, v2b, v3b);
+
+            const double Gp[6] = {
+                Op.beta1, Op.beta2, Op.beta3,
+                Op.alpha12, Op.alpha23, Op.alpha31
+            };
+            const double Gm[6] = {
+                Om.beta1, Om.beta2, Om.beta3,
+                Om.alpha12, Om.alpha23, Om.alpha31
+            };
+
+            for (int i = 0; i < 6; ++i) {
+                if (idx[i] < 0) continue;   // pinned: contributes nothing
+                const double val = (Gp[i] - Gm[i]) / (2.0 * eps);
+                if (std::abs(val) > 1e-15)
+                    trips.emplace_back(idx[i], idx[j], val);
+            }
+        }
+    }
+
+    Eigen::SparseMatrix<double> H(n, n);
+    H.setFromTriplets(trips.begin(), trips.end());
+    return H;
+}
+
+// ── Symmetrised block-FD Hessian ─────────────────────────────────────────────
+//
+// The per-face block is symmetric to within FD rounding, but the
+// accumulation may amplify tiny asymmetries.  This helper returns
+// (H + Hᵀ)/2, identical in spirit to `hyper_ideal_hessian_sym`.
+inline Eigen::SparseMatrix<double> hyper_ideal_hessian_block_fd_sym(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const HyperIdealMaps&      m,
+    double                     eps = 1e-5)
+{
+    auto H = hyper_ideal_hessian_block_fd(mesh, x, m, eps);
+    Eigen::SparseMatrix<double> Ht = H.transpose();
+    return (H + Ht) * 0.5;
+}
+
+} // namespace conformallab

code/include/hyper_ideal_utility.hpp

@@ -4,6 +4,7 @@
 // Ported from de.varylab.discreteconformal.functional.HyperIdealUtility (Java).
 
 #include "clausen.hpp"
+#include "constants.hpp"
 
 #include <Eigen/Dense>
 #include <cmath>
@@ -16,10 +17,10 @@ namespace conformallab {
 // Corresponds to Java HyperIdealUtility.calculateTetrahedronVolume().
 inline double calculateTetrahedronVolume(double A, double B, double C,
                                          double D, double E, double F) {
-    constexpr double pi = 3.14159265358979323846264338328;
+    // PI from constants.hpp (conformallab::PI)
 
     // Degenerate if any angle equals pi.
-    if (A == pi || B == pi || C == pi || D == pi || E == pi || F == pi)
+    if (A == PI || B == PI || C == PI || D == PI || E == PI || F == PI)
         return 0.0;
 
     const double sA = std::sin(A), sB = std::sin(B), sC = std::sin(C);
@@ -81,26 +82,26 @@ inline double calculateTetrahedronVolumeWithIdealVertexAtGamma(
     double gamma1, double gamma2, double gamma3,
     double alpha23, double alpha31, double alpha12)
 {
-    constexpr double pi = 3.14159265358979323846264338328;
+    // PI from constants.hpp (conformallab::PI)
     auto L = [](double x) { return Lobachevsky(x); };
 
     double result = L(gamma1) + L(gamma2) + L(gamma3);
 
-    result += L((pi + alpha31 - alpha12 - gamma1) / 2.0);
+    result += L((PI + alpha31 - alpha12 - gamma1) / 2.0);
-    result += L((pi + alpha12 - alpha23 - gamma2) / 2.0);
+    result += L((PI + alpha12 - alpha23 - gamma2) / 2.0);
-    result += L((pi + alpha23 - alpha31 - gamma3) / 2.0);
+    result += L((PI + alpha23 - alpha31 - gamma3) / 2.0);
 
-    result += L((pi - alpha31 + alpha12 - gamma1) / 2.0);
+    result += L((PI - alpha31 + alpha12 - gamma1) / 2.0);
-    result += L((pi - alpha12 + alpha23 - gamma2) / 2.0);
+    result += L((PI - alpha12 + alpha23 - gamma2) / 2.0);
-    result += L((pi - alpha23 + alpha31 - gamma3) / 2.0);
+    result += L((PI - alpha23 + alpha31 - gamma3) / 2.0);
 
-    result += L((pi + alpha31 + alpha12 - gamma1) / 2.0);
+    result += L((PI + alpha31 + alpha12 - gamma1) / 2.0);
-    result += L((pi + alpha12 + alpha23 - gamma2) / 2.0);
+    result += L((PI + alpha12 + alpha23 - gamma2) / 2.0);
-    result += L((pi + alpha23 + alpha31 - gamma3) / 2.0);
+    result += L((PI + alpha23 + alpha31 - gamma3) / 2.0);
 
-    result += L((pi - alpha31 - alpha12 - gamma1) / 2.0);
+    result += L((PI - alpha31 - alpha12 - gamma1) / 2.0);
-    result += L((pi - alpha12 - alpha23 - gamma2) / 2.0);
+    result += L((PI - alpha12 - alpha23 - gamma2) / 2.0);
-    result += L((pi - alpha23 - alpha31 - gamma3) / 2.0);
+    result += L((PI - alpha23 - alpha31 - gamma3) / 2.0);
 
     return result / 2.0;
 }

code/include/hyper_ideal_visualization_utility.hpp

@@ -0,0 +1,129 @@
+#pragma once
+// Port of the static helper
+//   HyperIdealVisualizationPlugin.getEuclideanCircleFromHyperbolic()
+// from de.varylab.discreteconformal.plugin.
+//
+// Converts a hyperbolic circle (center + radius in the hyperboloid model)
+// to its Euclidean representation (cx, cy, r) in the Poincaré disk model.
+//
+// Mathematical background
+// -----------------------
+// Hyperboloid model: points (x,y,z,w) with w²-x²-y²-z²=1, w>0.
+// Metric signature: g = diag(+1,+1,+1,−1)  (spatial-first, time-last).
+//
+// Hyperbolic translation from the origin e₄=(0,0,0,1) to p=(a,b,c,d):
+//   T = [ I₃ + p'·p'ᵀ/(d+1)   p' ]      p' = (a,b,c)
+//       [        p'ᵀ            d  ]
+// This is the standard Lorentz boost; it is in O(3,1) and maps e₄ → p.
+//
+// Poincaré disk projection (jReality convention):
+//   (x,y,z,w)  →  (x,y) / (w+1)
+//
+// The three reference points on the unit hyperbolic circle (at origin) are
+//   p1 = (sinh r, 0, 0, cosh r)
+//   p2 = (0, sinh r, 0, cosh r)
+//   p3 = (-sinh r, 0, 0, cosh r)
+// After translation and projection to the Poincaré disk their circumcircle
+// equals the image of the original hyperbolic circle.
+
+#include <Eigen/Dense>
+#include <array>
+#include <cmath>
+
+namespace conformallab {
+
+// ---------------------------------------------------------------------------
+// Circumcenter of three 2-D points
+// ---------------------------------------------------------------------------
+inline Eigen::Vector2d circumcenter2d(
+    const Eigen::Vector2d& a,
+    const Eigen::Vector2d& b,
+    const Eigen::Vector2d& c)
+{
+    double ax = a.x(), ay = a.y();
+    double bx = b.x(), by = b.y();
+    double cx = c.x(), cy = c.y();
+
+    double D = 2.0 * (ax*(by - cy) + bx*(cy - ay) + cx*(ay - by));
+    double a2 = ax*ax + ay*ay;
+    double b2 = bx*bx + by*by;
+    double c2 = cx*cx + cy*cy;
+
+    double ux = (a2*(by - cy) + b2*(cy - ay) + c2*(ay - by)) / D;
+    double uy = (a2*(cx - bx) + b2*(ax - cx) + c2*(bx - ax)) / D;
+    return {ux, uy};
+}
+
+// ---------------------------------------------------------------------------
+// 4×4 Lorentz boost: maps the hyperboloid origin e₄=(0,0,0,1) to `center`.
+// `center` must lie on the hyperboloid: center[3]² - ‖center.head<3>()‖² = 1.
+// ---------------------------------------------------------------------------
+inline Eigen::Matrix4d hyperboloidTranslation(const Eigen::Vector4d& center)
+{
+    Eigen::Vector3d p = center.head<3>();
+    double          d = center(3);
+
+    Eigen::Matrix4d T = Eigen::Matrix4d::Identity();
+    // Upper-left 3×3 block: I + p'·p'ᵀ / (d+1)
+    T.block<3,3>(0,0) += p * p.transpose() / (d + 1.0);
+    // Right column and bottom row
+    T.block<3,1>(0,3) = p;
+    T.block<1,3>(3,0) = p.transpose();
+    T(3,3) = d;
+    return T;
+}
+
+// ---------------------------------------------------------------------------
+// Project a hyperboloid point to the Poincaré disk (jReality convention:
+// add 1 to the w-coordinate, then dehomogenize the spatial part).
+// ---------------------------------------------------------------------------
+inline Eigen::Vector2d toPoincareDisk(const Eigen::Vector4d& x)
+{
+    double w = x(3) + 1.0;
+    return {x(0) / w, x(1) / w};
+}
+
+// ---------------------------------------------------------------------------
+// getEuclideanCircleFromHyperbolic
+//
+// Inputs
+//   center  – point on the hyperboloid, e.g. (0,0,0,1) for the origin
+//   radius  – hyperbolic radius (real number > 0)
+//
+// Output
+//   { euclidean_cx, euclidean_cy, euclidean_radius }
+//   describing the circle in the Poincaré disk that corresponds to the
+//   given hyperbolic circle.
+//
+// Port of HyperIdealVisualizationPlugin.getEuclideanCircleFromHyperbolic()
+// ---------------------------------------------------------------------------
+inline std::array<double,3> getEuclideanCircleFromHyperbolic(
+    const Eigen::Vector4d& center, double radius)
+{
+    const double s  = std::sinh(radius);
+    const double ch = std::cosh(radius);
+
+    // Three points on the hyperbolic circle centered at the origin
+    Eigen::Vector4d p1( s, 0.0, 0.0, ch);
+    Eigen::Vector4d p2(0.0,  s, 0.0, ch);
+    Eigen::Vector4d p3(-s, 0.0, 0.0, ch);
+
+    // Apply the hyperbolic translation to the target center
+    const Eigen::Matrix4d T = hyperboloidTranslation(center);
+    p1 = T * p1;
+    p2 = T * p2;
+    p3 = T * p3;
+
+    // Project to the Poincaré disk
+    const Eigen::Vector2d q1 = toPoincareDisk(p1);
+    const Eigen::Vector2d q2 = toPoincareDisk(p2);
+    const Eigen::Vector2d q3 = toPoincareDisk(p3);
+
+    // Euclidean circumcircle of the three projected points
+    const Eigen::Vector2d ec = circumcenter2d(q1, q2, q3);
+    const double           r  = (ec - q1).norm();
+
+    return {ec.x(), ec.y(), r};
+}
+
+} // namespace conformallab

code/include/inversive_distance_functional.hpp

@@ -0,0 +1,344 @@
+#pragma once
+// inversive_distance_functional.hpp
+//
+// Phase 9a.2 — Inversive-distance circle-packing functional (Luo 2004).
+//
+// VERTEX-based circle packing.  Each vertex carries a circle of radius
+// r_i = exp(u_i).  The inversive distance I_ij between two adjacent
+// circles is a constant of the edge, derived once from the initial
+// geometry via Bowers-Stephenson 2004.
+//
+// This is the FACE-DUAL of CPEuclideanFunctional (Phase 9a.1).  The
+// correspondence is I_ij = cos θ_e (Glickenstein 2011 §5).
+//
+// ┌──────────────────────────────────────────────────────────────────────────┐
+// │  Mathematical model                                                       │
+// │  ──────────────────                                                       │
+// │                                                                           │
+// │  Variables:    u_i = log r_i           (per vertex; r_i is the radius)   │
+// │  Constants:    I_ij                    (per edge; inversive distance)    │
+// │                Θ_v                     (per vertex; target cone angle)   │
+// │                                                                           │
+// │  Bowers-Stephenson (init from initial geometry):                          │
+// │    I_ij = ( ℓ_ij² − r_i² − r_j² ) / ( 2 r_i r_j )                        │
+// │                                                                           │
+// │  Edge length (Luo 2004 §3, Glickenstein 2011 eq. 2.1):                   │
+// │    ℓ_ij(u)²  =  exp(2 u_i) + exp(2 u_j) + 2 I_ij exp(u_i + u_j)         │
+// │             =  r_i² + r_j² + 2 I_ij r_i r_j                              │
+// │                                                                           │
+// │  Triangle angles:    same half-tangent law of cosines as the              │
+// │                      Euclidean functional (numerically stable).           │
+// │                                                                           │
+// │  Gradient  (Luo 2004 Lemma 3.1):                                          │
+// │    ∂E/∂u_v  =  Θ_v  −  Σ_{T ∋ v}  α_v(T)                                 │
+// │                                                                           │
+// │  Energy:     path integral E(u) = ∫₀¹ ⟨G(tu), u⟩ dt                      │
+// │              (Luo's 1-form is closed; we use 10-point Gauss-Legendre      │
+// │               quadrature, identical to euclidean_functional.hpp)          │
+// │                                                                           │
+// │  Hessian:    finite-difference for the MVP port; an analytic form is      │
+// │              given in Glickenstein 2011 eq. (4.6) and may be added        │
+// │              later for performance.                                       │
+// └──────────────────────────────────────────────────────────────────────────┘
+//
+// Relation to euclidean_functional.hpp
+// ────────────────────────────────────
+// The two are structurally identical in:
+//   • DOF layout (per vertex), DOF index sentinel (−1 = pinned)
+//   • Gradient pattern (Θ − Σ α)
+//   • Energy via path integral (same Gauss-Legendre constants)
+//   • Halfedge convention (h0/h1/h2, source pattern, α opposite-edge)
+//
+// They differ ONLY in:
+//   • Per-edge constant: λ°_ij (log²-length) vs I_ij (inversive distance)
+//   • Edge-length formula:
+//       Euclidean:           ℓ_ij  = exp((λ°_ij + u_i + u_j) / 2)
+//       Inversive distance:  ℓ_ij² = exp(2u_i) + exp(2u_j)
+//                                  + 2 I_ij exp(u_i + u_j)
+//
+// In particular at the tangential limit I_ij = 1 the inversive-distance length
+// reduces to (exp(u_i) + exp(u_j))² ⇒ ℓ_ij = r_i + r_j (tangential circles),
+// which is *different* from the Euclidean-conformal length even at the same
+// initial geometry.  The two functionals describe distinct geometric objects.
+//
+// Property-map name prefix: "iv:" (vertex) and "ie:" (edge).
+
+#include "conformal_mesh.hpp"
+#include "constants.hpp"
+#include "euclidean_geometry.hpp"   // euclidean_angles(λ12, λ23, λ31)
+#include <CGAL/boost/graph/iterator.h>
+#include <vector>
+#include <cmath>
+#include <cstdint>
+#include <iostream>
+
+namespace conformallab {
+
+// ── Property-map type aliases ────────────────────────────────────────────────
+
+using IDVMapI = ConformalMesh::Property_map<Vertex_index, int>;
+using IDVMapD = ConformalMesh::Property_map<Vertex_index, double>;
+using IDEMapD = ConformalMesh::Property_map<Edge_index, double>;
+
+// ── Persistent map bundle ─────────────────────────────────────────────────────
+
+struct InversiveDistanceMaps {
+    IDVMapI v_idx;       ///< DOF index per vertex (−1 = pinned / u_v = 0)
+    IDVMapD theta_v;     ///< target cone angle Θ_v   (default 2π)
+    IDVMapD r0;          ///< initial radius r_i^(0)  (default 1)
+    IDEMapD I_e;         ///< inversive distance I_ij (per edge, constant)
+};
+
+// Create the property maps with sensible defaults.
+inline InversiveDistanceMaps setup_inversive_distance_maps(ConformalMesh& mesh)
+{
+    InversiveDistanceMaps m;
+    m.v_idx   = mesh.add_property_map<Vertex_index, int>   ("iv:idx",   -1     ).first;
+    m.theta_v = mesh.add_property_map<Vertex_index, double>("iv:theta", TWO_PI ).first;
+    m.r0      = mesh.add_property_map<Vertex_index, double>("iv:r0",    1.0    ).first;
+    m.I_e     = mesh.add_property_map<Edge_index,   double>("ie:I",     1.0    ).first;
+    return m;
+}
+
+// Assign sequential DOF indices to all vertices (no gauge pinning here —
+// the caller should set one v_idx to −1 before assigning).
+inline int assign_inversive_distance_vertex_dof_indices(ConformalMesh&         mesh,
+                                                          InversiveDistanceMaps& m)
+{
+    int idx = 0;
+    for (auto v : mesh.vertices()) m.v_idx[v] = idx++;
+    return idx;
+}
+
+// Count free DOFs.
+inline int inversive_distance_dimension(const ConformalMesh&         mesh,
+                                          const InversiveDistanceMaps& m)
+{
+    int dim = 0;
+    for (auto v : mesh.vertices()) if (m.v_idx[v] >= 0) ++dim;
+    return dim;
+}
+
+// ── Initialisation from initial mesh geometry ────────────────────────────────
+//
+// Two-phase init mirroring "compute_lambda0" for euclidean_functional:
+//   1. Choose r_i^(0).  Simplest heuristic: r_i = (1/3)·(min adjacent ℓ).
+//      Other choices (max ℓ, mean ℓ, length-of-shortest-vertex-cycle) are
+//      possible; the user can override `m.r0[v]` between setup and init.
+//   2. Compute I_ij = ( ℓ² − r_i² − r_j² ) / ( 2 r_i r_j ) for each edge.
+//
+// Note: a valid inversive-distance packing requires I_ij > −1 on every edge,
+// and the triangle inequality must hold on every face under the resulting ℓ.
+// The choice r_i = ⅓·min(ℓ_e adj v) keeps I_ij safely positive for most
+// real meshes.
+inline void compute_inversive_distance_init_from_mesh(ConformalMesh&         mesh,
+                                                        InversiveDistanceMaps& m)
+{
+    // Phase 1: r_i = (1/3) · min adjacent edge length.
+    for (auto v : mesh.vertices()) {
+        double min_len = std::numeric_limits<double>::infinity();
+        for (auto h : CGAL::halfedges_around_target(v, mesh)) {
+            auto p1 = mesh.point(mesh.source(h));
+            auto p2 = mesh.point(mesh.target(h));
+            double dx = p1.x() - p2.x();
+            double dy = p1.y() - p2.y();
+            double dz = p1.z() - p2.z();
+            double len = std::sqrt(dx*dx + dy*dy + dz*dz);
+            if (len < min_len) min_len = len;
+        }
+        m.r0[v] = (std::isfinite(min_len) && min_len > 1e-15)
+                ? min_len / 3.0
+                : 1.0;
+    }
+
+    // Phase 2: I_ij from initial geometry.
+    for (auto e : mesh.edges()) {
+        auto h  = mesh.halfedge(e);
+        auto vi = mesh.source(h);
+        auto vj = mesh.target(h);
+        auto p1 = mesh.point(vi);
+        auto p2 = mesh.point(vj);
+        double dx = p1.x() - p2.x();
+        double dy = p1.y() - p2.y();
+        double dz = p1.z() - p2.z();
+        double l2 = dx*dx + dy*dy + dz*dz;
+        double ri = m.r0[vi];
+        double rj = m.r0[vj];
+        m.I_e[e] = (l2 - ri*ri - rj*rj) / (2.0 * ri * rj);
+    }
+}
+
+// ── Internal helpers ──────────────────────────────────────────────────────────
+
+namespace id_detail {
+
+inline double dof_val(int idx, const std::vector<double>& x) noexcept
+{
+    return idx >= 0 ? x[static_cast<std::size_t>(idx)] : 0.0;
+}
+
+inline std::size_t hidx(Halfedge_index h) noexcept
+{
+    return static_cast<std::size_t>(static_cast<std::uint32_t>(h));
+}
+
+// Inversive-distance edge length squared: ℓ² = exp(2u_i) + exp(2u_j) + 2 I r_i r_j
+// where r_i = exp(u_i), so:  ℓ² = r_i² + r_j² + 2 I r_i r_j.
+// Returns -1 if the result is non-positive (degenerate; the caller skips the face).
+inline double edge_length_squared(double u_i, double u_j, double I_ij) noexcept
+{
+    double ri = std::exp(u_i);
+    double rj = std::exp(u_j);
+    double l2 = ri*ri + rj*rj + 2.0 * I_ij * ri * rj;
+    return l2 > 0.0 ? l2 : -1.0;
+}
+
+} // namespace id_detail
+
+// ── Gradient ──────────────────────────────────────────────────────────────────
+//
+// G_v = Θ_v − Σ_{f ∋ v} α_v(f)
+//
+// Halfedge convention (identical to euclidean_functional.hpp):
+//   h_alpha[h] = corner angle OPPOSITE to the edge of halfedge h in its face.
+inline std::vector<double> inversive_distance_gradient(
+    const ConformalMesh&             mesh,
+    const std::vector<double>&       x,
+    const InversiveDistanceMaps&     m)
+{
+    const int n = inversive_distance_dimension(mesh, m);
+    std::vector<double> G(static_cast<std::size_t>(n), 0.0);
+
+    const std::size_t nh = mesh.number_of_halfedges();
+    std::vector<double> h_alpha(nh, 0.0);
+
+    // Pass 1 — per face, compute corner angles via the law of cosines.
+    // We reuse euclidean_angles(λ12, λ23, λ31) which takes 2·log(ℓ) per edge.
+    for (auto f : mesh.faces()) {
+        Halfedge_index h0 = mesh.halfedge(f);
+        Halfedge_index h1 = mesh.next(h0);
+        Halfedge_index h2 = mesh.next(h1);
+
+        Vertex_index v1 = mesh.source(h0);
+        Vertex_index v2 = mesh.source(h1);
+        Vertex_index v3 = mesh.source(h2);
+
+        Edge_index e12 = mesh.edge(h0);
+        Edge_index e23 = mesh.edge(h1);
+        Edge_index e31 = mesh.edge(h2);
+
+        double u1 = id_detail::dof_val(m.v_idx[v1], x);
+        double u2 = id_detail::dof_val(m.v_idx[v2], x);
+        double u3 = id_detail::dof_val(m.v_idx[v3], x);
+
+        double l12sq = id_detail::edge_length_squared(u1, u2, m.I_e[e12]);
+        double l23sq = id_detail::edge_length_squared(u2, u3, m.I_e[e23]);
+        double l31sq = id_detail::edge_length_squared(u3, u1, m.I_e[e31]);
+
+        if (l12sq <= 0 || l23sq <= 0 || l31sq <= 0) continue;
+
+        // euclidean_angles expects 2·log(ℓ) per edge — feed log(ℓ²).
+        auto fa = euclidean_angles(std::log(l12sq), std::log(l23sq), std::log(l31sq));
+        if (!fa.valid) continue;
+
+        h_alpha[id_detail::hidx(h0)] = fa.alpha3;
+        h_alpha[id_detail::hidx(h1)] = fa.alpha1;
+        h_alpha[id_detail::hidx(h2)] = fa.alpha2;
+    }
+
+    // Pass 2 — accumulate vertex gradient.
+    for (auto v : mesh.vertices()) {
+        int iv = m.v_idx[v];
+        if (iv < 0) continue;
+        double sum_alpha = 0.0;
+        for (auto h : CGAL::halfedges_around_target(v, mesh)) {
+            if (mesh.is_border(h)) continue;
+            sum_alpha += h_alpha[id_detail::hidx(mesh.prev(h))];
+        }
+        G[static_cast<std::size_t>(iv)] = m.theta_v[v] - sum_alpha;
+    }
+
+    return G;
+}
+
+// ── Energy via Gauss-Legendre path integral ─────────────────────────────────
+//
+// E(u) = ∫₀¹ ⟨G(tu), u⟩ dt   (10-point GL, identical constants to euclidean_functional)
+inline double inversive_distance_energy(
+    const ConformalMesh&             mesh,
+    const std::vector<double>&       x,
+    const InversiveDistanceMaps&     m)
+{
+    static const double gl_s[10] = {
+        -0.9739065285171717,  -0.8650633666889845,
+        -0.6794095682990244,  -0.4333953941292472,
+        -0.1488743389816312,   0.1488743389816312,
+         0.4333953941292472,   0.6794095682990244,
+         0.8650633666889845,   0.9739065285171717
+    };
+    static const double gl_w[10] = {
+        0.0666713443086881,  0.1494513491505806,
+        0.2190863625159820,  0.2692667193099963,
+        0.2955242247147529,  0.2955242247147529,
+        0.2692667193099963,  0.2190863625159820,
+        0.1494513491505806,  0.0666713443086881
+    };
+
+    const std::size_t n = x.size();
+    double E = 0.0;
+    std::vector<double> tx(n);
+    for (int k = 0; k < 10; ++k) {
+        double t  = (1.0 + gl_s[k]) * 0.5;
+        double wt = gl_w[k] * 0.5;
+        for (std::size_t i = 0; i < n; ++i) tx[i] = t * x[i];
+        auto G = inversive_distance_gradient(mesh, tx, m);
+        double dot = 0.0;
+        for (std::size_t i = 0; i < n; ++i) dot += G[i] * x[i];
+        E += wt * dot;
+    }
+    return E;
+}
+
+// ── Finite-difference gradient check ─────────────────────────────────────────
+inline bool gradient_check_inversive_distance(
+    const ConformalMesh&             mesh,
+    const std::vector<double>&       x,
+    const InversiveDistanceMaps&     m,
+    double eps = 1e-5,
+    double tol = 1e-6)
+{
+    auto G = inversive_distance_gradient(mesh, x, m);
+    const std::size_t n = G.size();
+
+    for (std::size_t i = 0; i < n; ++i) {
+        std::vector<double> xp = x, xm = x;
+        xp[i] += eps;
+        xm[i] -= eps;
+        double Ep = inversive_distance_energy(mesh, xp, m);
+        double Em = inversive_distance_energy(mesh, xm, m);
+        double fd = (Ep - Em) / (2.0 * eps);
+        if (std::abs(G[i] - fd) > tol) {
+            std::cerr << "[inversive-distance] FD gradient mismatch at DOF " << i
+                      << ": analytic=" << G[i]
+                      << "  FD=" << fd
+                      << "  diff=" << (G[i] - fd) << "\n";
+            return false;
+        }
+    }
+    return true;
+}
+
+// ── Newton equilibrium check: gradient vanishes at converged u ──────────────
+inline bool is_inversive_distance_equilibrium(
+    const ConformalMesh&             mesh,
+    const std::vector<double>&       x,
+    const InversiveDistanceMaps&     m,
+    double tol = 1e-8)
+{
+    auto G = inversive_distance_gradient(mesh, x, m);
+    for (double g : G)
+        if (std::abs(g) > tol) return false;
+    return true;
+}
+
+} // namespace conformallab

code/include/layout.hpp

@@ -0,0 +1,881 @@
+#pragma once
+// layout.hpp
+//
+// Phase 5/6/7 — Layout / embedding: DOF vector → vertex coordinates in the
+// target geometry via BFS-trilateration.
+//
+// ┌──────────────────────────────────────────────────────────────────────────┐
+// │  Algorithm (all three geometries)                                        │
+// │                                                                          │
+// │  1. For every edge e compute the updated length l_e from the DOF x.     │
+// │  2. Choose root face = largest 3-D area face (quality heuristic).       │
+// │  3. Priority BFS over the dual graph: faces are processed in order of   │
+// │     shortest distance (BFS depth) from the root face, minimising error  │
+// │     accumulation.  Equal depth: arbitrary tie-break.                    │
+// │  4. For each face: trilaterate the one unplaced vertex from the two     │
+// │     already-placed vertices on the shared edge.                          │
+// │  5. If a CutGraph is supplied, cut edges are treated as boundary;       │
+// │     holonomy is recorded for each cut edge.                              │
+// │  6. After the first connected component, unvisited faces start new      │
+// │     BFS sweeps (multi-component support).                               │
+// │                                                                          │
+// │  For open meshes: globally consistent placement.                        │
+// │  For closed meshes without CutGraph: first (shallowest) visit wins,    │
+// │     has_seam = true.                                                    │
+// │  For closed meshes with CutGraph: each vertex placed exactly once;     │
+// │     holonomy = translation (Euclidean) or Möbius map (hyperbolic).     │
+// └──────────────────────────────────────────────────────────────────────────┘
+//
+// Trilateration
+//   Euclidean   — exact (analytic formula in ℝ²)
+//   Spherical   — exact (spherical law of cosines on S²)
+//   Hyperbolic  — exact (hyperbolic law of cosines + Möbius maps,
+//                 Poincaré disk model)
+//
+// Normalisation
+//   normalise_euclidean(layout)   — centroid → origin, major axis → x-axis (PCA)
+//   normalise_hyperbolic(layout)  — face-area-weighted iterative Möbius centering
+//   normalise_spherical(layout)   — rotate centroid to north pole (Rodrigues)
+//
+// Holonomy (genus-g surfaces with CutGraph)
+//   HolonomyData.translations[i]  — translation ω_i   (Euclidean / spherical)
+//   HolonomyData.mobius_maps[i]   — Möbius map T_i    (hyperbolic)
+//   For a flat torus: ω_1, ω_2 are the lattice generators; τ = ω_2/ω_1 ∈ ℍ.
+//
+// Texture atlas
+//   Layout2D.halfedge_uv[h.idx()] — UV of source(h) as seen from face(h).
+//   At seam edges the two opposite halfedges carry different UV values,
+//   enabling proper per-halfedge UV for GPU texture atlasing.
+//
+// Möbius map
+//   MobiusMap::from_three(z1→w1, z2→w2, z3→w3) — fit a Möbius transformation
+//   to three point correspondences (via 3×3 complex linear system).
+
+#include "conformal_mesh.hpp"
+#include "euclidean_functional.hpp"
+#include "spherical_functional.hpp"
+#include "hyper_ideal_functional.hpp"
+#include "cut_graph.hpp"
+#include <Eigen/Dense>
+#include <vector>
+#include <queue>
+#include <complex>
+#include <cmath>
+#include <algorithm>
+#include <limits>
+#include <fstream>
+#include <string>
+
+namespace conformallab {
+
+// ── Möbius map ────────────────────────────────────────────────────────────────
+//
+// T(z) = (a·z + b) / (c·z + d)
+//
+// For hyperbolic holonomy the map is an orientation-preserving isometry of
+// the Poincaré disk (SU(1,1) element).
+// ─────────────────────────────────────────────────────────────────────────────
+struct MobiusMap {
+    using C = std::complex<double>;
+    C a{1.0, 0.0};
+    C b{0.0, 0.0};
+    C c{0.0, 0.0};
+    C d{1.0, 0.0};
+
+    C apply(C z) const { return (a * z + b) / (c * z + d); }
+
+    Eigen::Vector2d apply(const Eigen::Vector2d& p) const {
+        C w = apply(C(p.x(), p.y()));
+        return Eigen::Vector2d(w.real(), w.imag());
+    }
+
+    static MobiusMap identity() { return {C(1), C(0), C(0), C(1)}; }
+
+    MobiusMap inverse()               const { return {d, -b, -c, a}; }
+    MobiusMap compose(const MobiusMap& T) const {
+        return { a*T.a + b*T.c,  a*T.b + b*T.d,
+                 c*T.a + d*T.c,  c*T.b + d*T.d };
+    }
+
+    bool is_identity(double tol = 1e-9) const {
+        if (std::abs(d) < 1e-14) return false;
+        C a_ = a/d, b_ = b/d, c_ = c/d;
+        return std::abs(a_ - C(1)) < tol && std::abs(b_) < tol && std::abs(c_) < tol;
+    }
+
+    /// Fit T to three point correspondences  T(z_i) = w_i.
+    /// Sets d=1 and solves the 3×3 complex linear system.
+    /// Returns identity when the system is (near-)singular.
+    static MobiusMap from_three(C z1, C w1, C z2, C w2, C z3, C w3) {
+        Eigen::Matrix3cd M;
+        M << z1, C(1), -w1*z1,
+             z2, C(1), -w2*z2,
+             z3, C(1), -w3*z3;
+        Eigen::Vector3cd rhs; rhs << w1, w2, w3;
+        Eigen::ColPivHouseholderQR<Eigen::Matrix3cd> qr(M);
+        if (qr.rank() < 3) return identity();
+        Eigen::Vector3cd sol = qr.solve(rhs);
+        return { sol[0], sol[1], sol[2], C(1.0) };
+    }
+};
+
+// ── Result types ──────────────────────────────────────────────────────────────
+
+struct Layout2D {
+    /// uv[v.idx()] — primary 2-D position (first / shallowest-BFS-depth visit).
+    std::vector<Eigen::Vector2d> uv;
+
+    /// halfedge_uv[h.idx()] — UV of source(h) as seen from face(h).
+    ///
+    /// For interior (non-seam) halfedges: equals uv[source(h).idx()].
+    /// For seam halfedges:  carries the UV from the virtual unfolding across
+    ///   the cut (i.e. the trilaterated position that was NOT used as the
+    ///   primary uv).  This gives each face its own copy of a seam vertex,
+    ///   enabling a proper GPU texture atlas without vertex duplication.
+    ///
+    /// Size = mesh.number_of_halfedges().  Border halfedges = (0,0).
+    std::vector<Eigen::Vector2d> halfedge_uv;
+
+    bool success  = false;
+    bool has_seam = false;   ///< true when a vertex was reached via two paths
+};
+
+struct Layout3D {
+    std::vector<Eigen::Vector3d> pos;
+    bool success  = false;
+    bool has_seam = false;
+};
+
+/// Per-cut-edge holonomy.
+///
+/// Euclidean: translations[i] = ω_i  (translation vector for cut_edge_indices[i]).
+///   For a flat torus ω_1, ω_2 are the lattice generators; τ = ω_2/ω_1 ∈ ℍ.
+///
+/// Hyperbolic: mobius_maps[i] = T_i  (Möbius isometry of the Poincaré disk).
+///   T_i(p_actual) = p_virtual  — maps the placed vertex position to the
+///   trilaterated virtual position obtained by continuing the unfolding across
+///   the cut.
+struct HolonomyData {
+    std::vector<Eigen::Vector2d> translations;   ///< Euclidean / spherical
+    std::vector<MobiusMap>       mobius_maps;     ///< hyperbolic (Phase 7)
+    std::vector<std::size_t>     cut_edge_indices;
+};
+
+// ── Internal helpers ──────────────────────────────────────────────────────────
+
+namespace detail {
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Priority-BFS queue entry
+// Faces closer to the root (lower depth) are processed first, reducing the
+// accumulation of trilateration errors for later-placed vertices.
+// ─────────────────────────────────────────────────────────────────────────────
+struct BFSEntry {
+    int            depth;  ///< BFS depth = max(depth[v_src], depth[v_tgt]) + 1
+    Halfedge_index h;      ///< halfedge pointing INTO the face to be placed
+    /// min-heap: smaller depth = higher priority
+    bool operator>(const BFSEntry& o) const { return depth > o.depth; }
+};
+
+using BFSQueue = std::priority_queue<BFSEntry,
+                                     std::vector<BFSEntry>,
+                                     std::greater<BFSEntry>>;
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Euclidean trilateration — LEFT (CCW) side of p_a → p_b
+// ─────────────────────────────────────────────────────────────────────────────
+inline Eigen::Vector2d trilaterate_2d(
+    const Eigen::Vector2d& pa, const Eigen::Vector2d& pb,
+    double da, double db)
+{
+    Eigen::Vector2d ab = pb - pa;
+    double d = ab.norm();
+    if (d < 1e-14) return pa;
+    Eigen::Vector2d e1 = ab / d;
+    Eigen::Vector2d e2(-e1.y(), e1.x());
+    double t  = (da*da - db*db + d*d) / (2.0 * d);
+    double s2 = da*da - t*t;
+    double s  = (s2 > 0.0) ? std::sqrt(s2) : 0.0;
+    return pa + t*e1 + s*e2;
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Spherical trilateration (S²) — LEFT side of geodesic arc pa → pb
+// ─────────────────────────────────────────────────────────────────────────────
+inline Eigen::Vector3d trilaterate_sph(
+    const Eigen::Vector3d& pa, const Eigen::Vector3d& pb,
+    double da, double db)
+{
+    double c     = pa.dot(pb);
+    double denom = 1.0 - c*c;
+    if (denom < 1e-14) return pa;
+    double cda = std::cos(da), cdb = std::cos(db);
+    double alpha = (cda - c*cdb) / denom;
+    double beta  = (cdb - c*cda) / denom;
+    Eigen::Vector3d cross  = pa.cross(pb);
+    double          cn2    = cross.squaredNorm();
+    double gamma2 = 1.0 - alpha*alpha - beta*beta - 2.0*alpha*beta*c;
+    double gamma  = (gamma2 > 0.0 && cn2 > 1e-28) ? std::sqrt(gamma2 / cn2) : 0.0;
+    Eigen::Vector3d p = alpha*pa + beta*pb + gamma*cross;
+    double n = p.norm();
+    return (n > 1e-14) ? (p / n) : pa;
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Hyperbolic trilateration — exact via Möbius + hyperbolic law of cosines.
+// LEFT (CCW) side of pa → pb in the Poincaré disk.
+// ─────────────────────────────────────────────────────────────────────────────
+inline Eigen::Vector2d trilaterate_hyp(
+    const Eigen::Vector2d& pa, const Eigen::Vector2d& pb,
+    double D, double da, double db)
+{
+    using C = std::complex<double>;
+    if (D < 1e-12 || da < 1e-12) return pa;
+    if (std::sinh(da) * std::sinh(D) < 1e-14) return pa;
+    C a(pa.x(), pa.y()), b(pb.x(), pb.y());
+    auto fwd = [](C z, C c_) { return (z-c_)/(C(1)-std::conj(c_)*z); };
+    auto inv = [](C w, C c_) { return (w+c_)/(C(1)+std::conj(c_)*w); };
+    double theta_b  = std::arg(fwd(b, a));
+    double cos_alpha = (std::cosh(da)*std::cosh(D) - std::cosh(db))
+                     / (std::sinh(da)*std::sinh(D));
+    cos_alpha = std::max(-1.0, std::min(1.0, cos_alpha));
+    C pc_origin = std::tanh(da*0.5) * std::exp(C(0.0, theta_b + std::acos(cos_alpha)));
+    C pc_c = inv(pc_origin, a);
+    return Eigen::Vector2d(pc_c.real(), pc_c.imag());
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// 3-D face area (cross product / 2)
+// ─────────────────────────────────────────────────────────────────────────────
+inline double face_3d_area(const ConformalMesh& mesh, Face_index f)
+{
+    Halfedge_index h = mesh.halfedge(f);
+    auto pA = mesh.point(mesh.source(h));
+    auto pB = mesh.point(mesh.target(h));
+    auto pC = mesh.point(mesh.target(mesh.next(h)));
+    Eigen::Vector3d ab(pB.x()-pA.x(), pB.y()-pA.y(), pB.z()-pA.z());
+    Eigen::Vector3d ac(pC.x()-pA.x(), pC.y()-pA.y(), pC.z()-pA.z());
+    return 0.5 * ab.cross(ac).norm();
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Best root face: largest 3-D area, with 1.5× bonus for interior faces.
+// ─────────────────────────────────────────────────────────────────────────────
+inline Face_index best_root_face(const ConformalMesh& mesh)
+{
+    Face_index best;
+    double best_score = -1.0;
+    for (auto f : mesh.faces()) {
+        double area = face_3d_area(mesh, f);
+        bool interior = true;
+        for (auto h : CGAL::halfedges_around_face(mesh.halfedge(f), mesh))
+            if (mesh.is_border(mesh.opposite(h))) { interior = false; break; }
+        double score = area * (interior ? 1.5 : 1.0);
+        if (score > best_score) { best_score = score; best = f; }
+    }
+    return best;
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Euclidean bounding box of placed vertices (for multi-component offset)
+// ─────────────────────────────────────────────────────────────────────────────
+inline void bbox_2d(const std::vector<Eigen::Vector2d>& uv,
+                    const std::vector<bool>& placed,
+                    double& xmax)
+{
+    xmax = 0.0;
+    for (std::size_t i = 0; i < uv.size(); ++i)
+        if (placed[i]) xmax = std::max(xmax, uv[i].x());
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Möbius centering — unweighted (fallback / Phase 6 compat.)
+// ─────────────────────────────────────────────────────────────────────────────
+inline void center_poincare_disk(std::vector<Eigen::Vector2d>& uv)
+{
+    if (uv.empty()) return;
+    using C = std::complex<double>;
+    C c(0);
+    for (auto& p : uv) c += C(p.x(), p.y());
+    c /= static_cast<double>(uv.size());
+    double r = std::abs(c);
+    if (r > 0.999) c *= 0.999/r;
+    if (r < 1e-12) return;
+    for (auto& p : uv) {
+        C z(p.x(), p.y());
+        C w = (z-c)/(C(1)-std::conj(c)*z);
+        p = {w.real(), w.imag()};
+    }
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Face-area-weighted iterative Möbius centering (Fréchet mean, Phase 7).
+// weights[i] = Voronoi area of vertex i.
+// ─────────────────────────────────────────────────────────────────────────────
+inline void center_poincare_disk_weighted(
+    std::vector<Eigen::Vector2d>& uv, const std::vector<double>& weights)
+{
+    if (uv.empty()) return;
+    using C = std::complex<double>;
+    double total_w = 0.0; for (double w : weights) total_w += w;
+    if (total_w < 1e-14) { center_poincare_disk(uv); return; }
+    C c(0);
+    for (int iter = 0; iter < 30; ++iter) {
+        C mt(0);
+        for (std::size_t i = 0; i < uv.size(); ++i) {
+            C z(uv[i].x(), uv[i].y());
+            mt += weights[i] * (z-c)/(C(1)-std::conj(c)*z);
+        }
+        mt /= total_w;
+        C c_new = (mt+c)/(C(1)+std::conj(c)*mt);
+        if (std::abs(c_new-c) < 1e-12) { c = c_new; break; }
+        c = c_new;
+    }
+    double r = std::abs(c);
+    if (r > 0.999) c *= 0.999/r;
+    if (r < 1e-12) return;
+    for (auto& p : uv) {
+        C z(p.x(), p.y());
+        C w = (z-c)/(C(1)-std::conj(c)*z);
+        p = {w.real(), w.imag()};
+    }
+}
+
+} // namespace detail
+
+// ── Vertex Voronoi area weights ───────────────────────────────────────────────
+inline std::vector<double> compute_vertex_area_weights(const ConformalMesh& mesh)
+{
+    std::vector<double> w(mesh.number_of_vertices(), 0.0);
+    for (auto f : mesh.faces()) {
+        double area = detail::face_3d_area(mesh, f);
+        for (auto h : CGAL::halfedges_around_face(mesh.halfedge(f), mesh))
+            w[static_cast<std::size_t>(mesh.target(h).idx())] += area / 3.0;
+    }
+    return w;
+}
+
+// ── Layout normalisation ──────────────────────────────────────────────────────
+
+inline void normalise_euclidean(Layout2D& layout)
+{
+    if (!layout.success || layout.uv.empty()) return;
+    const std::size_t n = layout.uv.size();
+    Eigen::Vector2d mean = Eigen::Vector2d::Zero();
+    for (auto& p : layout.uv) mean += p;
+    mean /= static_cast<double>(n);
+    for (auto& p : layout.uv) p -= mean;
+    for (auto& p : layout.halfedge_uv) p -= mean;
+    Eigen::Matrix2d cov = Eigen::Matrix2d::Zero();
+    for (auto& p : layout.uv) cov += p * p.transpose();
+    cov /= static_cast<double>(n);
+    Eigen::SelfAdjointEigenSolver<Eigen::Matrix2d> eig(cov);
+    Eigen::Vector2d major = eig.eigenvectors().col(1);
+    double angle = -std::atan2(major.y(), major.x());
+    Eigen::Matrix2d R;
+    R << std::cos(angle), -std::sin(angle),
+         std::sin(angle),  std::cos(angle);
+    for (auto& p : layout.uv)         p = R * p;
+    for (auto& p : layout.halfedge_uv) p = R * p;
+}
+
+/// Hyperbolic — face-area-weighted iterative Möbius centering.
+inline void normalise_hyperbolic(Layout2D& layout, const ConformalMesh& mesh)
+{
+    if (!layout.success || layout.uv.empty()) return;
+    auto weights = compute_vertex_area_weights(mesh);
+    detail::center_poincare_disk_weighted(layout.uv, weights);
+    // halfedge_uv follows the same Möbius map
+    detail::center_poincare_disk(layout.halfedge_uv);
+}
+inline void normalise_hyperbolic(Layout2D& layout) // fallback without mesh
+{
+    if (!layout.success || layout.uv.empty()) return;
+    detail::center_poincare_disk(layout.uv);
+    detail::center_poincare_disk(layout.halfedge_uv);
+}
+
+inline void normalise_spherical(Layout3D& layout)
+{
+    if (!layout.success || layout.pos.empty()) return;
+    Eigen::Vector3d mean = Eigen::Vector3d::Zero();
+    for (auto& p : layout.pos) mean += p;
+    double len = mean.norm(); if (len < 1e-12) return;
+    mean /= len;
+    Eigen::Vector3d north(0, 0, 1);
+    Eigen::Vector3d axis = mean.cross(north);
+    double sin_a = axis.norm(), cos_a = mean.dot(north);
+    if (sin_a < 1e-12) return;
+    axis /= sin_a;
+    Eigen::Matrix3d K;
+    K <<    0, -axis.z(),  axis.y(),
+       axis.z(),       0, -axis.x(),
+      -axis.y(),  axis.x(),       0;
+    Eigen::Matrix3d Rot = Eigen::Matrix3d::Identity() + sin_a*K + (1-cos_a)*K*K;
+    for (auto& p : layout.pos) p = Rot * p;
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// BFS placement helpers shared across all three layout functions.
+//
+// set_face_huv: populate halfedge_uv for the three halfedges of face f,
+//               given that the face was entered via halfedge h_enter.
+//   h_enter: source=v_src, target=v_tgt   → huv = uv[v_src]
+//   next(h_enter): source=v_tgt, target=v_new → huv = uv[v_tgt]
+//   prev(h_enter): source=v_new, target=v_src → huv = p_new  (trilaterated)
+// ─────────────────────────────────────────────────────────────────────────────
+namespace detail {
+
+inline void set_face_huv_2d(
+    std::vector<Eigen::Vector2d>& huv,
+    const ConformalMesh& mesh,
+    Halfedge_index h,
+    const std::vector<Eigen::Vector2d>& uv,
+    const Eigen::Vector2d& p_new)
+{
+    auto s = [](std::size_t x){ return x; };
+    huv[s(static_cast<std::size_t>(h.idx()))]               = uv[static_cast<std::size_t>(mesh.source(h).idx())];
+    huv[s(static_cast<std::size_t>(mesh.next(h).idx()))]    = uv[static_cast<std::size_t>(mesh.target(h).idx())];
+    huv[s(static_cast<std::size_t>(mesh.prev(h).idx()))]    = p_new;
+}
+
+inline void set_root_huv_2d(
+    std::vector<Eigen::Vector2d>& huv,
+    const ConformalMesh& mesh,
+    Face_index f,
+    const std::vector<Eigen::Vector2d>& uv)
+{
+    for (auto hf : CGAL::halfedges_around_face(mesh.halfedge(f), mesh))
+        huv[static_cast<std::size_t>(hf.idx())] = uv[static_cast<std::size_t>(mesh.source(hf).idx())];
+}
+
+} // namespace detail
+
+// ── Euclidean layout ──────────────────────────────────────────────────────────
+
+/// Embed the mesh in ℝ² using the Euclidean metric encoded in x.
+///
+/// Runs priority-BFS trilateration: places vertices in order of BFS depth
+/// from the root face (largest 3D area), so errors accumulate last.
+/// For closed genus-g surfaces a CutGraph must be supplied — otherwise
+/// the layout will have a seam discontinuity (Layout2D::has_seam = true).
+///
+/// \param mesh      Input surface mesh.  Must have lambda0 and v_idx set in maps.
+/// \param x         DOF vector returned by newton_euclidean().
+/// \param maps      EuclideanMaps (lambda0, v_idx, e_idx).
+/// \param cut       Optional cut graph (compute_cut_graph()).  Pass nullptr for
+///                  open meshes or if seams are acceptable.
+/// \param holonomy  If non-null and cut != nullptr, receives the lattice
+///                  translations ω_i ∈ ℂ per cut edge.
+///                  Pass to compute_period_matrix() for the conformal modulus τ.
+/// \param normalise If true, calls normalise_euclidean() on the result:
+///                  centroid → origin, major axis → x-axis (PCA).
+/// \return          Layout2D with .uv[v] (per-vertex UV) and
+///                  .halfedge_uv[h] (per-halfedge UV for texture atlasing).
+///
+/// \note halfedge_uv differs from uv at seam edges: the two sides of a cut
+///       carry different UV coordinates for proper GPU texture atlasing.
+inline Layout2D euclidean_layout(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const EuclideanMaps&       maps,
+    const CutGraph*            cut      = nullptr,
+    HolonomyData*              holonomy  = nullptr,
+    bool                       normalise = false)
+{
+    const std::size_t nv = mesh.number_of_vertices();
+    const std::size_t nf = mesh.number_of_faces();
+    const std::size_t nh = mesh.number_of_halfedges();
+    Layout2D result;
+    result.uv.assign(nv, Eigen::Vector2d::Zero());
+    result.halfedge_uv.assign(nh, Eigen::Vector2d::Zero());
+    if (nf == 0) return result;
+
+    auto get_u  = [&](Vertex_index v) {
+        int iv = maps.v_idx[v]; return (iv>=0) ? x[static_cast<std::size_t>(iv)] : 0.0; };
+    auto get_ue = [&](Edge_index e) {
+        int ie = maps.e_idx[e]; return (ie>=0) ? x[static_cast<std::size_t>(ie)] : 0.0; };
+    auto edge_len = [&](Halfedge_index h) {
+        Edge_index e = mesh.edge(h);
+        double lam = maps.lambda0[e] + get_u(mesh.source(h)) + get_u(mesh.target(h)) + get_ue(e);
+        return std::exp(lam * 0.5); };
+
+    std::vector<bool> vertex_placed(nv, false);
+    std::vector<bool> face_placed(nf, false);
+    std::vector<int>  vertex_depth(nv, std::numeric_limits<int>::max());
+
+    auto place_component = [&](Face_index f_root, Eigen::Vector2d offset) {
+        Halfedge_index h0 = mesh.halfedge(f_root);
+        Halfedge_index h1 = mesh.next(h0), h2 = mesh.next(h1);
+        Vertex_index vA = mesh.source(h0), vB = mesh.source(h1), vC = mesh.source(h2);
+        double lAB = edge_len(h0), lBC = edge_len(h1), lCA = edge_len(h2);
+        result.uv[vA.idx()] = offset;
+        result.uv[vB.idx()] = offset + Eigen::Vector2d(lAB, 0.0);
+        result.uv[vC.idx()] = detail::trilaterate_2d(result.uv[vA.idx()], result.uv[vB.idx()], lCA, lBC);
+        vertex_placed[vA.idx()] = vertex_placed[vB.idx()] = vertex_placed[vC.idx()] = true;
+        vertex_depth[vA.idx()]  = vertex_depth[vB.idx()]  = vertex_depth[vC.idx()]  = 0;
+        face_placed[f_root.idx()] = true;
+        detail::set_root_huv_2d(result.halfedge_uv, mesh, f_root, result.uv);
+
+        detail::BFSQueue pq;
+        auto enqueue = [&](Face_index f) {
+            for (auto hf : CGAL::halfedges_around_face(mesh.halfedge(f), mesh)) {
+                Halfedge_index ho = mesh.opposite(hf);
+                if (mesh.is_border(ho)) continue;
+                if (cut && cut->is_cut(mesh.edge(hf))) continue;
+                Face_index fadj = mesh.face(ho);
+                if (!face_placed[fadj.idx()]) {
+                    int d = std::max(vertex_depth[mesh.source(ho).idx()],
+                                     vertex_depth[mesh.target(ho).idx()]) + 1;
+                    pq.push({d, ho});
+                }
+            }
+        };
+        enqueue(f_root);
+
+        while (!pq.empty()) {
+            auto [depth, h] = pq.top(); pq.pop();
+            Face_index f = mesh.face(h);
+            if (face_placed[f.idx()]) continue;
+
+            Vertex_index v_src = mesh.source(h), v_tgt = mesh.target(h);
+            Vertex_index v_new = mesh.target(mesh.next(h));
+            Eigen::Vector2d p = detail::trilaterate_2d(
+                result.uv[v_src.idx()], result.uv[v_tgt.idx()],
+                edge_len(mesh.prev(h)), edge_len(mesh.next(h)));
+
+            if (!vertex_placed[v_new.idx()]) {
+                result.uv[v_new.idx()]     = p;
+                vertex_placed[v_new.idx()] = true;
+                vertex_depth[v_new.idx()]  = depth;
+            } else {
+                result.has_seam = true;
+            }
+            // halfedge_uv: p is the UV of v_new in THIS face (may differ from uv[v_new] at seam)
+            detail::set_face_huv_2d(result.halfedge_uv, mesh, h, result.uv, p);
+            face_placed[f.idx()] = true;
+            enqueue(f);
+        }
+    };
+
+    place_component(detail::best_root_face(mesh), Eigen::Vector2d::Zero());
+    for (auto f : mesh.faces()) {
+        if (face_placed[f.idx()]) continue;
+        double xmax; detail::bbox_2d(result.uv, vertex_placed, xmax);
+        place_component(f, Eigen::Vector2d(xmax * 1.1 + 1.0, 0.0));
+    }
+    result.success = true;
+
+    // ── Holonomy ──────────────────────────────────────────────────────────────
+    if (cut && holonomy) {
+        holonomy->cut_edge_indices = cut->cut_edge_indices;
+        holonomy->translations.clear();
+        holonomy->translations.reserve(cut->cut_edge_indices.size());
+        holonomy->mobius_maps.clear();
+        for (std::size_t ce_idx : cut->cut_edge_indices) {
+            Edge_index e = *std::next(mesh.edges().begin(), static_cast<std::ptrdiff_t>(ce_idx));
+            Halfedge_index h  = mesh.halfedge(e);
+            Halfedge_index ho = mesh.opposite(h);
+            Halfedge_index hx = mesh.is_border(ho) ? h : ho;
+            if (mesh.is_border(hx)) { holonomy->translations.push_back(Eigen::Vector2d::Zero()); continue; }
+            Vertex_index vs = mesh.source(hx), vt = mesh.target(hx), vn = mesh.target(mesh.next(hx));
+            if (!vertex_placed[vn.idx()]) { holonomy->translations.push_back(Eigen::Vector2d::Zero()); continue; }
+            Eigen::Vector2d p_tri = detail::trilaterate_2d(
+                result.uv[vs.idx()], result.uv[vt.idx()],
+                edge_len(mesh.prev(hx)), edge_len(mesh.next(hx)));
+            // Store seam UV for the cut-crossing halfedges
+            detail::set_face_huv_2d(result.halfedge_uv, mesh, hx, result.uv, p_tri);
+            holonomy->translations.push_back(p_tri - result.uv[vn.idx()]);
+        }
+    }
+
+    if (normalise) normalise_euclidean(result);
+    return result;
+}
+
+// ── Spherical layout ──────────────────────────────────────────────────────────
+
+/// Embed the mesh on the unit sphere S² using the spherical metric encoded in x.
+///
+/// Runs priority-BFS trilateration using the spherical law of cosines.
+/// Typical use: genus-0 (sphere-like) surfaces after newton_spherical().
+///
+/// \param mesh      Input genus-0 surface mesh.
+/// \param x         DOF vector returned by newton_spherical().
+/// \param maps      SphericalMaps.
+/// \param cut       Optional cut graph (rarely needed for genus-0).
+/// \param holonomy  If non-null, receives rotational holonomies (spherical).
+/// \param normalise If true, calls normalise_spherical(): rotates the centroid
+///                  to the north pole (Rodrigues rotation formula).
+/// \return          Layout3D with .xyz[v] ∈ S² ⊂ ℝ³ for each vertex.
+inline Layout3D spherical_layout(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const SphericalMaps&       maps,
+    const CutGraph*            cut      = nullptr,
+    HolonomyData*              holonomy  = nullptr,
+    bool                       normalise = false)
+{
+    const std::size_t nv = mesh.number_of_vertices();
+    const std::size_t nf = mesh.number_of_faces();
+    Layout3D result;
+    result.pos.assign(nv, Eigen::Vector3d::Zero());
+    if (nf == 0) return result;
+
+    auto get_u  = [&](Vertex_index v) {
+        int iv = maps.v_idx[v]; return (iv>=0) ? x[static_cast<std::size_t>(iv)] : 0.0; };
+    auto get_ue = [&](Edge_index e) {
+        int ie = maps.e_idx[e]; return (ie>=0) ? x[static_cast<std::size_t>(ie)] : 0.0; };
+    auto arc_len = [&](Halfedge_index h) {
+        Edge_index e = mesh.edge(h);
+        double lam = maps.lambda0[e] + get_u(mesh.source(h)) + get_u(mesh.target(h)) + get_ue(e);
+        return 2.0 * std::asin(std::min(std::exp(lam*0.5), 1.0)); };
+
+    std::vector<bool> vertex_placed(nv, false);
+    std::vector<bool> face_placed(nf, false);
+    std::vector<int>  vertex_depth(nv, std::numeric_limits<int>::max());
+
+    auto place_component = [&](Face_index f_root) {
+        Halfedge_index h0 = mesh.halfedge(f_root);
+        Halfedge_index h1 = mesh.next(h0), h2 = mesh.next(h1);
+        Vertex_index vA = mesh.source(h0), vB = mesh.source(h1), vC = mesh.source(h2);
+        double lAB = arc_len(h0), lBC = arc_len(h1), lCA = arc_len(h2);
+        result.pos[vA.idx()] = Eigen::Vector3d(0, 0, 1);
+        result.pos[vB.idx()] = Eigen::Vector3d(std::sin(lAB), 0, std::cos(lAB));
+        result.pos[vC.idx()] = detail::trilaterate_sph(result.pos[vA.idx()], result.pos[vB.idx()], lCA, lBC);
+        vertex_placed[vA.idx()] = vertex_placed[vB.idx()] = vertex_placed[vC.idx()] = true;
+        vertex_depth[vA.idx()]  = vertex_depth[vB.idx()]  = vertex_depth[vC.idx()]  = 0;
+        face_placed[f_root.idx()] = true;
+
+        detail::BFSQueue pq;
+        auto enqueue = [&](Face_index f) {
+            for (auto hf : CGAL::halfedges_around_face(mesh.halfedge(f), mesh)) {
+                Halfedge_index ho = mesh.opposite(hf);
+                if (mesh.is_border(ho)) continue;
+                if (cut && cut->is_cut(mesh.edge(hf))) continue;
+                Face_index fadj = mesh.face(ho);
+                if (!face_placed[fadj.idx()]) {
+                    int d = std::max(vertex_depth[mesh.source(ho).idx()],
+                                     vertex_depth[mesh.target(ho).idx()]) + 1;
+                    pq.push({d, ho});
+                }
+            }
+        };
+        enqueue(f_root);
+
+        while (!pq.empty()) {
+            auto [depth, h] = pq.top(); pq.pop();
+            Face_index f = mesh.face(h);
+            if (face_placed[f.idx()]) continue;
+            Vertex_index vs = mesh.source(h), vt = mesh.target(h), vn = mesh.target(mesh.next(h));
+            Eigen::Vector3d p = detail::trilaterate_sph(
+                result.pos[vs.idx()], result.pos[vt.idx()],
+                arc_len(mesh.prev(h)), arc_len(mesh.next(h)));
+            if (!vertex_placed[vn.idx()]) {
+                result.pos[vn.idx()] = p; vertex_placed[vn.idx()] = true; vertex_depth[vn.idx()] = depth;
+            } else result.has_seam = true;
+            face_placed[f.idx()] = true; enqueue(f);
+        }
+    };
+
+    place_component(detail::best_root_face(mesh));
+    for (auto f : mesh.faces()) if (!face_placed[f.idx()]) place_component(f);
+    result.success = true;
+
+    if (cut && holonomy) {
+        holonomy->cut_edge_indices = cut->cut_edge_indices;
+        holonomy->translations.clear();
+        holonomy->translations.reserve(cut->cut_edge_indices.size());
+        holonomy->mobius_maps.clear();
+        for (std::size_t ce_idx : cut->cut_edge_indices) {
+            Edge_index e = *std::next(mesh.edges().begin(), static_cast<std::ptrdiff_t>(ce_idx));
+            Halfedge_index hx = mesh.is_border(mesh.opposite(mesh.halfedge(e)))
+                               ? mesh.halfedge(e) : mesh.opposite(mesh.halfedge(e));
+            if (mesh.is_border(hx)) { holonomy->translations.push_back(Eigen::Vector2d::Zero()); continue; }
+            Vertex_index vs = mesh.source(hx), vt = mesh.target(hx), vn = mesh.target(mesh.next(hx));
+            if (!vertex_placed[vn.idx()]) { holonomy->translations.push_back(Eigen::Vector2d::Zero()); continue; }
+            Eigen::Vector3d p_tri = detail::trilaterate_sph(
+                result.pos[vs.idx()], result.pos[vt.idx()],
+                arc_len(mesh.prev(hx)), arc_len(mesh.next(hx)));
+            Eigen::Vector3d diff = p_tri - result.pos[vn.idx()];
+            // Note: spherical holonomy is geometrically a 3-D rotation, not a 2-D
+            // translation.  The Vector2d here stores only the (x,y) component of the
+            // S²-position difference across the cut, which is an approximation.
+            // For accurate spherical holonomy (rotation axis + angle) use the full
+            // 3-D positions in result.pos[] directly.  Phase 10+ will replace this
+            // with a proper SO(3) representation.
+            holonomy->translations.push_back(Eigen::Vector2d(diff.x(), diff.y()));
+        }
+    }
+
+    if (normalise) normalise_spherical(result);
+    return result;
+}
+
+// ── HyperIdeal layout (Poincaré disk) — exact trilateration ──────────────────
+
+/// Embed the mesh in the Poincaré disk (H²) using the hyperbolic metric encoded in x.
+///
+/// Runs priority-BFS trilateration using exact Möbius-isometric placement:
+/// each new vertex is located by solving the hyperbolic law of cosines and
+/// applying a Möbius map to position it in the disk.
+/// For closed genus-g surfaces (g ≥ 1) a CutGraph is required.
+///
+/// \param mesh      Input genus-g surface mesh (g ≥ 1).
+/// \param x         DOF vector returned by newton_hyper_ideal()
+///                  (vertex b_v and edge a_e variables).
+/// \param maps      HyperIdealMaps (lambda0, v_idx, e_idx).
+/// \param cut       CutGraph from compute_cut_graph().  Required for closed surfaces.
+/// \param holonomy  If non-null, receives the Möbius maps T_i ∈ SU(1,1) per cut edge.
+///                  Pass to compute_period_matrix() for holonomy analysis.
+/// \param normalise If true, calls normalise_hyperbolic(): iterative face-area-weighted
+///                  Möbius centring (Fréchet mean, 30 iterations) → disk origin.
+/// \return          Layout2D with .uv[v] ∈ Poincaré disk (|uv| < 1) for each vertex,
+///                  and .halfedge_uv[h] for seam-aware texture atlasing.
+///
+/// \note All vertex positions satisfy |uv[v]| < 1 (inside the Poincaré disk)
+///       if the metric is hyperbolic.  Points on or outside the boundary indicate
+///       a non-hyperbolic metric (Gauss–Bonnet violation).
+inline Layout2D hyper_ideal_layout(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const HyperIdealMaps&      maps,
+    const CutGraph*            cut      = nullptr,
+    HolonomyData*              holonomy  = nullptr,
+    bool                       normalise = false)
+{
+    const std::size_t nv = mesh.number_of_vertices();
+    const std::size_t nf = mesh.number_of_faces();
+    const std::size_t nh = mesh.number_of_halfedges();
+    Layout2D result;
+    result.uv.assign(nv, Eigen::Vector2d::Zero());
+    result.halfedge_uv.assign(nh, Eigen::Vector2d::Zero());
+    if (nf == 0) return result;
+
+    auto get_b = [&](Vertex_index v) {
+        int iv = maps.v_idx[v]; return (iv>=0) ? x[static_cast<std::size_t>(iv)] : 0.0; };
+    auto get_a = [&](Edge_index e) {
+        int ie = maps.e_idx[e]; return (ie>=0) ? x[static_cast<std::size_t>(ie)] : 0.0; };
+    auto hyp_dist = [&](Halfedge_index h) {
+        double bi = get_b(mesh.source(h)), bj = get_b(mesh.target(h)), a = get_a(mesh.edge(h));
+        return std::acosh(std::max(1.0, std::cosh(bi+a*0.5)*std::cosh(bj+a*0.5)-std::sinh(bi)*std::sinh(bj))); };
+
+    std::vector<bool> vertex_placed(nv, false);
+    std::vector<bool> face_placed(nf, false);
+    std::vector<int>  vertex_depth(nv, std::numeric_limits<int>::max());
+
+    auto place_component = [&](Face_index f_root) {
+        Halfedge_index h0 = mesh.halfedge(f_root);
+        Halfedge_index h1 = mesh.next(h0), h2 = mesh.next(h1);
+        Vertex_index vA = mesh.source(h0), vB = mesh.source(h1), vC = mesh.source(h2);
+        double dAB = hyp_dist(h0), dBC = hyp_dist(h1), dCA = hyp_dist(h2);
+        result.uv[vA.idx()] = Eigen::Vector2d::Zero();
+        result.uv[vB.idx()] = Eigen::Vector2d(std::tanh(dAB*0.5), 0.0);
+        result.uv[vC.idx()] = detail::trilaterate_hyp(result.uv[vA.idx()], result.uv[vB.idx()], dAB, dCA, dBC);
+        vertex_placed[vA.idx()] = vertex_placed[vB.idx()] = vertex_placed[vC.idx()] = true;
+        vertex_depth[vA.idx()]  = vertex_depth[vB.idx()]  = vertex_depth[vC.idx()]  = 0;
+        face_placed[f_root.idx()] = true;
+        detail::set_root_huv_2d(result.halfedge_uv, mesh, f_root, result.uv);
+
+        detail::BFSQueue pq;
+        auto enqueue = [&](Face_index f) {
+            for (auto hf : CGAL::halfedges_around_face(mesh.halfedge(f), mesh)) {
+                Halfedge_index ho = mesh.opposite(hf);
+                if (mesh.is_border(ho)) continue;
+                if (cut && cut->is_cut(mesh.edge(hf))) continue;
+                Face_index fadj = mesh.face(ho);
+                if (!face_placed[fadj.idx()]) {
+                    int d = std::max(vertex_depth[mesh.source(ho).idx()],
+                                     vertex_depth[mesh.target(ho).idx()]) + 1;
+                    pq.push({d, ho});
+                }
+            }
+        };
+        enqueue(f_root);
+
+        while (!pq.empty()) {
+            auto [depth, h] = pq.top(); pq.pop();
+            Face_index f = mesh.face(h);
+            if (face_placed[f.idx()]) continue;
+            Vertex_index vs = mesh.source(h), vt = mesh.target(h), vn = mesh.target(mesh.next(h));
+            double D = hyp_dist(h), da = hyp_dist(mesh.prev(h)), db = hyp_dist(mesh.next(h));
+            Eigen::Vector2d p = detail::trilaterate_hyp(result.uv[vs.idx()], result.uv[vt.idx()], D, da, db);
+            if (!vertex_placed[vn.idx()]) {
+                result.uv[vn.idx()] = p; vertex_placed[vn.idx()] = true; vertex_depth[vn.idx()] = depth;
+            } else result.has_seam = true;
+            detail::set_face_huv_2d(result.halfedge_uv, mesh, h, result.uv, p);
+            face_placed[f.idx()] = true; enqueue(f);
+        }
+    };
+
+    place_component(detail::best_root_face(mesh));
+    for (auto f : mesh.faces()) if (!face_placed[f.idx()]) place_component(f);
+    result.success = true;
+
+    // ── Möbius-map holonomy ───────────────────────────────────────────────────
+    if (cut && holonomy) {
+        holonomy->cut_edge_indices = cut->cut_edge_indices;
+        holonomy->translations.clear();
+        holonomy->mobius_maps.clear();
+        holonomy->mobius_maps.reserve(cut->cut_edge_indices.size());
+        for (std::size_t ce_idx : cut->cut_edge_indices) {
+            Edge_index e = *std::next(mesh.edges().begin(), static_cast<std::ptrdiff_t>(ce_idx));
+            Halfedge_index hcut = mesh.halfedge(e), ho = mesh.opposite(hcut);
+            Halfedge_index hx   = mesh.is_border(ho) ? hcut : ho;
+            if (mesh.is_border(hx)) { holonomy->mobius_maps.push_back(MobiusMap::identity()); continue; }
+            Vertex_index vs = mesh.source(hx), vt = mesh.target(hx), vn = mesh.target(mesh.next(hx));
+            if (!vertex_placed[vn.idx()]) { holonomy->mobius_maps.push_back(MobiusMap::identity()); continue; }
+            double D = hyp_dist(hx), da = hyp_dist(mesh.prev(hx)), db = hyp_dist(mesh.next(hx));
+            Eigen::Vector2d p_tri = detail::trilaterate_hyp(result.uv[vs.idx()], result.uv[vt.idx()], D, da, db);
+            detail::set_face_huv_2d(result.halfedge_uv, mesh, hx, result.uv, p_tri);
+            using C = std::complex<double>;
+            // Möbius deck transformation T across cut edge (vs,vt):
+            // T is the unique Möbius isometry of the Poincaré disk that:
+            //   - fixes vs and vt  (z1=w1, z2=w2: the cut-edge endpoints are
+            //     identified across the seam, so T maps each to itself)
+            //   - maps vn (placed side) → p_tri (virtual side)
+            // This uniquely determines the hyperbolic translation/rotation
+            // along the geodesic through vs and vt.
+            holonomy->mobius_maps.push_back(MobiusMap::from_three(
+                C(result.uv[vs.idx()].x(), result.uv[vs.idx()].y()),
+                C(result.uv[vs.idx()].x(), result.uv[vs.idx()].y()),
+                C(result.uv[vt.idx()].x(), result.uv[vt.idx()].y()),
+                C(result.uv[vt.idx()].x(), result.uv[vt.idx()].y()),
+                C(result.uv[vn.idx()].x(), result.uv[vn.idx()].y()),
+                C(p_tri.x(), p_tri.y())));
+        }
+    }
+
+    if (normalise) normalise_hyperbolic(result, mesh);
+    return result;
+}
+
+// ── Convenience: save layout as OFF ──────────────────────────────────────────
+
+inline void save_layout_off(
+    const std::string& path, ConformalMesh& mesh, const Layout2D& layout)
+{
+    std::ofstream ofs(path);
+    ofs << "OFF\n" << mesh.number_of_vertices() << " " << mesh.number_of_faces() << " 0\n";
+    for (auto v : mesh.vertices()) { auto& p = layout.uv[v.idx()]; ofs << p.x() << " " << p.y() << " 0\n"; }
+    for (auto f : mesh.faces()) {
+        ofs << "3";
+        for (auto h : CGAL::halfedges_around_face(mesh.halfedge(f), mesh)) ofs << " " << mesh.target(h).idx();
+        ofs << "\n";
+    }
+}
+
+inline void save_layout_off(
+    const std::string& path, ConformalMesh& mesh, const Layout3D& layout)
+{
+    std::ofstream ofs(path);
+    ofs << "OFF\n" << mesh.number_of_vertices() << " " << mesh.number_of_faces() << " 0\n";
+    for (auto v : mesh.vertices()) { auto& p = layout.pos[v.idx()]; ofs << p.x() << " " << p.y() << " " << p.z() << "\n"; }
+    for (auto f : mesh.faces()) {
+        ofs << "3";
+        for (auto h : CGAL::halfedges_around_face(mesh.halfedge(f), mesh)) ofs << " " << mesh.target(h).idx();
+        ofs << "\n";
+    }
+}
+
+} // namespace conformallab

code/include/mesh_builder.hpp

@@ -0,0 +1,150 @@
+#pragma once
+// mesh_builder.hpp
+//
+// Factory functions that build simple reference meshes for testing and examples.
+// All functions return a ConformalMesh (CGAL::Surface_mesh<Point3>).
+//
+// Replaces Java mesh generators:
+//   CoHDS generators (convex hull, hyper-ideal generator) come later (Phase 3c/4).
+//   These builders cover the minimal meshes needed for functional unit tests.
+
+#include "conformal_mesh.hpp"
+#include "constants.hpp"
+#include <cmath>
+#include <vector>
+
+namespace conformallab {
+
+// ── Single triangle ──────────────────────────────────────────────────────────
+//
+//   v2
+//   |  \
+//   |    \
+//   v0 ─ v1
+//
+// Returns a mesh with 1 face, 3 vertices, 3 edges.
+// The triangle lies in the xy-plane with a right angle at v0.
+inline ConformalMesh make_triangle(
+    double x0=0, double y0=0,
+    double x1=1, double y1=0,
+    double x2=0, double y2=1)
+{
+    ConformalMesh mesh;
+    auto v0 = mesh.add_vertex(Point3(x0, y0, 0));
+    auto v1 = mesh.add_vertex(Point3(x1, y1, 0));
+    auto v2 = mesh.add_vertex(Point3(x2, y2, 0));
+    mesh.add_face(v0, v1, v2);
+    return mesh;
+}
+
+// ── Regular tetrahedron ──────────────────────────────────────────────────────
+//
+// 4 vertices, 4 faces, 6 edges.
+// Euler characteristic: V - E + F = 4 - 6 + 4 = 2  (sphere topology).
+// Used to test closed-surface traversal.
+inline ConformalMesh make_tetrahedron()
+{
+    ConformalMesh mesh;
+
+    // Vertices of a regular tetrahedron centred at origin, edge length √2·2
+    auto v0 = mesh.add_vertex(Point3( 1,  1,  1));
+    auto v1 = mesh.add_vertex(Point3( 1, -1, -1));
+    auto v2 = mesh.add_vertex(Point3(-1,  1, -1));
+    auto v3 = mesh.add_vertex(Point3(-1, -1,  1));
+
+    // 4 outward-facing triangles (consistent winding)
+    mesh.add_face(v0, v2, v1);  // bottom (z=-1 side)
+    mesh.add_face(v0, v1, v3);  // front  (y=-1 side)
+    mesh.add_face(v0, v3, v2);  // left   (x=-1 side)
+    mesh.add_face(v1, v2, v3);  // back
+
+    return mesh;
+}
+
+// ── Two-triangle strip ───────────────────────────────────────────────────────
+//
+//   v2 ─ v3
+//   |  \ |
+//   v0 ─ v1
+//
+// 4 vertices, 2 faces, 5 edges (1 interior edge v1–v2 shared by both faces).
+// Useful for testing edge-interior vs edge-boundary distinction.
+inline ConformalMesh make_quad_strip()
+{
+    ConformalMesh mesh;
+    auto v0 = mesh.add_vertex(Point3(0, 0, 0));
+    auto v1 = mesh.add_vertex(Point3(1, 0, 0));
+    auto v2 = mesh.add_vertex(Point3(0, 1, 0));
+    auto v3 = mesh.add_vertex(Point3(1, 1, 0));
+
+    mesh.add_face(v0, v1, v2);   // lower-left  triangle
+    mesh.add_face(v1, v3, v2);   // upper-right triangle  (shares edge v1–v2)
+
+    return mesh;
+}
+
+// ── Regular flat polygon fan ─────────────────────────────────────────────────
+//
+// n triangles sharing a central vertex; forms a disk topology (boundary).
+// Used to verify valence-n vertex traversal.
+inline ConformalMesh make_fan(int n)
+{
+    CGAL_precondition(n >= 3);
+    ConformalMesh mesh;
+
+    auto center = mesh.add_vertex(Point3(0, 0, 0));
+
+    const double dtheta = TWO_PI / n;
+    std::vector<Vertex_index> rim(n);
+    for (int i = 0; i < n; ++i) {
+        double a = i * dtheta;
+        rim[i] = mesh.add_vertex(Point3(std::cos(a), std::sin(a), 0));
+    }
+
+    for (int i = 0; i < n; ++i)
+        mesh.add_face(center, rim[i], rim[(i+1) % n]);
+
+    return mesh;
+}
+
+// ── Spherical tetrahedron (vertices on the unit sphere) ───────────────────────
+//
+// The four vertices of a regular tetrahedron projected onto the unit sphere.
+// Starting from (±1,±1,±1), dividing by √3 gives unit-length positions.
+// All edge lengths equal arccos(−1/3) ≈ 1.9106 radians.
+// Used for SphericalFunctional tests (all four faces are valid spherical triangles).
+inline ConformalMesh make_spherical_tetrahedron()
+{
+    ConformalMesh mesh;
+    const double s = 1.0 / std::sqrt(3.0);
+
+    auto v0 = mesh.add_vertex(Point3( s,  s,  s));
+    auto v1 = mesh.add_vertex(Point3( s, -s, -s));
+    auto v2 = mesh.add_vertex(Point3(-s,  s, -s));
+    auto v3 = mesh.add_vertex(Point3(-s, -s,  s));
+
+    mesh.add_face(v0, v2, v1);
+    mesh.add_face(v0, v1, v3);
+    mesh.add_face(v0, v3, v2);
+    mesh.add_face(v1, v2, v3);
+
+    return mesh;
+}
+
+// ── Octahedron face triangle (vertices on the unit sphere) ────────────────────
+//
+// One face of a regular octahedron: the triangle (1,0,0)→(0,1,0)→(0,0,1).
+// All edge lengths equal arccos(0) = π/2.
+// The corner angles are all π/2 (right-angled spherical triangle).
+// base log-length: λ° = 2·log(sin(π/4)) = 2·log(1/√2) = −log(2) ≈ −0.6931.
+inline ConformalMesh make_octahedron_face()
+{
+    ConformalMesh mesh;
+    auto v0 = mesh.add_vertex(Point3(1, 0, 0));
+    auto v1 = mesh.add_vertex(Point3(0, 1, 0));
+    auto v2 = mesh.add_vertex(Point3(0, 0, 1));
+    mesh.add_face(v0, v1, v2);
+    return mesh;
+}
+
+} // namespace conformallab

code/include/mesh_io.hpp

@@ -0,0 +1,65 @@
+#pragma once
+// mesh_io.hpp
+//
+// Phase 4b — CGAL::IO wrappers for ConformalMesh.
+//
+// Reads and writes ConformalMesh (CGAL::Surface_mesh) in standard polygon-mesh
+// formats using the CGAL Polygon Mesh I/O utilities (CGAL 5.4+).
+//
+// Supported formats (detected by file extension):
+//   .off   — Object File Format   (read + write)
+//   .obj   — Wavefront OBJ        (read + write)
+//   .ply   — Polygon File Format  (read + write)
+//
+// Usage:
+//   ConformalMesh mesh;
+//   if (!read_mesh("input.off", mesh))  throw std::runtime_error("read failed");
+//   // ... process mesh ...
+//   write_mesh("output.off", mesh);
+//
+// Note: property maps (lambda0, v_idx, etc.) are NOT serialised — they must
+// be re-initialised with setup_*_maps() + compute_lambda0_from_mesh() after
+// reading a file.
+
+#include "conformal_mesh.hpp"
+#include <CGAL/IO/polygon_mesh_io.h>
+#include <string>
+#include <stdexcept>
+
+namespace conformallab {
+
+// ── Read ──────────────────────────────────────────────────────────────────────
+//
+// Reads a polygon mesh from file into `mesh` (clears any existing content).
+// Returns true on success, false on failure.
+inline bool read_mesh(const std::string& filename, ConformalMesh& mesh)
+{
+    mesh.clear();
+    return CGAL::IO::read_polygon_mesh(filename, mesh);
+}
+
+// ── Write ─────────────────────────────────────────────────────────────────────
+//
+// Writes `mesh` to `filename`.  Returns true on success.
+inline bool write_mesh(const std::string& filename, const ConformalMesh& mesh)
+{
+    return CGAL::IO::write_polygon_mesh(filename, mesh);
+}
+
+// ── Convenience: throwing wrappers ────────────────────────────────────────────
+
+inline ConformalMesh load_mesh(const std::string& filename)
+{
+    ConformalMesh mesh;
+    if (!read_mesh(filename, mesh))
+        throw std::runtime_error("conformallab: failed to read mesh from " + filename);
+    return mesh;
+}
+
+inline void save_mesh(const std::string& filename, const ConformalMesh& mesh)
+{
+    if (!write_mesh(filename, mesh))
+        throw std::runtime_error("conformallab: failed to write mesh to " + filename);
+}
+
+} // namespace conformallab

code/include/mesh_utils.hpp

@@ -36,7 +36,7 @@ void simple_visualize_mesh(Eigen::MatrixXd& V, Eigen::MatrixXi& F) {
   viewer.data().set_mesh(V, F);
   viewer.launch();
 }
-// Zero-Copy Map für V (optional)
+// Zero-copy map for V (optional)
 template <typename Kernel>
 Eigen::Map<Eigen::Matrix<double, Eigen::Dynamic, 3, Eigen::RowMajor>>
 get_vertex_map(CGAL::Surface_mesh<typename Kernel::Point_3>& mesh) {

code/include/newton_solver.hpp

@@ -0,0 +1,563 @@
+#pragma once
+// newton_solver.hpp
+//
+// Phase 4a — Newton solver for all three discrete conformal functionals.
+//
+// Solves G(x) = 0 where G is the gradient of the discrete conformal energy.
+//
+// ┌──────────────────────────────────────────────────────────────────────────┐
+// │  Gradient sign conventions                                               │
+// │    Euclidean / Spherical:  G_v = Θ_v − Σ α_v    (target − actual)      │
+// │    HyperIdeal:             G_v = Σ β_v − Θ_v    (actual − target)      │
+// │                                                                          │
+// │  All solvers use the same Newton step  Δx = −H⁻¹·G                     │
+// │                                                                          │
+// │  Hessian sign at equilibrium                                             │
+// │    Euclidean:   H PSD  → SimplicialLDLT on  H                          │
+// │    Spherical:   H NSD  → SimplicialLDLT on −H  (solve (−H)Δx = G)      │
+// │    HyperIdeal:  H PSD  → SimplicialLDLT on  H  (analytical H: future)  │
+// └──────────────────────────────────────────────────────────────────────────┘
+//
+// SparseQR fallback:
+//   When SimplicialLDLT reports a failure (e.g. singular H on a closed mesh
+//   without a pinned vertex), the solver automatically retries with
+//   Eigen::SparseQR, which finds the minimum-norm Newton step orthogonal to
+//   the null space.  This handles the gauge mode on closed surfaces without
+//   requiring the caller to pin a vertex explicitly.
+//
+// Requires:
+//   Eigen::SimplicialLDLT, Eigen::SparseQR (Eigen sparse module)
+
+#include "euclidean_hessian.hpp"
+#include "spherical_hessian.hpp"
+#include "hyper_ideal_hessian.hpp"
+#include "cp_euclidean_functional.hpp"
+#include "inversive_distance_functional.hpp"
+#include <Eigen/SparseCholesky>
+#include <Eigen/SparseQR>
+#include <Eigen/OrderingMethods>
+#include <Eigen/Dense>
+#include <algorithm>
+#include <cmath>
+
+namespace conformallab {
+
+// ── Result ────────────────────────────────────────────────────────────────────
+
+struct NewtonResult {
+    std::vector<double> x;            ///< DOF vector at termination
+    int                 iterations;   ///< Newton steps taken
+    double              grad_inf_norm;///< max |G_i| at termination
+    bool                converged;    ///< true iff grad_inf_norm < tol
+};
+
+// ── Internal helpers ──────────────────────────────────────────────────────────
+
+namespace detail {
+
+// Solve A·Δx = rhs with SimplicialLDLT; on failure fall back to SparseQR.
+// Returns Δx.  ok is set to false only if both solvers fail.
+// If fallback_used is non-null, it is set to true iff SparseQR was needed.
+inline Eigen::VectorXd solve_with_fallback(
+    const Eigen::SparseMatrix<double>& A,
+    const Eigen::VectorXd&             rhs,
+    bool&                              ok,
+    bool*                              fallback_used = nullptr)
+{
+    if (fallback_used) *fallback_used = false;
+
+    Eigen::SimplicialLDLT<Eigen::SparseMatrix<double>> ldlt(A);
+    if (ldlt.info() == Eigen::Success) {
+        Eigen::VectorXd dx = ldlt.solve(rhs);
+        if (ldlt.info() == Eigen::Success) { ok = true; return dx; }
+    }
+    // Fallback: SparseQR — handles singular/rank-deficient H (gauge modes).
+    if (fallback_used) *fallback_used = true;
+    Eigen::SparseQR<Eigen::SparseMatrix<double>, Eigen::COLAMDOrdering<int>> qr(A);
+    if (qr.info() == Eigen::Success) {
+        Eigen::VectorXd dx = qr.solve(rhs);
+        if (qr.info() == Eigen::Success) { ok = true; return dx; }
+    }
+    ok = false;
+    return Eigen::VectorXd::Zero(rhs.size());
+}
+
+} // namespace detail
+
+// ── Public linear-system solver (SparseQR fallback) ──────────────────────────
+//
+// Solve A·x = rhs with Eigen::SimplicialLDLT; if that fails (singular or
+// rank-deficient A), retry with Eigen::SparseQR which finds the minimum-norm
+// solution orthogonal to the null space.
+//
+// This is the same primitive used internally by all three Newton solvers.
+// Exposing it publicly lets callers (tests, downstream code) reuse the logic
+// and — via the optional fallback_used pointer — verify which code path ran.
+//
+//   fallback_used  – if non-null, set to true iff SparseQR was invoked
+//   Returns Eigen::VectorXd::Zero(rhs.size()) if both solvers fail.
+inline Eigen::VectorXd solve_linear_system(
+    const Eigen::SparseMatrix<double>& A,
+    const Eigen::VectorXd&             rhs,
+    bool*                              fallback_used = nullptr)
+{
+    bool ok = false;
+    return detail::solve_with_fallback(A, rhs, ok, fallback_used);
+}
+
+namespace detail {   // re-open for the remaining helpers
+
+// Backtracking line search: find the largest α in {1, 0.5, 0.25, …} such that
+// ||G(x + α·Δx)||₂ < ||G(x)||₂.  Returns the accepted step (α may stay 1).
+template <typename GradFn>
+inline std::vector<double> line_search(
+    const std::vector<double>& x,
+    const Eigen::VectorXd&     dx,
+    double                     norm0,
+    GradFn&&                   grad_fn,
+    int                        max_halvings = 20)
+{
+    const int n = static_cast<int>(x.size());
+    double alpha = 1.0;
+    std::vector<double> xnew(static_cast<std::size_t>(n));
+
+    for (int ls = 0; ls < max_halvings; ++ls) {
+        for (int i = 0; i < n; ++i)
+            xnew[static_cast<std::size_t>(i)] = x[static_cast<std::size_t>(i)]
+                                                + alpha * dx[i];
+        auto Gnew = grad_fn(xnew);
+        double norm_new = 0.0;
+        for (double v : Gnew) norm_new += v * v;
+        norm_new = std::sqrt(norm_new);
+        if (norm_new < norm0) return xnew;
+        alpha *= 0.5;
+    }
+    // No improvement found — return best attempt (full step)
+    for (int i = 0; i < n; ++i)
+        xnew[static_cast<std::size_t>(i)] = x[static_cast<std::size_t>(i)] + dx[i];
+    return xnew;
+}
+
+} // namespace detail
+
+// ── Euclidean Newton solver ────────────────────────────────────────────────────
+
+/// Solve the Euclidean discrete conformal problem: find u ∈ ℝ^V such that
+/// Σ_{faces adj v} α_v(u) = Θ_v for all vertices v.
+///
+/// Starting from x0, Newton's method minimises E(u) (the Euclidean DCE energy,
+/// which is convex) by iterating  u ← u − H⁻¹·G  with backtracking line search.
+/// The Hessian H is the cotangent Laplacian — PSD with one zero eigenvalue on
+/// closed surfaces (gauge mode).  A SparseQR fallback handles this automatically.
+///
+/// \param mesh      Input triangulated surface (edges must carry lambda0 + theta_v).
+/// \param x0        Initial DOF vector (length = number of free vertices).
+///                  Pass all-zeros for a flat start (typical).
+/// \param m         EuclideanMaps: lambda0[e], theta_v[v], v_idx[v] must be set.
+///                  Call setup_euclidean_maps() + compute_euclidean_lambda0_from_mesh()
+///                  + enforce_gauss_bonnet() before passing here.
+/// \param tol       Convergence threshold on max |G_i|.  Default: 1e-8.
+/// \param max_iter  Maximum Newton iterations.  Default: 200.
+/// \return          NewtonResult{x*, iterations, grad_inf_norm, converged}.
+///
+/// \note On closed meshes without a pinned vertex, SimplicialLDLT detects the
+///       gauge singularity and falls back to SparseQR automatically.
+///
+/// \see doc/math/discrete-conformal-theory.md §3 for the mathematical background.
+inline NewtonResult newton_euclidean(
+    ConformalMesh&             mesh,
+    std::vector<double>        x0,
+    const EuclideanMaps&       m,
+    double                     tol      = 1e-8,
+    int                        max_iter = 200)
+{
+    std::vector<double> x = x0;
+    const int n = static_cast<int>(x.size());
+
+    NewtonResult res;
+    res.converged      = false;
+    res.iterations     = 0;
+    res.grad_inf_norm  = 0.0;
+
+    Eigen::SimplicialLDLT<Eigen::SparseMatrix<double>> solver;
+
+    for (int iter = 0; iter < max_iter; ++iter) {
+        // ── Gradient ──────────────────────────────────────────────────────────
+        auto G_std = euclidean_gradient(mesh, x, m);
+        Eigen::Map<const Eigen::VectorXd> G(G_std.data(), n);
+
+        double inf_norm = G.cwiseAbs().maxCoeff();
+        if (inf_norm < tol) {
+            res.converged     = true;
+            res.grad_inf_norm = inf_norm;
+            res.iterations    = iter;
+            res.x             = x;
+            return res;
+        }
+
+        // ── Hessian + solve H·Δx = −G  (SparseQR fallback for singular H) ──
+        auto H = euclidean_hessian(mesh, x, m);
+        bool ok = false;
+        Eigen::VectorXd dx = detail::solve_with_fallback(H, -G, ok);
+        if (!ok) break;
+
+        // ── Backtracking line search ──────────────────────────────────────────
+        double norm0 = G.norm();
+        x = detail::line_search(x, dx, norm0,
+            [&](const std::vector<double>& xnew) {
+                return euclidean_gradient(mesh, xnew, m);
+            });
+
+        res.iterations = iter + 1;
+    }
+
+    // Report final gradient norm
+    auto G_final = euclidean_gradient(mesh, x, m);
+    double inf_final = 0.0;
+    for (double v : G_final) inf_final = std::max(inf_final, std::abs(v));
+    res.grad_inf_norm = inf_final;
+    res.x             = x;
+    return res;
+}
+
+// ── Spherical Newton solver ───────────────────────────────────────────────────
+
+/// Solve the spherical discrete conformal problem: find u ∈ ℝ^V such that
+/// Σ_{faces adj v} α_v(u) = Θ_v for all vertices v (genus-0 / sphere-like surfaces).
+///
+/// The spherical DCE energy is *concave*, so the Hessian H is NSD at the solution.
+/// The solver factorises −H (which is PSD) and solves (−H)·Δx = G.
+/// A gauge vertex must be pinned (set v_idx = -1) to remove the rotational mode.
+///
+/// \param mesh      Input triangulated surface, genus 0.
+/// \param x0        Initial DOF vector (length = free vertices, excluding gauge_vertex).
+///                  All-zeros is a good start.
+/// \param m         SphericalMaps: lambda0[e], theta_v[v], v_idx[v], gauge_vertex set.
+///                  Call setup_spherical_maps() + compute_spherical_lambda0_from_mesh()
+///                  + enforce_gauss_bonnet() (checks Σ(2π-Θ) > 0) before passing here.
+/// \param tol       Convergence threshold on max |G_i|.  Default: 1e-8.
+/// \param max_iter  Maximum Newton iterations.  Default: 200.
+/// \return          NewtonResult{x*, iterations, grad_inf_norm, converged}.
+///
+/// \note Unlike the Euclidean solver, the spherical solver does NOT need a SparseQR
+///       fallback — the gauge vertex pins the null mode directly.
+///
+/// \see doc/math/geometry-modes.md §Spherical for sign-convention details.
+inline NewtonResult newton_spherical(
+    ConformalMesh&             mesh,
+    std::vector<double>        x0,
+    const SphericalMaps&       m,
+    double                     tol      = 1e-8,
+    int                        max_iter = 200)
+{
+    std::vector<double> x = x0;
+    const int n = static_cast<int>(x.size());
+
+    NewtonResult res;
+    res.converged      = false;
+    res.iterations     = 0;
+    res.grad_inf_norm  = 0.0;
+
+    for (int iter = 0; iter < max_iter; ++iter) {
+        // ── Gradient ──────────────────────────────────────────────────────────
+        auto G_std = spherical_gradient(mesh, x, m);
+        Eigen::Map<const Eigen::VectorXd> G(G_std.data(), n);
+
+        double inf_norm = G.cwiseAbs().maxCoeff();
+        if (inf_norm < tol) {
+            res.converged     = true;
+            res.grad_inf_norm = inf_norm;
+            res.iterations    = iter;
+            res.x             = x;
+            return res;
+        }
+
+        // ── Hessian: negate to get PSD; solve (−H)·Δx = G  ──────────────────
+        auto H    = spherical_hessian(mesh, x, m);
+        auto negH = Eigen::SparseMatrix<double>(-H);
+        bool ok   = false;
+        Eigen::VectorXd dx = detail::solve_with_fallback(negH, G, ok);
+        if (!ok) break;
+
+        // ── Backtracking line search ──────────────────────────────────────────
+        double norm0 = G.norm();
+        x = detail::line_search(x, dx, norm0,
+            [&](const std::vector<double>& xnew) {
+                return spherical_gradient(mesh, xnew, m);
+            });
+
+        res.iterations = iter + 1;
+    }
+
+    auto G_final = spherical_gradient(mesh, x, m);
+    double inf_final = 0.0;
+    for (double v : G_final) inf_final = std::max(inf_final, std::abs(v));
+    res.grad_inf_norm = inf_final;
+    res.x             = x;
+    return res;
+}
+
+// ── HyperIdeal Newton solver ──────────────────────────────────────────────────
+
+/// Solve the hyper-ideal discrete conformal problem: find (b, a) ∈ ℝ^{V+E} such that
+/// Σ β_v(b,a) = Θ_v  and  Σ α_e(b,a) = θ_e  for all vertices v and edges e.
+///
+/// Used for genus-g surfaces (g ≥ 1) under hyperbolic cone metrics.
+/// The energy is *strictly convex* (Springborn 2020, Theorem 1.3), so Newton
+/// converges globally from any starting point.
+///
+/// DOF layout: first V_free entries are vertex variables b_v (hyper-ideal radii),
+/// followed by E entries for edge variables a_e (intersection angles).
+/// Use assign_all_dof_indices(mesh, maps) to set v_idx and e_idx automatically —
+/// no vertex needs to be pinned.
+///
+/// \param mesh      Input triangulated surface, genus g ≥ 1.
+/// \param x0        Initial DOF vector (length = V + E).  All-zeros typical.
+/// \param m         HyperIdealMaps: lambda0[e], theta_v[v], v_idx[v], e_idx[e] set.
+///                  Call setup_hyper_ideal_maps() + compute_hyper_ideal_lambda0_from_mesh().
+/// \param tol       Convergence threshold on max |G_i|.  Default: 1e-8.
+/// \param max_iter  Maximum Newton iterations.  Default: 200.
+/// \param hess_eps  Finite-difference step for Hessian approximation.  Default: 1e-5.
+///                  (Phase 9b will replace this with an analytic Hessian.)
+/// \return          NewtonResult{x*, iterations, grad_inf_norm, converged}.
+///
+/// \see Springborn (2020), Theorem 1.3 for the strict convexity proof.
+/// \see doc/math/geometry-modes.md §Hyper-ideal for DOF layout details.
+inline NewtonResult newton_hyper_ideal(
+    ConformalMesh&             mesh,
+    std::vector<double>        x0,
+    const HyperIdealMaps&      m,
+    double                     tol      = 1e-8,
+    int                        max_iter = 200,
+    double                     hess_eps = 1e-5)
+{
+    std::vector<double> x = x0;
+    const int n = static_cast<int>(x.size());
+
+    NewtonResult res;
+    res.converged      = false;
+    res.iterations     = 0;
+    res.grad_inf_norm  = 0.0;
+
+    for (int iter = 0; iter < max_iter; ++iter) {
+        // ── Gradient ──────────────────────────────────────────────────────────
+        auto G_std = evaluate_hyper_ideal(mesh, x, m, /*energy=*/false).gradient;
+        Eigen::Map<const Eigen::VectorXd> G(G_std.data(), n);
+
+        double inf_norm = G.cwiseAbs().maxCoeff();
+        if (inf_norm < tol) {
+            res.converged     = true;
+            res.grad_inf_norm = inf_norm;
+            res.iterations    = iter;
+            res.x             = x;
+            return res;
+        }
+
+        // ── Hessian (numerical FD) + solve H·Δx = −G ─────────────────────────
+        auto H  = hyper_ideal_hessian_sym(mesh, x, m, hess_eps);
+        bool ok = false;
+        Eigen::VectorXd dx = detail::solve_with_fallback(H, -G, ok);
+        if (!ok) break;
+
+        // ── Backtracking line search ──────────────────────────────────────────
+        double norm0 = G.norm();
+        x = detail::line_search(x, dx, norm0,
+            [&](const std::vector<double>& xnew) {
+                return evaluate_hyper_ideal(mesh, xnew, m, false).gradient;
+            });
+
+        res.iterations = iter + 1;
+    }
+
+    auto G_final = evaluate_hyper_ideal(mesh, x, m, false).gradient;
+    double inf_final = 0.0;
+    for (double v : G_final) inf_final = std::max(inf_final, std::abs(v));
+    res.grad_inf_norm = inf_final;
+    res.x             = x;
+    return res;
+}
+
+// ── CP-Euclidean Newton solver (Phase 9a.1) ───────────────────────────────────
+
+/// Solve the CP-Euclidean circle-packing problem: find ρ ∈ ℝ^F such that the
+/// per-face angle sums match φ_f at every free face.
+///
+/// The CP-Euclidean energy (Bobenko-Pinkall-Springborn 2010 §6) is strictly
+/// convex on its open domain of validity, so the Hessian H is PSD and the
+/// solution is unique up to the gauge mode pinned by `f_idx == −1`.
+/// `cp_euclidean_hessian` provides the analytic 2×2-per-edge formula
+/// `h_jk = sin θ / (cosh Δρ − cos θ)`; no FD machinery is required.
+///
+/// \param mesh      Input triangle mesh (closed or with boundary).
+/// \param x0        Initial DOF vector (length = number of free faces).
+///                  All-zeros is a valid start.
+/// \param m         CPEuclideanMaps: f_idx must have one pinned face
+///                  (`f_idx[f0] == −1`); theta_e and phi_f set by the caller.
+/// \param tol       Convergence threshold on `‖G‖∞`.  Default: 1e-8.
+/// \param max_iter  Newton iteration limit.  Default: 200.
+/// \return          NewtonResult{x*, iterations, grad_inf_norm, converged}.
+///
+/// \note  Unlike the Euclidean solver, the CP-Euclidean Hessian is exact
+///        (analytic), so the SparseQR fallback only triggers in genuine
+///        gauge-singular situations (no pinned face).
+/// \see   doc/architecture/phase-9a-validation.md §1 for the BPS-2010 mapping.
+inline NewtonResult newton_cp_euclidean(
+    ConformalMesh&             mesh,
+    std::vector<double>        x0,
+    const CPEuclideanMaps&     m,
+    double                     tol      = 1e-8,
+    int                        max_iter = 200)
+{
+    std::vector<double> x = x0;
+    const int n = static_cast<int>(x.size());
+
+    NewtonResult res;
+    res.converged      = false;
+    res.iterations     = 0;
+    res.grad_inf_norm  = 0.0;
+
+    for (int iter = 0; iter < max_iter; ++iter) {
+        auto G_std = cp_euclidean_gradient(mesh, x, m);
+        Eigen::Map<const Eigen::VectorXd> G(G_std.data(), n);
+
+        double inf_norm = G.cwiseAbs().maxCoeff();
+        if (inf_norm < tol) {
+            res.converged     = true;
+            res.grad_inf_norm = inf_norm;
+            res.iterations    = iter;
+            res.x             = x;
+            return res;
+        }
+
+        auto H = cp_euclidean_hessian(mesh, x, m);
+        bool ok = false;
+        Eigen::VectorXd dx = detail::solve_with_fallback(H, -G, ok);
+        if (!ok) break;
+
+        double norm0 = G.norm();
+        x = detail::line_search(x, dx, norm0,
+            [&](const std::vector<double>& xnew) {
+                return cp_euclidean_gradient(mesh, xnew, m);
+            });
+
+        res.iterations = iter + 1;
+    }
+
+    auto G_final = cp_euclidean_gradient(mesh, x, m);
+    double inf_final = 0.0;
+    for (double v : G_final) inf_final = std::max(inf_final, std::abs(v));
+    res.grad_inf_norm = inf_final;
+    res.x             = x;
+    return res;
+}
+
+// ── Inversive-Distance Newton solver (Phase 9a.2) ─────────────────────────────
+
+/// Solve the inversive-distance circle-packing problem: find u ∈ ℝ^V such that
+/// Σ_{faces adj v} α_v(u) = Θ_v at every free vertex (Luo 2004 Lemma 3.1).
+///
+/// The inversive-distance energy is (locally) strictly convex on the open
+/// domain where every triangle satisfies the inequalities.  Luo's 1-form is
+/// closed there, so the path-integral energy is well-defined.
+///
+/// MVP implementation: the Hessian is computed by **finite differences** of
+/// the analytic gradient (same pattern as the Phase 4a HyperIdeal solver).
+/// An analytic Hessian via Glickenstein 2011 eq. (4.6) is tracked in
+/// `doc/roadmap/research-track.md` as Phase 9a.2-analytic.
+///
+/// \param mesh      Input triangle mesh.
+/// \param x0        Initial DOF vector (length = number of free vertices).
+/// \param m         InversiveDistanceMaps: v_idx has at least one pinned
+///                  vertex; I_e and r0 set by compute_inversive_distance_init.
+/// \param tol       Convergence threshold on `‖G‖∞`.  Default: 1e-8.
+/// \param max_iter  Newton iteration limit.  Default: 200.
+/// \param hess_eps  FD step size for the Hessian.  Default: 1e-5.
+/// \return          NewtonResult{x*, iterations, grad_inf_norm, converged}.
+///
+/// \note  Convergence is sensitive to the initial point: u = 0 is the
+///        natural choice when `compute_inversive_distance_init_from_mesh`
+///        has been called, since the Bowers-Stephenson identity reconstructs
+///        the input edge lengths at u = 0.
+inline NewtonResult newton_inversive_distance(
+    ConformalMesh&                mesh,
+    std::vector<double>           x0,
+    const InversiveDistanceMaps&  m,
+    double                        tol      = 1e-8,
+    int                           max_iter = 200,
+    double                        hess_eps = 1e-5)
+{
+    std::vector<double> x = x0;
+    const int n = static_cast<int>(x.size());
+
+    NewtonResult res;
+    res.converged      = false;
+    res.iterations     = 0;
+    res.grad_inf_norm  = 0.0;
+
+    // Local FD Hessian builder — n × (cost of gradient eval).
+    auto build_hessian = [&](const std::vector<double>& xc) -> Eigen::SparseMatrix<double> {
+        std::vector<Eigen::Triplet<double>> trips;
+        trips.reserve(static_cast<std::size_t>(n) * 16);   // sparse heuristic
+
+        std::vector<double> xp = xc, xm = xc;
+        for (int j = 0; j < n; ++j) {
+            const std::size_t sj = static_cast<std::size_t>(j);
+            xp[sj] = xc[sj] + hess_eps;
+            xm[sj] = xc[sj] - hess_eps;
+
+            auto Gp = inversive_distance_gradient(mesh, xp, m);
+            auto Gm = inversive_distance_gradient(mesh, xm, m);
+
+            xp[sj] = xm[sj] = xc[sj];   // restore
+
+            for (int i = 0; i < n; ++i) {
+                double val = (Gp[static_cast<std::size_t>(i)]
+                            - Gm[static_cast<std::size_t>(i)])
+                             / (2.0 * hess_eps);
+                if (std::abs(val) > 1e-15)
+                    trips.emplace_back(i, j, val);
+            }
+        }
+        Eigen::SparseMatrix<double> H(n, n);
+        H.setFromTriplets(trips.begin(), trips.end());
+        // Symmetrise — FD rounding may introduce tiny asymmetries.
+        Eigen::SparseMatrix<double> Ht = H.transpose();
+        return (H + Ht) * 0.5;
+    };
+
+    for (int iter = 0; iter < max_iter; ++iter) {
+        auto G_std = inversive_distance_gradient(mesh, x, m);
+        Eigen::Map<const Eigen::VectorXd> G(G_std.data(), n);
+
+        double inf_norm = G.cwiseAbs().maxCoeff();
+        if (inf_norm < tol) {
+            res.converged     = true;
+            res.grad_inf_norm = inf_norm;
+            res.iterations    = iter;
+            res.x             = x;
+            return res;
+        }
+
+        auto H = build_hessian(x);
+        bool ok = false;
+        Eigen::VectorXd dx = detail::solve_with_fallback(H, -G, ok);
+        if (!ok) break;
+
+        double norm0 = G.norm();
+        x = detail::line_search(x, dx, norm0,
+            [&](const std::vector<double>& xnew) {
+                return inversive_distance_gradient(mesh, xnew, m);
+            });
+
+        res.iterations = iter + 1;
+    }
+
+    auto G_final = inversive_distance_gradient(mesh, x, m);
+    double inf_final = 0.0;
+    for (double v : G_final) inf_final = std::max(inf_final, std::abs(v));
+    res.grad_inf_norm = inf_final;
+    res.x             = x;
+    return res;
+}
+
+} // namespace conformallab

code/include/p2_utility.hpp

@@ -0,0 +1,102 @@
+#pragma once
+
+// 2-D projective geometry utilities for the Euclidean signature.
+// Ported from de.jreality.math.P2 and de.varylab.discreteconformal.math.P2Big.
+//
+// Points and lines are represented as homogeneous 3-vectors (x, y, w).
+// In the Euclidean case a finite point (px, py) is stored as (px, py, 1).
+
+#include <Eigen/Dense>
+#include <cmath>
+
+namespace conformallab {
+
+// ── Point / line duality ──────────────────────────────────────────────────────
+
+// Intersection of two lines l1, l2  (or  line through two points p1, p2)
+// via the cross product.  Works for any P2 element.
+// Corresponds to Java P2.pointFromLines / P2.lineFromPoints.
+inline Eigen::Vector3d pointFromLines(const Eigen::Vector3d& l1,
+                                      const Eigen::Vector3d& l2) {
+    return l1.cross(l2);
+}
+
+// ── Euclidean perpendicular bisector ─────────────────────────────────────────
+
+// Returns the homogeneous line coordinates (a, b, c) of the perpendicular
+// bisector of the segment [p, q] in the Euclidean plane.
+// Coordinates: ax + by + c = 0  (after dehomogenizing p and q).
+//
+// Corresponds to Java P2.perpendicularBisector(p, q, Pn.EUCLIDEAN).
+inline Eigen::Vector3d perpendicularBisectorEuclidean(const Eigen::Vector3d& p_h,
+                                                      const Eigen::Vector3d& q_h) {
+    // Dehomogenize
+    Eigen::Vector2d p = p_h.head<2>() / p_h(2);
+    Eigen::Vector2d q = q_h.head<2>() / q_h(2);
+
+    // Direction vector (p → direction, matching jReality sign convention)
+    Eigen::Vector2d d = p - q;
+
+    // Midpoint
+    Eigen::Vector2d m = (p + q) * 0.5;
+
+    // Line: d[0]*(x - m[0]) + d[1]*(y - m[1]) = 0
+    //      = d[0]*x + d[1]*y - (d[0]*m[0] + d[1]*m[1])
+    double c = -(d(0) * m(0) + d(1) * m(1));
+    return {d(0), d(1), c};
+}
+
+// ── Euclidean distance between two P2 homogeneous points ─────────────────────
+
+inline double euclideanDistanceP2(const Eigen::Vector3d& p_h,
+                                  const Eigen::Vector3d& q_h) {
+    Eigen::Vector2d p = p_h.head<2>() / p_h(2);
+    Eigen::Vector2d q = q_h.head<2>() / q_h(2);
+    return (p - q).norm();
+}
+
+// ── Direct Euclidean isometry from two point-frames ──────────────────────────
+
+// Build the 3×3 projective matrix that represents the coordinate frame
+// anchored at p0 with p1 defining the positive x-direction.
+// Euclidean case: columns are [dehom(p0), unit_dir(p0→p1), perp_dir].
+//
+// Template parameter S allows float / double / long double.
+template <typename S>
+Eigen::Matrix<S, 3, 3> makeFrameMatrix(Eigen::Matrix<S, 3, 1> p0_h,
+                                        Eigen::Matrix<S, 3, 1> p1_h) {
+    // Dehomogenize
+    Eigen::Matrix<S, 3, 1> p0 = p0_h / p0_h(2);          // (px, py, 1)
+    Eigen::Matrix<S, 3, 1> p1_d = p1_h / p1_h(2);
+
+    // Unit direction p0 → p1
+    Eigen::Matrix<S, 2, 1> dir2 = (p1_d - p0).template head<2>();
+    dir2.normalize();
+    Eigen::Matrix<S, 3, 1> p1n(dir2(0), dir2(1), S(0));
+
+    // Perpendicular direction
+    Eigen::Matrix<S, 3, 1> p2(-dir2(1), dir2(0), S(0));
+
+    Eigen::Matrix<S, 3, 3> M;
+    M.col(0) = p0;
+    M.col(1) = p1n;
+    M.col(2) = p2;
+    return M;
+}
+
+// Find the 3×3 Euclidean isometry (as a projective matrix) that maps
+// the frame (s1, s2) to the frame (t1, t2).
+//
+// Corresponds to Java P2.makeDirectIsometryFromFrames(s1, s2, t1, t2, Pn.EUCLIDEAN)
+// and P2Big.makeDirectIsometryFromFrames(...) (the BigDecimal / high-precision variant).
+template <typename S>
+Eigen::Matrix<S, 3, 3> makeDirectIsometryFromFramesEuclidean(
+    Eigen::Matrix<S, 3, 1> s1, Eigen::Matrix<S, 3, 1> s2,
+    Eigen::Matrix<S, 3, 1> t1, Eigen::Matrix<S, 3, 1> t2)
+{
+    auto toS = makeFrameMatrix<S>(s1, s2);
+    auto toT = makeFrameMatrix<S>(t1, t2);
+    return toT * toS.inverse();
+}
+
+} // namespace conformallab

code/include/period_matrix.hpp

@@ -0,0 +1,152 @@
+#pragma once
+// period_matrix.hpp
+//
+// Phase 7 — Period matrix for closed surfaces with Euclidean (flat) metric.
+//
+// For a closed genus-g surface with Euclidean conformal structure the holonomy
+// group is generated by 2g translations ω_1, ..., ω_{2g} ∈ ℂ ≅ ℝ².
+//
+// ─── Genus-1 (flat torus) ────────────────────────────────────────────────────
+//
+//   The lattice Λ = ℤ·ω_1 ⊕ ℤ·ω_2 determines the conformal type.
+//
+//   Period ratio:  τ = ω_2 / ω_1   (as complex numbers)
+//
+//   By convention choose ω_1 such that Im(τ) > 0.
+//   The conformal modulus / Teichmüller parameter is the SL(2,ℤ)-orbit of τ.
+//
+//   Reduction to fundamental domain {|τ| ≥ 1, −½ ≤ Re(τ) < ½, Im(τ) > 0}:
+//     S:  τ ↦ −1/τ        (inversion)
+//     T:  τ ↦  τ + 1      (translation)
+//   Apply S and T repeatedly until τ is in the fundamental domain.
+//
+// ─── Genus g > 1 ─────────────────────────────────────────────────────────────
+//
+//   The full period matrix is a g×g complex symmetric matrix Ω with positive
+//   definite imaginary part (Siegel upper half-space H_g).
+//   Computing Ω from holonomy data requires integration of holomorphic
+//   differentials — not implemented here.  For g > 1, this function returns
+//   only the 2×2 block for the first pair of generators.
+//
+// ─── API ─────────────────────────────────────────────────────────────────────
+//
+//   PeriodData pd = compute_period_matrix(holonomy);
+//   pd.tau          — complex period ratio τ (genus 1)
+//   pd.omega        — holonomy generators as complex numbers (size = 2g)
+//   pd.in_fundamental_domain — whether τ has been reduced
+//
+//   std::complex<double> reduce_to_fundamental_domain(τ)  — apply SL(2,ℤ)
+
+#include "layout.hpp"
+#include <complex>
+#include <cmath>
+#include <vector>
+#include <stdexcept>
+#include <sstream>
+
+namespace conformallab {
+
+// ─────────────────────────────────────────────────────────────────────────────
+// PeriodData
+// ─────────────────────────────────────────────────────────────────────────────
+
+struct PeriodData {
+    /// Lattice generators as complex numbers (one per cut edge).
+    /// omega[i] = translations[i].x() + i·translations[i].y()
+    std::vector<std::complex<double>> omega;
+
+    /// Period ratio τ = omega[1] / omega[0]  (genus-1 only).
+    /// Undefined (NaN) for genus != 1 or if holonomy has fewer than 2 generators.
+    std::complex<double> tau = std::complex<double>(
+        std::numeric_limits<double>::quiet_NaN(), 0.0);
+
+    /// True if τ has been reduced to the standard fundamental domain.
+    bool in_fundamental_domain = false;
+
+    int genus() const { return static_cast<int>(omega.size()) / 2; }
+};
+
+// ─────────────────────────────────────────────────────────────────────────────
+// reduce_to_fundamental_domain
+//
+// Applies SL(2,ℤ) generators S: τ↦−1/τ and T: τ↦τ+1 to bring τ into
+//   F = { τ ∈ ℍ : |τ| ≥ 1,  −½ ≤ Re(τ) < ½ }
+//
+// Returns the reduced τ.  Throws if Im(τ) ≤ 0 (not in upper half-plane).
+// ─────────────────────────────────────────────────────────────────────────────
+inline std::complex<double> reduce_to_fundamental_domain(std::complex<double> tau)
+{
+    if (tau.imag() <= 0.0) {
+        std::ostringstream msg;
+        msg << "period_matrix: τ = " << tau.real() << " + " << tau.imag()
+            << "i is not in the upper half-plane (Im(τ) must be > 0).";
+        throw std::domain_error(msg.str());
+    }
+
+    // Iterate at most 200 times (convergence is rapid for well-conditioned τ)
+    for (int k = 0; k < 200; ++k) {
+        // T step: shift Re(τ) into [−½, ½)
+        double re = tau.real();
+        long   n  = static_cast<long>(std::floor(re + 0.5));
+        tau -= std::complex<double>(static_cast<double>(n), 0.0);
+
+        // S step: if |τ| < 1, apply τ ← −1/τ
+        if (std::abs(tau) < 1.0 - 1e-12) {
+            tau = -1.0 / tau;
+        } else {
+            break;
+        }
+    }
+    return tau;
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// is_in_fundamental_domain — check membership in F with tolerance tol.
+// ─────────────────────────────────────────────────────────────────────────────
+inline bool is_in_fundamental_domain(std::complex<double> tau, double tol = 1e-9)
+{
+    if (tau.imag() <= 0.0) return false;
+    if (std::abs(tau.real()) > 0.5 + tol) return false;
+    if (std::abs(tau) < 1.0 - tol) return false;
+    return true;
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// compute_period_matrix
+//
+// Computes the period data from the Euclidean holonomy translations.
+// For genus-1 surfaces, also reduces τ to the fundamental domain.
+// ─────────────────────────────────────────────────────────────────────────────
+inline PeriodData compute_period_matrix(const HolonomyData& hol, bool reduce = true)
+{
+    PeriodData pd;
+    pd.omega.reserve(hol.translations.size());
+    for (auto& t : hol.translations)
+        pd.omega.push_back(std::complex<double>(t.x(), t.y()));
+
+    if (pd.omega.size() < 2) return pd;   // need at least 2 generators
+
+    // τ = ω_2 / ω_1  — choose ω_1 such that Im(τ) > 0
+    std::complex<double> w1 = pd.omega[0];
+    std::complex<double> w2 = pd.omega[1];
+    if (std::abs(w1) < 1e-14) return pd;
+
+    std::complex<double> tau = w2 / w1;
+    if (tau.imag() < 0.0) {
+        tau = std::conj(tau);              // swap orientation
+        w1  = std::conj(w1);
+        w2  = std::conj(w2);
+        pd.omega[0] = w1;
+        pd.omega[1] = w2;
+    }
+    if (tau.imag() < 0.0) return pd;      // degenerate
+
+    if (reduce) {
+        tau = reduce_to_fundamental_domain(tau);
+        pd.in_fundamental_domain = true;
+    }
+    pd.tau = tau;
+    return pd;
+}
+
+} // namespace conformallab

code/include/serialization.hpp

@@ -0,0 +1,289 @@
+#pragma once
+// serialization.hpp
+//
+// Phase 5 — Save and load conformal map results in JSON and XML formats.
+//
+// JSON (nlohmann/json, bundled in deps/single_includes/json.hpp):
+//   save_result_json / load_result_json
+//
+// XML (hand-written, no external parser dependency):
+//   save_result_xml / load_result_xml
+//
+// Both formats store:
+//   - geometry type string ("euclidean" / "spherical" / "hyper_ideal")
+//   - mesh statistics (vertex/face count)
+//   - solver metadata (converged, iterations, grad_inf_norm)
+//   - DOF vector x
+//   - optional 2D or 3D layout positions
+//
+// XML schema (ConformalResult):
+//
+//   <?xml version="1.0" encoding="UTF-8"?>
+//   <ConformalResult geometry="euclidean" vertices="4" faces="2">
+//     <Solver converged="true" iterations="3" grad_inf_norm="1.43e-13"/>
+//     <DOFVector n="3">0.0 1.23e-13 2.87e-13</DOFVector>
+//     <Layout dim="2" n="4">
+//       0.0 0.0  1.0 0.0  0.5 0.866  1.5 0.866
+//     </Layout>
+//   </ConformalResult>
+
+#include "newton_solver.hpp"
+#include "layout.hpp"
+#include <json.hpp>
+#include <fstream>
+#include <sstream>
+#include <string>
+#include <vector>
+#include <stdexcept>
+#include <iomanip>
+
+namespace conformallab {
+
+// ════════════════════════════════════════════════════════════════════════════
+// JSON
+// ════════════════════════════════════════════════════════════════════════════
+
+// Save solver result (+ optional 2D layout) to a JSON file.
+inline void save_result_json(
+    const std::string& path,
+    const NewtonResult& res,
+    const std::string&  geometry,
+    int                 n_vertices,
+    int                 n_faces,
+    const Layout2D*     layout2d = nullptr,
+    const Layout3D*     layout3d = nullptr)
+{
+    using json = nlohmann::json;
+    json j;
+    j["geometry"]  = geometry;
+    j["mesh"]      = { {"vertices", n_vertices}, {"faces", n_faces} };
+    j["solver"]    = {
+        {"converged",      res.converged},
+        {"iterations",     res.iterations},
+        {"grad_inf_norm",  res.grad_inf_norm}
+    };
+    j["dof_vector"] = res.x;
+
+    if (layout2d && layout2d->success) {
+        json uv = json::array();
+        for (auto& p : layout2d->uv) uv.push_back({p.x(), p.y()});
+        j["layout"] = { {"dim", 2}, {"uv", uv}, {"has_seam", layout2d->has_seam} };
+    }
+    if (layout3d && layout3d->success) {
+        json pos = json::array();
+        for (auto& p : layout3d->pos) pos.push_back({p.x(), p.y(), p.z()});
+        j["layout"] = { {"dim", 3}, {"pos", pos}, {"has_seam", layout3d->has_seam} };
+    }
+
+    std::ofstream ofs(path);
+    if (!ofs) throw std::runtime_error("Cannot write: " + path);
+    ofs << std::setw(2) << j << "\n";
+}
+
+// Load DOF vector from a JSON result file.
+// Returns the DOF vector; fills out res fields if non-null.
+inline std::vector<double> load_result_json(
+    const std::string& path,
+    NewtonResult*       res     = nullptr,
+    std::string*        geom    = nullptr,
+    Layout2D*           layout2d = nullptr)
+{
+    using json = nlohmann::json;
+    std::ifstream ifs(path);
+    if (!ifs) throw std::runtime_error("Cannot open: " + path);
+    json j; ifs >> j;
+
+    if (geom && j.contains("geometry"))
+        *geom = j["geometry"].get<std::string>();
+
+    std::vector<double> x = j.at("dof_vector").get<std::vector<double>>();
+
+    if (res) {
+        res->x             = x;
+        res->converged     = j["solver"]["converged"].get<bool>();
+        res->iterations    = j["solver"]["iterations"].get<int>();
+        res->grad_inf_norm = j["solver"]["grad_inf_norm"].get<double>();
+    }
+
+    if (layout2d && j.contains("layout") && j["layout"]["dim"] == 2) {
+        auto uv = j["layout"]["uv"];
+        layout2d->uv.resize(uv.size());
+        for (std::size_t i = 0; i < uv.size(); ++i)
+            layout2d->uv[i] = { uv[i][0].get<double>(), uv[i][1].get<double>() };
+        layout2d->has_seam = j["layout"].value("has_seam", false);
+        layout2d->success  = true;
+    }
+
+    return x;
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// XML
+// ════════════════════════════════════════════════════════════════════════════
+
+namespace detail_xml {
+
+// Minimal XML attribute escaping
+inline std::string xml_attr(double v)
+{
+    std::ostringstream s;
+    s << std::setprecision(15) << v;
+    return s.str();
+}
+
+// Write a flat vector of doubles as space-separated values
+inline std::string flat_doubles(const std::vector<double>& v)
+{
+    std::ostringstream s;
+    s << std::setprecision(15);
+    for (std::size_t i = 0; i < v.size(); ++i) {
+        if (i) s << ' ';
+        s << v[i];
+    }
+    return s.str();
+}
+
+// Simple attribute parser: find value of key= in a tag string
+inline std::string xml_get_attr(const std::string& tag, const std::string& key)
+{
+    auto pos = tag.find(key + "=\"");
+    if (pos == std::string::npos) return {};
+    pos += key.size() + 2;
+    auto end = tag.find('"', pos);
+    return tag.substr(pos, end - pos);
+}
+
+// Read all text content between the current position and </tag>
+inline std::string xml_read_text(std::istream& is, const std::string& close_tag)
+{
+    std::string buf, line;
+    std::string ctag = "</" + close_tag + ">";
+    while (std::getline(is, line)) {
+        auto pos = line.find(ctag);
+        if (pos != std::string::npos) {
+            buf += line.substr(0, pos);
+            return buf;
+        }
+        buf += line + " ";
+    }
+    return buf;
+}
+
+// Parse space-separated doubles
+inline std::vector<double> parse_doubles(const std::string& s)
+{
+    std::vector<double> v;
+    std::istringstream ss(s);
+    double d;
+    while (ss >> d) v.push_back(d);
+    return v;
+}
+
+} // namespace detail_xml
+
+// Save solver result (+ optional layout) to an XML file.
+inline void save_result_xml(
+    const std::string& path,
+    const NewtonResult& res,
+    const std::string&  geometry,
+    int                 n_vertices,
+    int                 n_faces,
+    const Layout2D*     layout2d = nullptr,
+    const Layout3D*     layout3d = nullptr)
+{
+    std::ofstream ofs(path);
+    if (!ofs) throw std::runtime_error("Cannot write: " + path);
+    ofs << std::setprecision(15);
+
+    ofs << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";
+    ofs << "<ConformalResult"
+        << " geometry=\"" << geometry << "\""
+        << " vertices=\""  << n_vertices << "\""
+        << " faces=\""     << n_faces << "\">\n";
+
+    ofs << "  <Solver"
+        << " converged=\""     << (res.converged ? "true" : "false") << "\""
+        << " iterations=\""    << res.iterations << "\""
+        << " grad_inf_norm=\"" << detail_xml::xml_attr(res.grad_inf_norm) << "\""
+        << "/>\n";
+
+    ofs << "  <DOFVector n=\"" << res.x.size() << "\">"
+        << detail_xml::flat_doubles(res.x)
+        << "</DOFVector>\n";
+
+    if (layout2d && layout2d->success) {
+        ofs << "  <Layout dim=\"2\" n=\"" << layout2d->uv.size()
+            << "\" has_seam=\"" << (layout2d->has_seam ? "true" : "false") << "\">\n";
+        for (auto& p : layout2d->uv)
+            ofs << "    " << p.x() << " " << p.y() << "\n";
+        ofs << "  </Layout>\n";
+    }
+    if (layout3d && layout3d->success) {
+        ofs << "  <Layout dim=\"3\" n=\"" << layout3d->pos.size()
+            << "\" has_seam=\"" << (layout3d->has_seam ? "true" : "false") << "\">\n";
+        for (auto& p : layout3d->pos)
+            ofs << "    " << p.x() << " " << p.y() << " " << p.z() << "\n";
+        ofs << "  </Layout>\n";
+    }
+
+    ofs << "</ConformalResult>\n";
+}
+
+// Load solver result from an XML file written by save_result_xml.
+// Returns the DOF vector; fills res/geom/layout2d if non-null.
+inline std::vector<double> load_result_xml(
+    const std::string& path,
+    NewtonResult*       res      = nullptr,
+    std::string*        geom     = nullptr,
+    Layout2D*           layout2d = nullptr)
+{
+    std::ifstream ifs(path);
+    if (!ifs) throw std::runtime_error("Cannot open: " + path);
+
+    std::vector<double> x;
+    std::string line;
+
+    while (std::getline(ifs, line)) {
+        // Root element
+        if (line.find("<ConformalResult") != std::string::npos) {
+            if (geom) *geom = detail_xml::xml_get_attr(line, "geometry");
+        }
+        // Solver metadata
+        else if (line.find("<Solver") != std::string::npos) {
+            if (res) {
+                res->converged     = (detail_xml::xml_get_attr(line, "converged") == "true");
+                res->iterations    = std::stoi(detail_xml::xml_get_attr(line, "iterations"));
+                res->grad_inf_norm = std::stod(detail_xml::xml_get_attr(line, "grad_inf_norm"));
+            }
+        }
+        // DOF vector
+        else if (line.find("<DOFVector") != std::string::npos) {
+            // Text may be on same line: <DOFVector n="...">0 1 2...</DOFVector>
+            auto open_end = line.find('>');
+            auto close    = line.find("</DOFVector>");
+            std::string text;
+            if (close != std::string::npos) {
+                text = line.substr(open_end + 1, close - open_end - 1);
+            } else {
+                text = line.substr(open_end + 1);
+                text += detail_xml::xml_read_text(ifs, "DOFVector");
+            }
+            x = detail_xml::parse_doubles(text);
+            if (res) res->x = x;
+        }
+        // Layout
+        else if (layout2d && line.find("<Layout") != std::string::npos
+                          && detail_xml::xml_get_attr(line, "dim") == "2") {
+            layout2d->has_seam = (detail_xml::xml_get_attr(line, "has_seam") == "true");
+            std::string text   = detail_xml::xml_read_text(ifs, "Layout");
+            auto vals          = detail_xml::parse_doubles(text);
+            layout2d->uv.clear();
+            for (std::size_t i = 0; i + 1 < vals.size(); i += 2)
+                layout2d->uv.push_back({vals[i], vals[i+1]});
+            layout2d->success = true;
+        }
+    }
+    return x;
+}
+
+} // namespace conformallab

code/include/spherical_functional.hpp

@@ -0,0 +1,475 @@
+#pragma once
+// spherical_functional.hpp
+//
+// Energy and gradient of the spherical discrete conformal functional
+// evaluated on a ConformalMesh (CGAL::Surface_mesh).
+//
+// Ported from de.varylab.discreteconformal.functional.SphericalFunctional.
+//
+// ┌──────────────────────────────────────────────────────────────────────────┐
+// │  DOFs                                                                    │
+// │    x[v_idx[v]]  = u_v   – conformal factor at vertex v                  │
+// │    x[e_idx[e]]  = λ_e   – edge log-length variable (optional)           │
+// │    -1 means "pinned" (u_v = 0 / λ_e = λ°_e fixed)                      │
+// │                                                                          │
+// │  Effective log-length:  Λ_ij = λ°_ij + u_i + u_j                       │
+// │  Spherical arc length:  l_ij = 2·asin(min(exp(Λ_ij/2), 1))             │
+// │                                                                          │
+// │  Gradient:                                                               │
+// │    ∂E/∂u_v  = Θ_v  –  Σ_{faces adj. v} α_v(face)                       │
+// │    ∂E/∂λ_e  = α_opp(face⁺) + α_opp(face⁻) – π                         │
+// │                                                                          │
+// │  Energy:                                                                 │
+// │    Computed as the Schläfli path integral                                │
+// │      E(x) = ∫₀¹ ⟨G(tx), x⟩ dt                                          │
+// │    using 10-point Gauss-Legendre quadrature. This is mathematically      │
+// │    exact for any conservative (curl-free) gradient G and is numerically  │
+// │    accurate to ~10⁻¹⁰ for smooth angle functions. The gradient check     │
+// │    therefore tests curl-freeness of G, which is the key integrability    │
+// │    condition for the spherical discrete conformal functional.             │
+// └──────────────────────────────────────────────────────────────────────────┘
+
+#include "conformal_mesh.hpp"
+#include "spherical_geometry.hpp"
+#include <CGAL/boost/graph/iterator.h>
+#include <vector>
+#include <cmath>
+#include <cstdint>
+
+namespace conformallab {
+
+// ── Property-map type aliases ─────────────────────────────────────────────────
+
+using SpherVMapD = ConformalMesh::Property_map<Vertex_index, double>;
+using SpherVMapI = ConformalMesh::Property_map<Vertex_index, int>;
+using SpherEMapD = ConformalMesh::Property_map<Edge_index,   double>;
+using SpherEMapI = ConformalMesh::Property_map<Edge_index,   int>;
+
+// ── Persistent map bundle ─────────────────────────────────────────────────────
+
+struct SphericalMaps {
+    SpherVMapI v_idx;    // DOF index per vertex   (-1 = pinned / u_v = 0)
+    SpherEMapI e_idx;    // DOF index per edge      (-1 = no edge DOF)
+    SpherVMapD theta_v;  // target cone angle Θ_v  (default 2π)
+    SpherEMapD theta_e;  // target edge angle θ_e  (default π)
+    SpherEMapD lambda0;  // base log-length λ°_e   (default 0.0)
+};
+
+// Defaults: theta_v = 2π, theta_e = π, lambda0 = 0.
+// lambda0 = 0 means exp(λ°/2)=1, i.e., l=π — degenerate unless u_i<0.
+// For real meshes, set lambda0 from mesh geometry via
+// compute_lambda0_from_mesh() below.
+inline SphericalMaps setup_spherical_maps(ConformalMesh& mesh)
+{
+    SphericalMaps m;
+    m.v_idx  = mesh.add_property_map<Vertex_index, int>   ("sv:idx",    -1     ).first;
+    m.e_idx  = mesh.add_property_map<Edge_index,   int>   ("se:idx",    -1     ).first;
+    m.theta_v= mesh.add_property_map<Vertex_index, double>("sv:theta",  2.0*PI_SPHER).first;
+    m.theta_e= mesh.add_property_map<Edge_index,   double>("se:theta",  PI_SPHER    ).first;
+    m.lambda0= mesh.add_property_map<Edge_index,   double>("se:lam0",   0.0    ).first;
+    return m;
+}
+
+// Assign DOF indices 0..n-1 for all vertices (only vertex DOFs).
+inline int assign_vertex_dof_indices(ConformalMesh& mesh, SphericalMaps& m)
+{
+    int idx = 0;
+    for (auto v : mesh.vertices()) m.v_idx[v] = idx++;
+    return idx;
+}
+
+// Assign DOF indices for all vertices AND edges.
+inline int assign_all_spherical_dof_indices(ConformalMesh& mesh, SphericalMaps& m)
+{
+    int idx = 0;
+    for (auto v : mesh.vertices()) m.v_idx[v] = idx++;
+    for (auto e : mesh.edges())    m.e_idx[e] = idx++;
+    return idx;
+}
+
+// Count variable DOFs.
+inline int spherical_dimension(const ConformalMesh& mesh, const SphericalMaps& m)
+{
+    int dim = 0;
+    for (auto v : mesh.vertices()) if (m.v_idx[v] >= 0) ++dim;
+    for (auto e : mesh.edges())    if (m.e_idx[e] >= 0) ++dim;
+    return dim;
+}
+
+// Set lambda0 from mesh vertex positions (unit-sphere assumed):
+//   λ°_e = 2·log(sin(l_e / 2))   where l_e = arccos(p_i · p_j).
+// Requires vertices to lie on the unit sphere.
+inline void compute_lambda0_from_mesh(ConformalMesh& mesh, SphericalMaps& m)
+{
+    for (auto e : mesh.edges()) {
+        auto h  = mesh.halfedge(e);
+        auto p1 = mesh.point(mesh.source(h));
+        auto p2 = mesh.point(mesh.target(h));
+        // Dot product (works for unit-sphere vertices).
+        double dot = p1.x()*p2.x() + p1.y()*p2.y() + p1.z()*p2.z();
+        dot = std::max(-1.0, std::min(1.0, dot));
+        double l_e = std::acos(dot);              // spherical arc length
+        double half_sin = std::sin(l_e * 0.5);   // = exp(λ°/2)
+        if (half_sin > 1e-15)
+            m.lambda0[e] = 2.0 * std::log(half_sin);
+        else
+            m.lambda0[e] = -30.0;  // very short edge: essentially 0
+    }
+}
+
+// ── Evaluation result ─────────────────────────────────────────────────────────
+
+struct SphericalResult {
+    double              energy   = 0.0;
+    std::vector<double> gradient;
+};
+
+// ── Internal helpers ──────────────────────────────────────────────────────────
+
+static inline double spher_dof_val(int idx, const std::vector<double>& x)
+{
+    return idx >= 0 ? x[static_cast<std::size_t>(idx)] : 0.0;
+}
+
+static inline std::size_t spher_hidx(Halfedge_index h)
+{
+    return static_cast<std::size_t>(static_cast<std::uint32_t>(h));
+}
+
+// ── Gradient only (no energy) ─────────────────────────────────────────────────
+
+// Compute gradient G(x).
+// G_v = Θ_v − Σ_faces α_v(face)
+// G_e = α_opp(face+) + α_opp(face−) − θ_e
+//
+// The corner angle α_v is stored on halfedges using the convention:
+//   h_alpha[h] = corner angle at source(prev(h)) = corner angle at the vertex
+//                ACROSS FROM the edge of halfedge h in its face.
+// This convention makes both the vertex and edge gradient accumulators natural.
+inline std::vector<double> spherical_gradient(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const SphericalMaps&       m)
+{
+    const int n = spherical_dimension(mesh, m);
+    std::vector<double> G(static_cast<std::size_t>(n), 0.0);
+
+    // Temporary per-halfedge corner-angle storage.
+    // h_alpha[h] = corner angle at the vertex opposite to the edge of h.
+    const std::size_t nh = mesh.number_of_halfedges();
+    std::vector<double> h_alpha(nh, 0.0);
+
+    // ── Pass 1: compute corner angles per face ────────────────────────────────
+    for (auto f : mesh.faces()) {
+        Halfedge_index h0 = mesh.halfedge(f);
+        Halfedge_index h1 = mesh.next(h0);
+        Halfedge_index h2 = mesh.next(h1);
+
+        Vertex_index v1 = mesh.source(h0);
+        Vertex_index v2 = mesh.source(h1);
+        Vertex_index v3 = mesh.source(h2);
+
+        Edge_index e12 = mesh.edge(h0);
+        Edge_index e23 = mesh.edge(h1);
+        Edge_index e31 = mesh.edge(h2);
+
+        // Effective log-length Λ_ij = λ°_ij + u_i + u_j
+        double u1 = spher_dof_val(m.v_idx[v1], x);
+        double u2 = spher_dof_val(m.v_idx[v2], x);
+        double u3 = spher_dof_val(m.v_idx[v3], x);
+
+        double lam12 = m.lambda0[e12] + u1 + u2 + spher_dof_val(m.e_idx[e12], x);
+        double lam23 = m.lambda0[e23] + u2 + u3 + spher_dof_val(m.e_idx[e23], x);
+        double lam31 = m.lambda0[e31] + u3 + u1 + spher_dof_val(m.e_idx[e31], x);
+
+        double l12 = spherical_l(lam12);
+        double l23 = spherical_l(lam23);
+        double l31 = spherical_l(lam31);
+
+        SphericalFaceAngles fa = spherical_angles(l12, l23, l31);
+
+        if (!fa.valid) continue;   // degenerate face: contributes 0
+
+        // Store convention: h_alpha[h] = corner angle at source(prev(h))
+        //   h0 (e12): opposite vertex is v3 → source(prev(h0)) = source(h2) = v3 → α3
+        //   h1 (e23): opposite vertex is v1 → source(prev(h1)) = source(h0) = v1 → α1
+        //   h2 (e31): opposite vertex is v2 → source(prev(h2)) = source(h1) = v2 → α2
+        h_alpha[spher_hidx(h0)] = fa.alpha3;
+        h_alpha[spher_hidx(h1)] = fa.alpha1;
+        h_alpha[spher_hidx(h2)] = fa.alpha2;
+    }
+
+    // ── Pass 2: accumulate gradient ───────────────────────────────────────────
+
+    // Vertex: G_v = Θ_v − Σ h_alpha[prev(h)] for each incoming non-border h to v.
+    for (auto v : mesh.vertices()) {
+        int iv = m.v_idx[v];
+        if (iv < 0) continue;
+        double sum_alpha = 0.0;
+        for (auto h : CGAL::halfedges_around_target(v, mesh)) {
+            if (mesh.is_border(h)) continue;
+            sum_alpha += h_alpha[spher_hidx(mesh.prev(h))];
+        }
+        G[static_cast<std::size_t>(iv)] = m.theta_v[v] - sum_alpha;
+    }
+
+    // Edge: G_e for λ_e additive (Λ_ij = λ°_ij + u_i + u_j + λ_e).
+    //
+    // From the Schläfli identity applied to the spherical face,
+    // the contribution of edge DOF λ_e from face f is:
+    //   a_f = (2·α_opp − S_f) / 2    where S_f = Σ angles in face f.
+    //
+    // Summing over both adjacent faces:
+    //   G_e = a_f+ + a_f−
+    //       = α_opp⁺ + α_opp⁻ − (S_f⁺ + S_f⁻) / 2 − θ_e
+    //
+    // For flat (Euclidean) triangles S_f = π, recovering the familiar
+    // α_opp⁺ + α_opp⁻ − π formula.  For spherical triangles S_f > π.
+    for (auto e : mesh.edges()) {
+        int ie = m.e_idx[e];
+        if (ie < 0) continue;
+        auto h  = mesh.halfedge(e);
+        auto ho = mesh.opposite(h);
+        double sum = 0.0;
+        if (!mesh.is_border(h)) {
+            double alpha_opp = h_alpha[spher_hidx(h)];
+            double S_f = alpha_opp
+                       + h_alpha[spher_hidx(mesh.next(h))]
+                       + h_alpha[spher_hidx(mesh.prev(h))];
+            sum += (2.0 * alpha_opp - S_f) * 0.5;
+        }
+        if (!mesh.is_border(ho)) {
+            double alpha_opp = h_alpha[spher_hidx(ho)];
+            double S_f = alpha_opp
+                       + h_alpha[spher_hidx(mesh.next(ho))]
+                       + h_alpha[spher_hidx(mesh.prev(ho))];
+            sum += (2.0 * alpha_opp - S_f) * 0.5;
+        }
+        G[static_cast<std::size_t>(ie)] = sum - m.theta_e[e];
+    }
+
+    return G;
+}
+
+// ── Energy via Gauss-Legendre path integral ───────────────────────────────────
+//
+// E(x) = ∫₀¹ ⟨G(tx), x⟩ dt
+//
+// This is the correct potential for any conservative G = ∇E.
+// Uses 10-point Gauss-Legendre quadrature; error ≈ O(h²⁰) for smooth G.
+//
+// 10-point GL nodes and weights on [0, 1] (transformed from [-1, 1]):
+//   t_k = (1 + s_k) / 2,  w_k = w_GL_k / 2
+inline double spherical_energy(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const SphericalMaps&       m)
+{
+    // 10-point Gauss-Legendre nodes and weights on [-1, 1].
+    static const double gl_s[10] = {
+        -0.9739065285171717,  -0.8650633666889845,
+        -0.6794095682990244,  -0.4333953941292472,
+        -0.1488743389816312,   0.1488743389816312,
+         0.4333953941292472,   0.6794095682990244,
+         0.8650633666889845,   0.9739065285171717
+    };
+    static const double gl_w[10] = {
+        0.0666713443086881,  0.1494513491505806,
+        0.2190863625159820,  0.2692667193099963,
+        0.2955242247147529,  0.2955242247147529,
+        0.2692667193099963,  0.2190863625159820,
+        0.1494513491505806,  0.0666713443086881
+    };
+
+    const std::size_t n = x.size();
+    double E = 0.0;
+
+    for (int k = 0; k < 10; ++k) {
+        double t   = (1.0 + gl_s[k]) * 0.5;   // node on [0, 1]
+        double wt  = gl_w[k] * 0.5;            // weight on [0, 1]
+
+        // Evaluate G(t·x)
+        std::vector<double> tx(n);
+        for (std::size_t i = 0; i < n; ++i) tx[i] = t * x[i];
+
+        auto G = spherical_gradient(mesh, tx, m);
+
+        // Accumulate ⟨G(tx), x⟩ · wt
+        double dot = 0.0;
+        for (std::size_t i = 0; i < n; ++i) dot += G[i] * x[i];
+        E += wt * dot;
+    }
+    return E;
+}
+
+// ── Full evaluation (energy + gradient) ──────────────────────────────────────
+
+inline SphericalResult evaluate_spherical(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const SphericalMaps&       m,
+    bool                       need_energy   = true,
+    bool                       need_gradient = true)
+{
+    SphericalResult res;
+    if (need_gradient)
+        res.gradient = spherical_gradient(mesh, x, m);
+    if (need_energy)
+        res.energy = spherical_energy(mesh, x, m);
+    return res;
+}
+
+// ── Finite-difference gradient check ─────────────────────────────────────────
+//
+// Tests |G[i] − fd[i]| / max(1, |G[i]|) < tol for all DOFs.
+// Same defaults as the hyper-ideal gradient check (Java FunctionalTest).
+inline bool gradient_check_spherical(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x0,
+    const SphericalMaps&       m,
+    double                     eps = 1E-5,
+    double                     tol = 1E-4)
+{
+    auto G = spherical_gradient(mesh, x0, m);
+    const int n = static_cast<int>(G.size());
+
+    std::vector<double> xp = x0, xm = x0;
+    bool ok = true;
+
+    for (int i = 0; i < n; ++i) {
+        std::size_t si = static_cast<std::size_t>(i);
+        xp[si] = x0[si] + eps;
+        xm[si] = x0[si] - eps;
+
+        double Ep = spherical_energy(mesh, xp, m);
+        double Em = spherical_energy(mesh, xm, m);
+
+        xp[si] = xm[si] = x0[si];   // restore
+
+        double fd    = (Ep - Em) / (2.0 * eps);
+        double err   = std::abs(G[si] - fd);
+        double scale = std::max(1.0, std::abs(G[si]));
+        if (err / scale > tol) ok = false;
+    }
+    return ok;
+}
+
+// ── Gauge-fix for closed spherical surfaces ───────────────────────────────────
+//
+// On a closed (boundaryless) spherical surface the energy E(u + t·1) has a
+// unique maximum w.r.t. t ∈ ℝ (the "global scale" gauge mode).  Without
+// fixing this gauge the functional is unbounded below and no solver converges.
+//
+// This function returns the scalar shift t* such that
+//   Σ_v G_v(x + t*·1_v)  =  0
+// i.e., the derivative of E(u+t·1) w.r.t. t is zero at t = t*.
+//
+// Apply the shift by adding t* to every vertex DOF in x.
+//
+// Implementation: bisection on f(t) = Σ_v G_v(x + t·1_v).
+//   f is strictly monotone decreasing (second derivative < 0) for a convex
+//   functional, so bisection converges in O(log₂(2·bracket/tol)) iterations.
+//
+// Parameters:
+//   bracket   – initial search interval [−bracket, +bracket]   (default 50)
+//   tol       – absolute tolerance on t*                        (default 1e-8)
+//
+// Returns 0.0 if the zero cannot be bracketed (already at gauge maximum,
+// or open surface — no shift needed).
+inline double spherical_gauge_shift(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const SphericalMaps&       m,
+    double                     bracket = 50.0,
+    double                     tol     = 1e-8)
+{
+    // Helper: sum of all vertex gradient components at x + t·1_v.
+    auto sum_Gv = [&](double t) -> double {
+        // Build a shifted copy of x (only vertex DOFs shifted).
+        std::vector<double> xt = x;
+        for (auto v : mesh.vertices()) {
+            int iv = m.v_idx[v];
+            if (iv >= 0)
+                xt[static_cast<std::size_t>(iv)] += t;
+        }
+        auto G = spherical_gradient(mesh, xt, m);
+        double sum = 0.0;
+        for (auto v : mesh.vertices()) {
+            int iv = m.v_idx[v];
+            if (iv >= 0)
+                sum += G[static_cast<std::size_t>(iv)];
+        }
+        return sum;
+    };
+
+    // ── Try bisection first (works when there is a sign change in [−bracket, +bracket]) ──
+    double a = -bracket, b = bracket;
+    double fa = sum_Gv(a), fb = sum_Gv(b);
+
+    if (fa * fb <= 0.0) {
+        for (int iter = 0; iter < 120; ++iter) {
+            double c  = 0.5 * (a + b);
+            double fc = sum_Gv(c);
+            if (std::abs(fc) < tol || (b - a) < tol) return c;
+            if (fa * fc < 0.0) { b = c; fb = fc; }
+            else               { a = c; fa = fc; }
+        }
+        return 0.5 * (a + b);
+    }
+
+    // ── No sign change (zero may lie at a domain boundary).  ───────────────────
+    //    Use damped Newton's method with backtracking line search.
+    //    f'(t) estimated by forward finite difference.
+    //    When the Newton step overshoots the valid domain (ΣG_v jumps back up
+    //    because faces become degenerate), backtracking halves the step until
+    //    |f| strictly decreases.
+    const double fd_eps = 1e-5;
+    double t  = 0.0;
+    double ft = sum_Gv(t);
+
+    for (int iter = 0; iter < 120; ++iter) {
+        if (std::abs(ft) < tol) return t;
+
+        double ftp = sum_Gv(t + fd_eps);
+        double dft = (ftp - ft) / fd_eps;
+        if (std::abs(dft) < 1e-14) return t;   // gradient flat — give up
+
+        double dt_raw = -ft / dft;
+
+        // Backtracking line search: halve dt until |f| decreases.
+        double alpha = 1.0;
+        bool improved = false;
+        for (int back = 0; back < 40; ++back) {
+            double t_try  = t + alpha * dt_raw;
+            t_try = std::max(-bracket, std::min(bracket, t_try));
+            double ft_try = sum_Gv(t_try);
+            if (std::abs(ft_try) < std::abs(ft)) {
+                t  = t_try;
+                ft = ft_try;
+                improved = true;
+                break;
+            }
+            alpha *= 0.5;
+        }
+        if (!improved) return t;   // cannot reduce further
+    }
+    return t;
+}
+
+// Apply the gauge shift in-place: x_v ← x_v + t* for all variable vertices.
+inline void apply_spherical_gauge(
+    ConformalMesh&       mesh,
+    std::vector<double>& x,
+    const SphericalMaps& m,
+    double               bracket = 50.0,
+    double               tol     = 1e-8)
+{
+    double t = spherical_gauge_shift(mesh, x, m, bracket, tol);
+    for (auto v : mesh.vertices()) {
+        int iv = m.v_idx[v];
+        if (iv >= 0)
+            x[static_cast<std::size_t>(iv)] += t;
+    }
+}
+
+} // namespace conformallab

code/include/spherical_geometry.hpp

@@ -0,0 +1,84 @@
+#pragma once
+// spherical_geometry.hpp
+//
+// Pure-math building blocks for the spherical discrete conformal map.
+// Ported from de.varylab.discreteconformal.functional.SphericalFunctional
+// (the geometry helpers embedded there).
+//
+// Notation:
+//   u_i          – vertex conformal factor (DOF)
+//   λ°_e         – base log-length of edge e (fixed initial value)
+//   λ_ij         – effective log-length = λ°_ij + u_i + u_j
+//   l_ij         – spherical arc length = 2·asin(min(exp(λ_ij/2), 1))
+//   α_k          – interior angle of the spherical triangle at vertex k
+
+#include "constants.hpp"
+#include <cmath>
+#include <algorithm>
+
+namespace conformallab {
+
+/// Backward-compatible alias — prefer conformallab::PI in new code.
+constexpr double PI_SPHER = PI;
+
+// ── Effective spherical arc length ────────────────────────────────────────────
+
+// l(λ) = 2·asin(min(exp(λ/2), 1)).
+// Clamps exp(λ/2) to [0, 1] so the arcsin stays in domain.
+inline double spherical_l(double lambda)
+{
+    double half = std::exp(lambda * 0.5);
+    if (half >= 1.0) half = 1.0 - 1e-15;
+    if (half <= 0.0) return 0.0;
+    return 2.0 * std::asin(half);
+}
+
+// ── Interior angles of a spherical triangle ──────────────────────────────────
+
+struct SphericalFaceAngles {
+    double alpha1, alpha2, alpha3;  // corner angles at v1, v2, v3
+    bool   valid;                   // false when the three lengths fail the
+                                    // spherical triangle inequality
+};
+
+// Compute corner angles from spherical arc lengths using the half-angle formula.
+//
+// Convention (matching the halfedge cycle h0→v1→v2, h1→v2→v3, h2→v3→v1):
+//   l12 – arc length of edge opposite v3 (edge e12)
+//   l23 – arc length of edge opposite v1 (edge e23)
+//   l31 – arc length of edge opposite v2 (edge e31)
+//
+// Half-angle formula (spherical law of cosines):
+//   α_k = 2·atan2(sqrt(sin(s-a)·sin(s-b)), sqrt(sin(s)·sin(s-c)))
+// where a,b are the two edges ADJACENT to vertex k, c is the opposite edge.
+//
+// Equivalently (in terms of s-deficiencies):
+//   α1 = 2·atan2( sqrt(sin(s12)·sin(s31)), sqrt(sin(s)·sin(s23)) )
+//   α2 = 2·atan2( sqrt(sin(s12)·sin(s23)), sqrt(sin(s)·sin(s31)) )
+//   α3 = 2·atan2( sqrt(sin(s23)·sin(s31)), sqrt(sin(s)·sin(s12)) )
+//
+// where s = (l12+l23+l31)/2 and s_ij = s - l_ij.
+inline SphericalFaceAngles spherical_angles(double l12, double l23, double l31)
+{
+    double s   = (l12 + l23 + l31) * 0.5;
+    double s12 = s - l12;
+    double s23 = s - l23;
+    double s31 = s - l31;
+
+    // Spherical triangle inequalities: all s-deficiencies > 0 and s < π.
+    if (s12 <= 0.0 || s23 <= 0.0 || s31 <= 0.0 || s >= PI_SPHER)
+        return {0.0, 0.0, 0.0, false};
+
+    const double ss   = std::sin(s);
+    const double ss12 = std::sin(s12);
+    const double ss23 = std::sin(s23);
+    const double ss31 = std::sin(s31);
+
+    double a1 = 2.0 * std::atan2(std::sqrt(ss12 * ss31), std::sqrt(ss * ss23));
+    double a2 = 2.0 * std::atan2(std::sqrt(ss12 * ss23), std::sqrt(ss * ss31));
+    double a3 = 2.0 * std::atan2(std::sqrt(ss23 * ss31), std::sqrt(ss * ss12));
+
+    return {a1, a2, a3, true};
+}
+
+} // namespace conformallab

code/include/spherical_hessian.hpp

@@ -0,0 +1,256 @@
+#pragma once
+// spherical_hessian.hpp
+//
+// Analytical Hessian of the spherical discrete conformal energy —
+// the spherical cotangent-Laplace operator.
+//
+// Ported from de.varylab.discreteconformal.functional.SphericalFunctional
+// (the hessian() method, vertex DOFs).
+//
+// ┌──────────────────────────────────────────────────────────────────────────┐
+// │  Hessian formula (vertex DOFs only)                                      │
+// │                                                                          │
+// │  For a spherical face (v1, v2, v3) with vertex angles α1, α2, α3:       │
+// │                                                                          │
+// │    Spherical cotangent weight for edge (vi, vj) with opposite vk:        │
+// │      β_k = (π − αi − αj + αk) / 2                                       │
+// │      w_k  = cot(β_k)  =  1/tan(β_k)                                     │
+// │                                                                          │
+// │    Euclidean limit: α1+α2+α3 → π, β_k → αk, w_k → cot(αk). ✓          │
+// │                                                                          │
+// │  Hessian contributions per face:                                         │
+// │    H[vi, vi] += w_ij + w_ik      (diagonal: weights of incident edges)  │
+// │    H[vi, vj] -= w_ij             (off-diagonal: weight of edge ij)      │
+// │                                                                          │
+// │  where w_ij is the weight of the edge between vi and vj (opposite vk):  │
+// │    w_ij = cot(β_k)  with  β_k = (π − αi − αj + αk) / 2.               │
+// └──────────────────────────────────────────────────────────────────────────┘
+//
+// Requires Eigen (header-only).  Returns Eigen::SparseMatrix<double>.
+
+#include "spherical_functional.hpp"
+#include <Eigen/Sparse>
+#include <vector>
+#include <cmath>
+
+namespace conformallab {
+
+// ── Spherical cotangent weight helper ────────────────────────────────────────
+//
+// Given the three face angles α1, α2, α3 of a spherical triangle, return the
+// three edge cotangent weights w1 (edge opp v1), w2 (edge opp v2), w3 (edge opp v3).
+//
+// w_k = cot(β_k)  where β_k = (π − α_adj1 − α_adj2 + α_opp) / 2
+//                             = (π − αi − αj + αk) / 2  for edge (vi,vj), opposite vk
+//
+// Mapping in our CGAL halfedge convention:
+//   h0 = halfedge(f): edge v1-v2 → opposite v3 → w = cot(β3), β3=(π-α1-α2+α3)/2
+//   h1: edge v2-v3 → opposite v1 → w = cot(β1), β1=(π-α2-α3+α1)/2
+//   h2: edge v3-v1 → opposite v2 → w = cot(β2), β2=(π-α3-α1+α2)/2
+//
+// Returns valid=false if any β_k is out of range (degenerate face).
+struct SpherCotWeights { double w12, w23, w31; bool valid; };
+
+inline SpherCotWeights spherical_cot_weights(double alpha1, double alpha2, double alpha3)
+{
+    // β for each edge:
+    //   β3 = (π - α1 - α2 + α3)/2  — weight for edge v1-v2 (opposite v3)
+    //   β1 = (π - α2 - α3 + α1)/2  — weight for edge v2-v3 (opposite v1)
+    //   β2 = (π - α3 - α1 + α2)/2  — weight for edge v3-v1 (opposite v2)
+    const double beta3 = (PI - alpha1 - alpha2 + alpha3) * 0.5;
+    const double beta1 = (PI - alpha2 - alpha3 + alpha1) * 0.5;
+    const double beta2 = (PI - alpha3 - alpha1 + alpha2) * 0.5;
+
+    // Each β_k must be in (0, π/2] for the weight to be positive and well-defined.
+    // For degenerate or very flat triangles some β may be ≤ 0 or ≥ π/2.
+    if (beta1 <= 0.0 || beta2 <= 0.0 || beta3 <= 0.0) return {0.0, 0.0, 0.0, false};
+
+    const double tb1 = std::tan(beta1);
+    const double tb2 = std::tan(beta2);
+    const double tb3 = std::tan(beta3);
+
+    if (std::abs(tb1) < 1e-15 || std::abs(tb2) < 1e-15 || std::abs(tb3) < 1e-15)
+        return {0.0, 0.0, 0.0, false};
+
+    // w_ij = cot(β_k) where β_k is for the edge opposite vk.
+    // w12 is for edge v1-v2 (opposite v3): cot(β3)
+    // w23 is for edge v2-v3 (opposite v1): cot(β1)
+    // w31 is for edge v3-v1 (opposite v2): cot(β2)
+    return {1.0 / tb3, 1.0 / tb1, 1.0 / tb2, true};
+}
+
+// ── Analytical Hessian ────────────────────────────────────────────────────────
+//
+// Returns the n×n sparse Hessian matrix H where n = spherical_dimension(mesh, m).
+// x – current DOF vector.
+//
+// Derivation: G_v = θ_v − Σ_f α_v^f  →  H[i,j] = −Σ_f ∂α_i^f/∂u_j
+//
+// For a face (v1,v2,v3) with arc-lengths l12,l23,l31 and angles α1,α2,α3,
+// differentiating the spherical law of cosines
+//   cos(l_opp) = cos(l_a)cos(l_b) + sin(l_a)sin(l_b)cos(α)
+// gives:
+//   ∂α1/∂l12 = [cot(l12)cos(α1) − cot(l31)] / sin(α1)     (adjacent side)
+//   ∂α1/∂l31 = [cot(l31)cos(α1) − cot(l12)] / sin(α1)     (adjacent side)
+//   ∂α1/∂l23 = sin(l23) / [sin(l12)sin(l31)sin(α1)]        (opposite side)
+//
+// Chain rule with ∂l_ij/∂u_k = tan(l_ij/2)  (from l = 2·asin(exp(λ/2))):
+//   ∂α1/∂u1 = ∂α1/∂l12·t12 + ∂α1/∂l31·t31
+//   ∂α1/∂u2 = ∂α1/∂l12·t12 + ∂α1/∂l23·t23
+//   ∂α1/∂u3 = ∂α1/∂l23·t23 + ∂α1/∂l31·t31
+inline Eigen::SparseMatrix<double> spherical_hessian(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x,
+    const SphericalMaps&       m)
+{
+    const int n = spherical_dimension(mesh, m);
+
+    std::vector<Eigen::Triplet<double>> trips;
+    trips.reserve(static_cast<std::size_t>(n) * 9);
+
+    for (auto f : mesh.faces()) {
+        Halfedge_index h0 = mesh.halfedge(f);
+        Halfedge_index h1 = mesh.next(h0);
+        Halfedge_index h2 = mesh.next(h1);
+
+        Vertex_index v1 = mesh.source(h0);
+        Vertex_index v2 = mesh.source(h1);
+        Vertex_index v3 = mesh.source(h2);
+
+        Edge_index e12 = mesh.edge(h0);
+        Edge_index e23 = mesh.edge(h1);
+        Edge_index e31 = mesh.edge(h2);
+
+        // Effective log-lengths.
+        double u1 = spher_dof_val(m.v_idx[v1], x);
+        double u2 = spher_dof_val(m.v_idx[v2], x);
+        double u3 = spher_dof_val(m.v_idx[v3], x);
+
+        double lam12 = m.lambda0[e12] + u1 + u2 + spher_dof_val(m.e_idx[e12], x);
+        double lam23 = m.lambda0[e23] + u2 + u3 + spher_dof_val(m.e_idx[e23], x);
+        double lam31 = m.lambda0[e31] + u3 + u1 + spher_dof_val(m.e_idx[e31], x);
+
+        const double l12 = spherical_l(lam12);
+        const double l23 = spherical_l(lam23);
+        const double l31 = spherical_l(lam31);
+
+        SphericalFaceAngles fa = spherical_angles(l12, l23, l31);
+        if (!fa.valid) continue;
+
+        const double sinl12 = std::sin(l12), cosl12 = std::cos(l12);
+        const double sinl23 = std::sin(l23), cosl23 = std::cos(l23);
+        const double sinl31 = std::sin(l31), cosl31 = std::cos(l31);
+
+        if (sinl12 < 1e-15 || sinl23 < 1e-15 || sinl31 < 1e-15) continue;
+
+        const double cot12 = cosl12 / sinl12;
+        const double cot23 = cosl23 / sinl23;
+        const double cot31 = cosl31 / sinl31;
+
+        // ∂l_ij/∂λ_ij = tan(l_ij/2)
+        const double t12 = std::tan(l12 * 0.5);
+        const double t23 = std::tan(l23 * 0.5);
+        const double t31 = std::tan(l31 * 0.5);
+
+        const double sinA1 = std::sin(fa.alpha1), cosA1 = std::cos(fa.alpha1);
+        const double sinA2 = std::sin(fa.alpha2), cosA2 = std::cos(fa.alpha2);
+        const double sinA3 = std::sin(fa.alpha3), cosA3 = std::cos(fa.alpha3);
+
+        if (sinA1 < 1e-15 || sinA2 < 1e-15 || sinA3 < 1e-15) continue;
+
+        // ∂α1/∂l_jk  (α1 at v1; opposite l23, adjacent l12,l31)
+        const double dA1_dl12 = (cot12 * cosA1 - cot31) / sinA1;
+        const double dA1_dl31 = (cot31 * cosA1 - cot12) / sinA1;
+        const double dA1_dl23 = sinl23 / (sinl12 * sinl31 * sinA1);
+
+        // ∂α2/∂l_jk  (α2 at v2; opposite l31, adjacent l12,l23)
+        const double dA2_dl12 = (cot12 * cosA2 - cot23) / sinA2;
+        const double dA2_dl23 = (cot23 * cosA2 - cot12) / sinA2;
+        const double dA2_dl31 = sinl31 / (sinl12 * sinl23 * sinA2);
+
+        // ∂α3/∂l_jk  (α3 at v3; opposite l12, adjacent l23,l31)
+        const double dA3_dl23 = (cot23 * cosA3 - cot31) / sinA3;
+        const double dA3_dl31 = (cot31 * cosA3 - cot23) / sinA3;
+        const double dA3_dl12 = sinl12 / (sinl23 * sinl31 * sinA3);
+
+        // Chain rule: ∂α_i/∂u_j   (u1 affects l12,l31; u2 affects l12,l23; u3 affects l23,l31)
+        const double dA1_du1 = dA1_dl12 * t12 + dA1_dl31 * t31;
+        const double dA1_du2 = dA1_dl12 * t12 + dA1_dl23 * t23;
+        const double dA1_du3 = dA1_dl23 * t23 + dA1_dl31 * t31;
+
+        const double dA2_du1 = dA2_dl12 * t12 + dA2_dl31 * t31;
+        const double dA2_du2 = dA2_dl12 * t12 + dA2_dl23 * t23;
+        const double dA2_du3 = dA2_dl23 * t23 + dA2_dl31 * t31;
+
+        const double dA3_du1 = dA3_dl12 * t12 + dA3_dl31 * t31;
+        const double dA3_du2 = dA3_dl12 * t12 + dA3_dl23 * t23;
+        const double dA3_du3 = dA3_dl23 * t23 + dA3_dl31 * t31;
+
+        const int i1 = m.v_idx[v1];
+        const int i2 = m.v_idx[v2];
+        const int i3 = m.v_idx[v3];
+
+        // H[vi, vj] -= ∂α_i/∂u_j   (G_v = θ_v − Σ α_v, so ∂G_i/∂u_j = −∂α_i/∂u_j)
+        if (i1 >= 0) trips.emplace_back(i1, i1, -dA1_du1);
+        if (i2 >= 0) trips.emplace_back(i2, i2, -dA2_du2);
+        if (i3 >= 0) trips.emplace_back(i3, i3, -dA3_du3);
+
+        if (i1 >= 0 && i2 >= 0) {
+            trips.emplace_back(i1, i2, -dA1_du2);
+            trips.emplace_back(i2, i1, -dA2_du1);
+        }
+        if (i2 >= 0 && i3 >= 0) {
+            trips.emplace_back(i2, i3, -dA2_du3);
+            trips.emplace_back(i3, i2, -dA3_du2);
+        }
+        if (i3 >= 0 && i1 >= 0) {
+            trips.emplace_back(i3, i1, -dA3_du1);
+            trips.emplace_back(i1, i3, -dA1_du3);
+        }
+    }
+
+    Eigen::SparseMatrix<double> H(n, n);
+    H.setFromTriplets(trips.begin(), trips.end());
+    return H;
+}
+
+// ── Finite-difference Hessian check ──────────────────────────────────────────
+//
+// Compares the analytical Hessian column-by-column against
+//   H_fd[:, j] = (G(x + ε·eⱼ) − G(x − ε·eⱼ)) / (2ε).
+inline bool hessian_check_spherical(
+    ConformalMesh&             mesh,
+    const std::vector<double>& x0,
+    const SphericalMaps&       m,
+    double                     eps = 1e-5,
+    double                     tol = 1e-4)
+{
+    const int n = static_cast<int>(x0.size());
+    auto H = spherical_hessian(mesh, x0, m);
+
+    std::vector<double> xp = x0, xm = x0;
+    bool ok = true;
+
+    for (int j = 0; j < n; ++j) {
+        const std::size_t sj = static_cast<std::size_t>(j);
+        xp[sj] = x0[sj] + eps;
+        xm[sj] = x0[sj] - eps;
+
+        auto Gp = spherical_gradient(mesh, xp, m);
+        auto Gm = spherical_gradient(mesh, xm, m);
+
+        xp[sj] = xm[sj] = x0[sj];
+
+        for (int i = 0; i < n; ++i) {
+            double fd_ij = (Gp[static_cast<std::size_t>(i)]
+                          - Gm[static_cast<std::size_t>(i)]) / (2.0 * eps);
+            double H_ij  = H.coeff(i, j);
+            double err   = std::abs(H_ij - fd_ij);
+            double scale = std::max(1.0, std::abs(H_ij));
+            if (err / scale > tol) ok = false;
+        }
+    }
+    return ok;
+}
+
+} // namespace conformallab

code/src/apps/v0/conformallab_cli.cpp

@@ -1,107 +1,314 @@
-#include <CGAL/Simple_cartesian.h>
+// conformallab_cli.cpp
-#include <CGAL/Surface_mesh.h>
+//
-#include <CGAL/IO/polygon_mesh_io.h>
+// ConformalLab++ command-line interface.
-
+//
-
+// Usage:
-#include "viewer_utils.h"
+//   conformallab_core -i input.off [-o layout.off] [-g euclidean|spherical|hyper_ideal]
-#include "mesh_utils.hpp"
+//                    [-j result.json] [-x result.xml] [-s] [-v]
+//
+// The tool:
+//   1. Loads an OFF/OBJ/PLY mesh.
+//   2. Sets up DOF maps + computes λ° from the input geometry.
+//   3. Pins one vertex (Euclidean/Spherical) or uses all-free DOFs (HyperIdeal).
+//   4. Runs Newton until convergence.
+//   5. Computes a 2-D (Euclidean / HyperIdeal) or 3-D (Spherical) layout.
+//   6. Saves the layout as an OFF file and optionally serialises the result
+//      to JSON and/or XML.
+//   7. Optionally shows the input mesh in a viewer (-s flag, requires WITH_VIEWER).
 
+#include "conformal_mesh.hpp"
+#include "mesh_io.hpp"
+#include "euclidean_functional.hpp"
+#include "spherical_functional.hpp"
+#include "hyper_ideal_functional.hpp"
+#include "newton_solver.hpp"
+#include "layout.hpp"
+#include "serialization.hpp"
 
 #include <CLI11.hpp>
 #include <iostream>
+#include <iomanip>
 #include <string>
+#include <vector>
+#include <cmath>
+#include <algorithm>
 
-
+// Viewer is always available when WITH_CGAL=ON (implies WITH_VIEWER=ON)
+#include "viewer_utils.h"
 #include <Eigen/Core>
 
+namespace cl = conformallab;
+using cl::ConformalMesh;
+using cl::Vertex_index;
+using cl::Edge_index;
 
-using Kernel = CGAL::Simple_cartesian<double>;
+// ─────────────────────────────────────────────────────────────────────────────
-using Point  = Kernel::Point_3;
+// Shared helpers (mirroring test_pipeline.cpp patterns)
-using Mesh   = CGAL::Surface_mesh<Point>;
+// ─────────────────────────────────────────────────────────────────────────────
 
-
+// Pin vertex 0, assign 0..n-1 to the rest
-
+static int pin_first_vertex(ConformalMesh& mesh, cl::EuclideanMaps& maps)
-
+{
-bool parse_arguments(int argc, char* argv[], std::string& input_file, std::string& output_file, bool& show, bool& verbose) {
+    auto vit = mesh.vertices().begin();
-    CLI::App app{"demo of conformallab"};
+    Vertex_index v0 = *vit++;
-    app.add_option("-i,--input",  input_file,  "Input OFF file")->required();
+    maps.v_idx[v0] = -1;
-    app.add_option("-o,--output", output_file, "Output OFF file (optional)");
+    int idx = 0;
-    app.add_flag("-s,--show", show, "Show the input mesh in a viewer ");
+    for (; vit != mesh.vertices().end(); ++vit)
-    app.add_flag("-v,--verbose",verbose, "Enable verbose output");
+        maps.v_idx[*vit] = idx++;
-    app.set_help_flag("-h,--help", "Show this help message and exit");
+    return idx;
-
-    try {
-        CLI11_PARSE(app, argc, argv);
-    } catch (const CLI::ParseError &e) {
-        return false;
-    }
-
-    if (input_file.empty()) {
-        std::cerr << "Input file is required.\n";
-        return EXIT_FAILURE;
-    }
-
-    if (input_file.substr(input_file.find_last_of('.') + 1) != "off") {
-        std::cerr << "Unsupported file format. Please provide an OFF file.\n";
-        return EXIT_FAILURE;
-    }
-
-    std::cout << "Input file: " << input_file << "\n";
-    std::cout << "Output file: " << output_file << "\n";
-    std::cout << "Show mesh: " << (show ? "Yes" : "No") << "\n";
-    std::cout << "Verbose: " << (verbose ? "Yes" : "No") << "\n";
-
-
-
-    return true;
 }
 
+// Natural theta for Euclidean: make x=0 the equilibrium
+static void set_natural_euclidean_theta(ConformalMesh& mesh, cl::EuclideanMaps& maps, int n)
+{
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto G = cl::euclidean_gradient(mesh, x0, maps);
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv < 0) continue;
+        maps.theta_v[v] -= G[static_cast<std::size_t>(iv)];
+    }
+}
 
+// Natural theta for HyperIdeal at base point (b=1, a=0.5) to avoid x=0 singularity
+static std::vector<double> set_natural_hyper_ideal_targets(
+    ConformalMesh& mesh, cl::HyperIdealMaps& maps, int n)
+{
+    const auto sz = static_cast<std::size_t>(n);
+    std::vector<double> xbase(sz, 0.0);
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv >= 0) xbase[static_cast<std::size_t>(iv)] = 1.0;
+    }
+    for (auto e : mesh.edges()) {
+        int ie = maps.e_idx[e];
+        if (ie >= 0) xbase[static_cast<std::size_t>(ie)] = 0.5;
+    }
+    auto G = cl::evaluate_hyper_ideal(mesh, xbase, maps, false).gradient;
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv < 0) continue;
+        maps.theta_v[v] += G[static_cast<std::size_t>(iv)];
+    }
+    for (auto e : mesh.edges()) {
+        int ie = maps.e_idx[e];
+        if (ie < 0) continue;
+        maps.theta_e[e] += G[static_cast<std::size_t>(ie)];
+    }
+    return xbase;
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Euclidean pipeline
+// ─────────────────────────────────────────────────────────────────────────────
+static int run_euclidean(ConformalMesh& mesh,
+                         const std::string& out_layout,
+                         const std::string& out_json,
+                         const std::string& out_xml,
+                         bool verbose)
+{
+    // Setup
+    auto maps = cl::setup_euclidean_maps(mesh);
+    cl::compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    // DOF assignment: pin vertex 0
+    int n = pin_first_vertex(mesh, maps);
+    if (n <= 0) { std::cerr << "Error: mesh has only one vertex.\n"; return 1; }
+
+    // Natural target angles
+    set_natural_euclidean_theta(mesh, maps, n);
+
+    // Newton
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto res = cl::newton_euclidean(mesh, x0, maps);
+
+    if (!res.converged && verbose)
+        std::cerr << "[warn] Newton did not converge (|grad|=" << res.grad_inf_norm << ")\n";
+
+    // Layout
+    cl::Layout2D layout = cl::euclidean_layout(mesh, res.x, maps);
+
+    // Output
+    if (!out_layout.empty()) cl::save_layout_off(out_layout, mesh, layout);
+    if (!out_json.empty())
+        cl::save_result_json(out_json, res, "euclidean",
+                             static_cast<int>(mesh.number_of_vertices()),
+                             static_cast<int>(mesh.number_of_faces()),
+                             &layout);
+    if (!out_xml.empty())
+        cl::save_result_xml(out_xml, res, "euclidean",
+                            static_cast<int>(mesh.number_of_vertices()),
+                            static_cast<int>(mesh.number_of_faces()),
+                            &layout);
+
+    std::cout << std::fixed << std::setprecision(6);
+    std::cout << "Euclidean: converged=" << (res.converged ? "yes" : "no")
+              << "  iter=" << res.iterations
+              << "  |grad|_inf=" << std::scientific << std::setprecision(3)
+              << res.grad_inf_norm << "\n";
+    if (!out_layout.empty()) std::cout << "  layout → " << out_layout << "\n";
+    if (!out_json.empty())   std::cout << "  json   → " << out_json   << "\n";
+    if (!out_xml.empty())    std::cout << "  xml    → " << out_xml    << "\n";
+    return 0;
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Spherical pipeline
+// ─────────────────────────────────────────────────────────────────────────────
+static int run_spherical(ConformalMesh& mesh,
+                          const std::string& out_layout,
+                          const std::string& out_json,
+                          const std::string& out_xml,
+                          bool verbose)
+{
+    auto maps = cl::setup_spherical_maps(mesh);
+    cl::compute_lambda0_from_mesh(mesh, maps);
+    int n = cl::assign_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto res = cl::newton_spherical(mesh, x0, maps);
+
+    if (!res.converged && verbose)
+        std::cerr << "[warn] Newton did not converge (|grad|=" << res.grad_inf_norm << ")\n";
+
+    cl::Layout3D layout = cl::spherical_layout(mesh, res.x, maps);
+
+    if (!out_layout.empty()) cl::save_layout_off(out_layout, mesh, layout);
+    if (!out_json.empty())
+        cl::save_result_json(out_json, res, "spherical",
+                             static_cast<int>(mesh.number_of_vertices()),
+                             static_cast<int>(mesh.number_of_faces()),
+                             nullptr, &layout);
+    if (!out_xml.empty())
+        cl::save_result_xml(out_xml, res, "spherical",
+                            static_cast<int>(mesh.number_of_vertices()),
+                            static_cast<int>(mesh.number_of_faces()),
+                            nullptr, &layout);
+
+    std::cout << "Spherical: converged=" << (res.converged ? "yes" : "no")
+              << "  iter=" << res.iterations
+              << "  |grad|_inf=" << std::scientific << std::setprecision(3)
+              << res.grad_inf_norm << "\n";
+    if (!out_layout.empty()) std::cout << "  layout → " << out_layout << "\n";
+    if (!out_json.empty())   std::cout << "  json   → " << out_json   << "\n";
+    if (!out_xml.empty())    std::cout << "  xml    → " << out_xml    << "\n";
+    return 0;
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// HyperIdeal pipeline
+// ─────────────────────────────────────────────────────────────────────────────
+static int run_hyper_ideal(ConformalMesh& mesh,
+                            const std::string& out_layout,
+                            const std::string& out_json,
+                            const std::string& out_xml,
+                            bool verbose)
+{
+    auto maps = cl::setup_hyper_ideal_maps(mesh);
+    int n = cl::assign_all_dof_indices(mesh, maps);
+
+    // Natural targets at base point (b=1, a=0.5)
+    auto xbase = set_natural_hyper_ideal_targets(mesh, maps, n);
+
+    // Perturb slightly from the base and solve back
+    std::vector<double> x0 = xbase;
+    for (auto& v : x0) v += 0.3;
+
+    auto res = cl::newton_hyper_ideal(mesh, x0, maps);
+
+    if (!res.converged && verbose)
+        std::cerr << "[warn] Newton did not converge (|grad|=" << res.grad_inf_norm << ")\n";
+
+    cl::Layout2D layout = cl::hyper_ideal_layout(mesh, res.x, maps);
+
+    if (!out_layout.empty()) cl::save_layout_off(out_layout, mesh, layout);
+    if (!out_json.empty())
+        cl::save_result_json(out_json, res, "hyper_ideal",
+                             static_cast<int>(mesh.number_of_vertices()),
+                             static_cast<int>(mesh.number_of_faces()),
+                             &layout);
+    if (!out_xml.empty())
+        cl::save_result_xml(out_xml, res, "hyper_ideal",
+                            static_cast<int>(mesh.number_of_vertices()),
+                            static_cast<int>(mesh.number_of_faces()),
+                            &layout);
+
+    std::cout << "HyperIdeal: converged=" << (res.converged ? "yes" : "no")
+              << "  iter=" << res.iterations
+              << "  |grad|_inf=" << std::scientific << std::setprecision(3)
+              << res.grad_inf_norm << "\n";
+    if (!out_layout.empty()) std::cout << "  layout → " << out_layout << "\n";
+    if (!out_json.empty())   std::cout << "  json   → " << out_json   << "\n";
+    if (!out_xml.empty())    std::cout << "  xml    → " << out_xml    << "\n";
+    return 0;
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// main
+// ─────────────────────────────────────────────────────────────────────────────
 int main(int argc, char* argv[])
 {
+    CLI::App app{"ConformalLab++ — discrete conformal mapping"};
+
     std::string input_file;
-    std::string output_file; 
+    std::string out_layout;
-    bool show = false;
+    std::string out_json;
+    std::string out_xml;
+    std::string geometry = "euclidean";
+    bool show    = false;
     bool verbose = false;
 
-    if(!parse_arguments(argc, argv, input_file, output_file, show, verbose)) {
+    app.add_option("-i,--input",    input_file, "Input mesh (OFF/OBJ/PLY)")->required();
-        std::cerr << "Failed to parse arguments.\n";
+    app.add_option("-o,--output",   out_layout, "Output layout OFF file");
+    app.add_option("-j,--json",     out_json,   "Save result as JSON");
+    app.add_option("-x,--xml",      out_xml,    "Save result as XML");
+    app.add_option("-g,--geometry", geometry,   "Target geometry: euclidean|spherical|hyper_ideal")
+        ->check(CLI::IsMember({"euclidean", "spherical", "hyper_ideal"}));
+    app.add_flag("-s,--show",    show,    "Visualise input mesh (requires WITH_VIEWER)");
+    app.add_flag("-v,--verbose", verbose, "Verbose output");
+
+    CLI11_PARSE(app, argc, argv);
+
+    // ── Load mesh ─────────────────────────────────────────────────────────────
+    ConformalMesh mesh;
+    if (!cl::read_mesh(input_file, mesh) || mesh.is_empty()) {
+        std::cerr << "Error: cannot load mesh from '" << input_file << "'\n";
         return EXIT_FAILURE;
     }
+    std::cout << "Loaded: " << input_file
+              << "  (" << mesh.number_of_vertices() << "v, "
+              << mesh.number_of_faces() << "f)\n";
 
-    Mesh surface_mesh;
+    // ── Optional viewer ───────────────────────────────────────────────────────
-    if (!CGAL::IO::read_polygon_mesh(input_file, surface_mesh) || surface_mesh.is_empty()) {
+    if (show) {
-        std::cerr << "Invalid input file: " << input_file << "\n";
+        // Convert ConformalMesh to Eigen matrices for the libigl viewer
-        return EXIT_FAILURE;
+        const std::size_t nv = mesh.number_of_vertices();
-    }
+        const std::size_t nf = mesh.number_of_faces();
-  
+        Eigen::MatrixXd V(static_cast<Eigen::Index>(nv), 3);
+        Eigen::MatrixXi F(static_cast<Eigen::Index>(nf), 3);
 
-
+        for (auto v : mesh.vertices()) {
-    
+            auto p = mesh.point(v);
-    // #TODO: later: here we would call the unwrapping code, e.g.:
+            V.row(v.idx()) << p.x(), p.y(), p.z();
-    // UnwrapSettings settings;
+        }
-    // settings.target_geometry = TargetGeometry::Euclidean;
+        Eigen::Index fi = 0;
-    // UnwrapJob job(surface_mesh, settings);
+        for (auto f : mesh.faces()) {
-    // auto result = job.run();
+            int ci = 0;
-    // Mesh unwrapped = result.surface_unwrapped;
+            for (auto h : CGAL::halfedges_around_face(mesh.halfedge(f), mesh))
-
+                F(fi, ci++) = static_cast<int>(mesh.target(h).idx());
-
+            ++fi;
-    // CGAL -> libigl
+        }
-    if(show){
+        viewer_utils::simple_visualize(V, F);
-    std::cout << "Visualizing input mesh...\n";
-    Eigen::MatrixXd V;
-    Eigen::MatrixXi F;
-   
-    mesh_utils::cgal_to_eigen<Kernel>(surface_mesh, V, F);
-    viewer_utils::simple_visualize(V, F);
-
-    }
-    
-    // for now, we just write the input mesh to the output file as a placeholder
-    if (!output_file.empty() && !CGAL::IO::write_polygon_mesh(output_file, surface_mesh)) {
-        std::cerr << "Failed to write output file: " << output_file << "\n";
-        return EXIT_FAILURE;
     }
 
-    return EXIT_SUCCESS;
+    // ── Dispatch ──────────────────────────────────────────────────────────────
+    if (geometry == "euclidean")
+        return run_euclidean(mesh, out_layout, out_json, out_xml, verbose);
+    if (geometry == "spherical")
+        return run_spherical(mesh, out_layout, out_json, out_xml, verbose);
+    if (geometry == "hyper_ideal")
+        return run_hyper_ideal(mesh, out_layout, out_json, out_xml, verbose);
+
+    std::cerr << "Unknown geometry: " << geometry << "\n";
+    return EXIT_FAILURE;
 }

code/tests/CMakeLists.txt

@@ -1,8 +1,20 @@
 add_executable(conformallab_tests
+    # ── Pure-math test suite (no CGAL, no mesh — runs on every branch) ─────
     test_clausen.cpp
     test_hyper_ideal_utility.cpp
     test_matrix_utility.cpp
     test_surface_curve_utility.cpp
+    test_discrete_elliptic_utility.cpp
+    test_p2_utility.cpp
+    test_hyper_ideal_visualization_utility.cpp
+    #
+    # Stale stub files were removed in v0.9.0:
+    #   test_hyper_ideal_functional.cpp
+    #   test_hyper_ideal_hyperelliptic_utility.cpp
+    #   test_spherical_functional.cpp
+    # They referenced a "HDS port (Phase 4)" that never happened —
+    # CoHDS was intentionally replaced by CGAL::Surface_mesh, and the
+    # functionals + tests live in code/tests/cgal/test_*_functional.cpp.
 )
 
 target_include_directories(conformallab_tests SYSTEM PRIVATE
@@ -17,3 +29,10 @@ target_link_libraries(conformallab_tests PRIVATE GTest::gtest_main)
 
 include(GoogleTest)
 gtest_discover_tests(conformallab_tests DISCOVERY_TIMEOUT 60)
+
+# ── CGAL test suite ──────────────────────────────────────────────────────────
+# Built with -DWITH_CGAL_TESTS=ON (headless CI, no viewer) or
+#            -DWITH_CGAL=ON       (full build with viewer + CLI).
+if(WITH_CGAL OR WITH_CGAL_TESTS)
+    add_subdirectory(cgal)
+endif()

code/tests/cgal/CMakeLists.txt

@@ -0,0 +1,124 @@
+# tests/cgal/CMakeLists.txt
+#
+# CGAL-dependent test target.  Only built when -DWITH_CGAL=ON.
+# Requires Boost (find_package(Boost REQUIRED) is called in the root CMakeLists).
+#
+# Run with:
+#   cmake -S code -B build -DWITH_CGAL=ON
+#   cmake --build build --target conformallab_cgal_tests
+#   ctest --test-dir build -R cgal
+
+add_executable(conformallab_cgal_tests
+    # ── Phase 3a: mesh infrastructure ──────────────────────────────────────
+    test_conformal_mesh.cpp
+
+    # ── Phase 3b: HyperIdealFunctional ─────────────────────────────────────
+    test_hyper_ideal_functional.cpp
+
+    # ── Phase 3c + 3e: SphericalFunctional + gauge-fix ────────────────────
+    test_spherical_functional.cpp
+
+    # ── Phase 3d: EuclideanCyclicFunctional ───────────────────────────────
+    test_euclidean_functional.cpp
+
+    # ── Phase 3f: Hessians (cotangent Laplacian, spherical + Euclidean) ───
+    test_euclidean_hessian.cpp
+    test_spherical_hessian.cpp
+
+    # ── Phase 9b: Hyper-ideal Hessian — block-FD vs full-FD validation ───
+    #    Verifies the O(F·36) block-local Hessian agrees with the
+    #    O(F·n) full-FD baseline.  Java upstream has no Hessian at all
+    #    (HyperIdealFunctional.hasHessian() returns false) — both
+    #    variants are conformallab++ extensions beyond the port.
+    test_hyper_ideal_hessian.cpp
+
+    # ── Phase 4a: Newton solver ────────────────────────────────────────────
+    test_newton_solver.cpp
+
+    # ── Phase 4b: Mesh I/O (CGAL::IO) ─────────────────────────────────────
+    test_mesh_io.cpp
+
+    # ── Phase 4c: End-to-end pipeline + user examples ─────────────────────
+    test_pipeline.cpp
+
+    # ── Phase 5: Layout / embedding + serialization ────────────────────────
+    test_layout.cpp
+
+    # ── Phase 6: Gauss–Bonnet, cut graph, exact trilateration, normalisation
+    test_phase6.cpp
+
+    # ── Phase 7: Java-parity layout — MobiusMap, priority BFS, halfedge_uv,
+    #             period matrix, fundamental domain, tiling
+    test_phase7.cpp
+
+    # ── Java parity: geometry utility tests ──────────────────────────────────
+    #    Ported from CuttinUtilityTest, UnwrapUtilityTest,
+    #    ConvergenceUtilityTests, HomologyTest (tests 1–6).
+    #    Test 7 (genus-2 homology) as GTEST_SKIP stub until Phase 8.
+    test_geometry_utils.cpp
+
+    # ── Scalability smoke tests ────────────────────────────────────────────────
+    #    Newton convergence on large real-world meshes (cathead, brezel, brezel2).
+    #    Assert correctness only (< 30 iterations, ||G|| < 1e-8).
+    #    Wall-clock time is printed for documentation but NOT asserted,
+    #    so the tests remain stable on slow CI hardware (Raspberry Pi ARM64).
+    test_scalability_smoke.cpp
+
+    # ── Phase 8 MVP: new CGAL-style public API ────────────────────────────────
+    #    First client of Conformal_map_traits.h + Discrete_conformal_map.h.
+    #    Acceptance probe before Phase 9a (Inversive-Distance) lands.
+    test_cgal_traits_mvp.cpp
+
+    # ── Phase 9a.1: CPEuclideanFunctional (BPS 2010 circle packing) ──────────
+    #    Face-based circle-packing functional ported from
+    #    CPEuclideanFunctional.java.  Reference: Bobenko-Pinkall-Springborn 2010.
+    test_cp_euclidean_functional.cpp
+
+    # ── Phase 9a.2: InversiveDistance (Luo 2004 + Glickenstein 2011) ─────────
+    #    Vertex-based inversive-distance circle-packing functional.  No Java
+    #    reference; implemented from the literature.  Cross-validated against
+    #    EuclideanCyclicFunctional at the natural initial geometry (u = 0).
+    test_inversive_distance_functional.cpp
+
+    # ── Phase 9a: Newton solvers for the two new circle-packing functionals ──
+    #    Convergence tests for newton_cp_euclidean (analytic Hessian) and
+    #    newton_inversive_distance (FD Hessian).
+    test_newton_phase9a.cpp
+
+    # ── Phase 8b-Lite: CGAL entry wrappers for the 4 non-Euclidean modes ─────
+    #    Spherical, HyperIdeal, CircleP-Euclidean, Inversive-Distance via
+    #    <CGAL/Discrete_*.h> public API + Conformal_layout.h wrapper.
+    test_cgal_phase8b_lite.cpp
+)
+
+target_include_directories(conformallab_cgal_tests SYSTEM PRIVATE
+    ${CMAKE_SOURCE_DIR}/deps/eigen-3.4.0
+    ${CMAKE_SOURCE_DIR}/deps/CGAL-6.1.1/include
+    ${CMAKE_SOURCE_DIR}/deps/single_includes
+    ${Boost_INCLUDE_DIRS}
+)
+
+target_include_directories(conformallab_cgal_tests PRIVATE
+    ${CMAKE_SOURCE_DIR}/include
+)
+
+target_compile_definitions(conformallab_cgal_tests PRIVATE
+    CGAL_DISABLE_GMP
+    CGAL_DISABLE_MPFR
+    # Data directory — absolute path to code/data/ at build time.
+    # Used by tests that load real mesh files (cathead.obj, brezel2.obj, …).
+    CONFORMALLAB_DATA_DIR="${CMAKE_SOURCE_DIR}/data"
+)
+
+# Suppress warnings from CGAL/Boost headers
+target_compile_options(conformallab_cgal_tests PRIVATE
+    $<$<CXX_COMPILER_ID:GNU,Clang,AppleClang>:-Wno-unused-parameter>
+)
+
+target_link_libraries(conformallab_cgal_tests PRIVATE GTest::gtest_main)
+
+include(GoogleTest)
+gtest_discover_tests(conformallab_cgal_tests
+    TEST_PREFIX "cgal."
+    DISCOVERY_TIMEOUT 60
+)

code/tests/cgal/test_cgal_phase8b_lite.cpp

@@ -0,0 +1,188 @@
+// test_cgal_phase8b_lite.cpp
+//
+// Phase 8b-Lite — Smoke tests for the four new CGAL-style entry functions
+// added on top of the Phase 8a MVP (`discrete_conformal_map_euclidean`).
+//
+// All entries are thin wrappers around the legacy Newton solvers; the
+// purpose of these tests is to verify:
+//   • the wrapper compiles + dispatches correctly
+//   • named parameters pass through (gradient_tolerance, max_iterations)
+//   • the returned Result struct contains the expected DOF vector
+//   • Newton convergence happens end-to-end via the public API
+
+#include <CGAL/Discrete_conformal_map.h>
+#include <CGAL/Discrete_circle_packing.h>
+#include <CGAL/Discrete_inversive_distance.h>
+#include <CGAL/Conformal_layout.h>
+
+#include "mesh_builder.hpp"
+#include "conformal_mesh.hpp"
+
+#include <gtest/gtest.h>
+#include <cmath>
+
+using namespace conformallab;
+
+namespace {
+
+// Mesh helper — closed regular tetrahedron, used for spherical / hyper-ideal /
+// circle-packing tests.
+inline ConformalMesh make_closed_tet() { return make_tetrahedron(); }
+
+// Open 3-face tetrahedron-minus-face, for layout testing.
+inline ConformalMesh make_open_3face()
+{
+    ConformalMesh mesh;
+    auto v0 = mesh.add_vertex(Point3( 1,  1,  1));
+    auto v1 = mesh.add_vertex(Point3( 1, -1, -1));
+    auto v2 = mesh.add_vertex(Point3(-1,  1, -1));
+    auto v3 = mesh.add_vertex(Point3(-1, -1,  1));
+    mesh.add_face(v0, v2, v1);
+    mesh.add_face(v0, v1, v3);
+    mesh.add_face(v0, v3, v2);
+    return mesh;
+}
+
+} // anonymous
+
+// ════════════════════════════════════════════════════════════════════════════
+// 1.  Spherical entry — closed genus-0 tetrahedron, natural-theta default
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CGALPhase8bLite, Spherical_ClosedTetrahedron_NaturalThetaConverges)
+{
+    auto mesh = make_closed_tet();
+    auto res  = CGAL::discrete_conformal_map_spherical(mesh);
+
+    EXPECT_TRUE(res.converged);
+    EXPECT_LT(res.gradient_norm, 1e-8);
+    EXPECT_EQ(res.u_per_vertex.size(), num_vertices(mesh));
+    // Natural-theta ⇒ u = 0 is the equilibrium ⇒ all values ≈ 0.
+    for (double u : res.u_per_vertex) EXPECT_NEAR(u, 0.0, 1e-8);
+}
+
+TEST(CGALPhase8bLite, Spherical_NamedParametersTakeEffect)
+{
+    auto mesh = make_closed_tet();
+    auto res  = CGAL::discrete_conformal_map_spherical(
+        mesh,
+        CGAL::parameters::max_iterations(0));
+    EXPECT_EQ(res.iterations, 0);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 2.  Hyper-ideal entry — wrapper compiles + runs, returns both b_v and a_e
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CGALPhase8bLite, HyperIdeal_Tetrahedron_ReturnsBothVertexAndEdgeDOFs)
+{
+    auto mesh = make_closed_tet();
+    auto res  = CGAL::discrete_conformal_map_hyper_ideal(
+        mesh,
+        CGAL::parameters::max_iterations(20));
+
+    // Newton on default targets (Θ=2π, θ=π) from the "natural" b=1, a=0.5
+    // start may or may not converge in 20 iterations — but the wrapper must
+    // populate the result struct in any case.
+    EXPECT_EQ(res.b_per_vertex.size(), num_vertices(mesh));
+    EXPECT_EQ(res.a_per_edge.size(),    num_edges   (mesh));
+    EXPECT_GE(res.iterations, 0);
+    EXPECT_TRUE(std::isfinite(res.gradient_norm));
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 3.  Circle-packing (face-based) entry — natural-phi convergence
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CGALPhase8bLite, CirclePacking_ClosedTetrahedron_NaturalPhiConverges)
+{
+    auto mesh = make_closed_tet();
+    auto res  = CGAL::discrete_circle_packing_euclidean(mesh);
+
+    EXPECT_TRUE(res.converged);
+    EXPECT_LT(res.gradient_norm, 1e-8);
+    EXPECT_EQ(res.rho_per_face.size(), num_faces(mesh));
+    // Pinned face is at index 0 (first iterated face); its ρ is 0 by gauge.
+    // After natural-phi the equilibrium is ρ_f = 0 for every face.
+    for (double r : res.rho_per_face) EXPECT_NEAR(r, 0.0, 1e-8);
+}
+
+TEST(CGALPhase8bLite, CirclePacking_GradientToleranceTakesEffect)
+{
+    auto mesh = make_closed_tet();
+    auto res_loose = CGAL::discrete_circle_packing_euclidean(
+        mesh,
+        CGAL::parameters::gradient_tolerance(1e-4));
+    EXPECT_TRUE(res_loose.converged);
+
+    auto mesh2 = make_closed_tet();
+    auto res_strict = CGAL::discrete_circle_packing_euclidean(
+        mesh2,
+        CGAL::parameters::gradient_tolerance(1e-12));
+    EXPECT_TRUE(res_strict.converged);
+    EXPECT_LT(res_strict.gradient_norm, 1e-10);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 4.  Inversive-distance (vertex-based) entry — natural-theta convergence
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CGALPhase8bLite, InversiveDistance_Triangle_NaturalThetaConverges)
+{
+    auto mesh = make_triangle();
+    auto res  = CGAL::discrete_inversive_distance_map(mesh);
+
+    EXPECT_TRUE(res.converged);
+    EXPECT_LT(res.gradient_norm, 1e-8);
+    EXPECT_EQ(res.u_per_vertex.size(), num_vertices(mesh));
+    for (double u : res.u_per_vertex) EXPECT_NEAR(u, 0.0, 1e-8);
+}
+
+TEST(CGALPhase8bLite, InversiveDistance_QuadStrip_NamedParametersWork)
+{
+    auto mesh = make_quad_strip();
+    // Named-parameter chaining (`a.b().c()`) is not currently supported on
+    // the package-local tags; pass one parameter per call instead.
+    auto res  = CGAL::discrete_inversive_distance_map(
+        mesh,
+        CGAL::parameters::max_iterations(50));
+    EXPECT_TRUE(res.converged);
+    EXPECT_LE(res.iterations, 50);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 5.  Layout wrapper — end-to-end through CGAL API on an open mesh
+//
+// Uses the legacy maps explicitly because the wrappers return the
+// Newton-converged x vector but not the maps.  This exercises that the
+// `CGAL::euclidean_layout` shim works as expected.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CGALPhase8bLite, Layout_EuclideanWrapper_RoundTrip)
+{
+    auto mesh = make_open_3face();
+
+    // Set up the maps + run Newton via the CGAL Euclidean entry.
+    auto res  = CGAL::discrete_conformal_map_euclidean(mesh);
+    ASSERT_TRUE(res.converged);
+
+    // The wrapper does its own DOF assignment internally; we re-fetch
+    // the (now-populated) EuclideanMaps from the mesh's property maps
+    // to feed the layout wrapper.
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    // Pin first vertex (mirrors the wrapper's gauge choice).
+    auto vit = mesh.vertices().begin();
+    maps.v_idx[*vit++] = -1;
+    int idx = 0;
+    for (; vit != mesh.vertices().end(); ++vit) maps.v_idx[*vit] = idx++;
+    std::vector<double> x(idx, 0.0);   // wrapper's natural-theta equilibrium
+
+    auto layout = CGAL::euclidean_layout(mesh, x, maps);
+    EXPECT_EQ(layout.uv.size(), num_vertices(mesh));
+    // All UVs finite — basic sanity that the layout ran.
+    for (auto& uv : layout.uv) {
+        EXPECT_TRUE(std::isfinite(uv.x()));
+        EXPECT_TRUE(std::isfinite(uv.y()));
+    }
+}

code/tests/cgal/test_cgal_traits_mvp.cpp

@@ -0,0 +1,246 @@
+// test_cgal_traits_mvp.cpp
+//
+// Phase 8 MVP — first tests for the new CGAL-style public API.
+//
+// Validates:
+//   1. Default_conformal_map_traits<Surface_mesh, K> compiles and
+//      provides all advertised types and property-map accessors.
+//   2. discrete_conformal_map_euclidean() runs end-to-end on a small mesh.
+//   3. Named-parameter overrides (gradient_tolerance, max_iterations)
+//      change the Newton behaviour as expected.
+//   4. The result agrees with the legacy newton_euclidean() at the
+//      same DOF assignment — proving the wrapper is non-destructive.
+//
+// These tests are the Phase 8 MVP acceptance probe.  Phase 9a
+// (Inversive-Distance) will become the next, deeper validation by
+// implementing a new functional against this same trait API.
+
+#include <CGAL/Conformal_map_traits.h>
+#include <CGAL/Discrete_conformal_map.h>
+#include <CGAL/Kernel_traits.h>
+
+#include "mesh_builder.hpp"   // make_triangle, make_quad_strip, make_tetrahedron
+#include "euclidean_functional.hpp"
+#include "newton_solver.hpp"
+
+#include <gtest/gtest.h>
+#include <cmath>
+
+using namespace conformallab;
+
+// ════════════════════════════════════════════════════════════════════════════
+// 1.  Traits class: compile-time type sanity
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CGALConformalTraits, DefaultTraitsTypes)
+{
+    using K     = CGAL::Simple_cartesian<double>;
+    using Mesh  = CGAL::Surface_mesh<K::Point_3>;
+    using Tr    = CGAL::Default_conformal_map_traits<Mesh, K>;
+
+    // FT comes from the kernel.
+    static_assert(std::is_same_v<typename Tr::FT, double>);
+
+    // Descriptors come from boost::graph_traits, not from Surface_mesh directly.
+    static_assert(std::is_same_v<typename Tr::Triangle_mesh, Mesh>);
+    static_assert(std::is_same_v<
+        typename Tr::Vertex_descriptor,
+        typename boost::graph_traits<Mesh>::vertex_descriptor>);
+
+    // Property-map types should match Surface_mesh::Property_map for the
+    // appropriate key.
+    static_assert(std::is_same_v<
+        typename Tr::Theta_pmap,
+        typename Mesh::template Property_map<typename Tr::Vertex_descriptor, double>>);
+    static_assert(std::is_same_v<
+        typename Tr::Vertex_index_pmap,
+        typename Mesh::template Property_map<typename Tr::Vertex_descriptor, int>>);
+    static_assert(std::is_same_v<
+        typename Tr::Lambda0_pmap,
+        typename Mesh::template Property_map<typename Tr::Edge_descriptor, double>>);
+
+    // Default kernel: Simple_cartesian<double>.
+    using TrDefault = CGAL::Default_conformal_map_traits<Mesh>;
+    static_assert(std::is_same_v<typename TrDefault::Kernel,
+                                 CGAL::Simple_cartesian<double>>);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 2.  Traits property-map accessors are non-destructive
+//
+// Setting up via the trait helpers and via setup_euclidean_maps() must yield
+// the same property map (Surface_mesh deduplicates by name).
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CGALConformalTraits, AccessorsReuseExistingMaps)
+{
+    using K   = CGAL::Simple_cartesian<double>;
+    using Mesh = CGAL::Surface_mesh<K::Point_3>;
+    using Tr  = CGAL::Default_conformal_map_traits<Mesh, K>;
+
+    auto mesh = make_triangle();
+    auto maps = setup_euclidean_maps(mesh);
+
+    auto theta_via_traits   = Tr::theta_map(mesh);
+    auto idx_via_traits     = Tr::vertex_index_map(mesh);
+    auto lambda0_via_traits = Tr::lambda0_map(mesh);
+
+    // Surface_mesh property maps with the same key type are equality-comparable
+    // by name lookup — accessing through the traits class must return the
+    // same map that setup_euclidean_maps() created.
+    EXPECT_EQ(theta_via_traits,   maps.theta_v);
+    EXPECT_EQ(idx_via_traits,     maps.v_idx);
+    EXPECT_EQ(lambda0_via_traits, maps.lambda0);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 3.  End-to-end: discrete_conformal_map_euclidean() on a small open mesh
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CGALDiscreteConformalMap, SingleTriangleConverges)
+{
+    auto mesh = make_triangle();
+    auto result = CGAL::discrete_conformal_map_euclidean(mesh);
+
+    EXPECT_TRUE(result.converged)
+        << "Newton did not converge on a single triangle";
+    EXPECT_LT(result.gradient_norm, 1e-8);
+    EXPECT_GE(result.iterations, 0);
+    EXPECT_EQ(result.u_per_vertex.size(), num_vertices(mesh));
+
+    // With the default flat-disc target curvature and the first vertex pinned,
+    // the natural-theta equilibrium is at u = 0 — Newton should accept x0=0.
+    for (double u : result.u_per_vertex)
+        EXPECT_NEAR(u, 0.0, 1e-8);
+}
+
+TEST(CGALDiscreteConformalMap, QuadStripConverges)
+{
+    auto mesh = make_quad_strip();
+    auto result = CGAL::discrete_conformal_map_euclidean(mesh);
+
+    EXPECT_TRUE(result.converged);
+    EXPECT_LT(result.gradient_norm, 1e-8);
+    EXPECT_EQ(result.u_per_vertex.size(), num_vertices(mesh));
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 4.  Named-parameter overrides take effect
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CGALDiscreteConformalMap, MaxIterationsTakesEffect)
+{
+    auto mesh = make_triangle();
+
+    // max_iterations(0) forces Newton to give up immediately.
+    auto result = CGAL::discrete_conformal_map_euclidean(
+        mesh,
+        CGAL::parameters::max_iterations(0));
+
+    EXPECT_EQ(result.iterations, 0);
+    // Trivial natural-theta case: gradient is already zero at x=0,
+    // so even 0 iterations may report "converged" depending on the
+    // initial gradient check.  The point is just that the parameter
+    // was *read* — verified by EXPECT_EQ on iterations above.
+}
+
+TEST(CGALDiscreteConformalMap, GradientToleranceTakesEffect)
+{
+    auto mesh = make_quad_strip();
+
+    // Loose tolerance — must still converge, but possibly in fewer steps.
+    auto result_loose = CGAL::discrete_conformal_map_euclidean(
+        mesh,
+        CGAL::parameters::gradient_tolerance(1e-4));
+    EXPECT_TRUE(result_loose.converged);
+
+    // Strict tolerance — also must converge, gradient norm must be tighter.
+    auto result_strict = CGAL::discrete_conformal_map_euclidean(
+        mesh,
+        CGAL::parameters::gradient_tolerance(1e-12));
+    EXPECT_TRUE(result_strict.converged);
+    EXPECT_LT(result_strict.gradient_norm, 1e-10);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 5.  Wrapper agrees with the legacy newton_euclidean() at the same setup
+//
+// This is the cross-API consistency check: same mesh, same default settings
+// (first vertex pinned, x0=0) — the u-vector returned by the wrapper must
+// match what newton_euclidean produces directly.
+// ════════════════════════════════════════════════════════════════════════════
+
+// ════════════════════════════════════════════════════════════════════════════
+// 6.  Kernel deduction: the wrapper must NOT hard-code Simple_cartesian
+//
+// Regression guard: the wrapper deduces its kernel from the mesh point type
+// via `CGAL::Kernel_traits`.  If anyone re-introduces a hard-coded
+// `Simple_cartesian<double>` in the wrapper, the static_asserts here still
+// pass (the legacy ConformalMesh uses that kernel) — but Phase 9a or any
+// user with a different kernel-backed Surface_mesh would fail to compile.
+// This test pins the deduction *contract* explicitly.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CGALDiscreteConformalMap, KernelIsDeducedFromMeshPointType)
+{
+    using Mesh = ConformalMesh;
+    using P    = typename Mesh::Point;
+
+    using DeducedKernel = typename CGAL::Kernel_traits<P>::Kernel;
+    using DeducedTraits = CGAL::Default_conformal_map_traits<Mesh, DeducedKernel>;
+
+    static_assert(std::is_same_v<DeducedKernel, CGAL::Simple_cartesian<double>>,
+                  "ConformalMesh point type must deduce to Simple_cartesian<double>");
+    static_assert(std::is_same_v<typename DeducedTraits::FT, double>);
+    static_assert(std::is_same_v<typename DeducedTraits::Triangle_mesh, Mesh>);
+
+    // Run-time sanity: the wrapper accepts the deduced-kernel mesh end-to-end.
+    auto mesh = make_quad_strip();
+    auto result = CGAL::discrete_conformal_map_euclidean(mesh);
+    EXPECT_TRUE(result.converged);
+}
+
+TEST(CGALDiscreteConformalMap, WrapperMatchesLegacyAPI)
+{
+    auto mesh = make_quad_strip();
+
+    // ── New API: applies natural-theta automatically ───────────────────────
+    auto result_new = CGAL::discrete_conformal_map_euclidean(mesh);
+
+    // ── Legacy API on a fresh mesh — must replicate the *same* preparation
+    //    that the wrapper performs internally (pin first vertex, assign
+    //    DOFs, apply natural-theta).  Otherwise the comparison is unfair
+    //    (Newton would diverge without natural-theta on these meshes). ────
+    auto mesh_legacy = make_quad_strip();
+    auto maps = setup_euclidean_maps(mesh_legacy);
+    compute_euclidean_lambda0_from_mesh(mesh_legacy, maps);
+
+    // Pin first vertex (gauge), assign sequential DOFs to the rest.
+    auto vit = mesh_legacy.vertices().begin();
+    maps.v_idx[*vit++] = -1;
+    int idx = 0;
+    for (; vit != mesh_legacy.vertices().end(); ++vit)
+        maps.v_idx[*vit] = idx++;
+
+    // Natural-theta: shift Θ so that x = 0 is the natural equilibrium.
+    std::vector<double> x0(idx, 0.0);
+    auto G0 = euclidean_gradient(mesh_legacy, x0, maps);
+    for (auto v : mesh_legacy.vertices()) {
+        int j = maps.v_idx[v];
+        if (j >= 0) maps.theta_v[v] -= G0[static_cast<std::size_t>(j)];
+    }
+
+    auto nr = newton_euclidean(mesh_legacy, x0, maps, 1e-10, 200);
+
+    // ── Compare ────────────────────────────────────────────────────────────
+    EXPECT_EQ(result_new.converged, nr.converged);
+    EXPECT_NEAR(result_new.gradient_norm, nr.grad_inf_norm, 1e-12);
+
+    // Pinned vertex u is 0 in both; for the rest the values agree.
+    for (auto v : mesh_legacy.vertices()) {
+        int j = maps.v_idx[v];
+        double u_legacy = (j >= 0) ? nr.x[static_cast<std::size_t>(j)] : 0.0;
+        EXPECT_NEAR(result_new.u_per_vertex[v.idx()], u_legacy, 1e-10)
+            << "Wrapper diverges from legacy for vertex " << v.idx();
+    }
+}

code/tests/cgal/test_conformal_mesh.cpp

@@ -0,0 +1,240 @@
+// test_conformal_mesh.cpp
+//
+// Phase 3a — CGAL Surface_mesh infrastructure tests.
+//
+// Verifies that ConformalMesh (CGAL::Surface_mesh<Point3>) and the
+// mesh_builder factories behave correctly before we build the functionals
+// on top of them (Phase 3b).
+//
+// Test groups
+// ───────────
+//   Topology     – vertex/edge/face counts, Euler characteristic
+//   Traversal    – halfedge iteration around vertex / face / edge
+//   PropertyMaps – read/write of lambda, theta, idx, alpha, f:type
+//   Validity     – all make_* factories produce valid, consistent meshes
+
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include <CGAL/boost/graph/iterator.h>
+#include <gtest/gtest.h>
+#include <cmath>
+
+using namespace conformallab;
+
+// ════════════════════════════════════════════════════════════
+// Topology
+// ════════════════════════════════════════════════════════════
+
+// Single triangle: 3 vertices, 1 face, 3 edges.
+TEST(ConformalMeshTopology, SingleTriangle)
+{
+    auto mesh = make_triangle();
+    EXPECT_EQ(3u, mesh.number_of_vertices());
+    EXPECT_EQ(1u, mesh.number_of_faces());
+    EXPECT_EQ(3u, mesh.number_of_edges());
+}
+
+// Tetrahedron: V=4, E=6, F=4  →  Euler = 2 (sphere topology).
+TEST(ConformalMeshTopology, TetrahedronEuler)
+{
+    auto mesh = make_tetrahedron();
+    EXPECT_EQ(4u, mesh.number_of_vertices());
+    EXPECT_EQ(6u, mesh.number_of_edges());
+    EXPECT_EQ(4u, mesh.number_of_faces());
+
+    int euler = (int)mesh.number_of_vertices()
+              - (int)mesh.number_of_edges()
+              + (int)mesh.number_of_faces();
+    EXPECT_EQ(2, euler) << "Euler characteristic of closed sphere must be 2";
+}
+
+// Two-triangle strip: V=4, E=5, F=2.
+// The interior edge (shared diagonal) has no border halfedge.
+TEST(ConformalMeshTopology, QuadStrip)
+{
+    auto mesh = make_quad_strip();
+    EXPECT_EQ(4u, mesh.number_of_vertices());
+    EXPECT_EQ(5u, mesh.number_of_edges());
+    EXPECT_EQ(2u, mesh.number_of_faces());
+
+    // Count interior (non-boundary) edges
+    int interior = 0;
+    for (auto e : mesh.edges())
+        if (!mesh.is_border(e)) ++interior;
+    EXPECT_EQ(1, interior) << "Only the shared diagonal should be interior";
+}
+
+// Fan with n triangles: V=n+1, E=2n, F=n.
+TEST(ConformalMeshTopology, FanCounts)
+{
+    for (int n : {3, 4, 6, 8}) {
+        auto mesh = make_fan(n);
+        EXPECT_EQ((std::size_t)(n + 1), mesh.number_of_vertices());
+        EXPECT_EQ((std::size_t)(2 * n),  mesh.number_of_edges());
+        EXPECT_EQ((std::size_t)(n),       mesh.number_of_faces());
+    }
+}
+
+// ════════════════════════════════════════════════════════════
+// Halfedge Traversal
+// ════════════════════════════════════════════════════════════
+
+// For a regular tetrahedron every vertex has valence 3.
+TEST(ConformalMeshTraversal, TetrahedronVertexValence)
+{
+    auto mesh = make_tetrahedron();
+    for (auto v : mesh.vertices()) {
+        int degree = 0;
+        for (auto h : CGAL::halfedges_around_target(v, mesh))
+            { (void)h; ++degree; }
+        EXPECT_EQ(3, degree) << "Each tetrahedron vertex has degree 3";
+    }
+}
+
+// For a fan with n triangles the center vertex has valence n.
+TEST(ConformalMeshTraversal, FanCenterValence)
+{
+    for (int n : {3, 5, 7}) {
+        auto mesh = make_fan(n);
+
+        // Center vertex is always the first one added (index 0).
+        auto center = *mesh.vertices().begin();
+        int degree = 0;
+        for (auto h : CGAL::halfedges_around_target(center, mesh))
+            { (void)h; ++degree; }
+        EXPECT_EQ(n, degree)
+            << "Fan center vertex must have valence == n=" << n;
+    }
+}
+
+// Every face of the tetrahedron has exactly 3 halfedges.
+TEST(ConformalMeshTraversal, FaceHalfedgeCount)
+{
+    auto mesh = make_tetrahedron();
+    for (auto f : mesh.faces()) {
+        int count = 0;
+        for (auto h : CGAL::halfedges_around_face(mesh.halfedge(f), mesh))
+            { (void)h; ++count; }
+        EXPECT_EQ(3, count) << "Each triangular face must have exactly 3 halfedges";
+    }
+}
+
+// opposite(h) and h share the same edge; opposite(opposite(h)) == h.
+TEST(ConformalMeshTraversal, OppositeHalfedgeConsistency)
+{
+    auto mesh = make_tetrahedron();
+    for (auto h : mesh.halfedges()) {
+        auto opp = mesh.opposite(h);
+        EXPECT_EQ(mesh.edge(h), mesh.edge(opp))
+            << "h and opposite(h) must share the same edge";
+        EXPECT_EQ(h, mesh.opposite(opp))
+            << "opposite(opposite(h)) must equal h";
+    }
+}
+
+// ════════════════════════════════════════════════════════════
+// Property Maps
+// ════════════════════════════════════════════════════════════
+
+// The conformal variable lambda can be written and read back per vertex.
+TEST(ConformalMeshProperties, VertexLambdaReadWrite)
+{
+    auto mesh = make_tetrahedron();
+    auto [lambda, theta, idx] = add_vertex_properties(mesh);
+
+    double value = 0.0;
+    for (auto v : mesh.vertices()) {
+        lambda[v] = value;
+        value    += 1.0;
+    }
+
+    value = 0.0;
+    for (auto v : mesh.vertices()) {
+        EXPECT_DOUBLE_EQ(value, lambda[v]);
+        value += 1.0;
+    }
+}
+
+// Default solver index is -1 (pinned); can be overwritten.
+TEST(ConformalMeshProperties, VertexSolverIndex)
+{
+    auto mesh = make_tetrahedron();
+    auto [lambda, theta, idx] = add_vertex_properties(mesh);
+
+    // All vertices start at -1 (pinned / boundary)
+    for (auto v : mesh.vertices())
+        EXPECT_EQ(-1, idx[v]) << "Default solver index must be -1";
+
+    // Assign sequential indices
+    int i = 0;
+    for (auto v : mesh.vertices())
+        idx[v] = i++;
+
+    i = 0;
+    for (auto v : mesh.vertices())
+        EXPECT_EQ(i++, idx[v]);
+}
+
+// Edge alpha (intersection angle): set and retrieve per edge.
+TEST(ConformalMeshProperties, EdgeAlpha)
+{
+    auto mesh = make_quad_strip();
+    auto alpha = add_edge_properties(mesh);
+
+    const double kAlpha = M_PI / 3.0;   // 60°
+    for (auto e : mesh.edges())
+        alpha[e] = kAlpha;
+
+    for (auto e : mesh.edges())
+        EXPECT_DOUBLE_EQ(kAlpha, alpha[e]);
+
+    EXPECT_EQ(5u, mesh.number_of_edges());
+}
+
+// Face geometry type: Euclidean by default, switchable to Hyperbolic.
+TEST(ConformalMeshProperties, FaceGeometryType)
+{
+    auto mesh = make_tetrahedron();
+    auto ftype = add_face_properties(mesh);
+
+    // Default: Euclidean
+    for (auto f : mesh.faces())
+        EXPECT_EQ(static_cast<int>(GeometryType::Euclidean), ftype[f]);
+
+    // Switch all to Hyperbolic
+    for (auto f : mesh.faces())
+        ftype[f] = static_cast<int>(GeometryType::Hyperbolic);
+
+    for (auto f : mesh.faces())
+        EXPECT_EQ(static_cast<int>(GeometryType::Hyperbolic), ftype[f]);
+}
+
+// Adding the same named property map twice: second call returns ok=false
+// and both handles alias the same storage.
+TEST(ConformalMeshProperties, PropertyMapIdempotent)
+{
+    auto mesh = make_triangle();
+    auto [pm1, ok1] = mesh.add_property_map<Vertex_index, double>("v:lambda", 0.0);
+    auto [pm2, ok2] = mesh.add_property_map<Vertex_index, double>("v:lambda", 0.0);
+
+    EXPECT_TRUE(ok1)  << "First add_property_map must succeed";
+    EXPECT_FALSE(ok2) << "Second add_property_map on existing name must return ok=false";
+
+    // Both handles must alias the same storage
+    auto v = *mesh.vertices().begin();
+    pm1[v] = 42.0;
+    EXPECT_DOUBLE_EQ(42.0, pm2[v]) << "Both handles must alias the same storage";
+}
+
+// ════════════════════════════════════════════════════════════
+// Mesh validity
+// ════════════════════════════════════════════════════════════
+
+// CGAL's built-in validity check must pass for all factory meshes.
+TEST(ConformalMeshValidity, AllBuilders)
+{
+    EXPECT_TRUE(make_triangle().is_valid())    << "triangle mesh invalid";
+    EXPECT_TRUE(make_tetrahedron().is_valid()) << "tetrahedron mesh invalid";
+    EXPECT_TRUE(make_quad_strip().is_valid())  << "quad strip mesh invalid";
+    EXPECT_TRUE(make_fan(6).is_valid())        << "fan-6 mesh invalid";
+}

code/tests/cgal/test_cp_euclidean_functional.cpp

@@ -0,0 +1,267 @@
+// test_cp_euclidean_functional.cpp
+//
+// Phase 9a.1 — CPEuclideanFunctional (BPS 2010) tests.
+//
+// Replicates de.varylab.discreteconformal.functional.CPEuclideanFunctionalTest
+// (88 lines) and adds boundary-edge coverage plus a closed-mesh case.
+//
+// Java test pattern (lines 50-87):
+//   1. Build dodecahedron via HalfEdgeUtils.addDodecahedron.
+//   2. Remove face 0 to produce an open mesh.
+//   3. theta_e = π/2 for every edge.       (orthogonal circle packing)
+//   4. phi_f   = 2π for every face.        (flat target)
+//   5. Random ρ ∈ [−0.5, 0.5] (seed 1).
+//   6. FunctionalTest.setXGradient(ρ) → FD-vs-analytic gradient check.
+//   7. FunctionalTest.setXHessian(ρ)  → FD-vs-analytic Hessian  check.
+//
+// C++ port uses the tetrahedron (4 faces) instead of the dodecahedron (12 faces)
+// because the analytic structure is identical and the smaller mesh keeps the
+// test fast and human-inspectable.  We exercise the boundary-edge code path
+// by additionally testing a tetrahedron with one face removed (3 faces, 3
+// boundary edges, 3 interior edges).
+
+#include "cp_euclidean_functional.hpp"
+#include "mesh_builder.hpp"
+#include "conformal_mesh.hpp"
+
+#include <Eigen/Eigenvalues>
+#include <gtest/gtest.h>
+#include <vector>
+#include <random>
+
+using namespace conformallab;
+
+// ════════════════════════════════════════════════════════════════════════════
+// 1.  Helper: explicit values for p(θ*, Δρ) at known inputs
+//
+// p(θ*, 0) = 0          (tanh 0 = 0)
+// p(π, Δρ) = π·sign(Δρ) (tan(π/2) = ∞, atan saturates to ±π/2)
+// p(0, Δρ) = 0          (tan(0) = 0)
+// p odd in Δρ (tanh is odd).
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CPEuclideanFunctional, PFunctionKnownValues)
+{
+    using cp_detail::p_function;
+    constexpr double PI_ = 3.14159265358979323846;
+
+    // p(any, 0) = 0
+    EXPECT_NEAR(p_function(PI_ / 4,  0.0), 0.0, 1e-15);
+    EXPECT_NEAR(p_function(PI_ / 2,  0.0), 0.0, 1e-15);
+
+    // Odd in Δρ
+    const double thStar = PI_ / 3;
+    for (double dr : {0.1, 0.5, 1.0, 2.0}) {
+        EXPECT_NEAR(p_function(thStar, dr) + p_function(thStar, -dr), 0.0, 1e-12)
+            << "p(θ*, Δρ) should be odd in Δρ";
+    }
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 2.  Property-map setup defaults
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CPEuclideanFunctional, SetupDefaults)
+{
+    auto mesh = make_tetrahedron();
+    auto m = setup_cp_euclidean_maps(mesh);
+
+    constexpr double PI_ = 3.14159265358979323846;
+    for (auto e : mesh.edges()) EXPECT_NEAR(m.theta_e[e], PI_ / 2, 1e-15);
+    for (auto f : mesh.faces()) EXPECT_NEAR(m.phi_f[f],   2.0 * PI_, 1e-15);
+    for (auto f : mesh.faces()) EXPECT_EQ(m.f_idx[f], -1) << "all faces start pinned";
+}
+
+TEST(CPEuclideanFunctional, AssignDofIndices_PinsOneFace)
+{
+    auto mesh = make_tetrahedron();
+    auto m    = setup_cp_euclidean_maps(mesh);
+    const int n = assign_cp_euclidean_face_dof_indices(mesh, m);
+
+    EXPECT_EQ(n, 3) << "tetrahedron has 4 faces; 1 pinned ⇒ 3 free DOFs";
+
+    int pinned_count = 0;
+    int max_idx = -1;
+    for (auto f : mesh.faces()) {
+        if (m.f_idx[f] == -1) ++pinned_count;
+        else max_idx = std::max(max_idx, m.f_idx[f]);
+    }
+    EXPECT_EQ(pinned_count, 1);
+    EXPECT_EQ(max_idx, 2);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 3.  Tangential limit (θ = 0): p = 0, energy collapses, gradient = φ_f
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CPEuclideanFunctional, TangentialLimitGradientEqualsPhi)
+{
+    auto mesh = make_tetrahedron();
+    auto m    = setup_cp_euclidean_maps(mesh);
+    for (auto e : mesh.edges()) m.theta_e[e] = 0.0;   // tangential limit
+    const int n = assign_cp_euclidean_face_dof_indices(mesh, m);
+
+    // At θ = 0: θ* = π.  Interior edge contribution: −(p+θ*) where p = π·sign(Δρ).
+    // Boundary contribution: −2π.  At ρ = 0, Δρ = 0 so p = 0; each interior face
+    // contributes −π per incident interior halfedge; for a tetrahedron each face
+    // has 3 interior halfedges ⇒ −3π.  Net gradient: 2π − 3π = −π per free face.
+    std::vector<double> x(static_cast<std::size_t>(n), 0.0);
+    auto G = cp_euclidean_gradient(mesh, x, m);
+
+    constexpr double PI_ = 3.14159265358979323846;
+    for (double g : G) EXPECT_NEAR(g, -PI_, 1e-10);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 4.  FD gradient check on closed tetrahedron at random ρ
+//
+// Java parity: this is exactly the structure of CPEuclideanFunctionalTest.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CPEuclideanFunctional, FDGradientCheck_ClosedTetrahedron_RandomRho)
+{
+    auto mesh = make_tetrahedron();
+    auto m    = setup_cp_euclidean_maps(mesh);
+    const int n = assign_cp_euclidean_face_dof_indices(mesh, m);
+
+    // Java: rnd.setSeed(1);  rho_i = rnd.nextDouble() − 0.5
+    std::mt19937 rng(1);
+    std::uniform_real_distribution<double> u(-0.5, 0.5);
+    std::vector<double> rho(static_cast<std::size_t>(n));
+    for (auto& r : rho) r = u(rng);
+
+    EXPECT_TRUE(gradient_check_cp_euclidean(mesh, rho, m))
+        << "FD vs analytic gradient mismatch on closed tetrahedron";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 5.  FD Hessian check on closed tetrahedron at random ρ
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CPEuclideanFunctional, FDHessianCheck_ClosedTetrahedron_RandomRho)
+{
+    auto mesh = make_tetrahedron();
+    auto m    = setup_cp_euclidean_maps(mesh);
+    const int n = assign_cp_euclidean_face_dof_indices(mesh, m);
+
+    std::mt19937 rng(1);
+    std::uniform_real_distribution<double> u(-0.5, 0.5);
+    std::vector<double> rho(static_cast<std::size_t>(n));
+    for (auto& r : rho) r = u(rng);
+
+    EXPECT_TRUE(hessian_check_cp_euclidean(mesh, rho, m))
+        << "FD vs analytic Hessian mismatch on closed tetrahedron";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 6.  Boundary-edge coverage: open mesh (tetrahedron with one face removed)
+//
+// Java test does this via `hds.removeFace(hds.getFace(0))`.  In CGAL we get
+// an equivalent open mesh by skipping the construction of one face.
+// ════════════════════════════════════════════════════════════════════════════
+
+inline ConformalMesh make_open_tetrahedron()
+{
+    ConformalMesh mesh;
+    auto v0 = mesh.add_vertex(Point3( 1,  1,  1));
+    auto v1 = mesh.add_vertex(Point3( 1, -1, -1));
+    auto v2 = mesh.add_vertex(Point3(-1,  1, -1));
+    auto v3 = mesh.add_vertex(Point3(-1, -1,  1));
+    // Three faces (omit the one opposite v0):
+    mesh.add_face(v0, v2, v1);
+    mesh.add_face(v0, v1, v3);
+    mesh.add_face(v0, v3, v2);
+    return mesh;
+}
+
+TEST(CPEuclideanFunctional, FDGradientCheck_OpenTetrahedron_RandomRho)
+{
+    auto mesh = make_open_tetrahedron();
+    auto m    = setup_cp_euclidean_maps(mesh);
+    const int n = assign_cp_euclidean_face_dof_indices(mesh, m);
+
+    EXPECT_EQ(n, 2);   // 3 faces, 1 pinned ⇒ 2 free DOFs
+
+    std::mt19937 rng(1);
+    std::uniform_real_distribution<double> u(-0.5, 0.5);
+    std::vector<double> rho(static_cast<std::size_t>(n));
+    for (auto& r : rho) r = u(rng);
+
+    EXPECT_TRUE(gradient_check_cp_euclidean(mesh, rho, m))
+        << "FD vs analytic gradient mismatch on open tetrahedron";
+}
+
+TEST(CPEuclideanFunctional, FDHessianCheck_OpenTetrahedron_RandomRho)
+{
+    auto mesh = make_open_tetrahedron();
+    auto m    = setup_cp_euclidean_maps(mesh);
+    const int n = assign_cp_euclidean_face_dof_indices(mesh, m);
+
+    std::mt19937 rng(1);
+    std::uniform_real_distribution<double> u(-0.5, 0.5);
+    std::vector<double> rho(static_cast<std::size_t>(n));
+    for (auto& r : rho) r = u(rng);
+
+    EXPECT_TRUE(hessian_check_cp_euclidean(mesh, rho, m))
+        << "FD vs analytic Hessian mismatch on open tetrahedron";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 7.  Hessian is symmetric positive-semidefinite (BPS-2010 §6 convexity)
+//
+// The energy is convex in ρ on its domain of validity.  Hence H is PSD with
+// a 1-dim null space (constant shift of all ρ, removed by gauge pin).
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CPEuclideanFunctional, HessianIsPSD)
+{
+    auto mesh = make_tetrahedron();
+    auto m    = setup_cp_euclidean_maps(mesh);
+    const int n = assign_cp_euclidean_face_dof_indices(mesh, m);
+
+    std::vector<double> rho(static_cast<std::size_t>(n), 0.1);
+    auto H = cp_euclidean_hessian(mesh, rho, m);
+
+    // Symmetry
+    Eigen::MatrixXd Hd(H);
+    EXPECT_NEAR((Hd - Hd.transpose()).cwiseAbs().maxCoeff(), 0.0, 1e-15);
+
+    // Smallest eigenvalue ≥ 0 (PSD)
+    Eigen::SelfAdjointEigenSolver<Eigen::MatrixXd> es(Hd);
+    EXPECT_GE(es.eigenvalues().minCoeff(), -1e-12)
+        << "Hessian must be PSD (BPS-2010 §6)";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 8.  At equilibrium (Newton-converged ρ*), the gradient is zero by construction
+//
+// We do not run a full Newton solver here; we set up the "natural-theta" trick:
+// adjust φ_f so that ρ = 0 is the equilibrium.  This is the analog of the
+// natural-theta convention already used in euclidean_functional tests
+// (see test_euclidean_functional.cpp lines 159-189).
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CPEuclideanFunctional, NaturalPhiMakesZeroTheEquilibrium)
+{
+    auto mesh = make_tetrahedron();
+    auto m    = setup_cp_euclidean_maps(mesh);
+    const int n = assign_cp_euclidean_face_dof_indices(mesh, m);
+
+    std::vector<double> rho(static_cast<std::size_t>(n), 0.0);
+
+    // Step 1: gradient at ρ = 0 with default φ.
+    auto G0 = cp_euclidean_gradient(mesh, rho, m);
+
+    // Step 2: adjust φ_f so the new gradient at ρ = 0 is zero.
+    //   ∂E/∂ρ_f = φ_f − (sum of edge contributions)
+    //   To zero G_f: subtract G_f from φ_f.
+    for (auto f : mesh.faces()) {
+        int i = m.f_idx[f];
+        if (i < 0) continue;
+        m.phi_f[f] -= G0[static_cast<std::size_t>(i)];
+    }
+
+    // Step 3: gradient at ρ = 0 should now be ~zero.
+    auto G_eq = cp_euclidean_gradient(mesh, rho, m);
+    for (double g : G_eq) EXPECT_NEAR(g, 0.0, 1e-13);
+}

code/tests/cgal/test_euclidean_functional.cpp

@@ -0,0 +1,289 @@
+// test_euclidean_functional.cpp
+//
+// Phase 3d — EuclideanCyclicFunctional ported to ConformalMesh.
+//
+// Corresponds to de.varylab.discreteconformal.functional.EuclideanCyclicFunctionalTest.
+//
+// Test map (Java → C++)
+// ──────────────────────
+//   testHessian (Ignored)       → GradientCheck_Hessian           (ported)
+//   testGradient…Triangle       → GradientCheck_TriangleVertex     (ported)
+//   testGradient…QuadStrip      → GradientCheck_QuadStripVertex    (ported)
+//   testGradient…Tetrahedron    → GradientCheck_TetrahedronVertex  (ported)
+//   testGradient…AllDofs        → GradientCheck_TetrahedronAllDofs (ported)
+//   testFunctionalAtNaNValue    → AnglesFiniteAtKnownPoint          (ported)
+//
+// Energy model
+// ────────────
+//   Uses the Schläfli path integral E(x) = ∫₀¹⟨G(tx),x⟩dt (10-point GL).
+//   The gradient check verifies G is curl-free.
+
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include "euclidean_geometry.hpp"
+#include "euclidean_functional.hpp"
+#include "euclidean_hessian.hpp"
+#include <gtest/gtest.h>
+#include <cmath>
+#include <vector>
+
+using namespace conformallab;
+
+// ════════════════════════════════════════════════════════════════════════════
+// Cross-module Hessian check: euclidean_gradient() ↔ euclidean_hessian()
+//
+// Java @Ignore reason: "no Hessian implemented yet" — the Java functional
+// test was written before the Hessian existed.  In C++ the analytic
+// cotangent-Laplace Hessian (euclidean_hessian.hpp, Phase 3f) is complete.
+//
+// This test verifies cross-module consistency:
+//   H[i,j] ≈ (G_i(x+ε·eⱼ) − G_i(x−ε·eⱼ)) / (2ε)
+// using the gradient from euclidean_functional.hpp and the Hessian from
+// euclidean_hessian.hpp.  A bug in DOF-index mapping or sign convention
+// that affects both modules independently would only be caught here.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanFunctional, GradientCheck_Hessian)
+{
+    auto mesh = make_triangle();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    int n = assign_euclidean_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), -0.1);
+
+    // hessian_check_euclidean: H[i,j] ≈ FD(G)[i,j] using euclidean_gradient()
+    EXPECT_TRUE(hessian_check_euclidean(mesh, x, maps))
+        << "Cross-module: euclidean_gradient() and euclidean_hessian() are inconsistent";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Angle formula: equilateral triangle → all angles = π/3
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanFunctional, EquilateralTriangleAnglesArePiOver3)
+{
+    // All sides equal: l = 1.0, log-length = 0.
+    auto fa = euclidean_angles(0.0, 0.0, 0.0);
+
+    ASSERT_TRUE(fa.valid) << "Equilateral triangle must be valid";
+    constexpr double PI_3 = 3.14159265358979323846 / 3.0;
+    EXPECT_NEAR(fa.alpha1, PI_3, 1e-12);
+    EXPECT_NEAR(fa.alpha2, PI_3, 1e-12);
+    EXPECT_NEAR(fa.alpha3, PI_3, 1e-12);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Angle formula: right isosceles triangle (legs 1, hypotenuse √2)
+//
+// For the 45-45-90 triangle: angles are π/4, π/4, π/2.
+// From make_triangle default (v0=(0,0), v1=(1,0), v2=(0,1)):
+//   e01: l=1, λ°=0
+//   e12: l=√2, λ°=log(2)
+//   e02: l=1, λ°=0
+// Angle at v0 (opposite e12) = π/2.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanFunctional, RightIsoscelesTriangleAnglesCorrect)
+{
+    const double log2 = std::log(2.0);
+    // lam12 = 0 (v0-v1, length 1), lam23 = log(2) (v1-v2, length √2), lam31 = 0 (v2-v0, length 1)
+    // v1 = v0 in our ordering → remap: l01=1, l12=√2, l20=1
+    // Using euclidean_angles(lam_v1v2, lam_v2v3, lam_v3v1):
+    //   v1=(0,0), v2=(1,0), v3=(0,1)
+    //   lam12 = log(1²) = 0, lam23 = log(√2 ²) = log2, lam31 = log(1²) = 0
+    auto fa = euclidean_angles(0.0, log2, 0.0);
+
+    ASSERT_TRUE(fa.valid);
+    constexpr double PI   = 3.14159265358979323846;
+    // v1=(0,0) is at the right-angle corner (opposite the hypotenuse l23=√2) → α1 = 90°.
+    // v2=(1,0) and v3=(0,1) are the 45° corners (each opposite a leg of length 1).
+    EXPECT_NEAR(fa.alpha1, PI / 2.0, 1e-12);   // angle at v1 (opposite l23=√2): 90°
+    EXPECT_NEAR(fa.alpha2, PI / 4.0, 1e-12);   // angle at v2 (opposite l31=1):  45°
+    EXPECT_NEAR(fa.alpha3, PI / 4.0, 1e-12);   // angle at v3 (opposite l12=1):  45°
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Angle sum = π for any valid Euclidean triangle
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanFunctional, AngleSumEqualsPi)
+{
+    // Scalene triangle with log-lengths (0, 0.5, -0.3).
+    auto fa = euclidean_angles(0.0, 0.5, -0.3);
+
+    ASSERT_TRUE(fa.valid);
+    constexpr double PI = 3.14159265358979323846;
+    EXPECT_NEAR(fa.alpha1 + fa.alpha2 + fa.alpha3, PI, 1e-12);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Degenerate triangle → valid = false
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanFunctional, DegenerateTriangleReturnsFalse)
+{
+    // l12 = l23 = 1, l31 = 3 → violates triangle inequality.
+    auto fa = euclidean_angles_from_lengths(1.0, 1.0, 3.0);
+    EXPECT_FALSE(fa.valid);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Gradient check: default right-isosceles triangle, vertex DOFs only
+//
+// Mirrors Java testGradient…SingleTriangle.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanFunctional, GradientCheck_TriangleVertex)
+{
+    auto mesh = make_triangle();   // (0,0)–(1,0)–(0,1)
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    int n = assign_euclidean_vertex_dof_indices(mesh, maps);
+
+    // Small uniform conformal perturbation.
+    std::vector<double> x(static_cast<std::size_t>(n), -0.1);
+
+    EXPECT_TRUE(gradient_check_euclidean(mesh, x, maps))
+        << "Gradient check failed on right-isosceles triangle (vertex DOFs)";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Gradient check: quad strip (2 triangles, 1 interior edge), vertex DOFs only
+//
+// Mirrors Java testGradient…QuadStrip / testGradientInExtendedDomain.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanFunctional, GradientCheck_QuadStripVertex)
+{
+    auto mesh = make_quad_strip();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    int n = assign_euclidean_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), -0.2);
+
+    EXPECT_TRUE(gradient_check_euclidean(mesh, x, maps))
+        << "Gradient check failed on quad strip (vertex DOFs)";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Gradient check: regular tetrahedron, vertex DOFs only
+//
+// Closed surface (4 faces, 4 vertices, 6 interior edges).
+// Exercises per-vertex angle-sum accumulation on multiple faces.
+// Mirrors Java testGradient…Tetrahedron / testGradientWithHyperIdeal…
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanFunctional, GradientCheck_TetrahedronVertex)
+{
+    auto mesh = make_tetrahedron();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    int n = assign_euclidean_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), -0.15);
+
+    EXPECT_TRUE(gradient_check_euclidean(mesh, x, maps))
+        << "Gradient check failed on regular tetrahedron (vertex DOFs)";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Gradient check: tetrahedron, all DOFs (vertex + edge)
+//
+// Exercises the edge-gradient branch G_e = α_opp⁺ + α_opp⁻ − π.
+// Mirrors Java testGradientWithHyperellipticCurve.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanFunctional, GradientCheck_TetrahedronAllDofs)
+{
+    auto mesh = make_tetrahedron();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    int n = assign_euclidean_all_dof_indices(mesh, maps);
+
+    // 4 vertex DOFs + 6 edge DOFs = 10 total.
+    std::vector<double> x(static_cast<std::size_t>(n), 0.0);
+    // Set vertex DOFs slightly negative to keep triangles non-degenerate.
+    for (int i = 0; i < 4; ++i)
+        x[static_cast<std::size_t>(i)] = -0.15;
+
+    EXPECT_TRUE(gradient_check_euclidean(mesh, x, maps))
+        << "Gradient check failed on regular tetrahedron (all DOFs)";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Angles are finite at a known interior point
+//
+// Mirrors Java testFunctionalAtNaNValue: stress-test the angle formula with
+// large negative conformal factors (compressed triangle) to ensure no NaN/Inf.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanFunctional, AnglesFiniteAtKnownPoint)
+{
+    auto mesh = make_tetrahedron();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    int n = assign_euclidean_vertex_dof_indices(mesh, maps);
+
+    // Very compressed: u_i = -3 (all sides shrunk by exp(-3) ≈ 0.05).
+    // Triangle stays well-formed (equilateral shrinks uniformly).
+    std::vector<double> x(static_cast<std::size_t>(n), -3.0);
+    auto G = euclidean_gradient(mesh, x, maps);
+
+    for (std::size_t i = 0; i < G.size(); ++i) {
+        EXPECT_FALSE(std::isnan(G[i])) << "Gradient component " << i << " is NaN";
+        EXPECT_FALSE(std::isinf(G[i])) << "Gradient component " << i << " is Inf";
+    }
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Gradient check: fan of 5 flat triangles, vertex DOFs only
+//
+// High-valence central vertex: exercises per-vertex angle accumulation
+// across 5 incident faces.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanFunctional, GradientCheck_Fan5Vertex)
+{
+    auto mesh = make_fan(5);
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    int n = assign_euclidean_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), -0.05);
+
+    EXPECT_TRUE(gradient_check_euclidean(mesh, x, maps))
+        << "Gradient check failed on flat fan-5 mesh";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Gradient check: mixed pinned/variable vertices
+//
+// Pins the first vertex (u_v0 = 0 fixed), lets the rest be variable.
+// Verifies that the gradient accumulator skips pinned vertices correctly.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanFunctional, GradientCheck_MixedPinnedVertices)
+{
+    auto mesh = make_quad_strip();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    // Manually pin v0; assign v1, v2, v3 as DOFs 0, 1, 2.
+    auto vit = mesh.vertices().begin();
+    Vertex_index v0 = *vit++;
+    Vertex_index v1 = *vit++;
+    Vertex_index v2 = *vit++;
+    Vertex_index v3 = *vit;
+
+    maps.v_idx[v0] = -1;   // pinned
+    maps.v_idx[v1] = 0;
+    maps.v_idx[v2] = 1;
+    maps.v_idx[v3] = 2;
+
+    std::vector<double> x = {-0.1, -0.3, -0.2};
+
+    EXPECT_TRUE(gradient_check_euclidean(mesh, x, maps))
+        << "Gradient check failed for mixed pinned/variable vertices";
+}

code/tests/cgal/test_euclidean_hessian.cpp

@@ -0,0 +1,213 @@
+// test_euclidean_hessian.cpp
+//
+// Phase 3f — Euclidean cotangent-Laplace Hessian.
+//
+// The Hessian of the Euclidean discrete conformal energy is the well-known
+// cotangent-Laplace operator (Pinkall–Polthier 1993, Springborn 2008).
+//
+// Tests:
+//   1. Cotangent weights are analytically correct for simple triangles.
+//   2. Hessian is symmetric.
+//   3. Hessian has the null-space property H·1 = 0 (uniform-shift mode).
+//   4. Hessian is positive semi-definite (all eigenvalues ≥ 0).
+//   5. Finite-difference check H[i,j] ≈ (G_i(x+ε·eⱼ)−G_i(x−ε·eⱼ))/(2ε).
+//
+// All tests use meshes and maps built with Phase-3d infrastructure.
+
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include "euclidean_hessian.hpp"
+#include <gtest/gtest.h>
+#include <Eigen/Dense>   // for dense conversion and eigenvalue solver
+#include <cmath>
+#include <vector>
+
+using namespace conformallab;
+
+// ════════════════════════════════════════════════════════════════════════════
+// Cotangent weight: equilateral triangle → all cots = 1/√3 = cot(60°)
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanHessian, CotWeights_EquilateralTriangle)
+{
+    // Equilateral triangle with l = 1 (all log-lengths = 0).
+    auto cw = euclidean_cot_weights(1.0, 1.0, 1.0);
+
+    ASSERT_TRUE(cw.valid);
+    const double expected = 1.0 / std::sqrt(3.0);   // cot(60°)
+    EXPECT_NEAR(cw.cot1, expected, 1e-12);
+    EXPECT_NEAR(cw.cot2, expected, 1e-12);
+    EXPECT_NEAR(cw.cot3, expected, 1e-12);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Cotangent weight: right-isosceles triangle (legs 1, hypotenuse √2)
+//
+//   v1=(0,0): right angle → cot(90°) = 0
+//   v2=(1,0), v3=(0,1): 45° angles → cot(45°) = 1
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanHessian, CotWeights_RightIsoscelesTriangle)
+{
+    // l12=1, l23=√2, l31=1
+    auto cw = euclidean_cot_weights(1.0, std::sqrt(2.0), 1.0);
+
+    ASSERT_TRUE(cw.valid);
+    EXPECT_NEAR(cw.cot1, 0.0, 1e-12);   // right angle at v1
+    EXPECT_NEAR(cw.cot2, 1.0, 1e-12);   // 45° at v2
+    EXPECT_NEAR(cw.cot3, 1.0, 1e-12);   // 45° at v3
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Hessian is symmetric: H[i,j] == H[j,i]
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanHessian, HessianIsSymmetric)
+{
+    auto mesh = make_quad_strip();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    int n = assign_euclidean_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), -0.1);
+    auto H = euclidean_hessian(mesh, x, maps);
+
+    Eigen::MatrixXd Hd = Eigen::MatrixXd(H);
+    EXPECT_NEAR((Hd - Hd.transpose()).norm(), 0.0, 1e-12)
+        << "Hessian must be symmetric";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Null-space property: H·1 = 0 for a closed surface (regular tetrahedron)
+//
+// The cotangent Laplacian on a closed mesh has the constant vector in its
+// null space (each row sums to zero).
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanHessian, NullSpaceIsConstantVector_ClosedMesh)
+{
+    auto mesh = make_tetrahedron();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    int n = assign_euclidean_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), 0.0);
+    auto H = euclidean_hessian(mesh, x, maps);
+
+    // 1-vector
+    Eigen::VectorXd ones = Eigen::VectorXd::Ones(n);
+    Eigen::VectorXd Hones = H * ones;
+
+    EXPECT_NEAR(Hones.norm(), 0.0, 1e-10)
+        << "H·1 must be zero on a closed mesh (cotangent Laplacian null-space)";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Hessian is positive semi-definite: all eigenvalues ≥ 0
+//
+// Checked on a small mesh (regular tetrahedron, 4 vertices) using dense
+// self-adjoint eigenvalue decomposition (only feasible for small n).
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanHessian, HessianIsPositiveSemiDefinite)
+{
+    auto mesh = make_tetrahedron();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    int n = assign_euclidean_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), 0.0);
+    auto H = euclidean_hessian(mesh, x, maps);
+
+    Eigen::MatrixXd Hd = Eigen::MatrixXd(H);
+    Eigen::SelfAdjointEigenSolver<Eigen::MatrixXd> es(Hd);
+    double min_ev = es.eigenvalues().minCoeff();
+
+    EXPECT_GE(min_ev, -1e-10)
+        << "All eigenvalues of the cotangent Laplacian must be ≥ 0; "
+           "smallest = " << min_ev;
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Finite-difference Hessian check: single right-isosceles triangle
+//
+// H[i,j] ≈ (G_i(x+ε·eⱼ) − G_i(x−ε·eⱼ)) / (2ε)
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanHessian, FDCheck_Triangle)
+{
+    auto mesh = make_triangle();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    int n = assign_euclidean_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), -0.1);
+
+    EXPECT_TRUE(hessian_check_euclidean(mesh, x, maps))
+        << "FD Hessian check failed on right-isosceles triangle";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Finite-difference Hessian check: quad strip (2 triangles, 1 interior edge)
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanHessian, FDCheck_QuadStrip)
+{
+    auto mesh = make_quad_strip();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    int n = assign_euclidean_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), -0.1);
+
+    EXPECT_TRUE(hessian_check_euclidean(mesh, x, maps))
+        << "FD Hessian check failed on quad strip";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Finite-difference Hessian check: regular tetrahedron (closed, 4 faces)
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanHessian, FDCheck_Tetrahedron)
+{
+    auto mesh = make_tetrahedron();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    int n = assign_euclidean_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), -0.15);
+
+    EXPECT_TRUE(hessian_check_euclidean(mesh, x, maps))
+        << "FD Hessian check failed on regular tetrahedron";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Finite-difference Hessian check: with mixed pinned/variable vertices
+//
+// One vertex pinned: the corresponding row/column must be absent from H
+// while the diagonal of neighbouring variable vertices still gets the full
+// cotangent contribution.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanHessian, FDCheck_MixedPinnedVertices)
+{
+    auto mesh = make_quad_strip();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    auto vit = mesh.vertices().begin();
+    Vertex_index v0 = *vit++;
+    Vertex_index v1 = *vit++;
+    Vertex_index v2 = *vit++;
+    Vertex_index v3 = *vit;
+
+    maps.v_idx[v0] = -1;   // pinned
+    maps.v_idx[v1] = 0;
+    maps.v_idx[v2] = 1;
+    maps.v_idx[v3] = 2;
+
+    std::vector<double> x = {-0.1, -0.2, -0.15};
+
+    EXPECT_TRUE(hessian_check_euclidean(mesh, x, maps))
+        << "FD Hessian check failed for mixed pinned/variable vertices";
+}

code/tests/cgal/test_geometry_utils.cpp

@@ -0,0 +1,510 @@
+// test_geometry_utils.cpp
+//
+// Port of the Java ConformalLab geometry utility tests.
+//
+// Java source                           Java test method                         Status
+// ─────────────────────────────────────────────────────────────────────────────────────
+// CuttinUtilityTest.java                testIsInConvexTextureFace_False          PORTED
+// CuttinUtilityTest.java                testIsInConvexTextureFace_True           PORTED
+// UnwrapUtilityTest.java                testGetAngleReturnsPI                    PORTED
+// ConvergenceUtilityTests.java          testGetTextureCircumRadius               PORTED
+// ConvergenceUtilityTests.java          testGetTextureTriangleArea               PORTED
+// ConvergenceUtilityTests.java          testScaleInvariantCircumCircleRadius     PORTED
+// HomologyTest.java                     testHomology                             PORTED
+// EuclideanLayoutTest.java              testDoLayout                             PORTED
+// EuclideanCyclicConvergenceTest.java   testEuclideanConvergence                 PORTED
+// SphericalConvergenceTest.java         testSphericalConvergence                 PORTED
+//
+// ─── Geometric background ────────────────────────────────────────────────────────────
+//
+//  Tests 1–2  Point-in-convex-triangle (2D UV space, barycentric sign method)
+//             Java: CuttingUtility.isInConvexTextureFace(pp, face, adapters)
+//             Note: Java test 2 has a 5-element T-array with w=0 (point at
+//             infinity), which is a typo in the original. Equivalent, well-formed
+//             coordinates are used here instead.
+//
+//  Test 3     Corner angle for collinear vertices via the law of cosines.
+//             Java: UnwrapUtility.getAngle(edge, adapters) — returns the angle at
+//             the target vertex. For v0=(-1,0,0), v1=(0,0,0), v2=(1,0,0) the
+//             angle at v1 is exactly π (degenerate triangle inequality).
+//
+//  Tests 4–5  2D circumradius and triangle area.
+//             Java: ConvergenceUtility.getTextureCircumCircleRadius(face)
+//                   ConvergenceUtility.getTextureTriangleArea(face)
+//             Formulas: Area = |det([B-A, C-A])| / 2
+//                       R    = (a·b·c) / (4·Area)
+//
+//  Test 6     Scale-invariant circumradius over a mesh.
+//             Java: ConvergenceUtility.getMaxMeanSumScaleInvariantCircumRadius(hds)
+//             Returns [max, mean, sum] of  R_f / sqrt(total_texture_area).
+//             Invariant under uniform scaling of texture coordinates (tested with
+//             homogeneous weight w: position = (T[0]/w, T[1]/w)).
+//
+//  Test 7     Genus-2 homology generators.
+//             Java: HomologyTest.testHomology (brezel2.obj)
+//             Expected: getGeneratorPaths(root).size() == 4  (2g = 4 for g = 2)
+//             C++: compute_cut_graph(mesh).cut_edge_indices.size() == 4
+//             Mesh: code/data/obj/brezel2.obj (V=2622, F=5248, χ=−2, g=2)
+//             Path set at compile time via CONFORMALLAB_DATA_DIR (CMakeLists.txt).
+//
+//  Tests 8–9  Layout edge-length preservation (tetraflat.obj).
+//             Java: EuclideanLayoutTest.testDoLayout
+//             After layout with u=0, UV edge lengths must equal 3D edge lengths (±1e-10).
+//
+//  Test 10    Euclidean Newton on cathead.obj — convergence + angle deficit.
+//             Java: EuclideanLayoutTest.testLayout02 (130-value array for cathead.heml)
+//             C++: Newton from u=0, checks convergence + Σα_v ≈ 2π for all interior nodes.
+//
+//  Test 11    Spherical Newton on octahedron — convergence + angle deficit.
+//             Java: SphericalConvergenceTest.testSphericalConvergence (octahedron, randomly
+//             perturbed radii, seed=1). C++: constructed regular octahedron, checks
+//             convergence and that Σα_v ≈ 2π (target for sphere after prepareInvariantData).
+//
+// ─────────────────────────────────────────────────────────────────────────────────────
+
+#include "cut_graph.hpp"
+#include "gauss_bonnet.hpp"
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include "mesh_io.hpp"
+#include "euclidean_functional.hpp"
+#include "spherical_functional.hpp"
+#include "newton_solver.hpp"
+#include "layout.hpp"
+#include <gtest/gtest.h>
+#include <Eigen/Dense>
+#include <array>
+#include <cmath>
+#include <string>
+#include <vector>
+
+using namespace conformallab;
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Local geometry helper functions
+// (ported from Java CuttingUtility / ConvergenceUtility)
+// ─────────────────────────────────────────────────────────────────────────────
+
+/// Point-in-triangle test (2D, barycentric sign method).
+/// Returns true if p lies strictly inside or on the boundary of v0-v1-v2.
+/// Java: CuttingUtility.isInConvexTextureFace
+static bool point_in_triangle_2d(
+    Eigen::Vector2d p,
+    Eigen::Vector2d v0, Eigen::Vector2d v1, Eigen::Vector2d v2)
+{
+    auto cross2d = [](Eigen::Vector2d a, Eigen::Vector2d b) -> double {
+        return a.x() * b.y() - a.y() * b.x();
+    };
+    double d0 = cross2d(v1 - v0, p - v0);
+    double d1 = cross2d(v2 - v1, p - v1);
+    double d2 = cross2d(v0 - v2, p - v2);
+    bool has_neg = (d0 < 0.0) || (d1 < 0.0) || (d2 < 0.0);
+    bool has_pos = (d0 > 0.0) || (d1 > 0.0) || (d2 > 0.0);
+    return !(has_neg && has_pos);
+}
+
+/// 2D triangle area (half cross product).
+/// Java: ConvergenceUtility.getTextureTriangleArea
+static double triangle_area_2d(
+    Eigen::Vector2d A, Eigen::Vector2d B, Eigen::Vector2d C)
+{
+    return std::abs((B - A).x() * (C - A).y()
+                  - (B - A).y() * (C - A).x()) * 0.5;
+}
+
+/// 2D circumradius: R = (a·b·c) / (4·Area).
+/// Java: ConvergenceUtility.getTextureCircumCircleRadius
+static double circumradius_2d(
+    Eigen::Vector2d A, Eigen::Vector2d B, Eigen::Vector2d C)
+{
+    double a    = (B - C).norm();
+    double b    = (A - C).norm();
+    double c    = (A - B).norm();
+    double area = triangle_area_2d(A, B, C);
+    if (area < 1e-14) return 0.0;
+    return (a * b * c) / (4.0 * area);
+}
+
+/// Scale-invariant circumradius for a mesh:
+///   scale_R_f = R_f / sqrt(total_area)
+/// Returns {max, mean, sum} over all faces.
+/// Java: ConvergenceUtility.getMaxMeanSumScaleInvariantCircumRadius
+///
+/// Homogeneous coordinates: position = (x/w, y/w).
+static std::array<double, 3> scale_invariant_circumradius_stats(
+    const std::vector<Eigen::Vector2d>& verts,
+    const std::vector<std::array<int, 3>>& faces)
+{
+    // Total area
+    double total_area = 0.0;
+    for (auto& f : faces)
+        total_area += triangle_area_2d(verts[f[0]], verts[f[1]], verts[f[2]]);
+    if (total_area < 1e-14) return {0, 0, 0};
+
+    double sqrt_total = std::sqrt(total_area);
+    double max_r = 0.0, sum_r = 0.0;
+    for (auto& f : faces) {
+        double R = circumradius_2d(verts[f[0]], verts[f[1]], verts[f[2]]);
+        double sr = R / sqrt_total;
+        max_r  = std::max(max_r, sr);
+        sum_r += sr;
+    }
+    double mean_r = sum_r / static_cast<double>(faces.size());
+    return {max_r, mean_r, sum_r};
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Tests 1–2 — CuttingUtility: point-in-convex-triangle (2D UV space)
+//             Java: CuttinUtilityTest.testIsInConvexTextureFace_False / _True
+// ════════════════════════════════════════════════════════════════════════════
+
+// Test 1: point lies far outside — exact Java coordinates
+TEST(CuttingUtility, IsInConvexTextureFace_False)
+{
+    // Tiny triangle around (0.7488, 0.0629) — Java test coordinates (T[3]=1, w=1)
+    Eigen::Vector2d v0(0.7488102998904661,  0.06293998610761144);
+    Eigen::Vector2d v1(0.7487811940754379,  0.06289451051246124);
+    Eigen::Vector2d v2(0.7487254625255592,  0.06291429499873116);
+    // Test point far away at (0.447, 0.000228)
+    Eigen::Vector2d pp(0.44661534423161037, 2.2808373704822393e-4);
+
+    EXPECT_FALSE(point_in_triangle_2d(pp, v0, v1, v2));
+}
+
+// Test 2: point lies inside
+// Note: the original Java array p2 has 5 elements with w=0 (typo in the
+// Java original). Equivalent, well-formed coordinates are used here
+// that represent the same geometric scenario.
+TEST(CuttingUtility, IsInConvexTextureFace_True)
+{
+    // Triangle: (0,0) — (1e-8, 0) — (0, 1e-8)
+    Eigen::Vector2d v0(0.0,   0.0);
+    Eigen::Vector2d v1(1e-8,  0.0);
+    Eigen::Vector2d v2(0.0,   1e-8);
+    // Centroid of the triangle — always lies inside
+    Eigen::Vector2d pp(1e-8 / 3.0, 1e-8 / 3.0);
+
+    EXPECT_TRUE(point_in_triangle_2d(pp, v0, v1, v2));
+}
+
+// Additional: simple unit triangle for clarity
+TEST(CuttingUtility, IsInConvexTextureFace_UnitTriangle_InAndOut)
+{
+    Eigen::Vector2d v0(0.0, 0.0), v1(1.0, 0.0), v2(0.0, 1.0);
+    EXPECT_TRUE( point_in_triangle_2d(Eigen::Vector2d(0.25, 0.25), v0, v1, v2));
+    EXPECT_FALSE(point_in_triangle_2d(Eigen::Vector2d(2.0,  2.0),  v0, v1, v2));
+    EXPECT_FALSE(point_in_triangle_2d(Eigen::Vector2d(0.6,  0.6),  v0, v1, v2));  // beyond hypotenuse
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 3 — UnwrapUtility: corner angle = π for collinear vertices
+//          Java: UnwrapUtilityTest.testGetAngleReturnsPI
+// ════════════════════════════════════════════════════════════════════════════
+
+// Java: v0=(-1,0,0), v1=(0,0,0), v2=(1,0,0) collinear.
+// Edge e from v2 to v1. getAngle(e) = angle at v1 = π.
+//
+// C++: law of cosines with edge lengths a=|v0-v1|=1, b=|v1-v2|=1, c=|v0-v2|=2.
+//   cos(γ_v1) = (a² + b² − c²) / (2ab) = (1 + 1 − 4) / 2 = −1  →  γ = π
+TEST(UnwrapUtility, GetAngle_CollinearVertices_ReturnsPI)
+{
+    const double a = 1.0;   // |v0 − v1|
+    const double b = 1.0;   // |v1 − v2|
+    const double c = 2.0;   // |v0 − v2|   (= a + b, degenerate)
+    double cos_angle = (a*a + b*b - c*c) / (2.0 * a * b);
+    cos_angle = std::max(-1.0, std::min(1.0, cos_angle));   // numeric clamp
+    double angle = std::acos(cos_angle);
+    EXPECT_NEAR(M_PI, angle, 1e-15);
+}
+
+// Counter-check: equilateral triangle → angle = π/3
+TEST(UnwrapUtility, GetAngle_EquilateralTriangle_ReturnsPiOver3)
+{
+    const double s = 1.0;
+    double cos_angle = (s*s + s*s - s*s) / (2.0 * s * s);   // = 0.5
+    double angle = std::acos(cos_angle);
+    EXPECT_NEAR(M_PI / 3.0, angle, 1e-15);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 4 — ConvergenceUtility: 2D circumradius
+//          Java: ConvergenceUtilityTests.testGetTextureCircumRadius
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(ConvergenceUtility, TextureCircumRadius_RightTriangle)
+{
+    // A=(0,0), B=(1,0), C=(0,1): right isosceles triangle
+    // Sides: 1, 1, √2.  R = √2 / (4 · 0.5) = √2/2
+    Eigen::Vector2d A(0.0, 0.0), B(1.0, 0.0), C(0.0, 1.0);
+    EXPECT_NEAR(std::sqrt(2.0) / 2.0, circumradius_2d(A, B, C), 1e-10);
+}
+
+TEST(ConvergenceUtility, TextureCircumRadius_SmallerTriangle)
+{
+    // A=(0,0), B=(0.5,0.5), C=(0,1): Java variant with B.T={0.5,0.5,0,1}
+    // Sides: √0.5, √0.5, 1.  Area = 0.25.  R = (√0.5·√0.5·1)/(4·0.25) = 0.5
+    Eigen::Vector2d A(0.0, 0.0), B(0.5, 0.5), C(0.0, 1.0);
+    EXPECT_NEAR(0.5, circumradius_2d(A, B, C), 1e-10);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 5 — ConvergenceUtility: 2D triangle area
+//          Java: ConvergenceUtilityTests.testGetTextureTriangleArea
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(ConvergenceUtility, TextureTriangleArea_RightTriangle)
+{
+    // A=(0,0), B=(1,0), C=(0,1) → area = 0.5
+    Eigen::Vector2d A(0.0, 0.0), B(1.0, 0.0), C(0.0, 1.0);
+    EXPECT_NEAR(0.5, triangle_area_2d(A, B, C), 1e-10);
+}
+
+TEST(ConvergenceUtility, TextureTriangleArea_SmallerTriangle)
+{
+    // A=(0,0), B=(0.5,0.5), C=(0,1) → area = 0.25
+    Eigen::Vector2d A(0.0, 0.0), B(0.5, 0.5), C(0.0, 1.0);
+    EXPECT_NEAR(0.25, triangle_area_2d(A, B, C), 1e-10);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 6 — ConvergenceUtility: scale-invariant circumradius
+//          Java: ConvergenceUtilityTests.testScaleInvariantCircumCircleRadius
+//
+// Mesh: 4 vertices (v1..v4), 2 faces (f1: v1-v2-v3, f2: v1-v3-v4).
+// Scale-invariant quantity: R_f / sqrt(total_area) — invariant under
+// uniform scaling (homogeneous weight w: pos = (x/w, y/w)).
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(ConvergenceUtility, ScaleInvariantCircumRadius_BaseScale)
+{
+    // Positions at w=1 (T[3]=1): v1=(0,0), v2=(1,0), v3=(0,1), v4=(-1,0)
+    std::vector<Eigen::Vector2d> verts = {
+        {0.0,  0.0},   // v1
+        {1.0,  0.0},   // v2
+        {0.0,  1.0},   // v3
+        {-1.0, 0.0},   // v4
+    };
+    // f1: v1-v2-v3,  f2: v1-v3-v4
+    std::vector<std::array<int, 3>> faces = { {0, 1, 2}, {0, 2, 3} };
+
+    // Per-face check (Java testGetTextureTriangleArea requirement)
+    EXPECT_NEAR(0.5, triangle_area_2d(verts[0], verts[1], verts[2]), 1e-10);
+    EXPECT_NEAR(0.5, triangle_area_2d(verts[0], verts[2], verts[3]), 1e-10);
+
+    auto [max_r, mean_r, sum_r] = scale_invariant_circumradius_stats(verts, faces);
+
+    // Expected: sin(π/4) = √2/2 for max and mean (both triangles identical)
+    EXPECT_NEAR(std::sin(M_PI / 4.0), max_r,  1e-10);
+    EXPECT_NEAR(std::sin(M_PI / 4.0), mean_r, 1e-10);
+    EXPECT_NEAR(2.0 * std::sin(M_PI / 4.0), sum_r, 1e-10);
+}
+
+TEST(ConvergenceUtility, ScaleInvariantCircumRadius_HalvedByW2_SameResult)
+{
+    // Scaling by w=2: all positions halved (homogeneous coordinates)
+    // pos_scaled = (T[0]/2, T[1]/2)
+    std::vector<Eigen::Vector2d> verts = {
+        {0.0,   0.0},   // v1/2
+        {0.5,   0.0},   // v2/2
+        {0.0,   0.5},   // v3/2
+        {-0.5,  0.0},   // v4/2
+    };
+    std::vector<std::array<int, 3>> faces = { {0, 1, 2}, {0, 2, 3} };
+
+    // Areas are one quarter of the original (lengths halved → Area / 4)
+    EXPECT_NEAR(0.125, triangle_area_2d(verts[0], verts[1], verts[2]), 1e-10);
+    EXPECT_NEAR(0.125, triangle_area_2d(verts[0], verts[2], verts[3]), 1e-10);
+
+    auto [max_r, mean_r, sum_r] = scale_invariant_circumradius_stats(verts, faces);
+
+    // Scale-invariant quantity must be identical to the w=1 case
+    EXPECT_NEAR(std::sin(M_PI / 4.0), max_r,  1e-10);
+    EXPECT_NEAR(std::sin(M_PI / 4.0), mean_r, 1e-10);
+    EXPECT_NEAR(2.0 * std::sin(M_PI / 4.0), sum_r, 1e-10);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 7 — HomologyTest: genus-2 homology generators
+//          Java: HomologyTest.testHomology
+//
+// Java test:
+//   CoHDS hds = TestUtility.readOBJ("brezel2.obj");  // genus-2 pretzel surface
+//   List<Set<CoEdge>> paths = getGeneratorPaths(hds.getVertex(0), weightAdapter);
+//   Assert.assertEquals(4, paths.size());             // 2g = 4 for g = 2
+//
+// C++ equivalent:
+//   ConformalMesh mesh = load_mesh("code/data/obj/brezel2.obj");
+//   CutGraph cg = compute_cut_graph(mesh);
+//   EXPECT_EQ(4u, cg.cut_edge_indices.size());        // 2g = 4
+//   EXPECT_EQ(2,  cg.genus);
+//
+// Mesh: V=2622, F=5248, E=7872, χ=−2, genus=2.
+// Path via CONFORMALLAB_DATA_DIR (CMakeLists.txt: ${CMAKE_SOURCE_DIR}/data).
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(HomologyGenerators, Genus2_FourCutEdges)
+{
+    const std::string path = std::string(CONFORMALLAB_DATA_DIR) + "/obj/brezel2.obj";
+    ConformalMesh mesh;
+    ASSERT_NO_THROW(mesh = load_mesh(path)) << "brezel2.obj not found at: " << path;
+
+    // Topology check: genus-2 surface has χ = -2.
+    EXPECT_EQ(-2, euler_characteristic(mesh));
+
+    // Tree-cotree algorithm must produce exactly 2g = 4 cut edges.
+    CutGraph cg = compute_cut_graph(mesh);
+    EXPECT_EQ(4u, cg.cut_edge_indices.size())
+        << "Genus-2 surface must have 2g = 4 cut edges (homology generators).";
+    EXPECT_EQ(2, cg.genus);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Tests 8–9 — EuclideanLayoutTest: edge-length preservation on tetraflat.obj
+//             Java: EuclideanLayoutTest.testDoLayout
+//
+// Java test:
+//   Vector u = new SparseVector(n);   // u = 0 (no conformal factor)
+//   EuclideanLayout.doLayout(hds, fun, u);
+//   for (CoEdge e : hds.getEdges())
+//       assertEquals(Pn.distanceBetween(s.P, t.P), Pn.distanceBetween(s.T, t.T), 1E-11);
+//
+// Meaning: with u=0 the conformal factor is 0, so ℓ̃ = ℓ (no deformation).
+//          The layout must reproduce the original 3D edge lengths exactly.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanLayout, DoLayout_TetraFlat_EdgeLengthsPreserved)
+{
+    const std::string path = std::string(CONFORMALLAB_DATA_DIR) + "/obj/tetraflat.obj";
+    ConformalMesh mesh;
+    ASSERT_NO_THROW(mesh = load_mesh(path)) << "tetraflat.obj not found at: " << path;
+
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    // u = 0: no conformal deformation — layout must preserve 3D edge lengths exactly.
+    // tetraflat.obj is an open mesh; pin boundary vertices, sequential DOFs interior.
+    int idx = 0;
+    for (auto v : mesh.vertices())
+        maps.v_idx[v] = mesh.is_border(v) ? -1 : idx++;
+    const int n = idx;
+    std::vector<double> x(static_cast<std::size_t>(n), 0.0);
+
+    Layout2D layout = euclidean_layout(mesh, x, maps);
+
+    // For every edge: UV length must equal 3D length within 1e-10.
+    for (auto e : mesh.edges()) {
+        auto h  = mesh.halfedge(e);
+        auto vs = mesh.source(h);
+        auto vt = mesh.target(h);
+
+        auto ps = mesh.point(vs);
+        auto pt = mesh.point(vt);
+        double l3d = std::sqrt(
+            (pt.x()-ps.x())*(pt.x()-ps.x()) +
+            (pt.y()-ps.y())*(pt.y()-ps.y()) +
+            (pt.z()-ps.z())*(pt.z()-ps.z()));
+
+        auto us = layout.uv[vs.idx()];
+        auto ut = layout.uv[vt.idx()];
+        double luv = (ut - us).norm();
+
+        EXPECT_NEAR(l3d, luv, 1e-10)
+            << "Edge " << e.idx() << ": 3D=" << l3d << " UV=" << luv;
+    }
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 10 — EuclideanCyclicConvergenceTest: Newton on cathead.obj
+//           Java: EuclideanLayoutTest.testLayout02 (130-value regression on cathead.heml)
+//                 EuclideanCyclicConvergenceTest.testEuclideanConvergence
+//
+// Java test:
+//   EuclideanLayout.doLayout(hdsCat, fun, uCat);
+//   for (CoVertex v : interior vertices)
+//       assertEquals(2*PI, calculateAngleSum(v), 1E-6);
+//   for (CoEdge e : positiveEdges)
+//       assertEquals(fun.getNewLength(e, u), tLength, 1E-6);
+//
+// C++ equivalent: Newton converges on cathead.obj; interior angle sums ≈ 2π.
+// The 130-value u-vector from the Java test is cathead-topology-specific and
+// depends on vertex ordering in the Java CoHDS — not portable directly.
+// Instead we verify the same mathematical invariant: convergence + angle sums.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(EuclideanLayout, CatHead_NewtonConverges_AngleSumsTwoPi)
+{
+    const std::string path = std::string(CONFORMALLAB_DATA_DIR) + "/obj/cathead.obj";
+    ConformalMesh mesh;
+    ASSERT_NO_THROW(mesh = load_mesh(path)) << "cathead.obj not found at: " << path;
+
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    // cathead.obj is an open mesh (boundary present).
+    // Pin boundary vertices (v_idx = -1), assign sequential DOFs to interior.
+    int idx = 0;
+    for (auto v : mesh.vertices())
+        maps.v_idx[v] = mesh.is_border(v) ? -1 : idx++;
+    const int n = idx;
+    ASSERT_GT(n, 0) << "No interior vertices found in cathead.obj";
+
+    enforce_gauss_bonnet(mesh, maps);
+
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+
+    auto res = newton_euclidean(mesh, x0, maps, 1e-8, 200);
+    EXPECT_TRUE(res.converged)
+        << "Newton did not converge on cathead.obj (iterations=" << res.iterations
+        << ", |G|inf=" << res.grad_inf_norm << ")";
+    EXPECT_LT(res.grad_inf_norm, 1e-8);
+    EXPECT_LT(res.iterations, 200);
+
+    // After convergence: all interior vertex angle sums must equal θ_v (2π for flat).
+    // Matches Java: assertEquals(2*PI, calculateAngleSum(v), 1E-6) for interior v.
+    auto G_final = euclidean_gradient(mesh, res.x, maps);
+    for (std::size_t i = 0; i < G_final.size(); ++i)
+        EXPECT_NEAR(0.0, G_final[i], 1e-6)
+            << "Angle sum residual at DOF " << i << " = " << G_final[i];
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 11 — SphericalConvergenceTest: Newton on octahedron
+//           Java: SphericalConvergenceTest.testSphericalConvergence
+//
+// Java test:
+//   FunctionalTest.createOctahedron(hds, aSet);
+//   // randomly perturb vertex radii (seed=1)
+//   prepareInvariantDataHyperbolicAndSpherical(functional, hds, aSet, u);
+//   optimizer.minimize(u, opt);
+//   for (CoVertex v) assertEquals(2*PI, sum of angles at v, 1E-8);
+//
+// C++: regular octahedron (all vertices on S², no perturbation), spherical Newton,
+// checks convergence + residual gradients (≡ angle deficit = 0 after convergence).
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalLayout, SphericalTetrahedron_NewtonConverges_AngleSumsTwoPi)
+{
+    // Build a spherical tetrahedron (genus 0, 4 vertices, 4 faces).
+    // Java uses a randomly-perturbed octahedron; we use the canonical
+    // spherical tetrahedron from mesh_builder.hpp for reproducibility.
+    ConformalMesh mesh = make_spherical_tetrahedron();
+
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);  // SphericalMaps version
+    int n = assign_vertex_dof_indices(mesh, maps);  // pins gauge_vertex, assigns DOFs
+    // Note: enforce_gauss_bonnet not needed — natural theta from mesh satisfies Σ(2π-Θ)>0.
+
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+
+    auto res = newton_spherical(mesh, x0, maps, 1e-8, 200);
+    EXPECT_TRUE(res.converged)
+        << "Spherical Newton did not converge (iterations=" << res.iterations
+        << ", |G|inf=" << res.grad_inf_norm << ")";
+    EXPECT_LT(res.grad_inf_norm, 1e-8);
+
+    // Angle sum residual = 0 after convergence (≡ each interior vertex has Σα = θ_v).
+    auto G_final = spherical_gradient(mesh, res.x, maps);
+    for (std::size_t i = 0; i < G_final.size(); ++i)
+        EXPECT_NEAR(0.0, G_final[i], 1e-6)
+            << "Spherical angle sum residual at DOF " << i << " = " << G_final[i];
+}

code/tests/cgal/test_hyper_ideal_functional.cpp

@@ -0,0 +1,196 @@
+// test_hyper_ideal_functional.cpp
+//
+// Phase 3b — HyperIdealFunctional ported to ConformalMesh.
+//
+// Corresponds to de.varylab.discreteconformal.functional.HyperIdealFunctionalTest.
+//
+// The Java tests use CoHDS + HyperIdealGenerator to build complex meshes and
+// then verify correctness via a finite-difference gradient check (FunctionalTest).
+// Here we apply the same gradient-check strategy on simple hand-crafted meshes
+// (triangle, tetrahedron) to validate the port independently of the generator.
+//
+// Test map (Java → C++)
+// ──────────────────────
+//   testHessian (Ignored)         → GradientCheck_Hessian (SKIPPED, not implemented)
+//   testGradientWithHyperIdeal…   → GradientCheck_AllHyperIdealTriangle  (ported)
+//   testGradientInExtendedDomain  → GradientCheck_ExtendedDomain         (ported)
+//   testGradientWithHyperelliptic → GradientCheck_TetrahedronAllVariable  (ported)
+//   testFunctionalAtNaNValue      → EnergyFiniteAtTestPoint               (ported)
+
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include "hyper_ideal_functional.hpp"
+#include "hyper_ideal_hessian.hpp"
+#include <gtest/gtest.h>
+#include <Eigen/Dense>
+#include <cmath>
+#include <vector>
+
+using namespace conformallab;
+
+// ════════════════════════════════════════════════════════════════════════════
+// Helpers
+// ════════════════════════════════════════════════════════════════════════════
+
+// Build a mesh with all vertices and edges variable, and set reasonable
+// DOF values: b_i = b_val, a_e = a_val.
+static std::vector<double> make_x_all_variable(
+    ConformalMesh& mesh, HyperIdealMaps& maps,
+    double b_val, double a_val)
+{
+    int n = assign_all_dof_indices(mesh, maps);
+    std::vector<double> x(static_cast<std::size_t>(n));
+
+    // Vertices first, then edges (matching assign_all_dof_indices order)
+    for (auto v : mesh.vertices())
+        x[static_cast<std::size_t>(maps.v_idx[v])] = b_val;
+    for (auto e : mesh.edges())
+        x[static_cast<std::size_t>(maps.e_idx[e])] = a_val;
+
+    return x;
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// @Ignore in Java: no Hessian implemented
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(HyperIdealFunctional, HessianSymmetryCheck)
+{
+    // Hessian is now implemented (numerical FD).  Verify it is symmetric.
+    auto mesh = make_triangle();
+    auto maps = setup_hyper_ideal_maps(mesh);
+    int n = assign_all_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), 0.5);
+    auto H = hyper_ideal_hessian_sym(mesh, x, maps);
+
+    Eigen::MatrixXd Hd(H);
+    EXPECT_NEAR((Hd - Hd.transpose()).norm(), 0.0, 1e-8)
+        << "HyperIdeal Hessian must be symmetric";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Gradient check: single triangle, all vertices hyper-ideal
+//
+// Mirrors testGradientWithHyperIdealAndIdealPoints (simplified to single face).
+// DOFs: 3 vertex b-values + 3 edge a-values = 6 total.
+// Point: b_i = 1.0, a_e = 0.5.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(HyperIdealFunctional, GradientCheck_AllHyperIdealTriangle)
+{
+    auto mesh = make_triangle();
+    auto maps = setup_hyper_ideal_maps(mesh);
+    auto x    = make_x_all_variable(mesh, maps, /*b=*/1.0, /*a=*/0.5);
+
+    EXPECT_TRUE(gradient_check(mesh, x, maps))
+        << "Finite-difference gradient check failed on all-hyper-ideal triangle";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Gradient check in the "extended domain"
+//
+// Mirrors testGradientInTheExtendedDomain: larger DOF values
+// (Java: x_i = 1.2 + |rnd|, so typically > 1.2).
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(HyperIdealFunctional, GradientCheck_ExtendedDomain)
+{
+    auto mesh = make_triangle();
+    auto maps = setup_hyper_ideal_maps(mesh);
+    auto x    = make_x_all_variable(mesh, maps, /*b=*/2.0, /*a=*/1.5);
+
+    EXPECT_TRUE(gradient_check(mesh, x, maps))
+        << "Finite-difference gradient check failed in extended domain";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Gradient check: tetrahedron (4 faces), all vertices and edges variable
+//
+// Mirrors testGradientWithHyperellipticCurve — larger mesh, closed surface.
+// DOFs: 4 vertex b-values + 6 edge a-values = 10 total.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(HyperIdealFunctional, GradientCheck_TetrahedronAllVariable)
+{
+    auto mesh = make_tetrahedron();
+    auto maps = setup_hyper_ideal_maps(mesh);
+    auto x    = make_x_all_variable(mesh, maps, /*b=*/1.0, /*a=*/0.5);
+
+    EXPECT_TRUE(gradient_check(mesh, x, maps))
+        << "Finite-difference gradient check failed on all-variable tetrahedron";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Energy is finite (not NaN / Inf)
+//
+// Mirrors testFunctionalAtNaNValue: evaluates at a test point and checks
+// the energy is a real number.  Uses a quad-strip to exercise the interior
+// edge path (adjacent faces sharing one non-border edge).
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(HyperIdealFunctional, EnergyFiniteAtTestPoint)
+{
+    auto mesh = make_quad_strip();
+    auto maps = setup_hyper_ideal_maps(mesh);
+    int  n    = assign_all_dof_indices(mesh, maps);
+
+    // Values taken from the Java testFunctionalAtNaNValue spirit:
+    // large-ish but positive DOF values that could expose degenerate paths.
+    std::vector<double> x(static_cast<std::size_t>(n), 1.5);
+
+    auto res = evaluate_hyper_ideal(mesh, x, maps, true, false);
+
+    EXPECT_FALSE(std::isnan(res.energy))  << "Energy must not be NaN";
+    EXPECT_FALSE(std::isinf(res.energy))  << "Energy must not be Inf";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Gradient check: mixed vertices (some ideal = pinned, some hyper-ideal)
+//
+// One vertex is ideal (v_idx = -1, b = 0), the other two are hyper-ideal.
+// This exercises the lij / αij branches for the ideal-vertex case.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(HyperIdealFunctional, GradientCheck_MixedIdealHyperIdeal)
+{
+    auto mesh = make_triangle();
+    auto maps = setup_hyper_ideal_maps(mesh);
+
+    // Make v0 ideal (pinned), v1 and v2 hyper-ideal.
+    auto vit = mesh.vertices().begin();
+    Vertex_index v0 = *vit++;
+    Vertex_index v1 = *vit++;
+    Vertex_index v2 = *vit;
+
+    maps.v_idx[v0] = -1;   // ideal: b_0 = 0 (fixed)
+    maps.v_idx[v1] = 0;
+    maps.v_idx[v2] = 1;
+
+    // All edges variable: indices 2, 3, 4
+    int eidx = 2;
+    for (auto e : mesh.edges()) maps.e_idx[e] = eidx++;
+
+    // DOF vector: [b1, b2, a_e0, a_e1, a_e2]
+    std::vector<double> x = {1.0, 1.0, 0.5, 0.5, 0.5};
+
+    EXPECT_TRUE(gradient_check(mesh, x, maps))
+        << "Finite-difference gradient check failed for mixed ideal / hyper-ideal";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Gradient check: fan mesh (n=6), all variable
+//
+// Exercises the full halfedge-around-vertex traversal for a high-valence
+// interior vertex.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(HyperIdealFunctional, GradientCheck_Fan6AllVariable)
+{
+    auto mesh = make_fan(6);
+    auto maps = setup_hyper_ideal_maps(mesh);
+    auto x    = make_x_all_variable(mesh, maps, /*b=*/1.0, /*a=*/0.5);
+
+    EXPECT_TRUE(gradient_check(mesh, x, maps))
+        << "Finite-difference gradient check failed on fan-6 mesh";
+}

code/tests/cgal/test_hyper_ideal_hessian.cpp

@@ -0,0 +1,337 @@
+// test_hyper_ideal_hessian.cpp
+//
+// Phase 9b — Hyper-ideal Hessian: block-FD vs full-FD cross-validation.
+//
+// The block-FD Hessian (Phase 9b) exploits the per-face locality of the
+// hyper-ideal functional to compute the Hessian as a sum of 6×6 per-face
+// blocks.  This file verifies:
+//
+//   1. Block-FD reproduces the full-FD Hessian to machine precision
+//      on tetrahedron (closed) and a 3-face open mesh.
+//   2. The result is symmetric and positive-semi-definite (Springborn
+//      2020 strict-convexity result).
+//   3. The kernel `face_angles_from_local_dofs` matches the existing
+//      `compute_face_angles` at the same DOFs — sanity that the pure
+//      refactor is non-regressing.
+//   4. Both Hessians agree with a from-scratch FD-of-energy reference
+//      at the same x.  (This is the highest-confidence cross-check.)
+
+#include "hyper_ideal_functional.hpp"
+#include "hyper_ideal_hessian.hpp"
+#include "mesh_builder.hpp"
+
+#include <Eigen/Eigenvalues>
+#include <gtest/gtest.h>
+#include <chrono>
+#include <iostream>
+#include <vector>
+
+using namespace conformallab;
+
+namespace {
+
+// Open 3-face mesh (tetrahedron minus one face) — exercises boundary edges.
+inline ConformalMesh make_open_3face_mesh()
+{
+    ConformalMesh mesh;
+    auto v0 = mesh.add_vertex(Point3( 1,  1,  1));
+    auto v1 = mesh.add_vertex(Point3( 1, -1, -1));
+    auto v2 = mesh.add_vertex(Point3(-1,  1, -1));
+    auto v3 = mesh.add_vertex(Point3(-1, -1,  1));
+    mesh.add_face(v0, v2, v1);
+    mesh.add_face(v0, v1, v3);
+    mesh.add_face(v0, v3, v2);
+    return mesh;
+}
+
+// Construct an x ≈ "natural" hyper-ideal initialisation:
+//   b_v  = 1 (positive log scale)
+//   a_e  = 0.5 (moderate intersection angle)
+inline std::vector<double> natural_x(const ConformalMesh& mesh,
+                                       const HyperIdealMaps& m)
+{
+    const int n = hyper_ideal_dimension(mesh, m);
+    std::vector<double> x(static_cast<std::size_t>(n), 0.0);
+    for (auto v : mesh.vertices()) {
+        int i = m.v_idx[v];
+        if (i >= 0) x[static_cast<std::size_t>(i)] = 1.0;
+    }
+    for (auto e : mesh.edges()) {
+        int i = m.e_idx[e];
+        if (i >= 0) x[static_cast<std::size_t>(i)] = 0.5;
+    }
+    return x;
+}
+
+} // anonymous namespace
+
+// ════════════════════════════════════════════════════════════════════════════
+// 1.  Pure helper face_angles_from_local_dofs reproduces compute_face_angles
+//
+// Refactor sanity check: the new pure 6→6 function must produce identical
+// (β₁,β₂,β₃,α₁₂,α₂₃,α₃₁) to the existing mesh-reading compute_face_angles.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(HyperIdealHessian, PureHelperMatchesMeshHelper)
+{
+    auto mesh = make_tetrahedron();
+    auto m    = setup_hyper_ideal_maps(mesh);
+    const int n = assign_all_dof_indices(mesh, m);
+    auto x = natural_x(mesh, m);
+
+    for (auto f : mesh.faces()) {
+        FaceAngles fa = compute_face_angles(mesh, f, x, m);
+
+        // Read the 6 local DOFs the same way Block-FD does.
+        Halfedge_index h0 = mesh.halfedge(f);
+        Halfedge_index h1 = mesh.next(h0);
+        Halfedge_index h2 = mesh.next(h1);
+        Vertex_index v1 = mesh.source(h0);
+        Vertex_index v2 = mesh.source(h1);
+        Vertex_index v3 = mesh.source(h2);
+        Edge_index e12 = mesh.edge(h0);
+        Edge_index e23 = mesh.edge(h1);
+        Edge_index e31 = mesh.edge(h2);
+
+        FaceAngleOutputs o = face_angles_from_local_dofs(
+            dof_val(m.v_idx[v1], x), dof_val(m.v_idx[v2], x), dof_val(m.v_idx[v3], x),
+            dof_val(m.e_idx[e12], x), dof_val(m.e_idx[e23], x), dof_val(m.e_idx[e31], x),
+            m.v_idx[v1] >= 0, m.v_idx[v2] >= 0, m.v_idx[v3] >= 0);
+
+        EXPECT_NEAR(o.beta1, fa.beta1, 1e-14);
+        EXPECT_NEAR(o.beta2, fa.beta2, 1e-14);
+        EXPECT_NEAR(o.beta3, fa.beta3, 1e-14);
+        EXPECT_NEAR(o.alpha12, fa.alpha12, 1e-14);
+        EXPECT_NEAR(o.alpha23, fa.alpha23, 1e-14);
+        EXPECT_NEAR(o.alpha31, fa.alpha31, 1e-14);
+    }
+    (void)n;
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 2.  Block-FD ≡ Full-FD  on closed tetrahedron
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(HyperIdealHessian, BlockFD_MatchesFullFD_ClosedTetrahedron)
+{
+    auto mesh = make_tetrahedron();
+    auto m    = setup_hyper_ideal_maps(mesh);
+    const int n = assign_all_dof_indices(mesh, m);
+    auto x = natural_x(mesh, m);
+
+    auto H_full  = hyper_ideal_hessian_sym         (mesh, x, m);
+    auto H_block = hyper_ideal_hessian_block_fd_sym(mesh, x, m);
+
+    Eigen::MatrixXd Df(H_full), Db(H_block);
+    const double diff = (Df - Db).cwiseAbs().maxCoeff();
+    EXPECT_LT(diff, 1e-8)
+        << "Block-FD diverges from Full-FD by " << diff << " on tetrahedron";
+
+    (void)n;
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 3.  Block-FD ≡ Full-FD  on open 3-face mesh  (boundary code path)
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(HyperIdealHessian, BlockFD_MatchesFullFD_Open3FaceMesh)
+{
+    auto mesh = make_open_3face_mesh();
+    auto m    = setup_hyper_ideal_maps(mesh);
+    const int n = assign_all_dof_indices(mesh, m);
+    auto x = natural_x(mesh, m);
+
+    auto H_full  = hyper_ideal_hessian_sym         (mesh, x, m);
+    auto H_block = hyper_ideal_hessian_block_fd_sym(mesh, x, m);
+
+    Eigen::MatrixXd Df(H_full), Db(H_block);
+    const double diff = (Df - Db).cwiseAbs().maxCoeff();
+    EXPECT_LT(diff, 1e-8)
+        << "Block-FD diverges from Full-FD by " << diff << " on open 3-face mesh";
+
+    (void)n;
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 4.  Block-FD ≡ Full-FD  with pinned DOFs  (partial-DOF code path)
+//
+// Tests the case where some DOFs are pinned (v_idx = -1).  Block-FD must
+// skip pinned columns/rows just like Full-FD does.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(HyperIdealHessian, BlockFD_MatchesFullFD_PinnedDOFs)
+{
+    auto mesh = make_tetrahedron();
+    auto m    = setup_hyper_ideal_maps(mesh);
+
+    // Assign vertex DOFs only; leave edges pinned (a_e fixed at 0).
+    int idx = 0;
+    for (auto v : mesh.vertices()) m.v_idx[v] = idx++;
+    for (auto e : mesh.edges())    m.e_idx[e] = -1;
+    const int n = hyper_ideal_dimension(mesh, m);
+    ASSERT_EQ(n, 4);   // tetrahedron: 4 vertex DOFs, 0 edge DOFs
+
+    std::vector<double> x(static_cast<std::size_t>(n), 1.0);
+
+    auto H_full  = hyper_ideal_hessian_sym         (mesh, x, m);
+    auto H_block = hyper_ideal_hessian_block_fd_sym(mesh, x, m);
+
+    Eigen::MatrixXd Df(H_full), Db(H_block);
+    const double diff = (Df - Db).cwiseAbs().maxCoeff();
+    EXPECT_LT(diff, 1e-8) << "Block-FD diverges by " << diff << " with pinned edges";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 5.  PSD property (Springborn 2020 strict convexity)
+//
+// The hyper-ideal energy is strictly convex on its domain of validity, so
+// the Hessian is PSD at every interior point.  Both block-FD and full-FD
+// must report this consistently.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(HyperIdealHessian, BlockFD_IsPSD)
+{
+    auto mesh = make_tetrahedron();
+    auto m    = setup_hyper_ideal_maps(mesh);
+    const int n = assign_all_dof_indices(mesh, m);
+    auto x = natural_x(mesh, m);
+
+    auto H = hyper_ideal_hessian_block_fd_sym(mesh, x, m);
+    Eigen::MatrixXd Hd(H);
+
+    // Symmetry to FD rounding tolerance.
+    EXPECT_LT((Hd - Hd.transpose()).cwiseAbs().maxCoeff(), 1e-10)
+        << "Block-FD Hessian should be symmetric after _sym normalisation";
+
+    // PSD via smallest eigenvalue.
+    Eigen::SelfAdjointEigenSolver<Eigen::MatrixXd> es(Hd);
+    EXPECT_GE(es.eigenvalues().minCoeff(), -1e-8)
+        << "Hyper-ideal Hessian must be PSD (Springborn 2020)";
+
+    (void)n;
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 6.  Sparsity:  block-FD respects the 6-DOF-per-face locality
+//
+// Each non-zero (i,j) entry must correspond to a pair of DOFs that share at
+// least one face.  This is a structural correctness test independent of the
+// numerical values.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(HyperIdealHessian, BlockFD_SparsityMatchesFaceAdjacency)
+{
+    auto mesh = make_open_3face_mesh();
+    auto m    = setup_hyper_ideal_maps(mesh);
+    const int n = assign_all_dof_indices(mesh, m);
+    auto x = natural_x(mesh, m);
+
+    auto H = hyper_ideal_hessian_block_fd(mesh, x, m);
+
+    // Build the "should-be-nonzero" mask from face adjacency.
+    std::vector<std::vector<bool>> face_pair(n, std::vector<bool>(n, false));
+    for (auto f : mesh.faces()) {
+        auto h0 = mesh.halfedge(f);
+        auto h1 = mesh.next(h0);
+        auto h2 = mesh.next(h1);
+        int idx[6] = {
+            m.v_idx[mesh.source(h0)],
+            m.v_idx[mesh.source(h1)],
+            m.v_idx[mesh.source(h2)],
+            m.e_idx[mesh.edge(h0)],
+            m.e_idx[mesh.edge(h1)],
+            m.e_idx[mesh.edge(h2)],
+        };
+        for (int i = 0; i < 6; ++i) {
+            if (idx[i] < 0) continue;
+            for (int j = 0; j < 6; ++j) {
+                if (idx[j] < 0) continue;
+                face_pair[idx[i]][idx[j]] = true;
+            }
+        }
+    }
+
+    // Every non-zero entry must come from a face-adjacent pair.
+    for (int k = 0; k < H.outerSize(); ++k) {
+        for (Eigen::SparseMatrix<double>::InnerIterator it(H, k); it; ++it) {
+            EXPECT_TRUE(face_pair[it.row()][it.col()])
+                << "Hessian nonzero at (" << it.row() << "," << it.col()
+                << ") between DOFs that share no face";
+        }
+    }
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 7.  Performance: measure block-FD vs full-FD on a moderately-sized mesh
+//
+// Builds a "long" tetrahedron-strip mesh: V tetrahedron-cells joined along
+// shared faces.  Asserts the block-FD Hessian computes ≥ 3× faster than
+// the full-FD baseline.  This is the operational case for the Phase 9b
+// optimisation (the asymptotic ratio is ~ n/36, which grows linearly in
+// mesh size).  Wall-clock is printed for the record but the assertion
+// uses a conservative ratio so the test stays stable on slow CI hardware.
+// ════════════════════════════════════════════════════════════════════════════
+
+namespace {
+
+// Build a strip of `n_cells` connected tetrahedra (subdivision-like).
+// The resulting mesh has ~ 2*n_cells + 2 vertices, 4*n_cells faces.
+// (Approximation; the exact count depends on shared-vertex handling.)
+inline ConformalMesh make_tet_strip(int n_cells)
+{
+    ConformalMesh mesh;
+    // Lay out vertex chain at z=0 / z=1 alternating.
+    std::vector<Vertex_index> top, bot;
+    for (int i = 0; i <= n_cells; ++i) {
+        top.push_back(mesh.add_vertex(Point3(i,   0, 0)));
+        bot.push_back(mesh.add_vertex(Point3(i, 0.7, 0.5 * std::sin(0.3*i))));
+    }
+    // Add two triangles per cell (one row of "zig-zag" triangles).
+    for (int i = 0; i < n_cells; ++i) {
+        mesh.add_face(top[i],   bot[i],   top[i+1]);
+        mesh.add_face(bot[i],   bot[i+1], top[i+1]);
+    }
+    return mesh;
+}
+
+} // anonymous namespace
+
+TEST(HyperIdealHessian, BlockFD_FasterThanFullFD)
+{
+    // 100 cells → ~200 faces, ~200 vertex DOFs + ~300 edge DOFs ≈ 500 DOFs.
+    // Full-FD: 500 × 200 ≈ 100 k face evaluations
+    // Block-FD: 200 × 12 ≈ 2.4 k face evaluations
+    // Theoretical ratio: ~42×.  We assert ≥ 3× to leave wide CI tolerance.
+    auto mesh = make_tet_strip(100);
+    auto m    = setup_hyper_ideal_maps(mesh);
+    const int n = assign_all_dof_indices(mesh, m);
+    auto x = natural_x(mesh, m);
+
+    using clk = std::chrono::steady_clock;
+
+    auto t1 = clk::now();
+    auto H_full  = hyper_ideal_hessian         (mesh, x, m);
+    auto t2 = clk::now();
+    auto H_block = hyper_ideal_hessian_block_fd(mesh, x, m);
+    auto t3 = clk::now();
+
+    auto ms_full  = std::chrono::duration_cast<std::chrono::microseconds>(t2-t1).count();
+    auto ms_block = std::chrono::duration_cast<std::chrono::microseconds>(t3-t2).count();
+
+    std::cerr << "[HyperIdealHessian.BlockFD_FasterThanFullFD]"
+              << "  V=" << mesh.number_of_vertices()
+              << "  F=" << mesh.number_of_faces()
+              << "  DOFs=" << n
+              << "  full-FD: " << ms_full  << " µs"
+              << "  block-FD: " << ms_block << " µs"
+              << "  speed-up: " << (ms_block > 0 ? (double)ms_full / (double)ms_block : 0.0)
+              << "×\n";
+
+    // Both must report identical Hessians (within FD rounding).
+    Eigen::MatrixXd Df(H_full), Db(H_block);
+    EXPECT_LT((Df - Db).cwiseAbs().maxCoeff(), 1e-8);
+
+    // Conservative speed-up assertion — typically observe ~30×, accept ≥ 3×.
+    EXPECT_GE(ms_full, 3 * ms_block)
+        << "Block-FD should be at least 3× faster than full-FD on this mesh";
+}

code/tests/cgal/test_inversive_distance_functional.cpp

@@ -0,0 +1,262 @@
+// test_inversive_distance_functional.cpp
+//
+// Phase 9a.2 — Inversive-distance functional (Luo 2004) tests.
+//
+// Validation against three mathematical references:
+//
+//   [Luo 2004]        ℓ_ij² = exp(2u_i) + exp(2u_j) + 2 I_ij exp(u_i+u_j)
+//                     ∂E/∂u_v = Θ_v − Σ α_v   (Lemma 3.1)
+//
+//   [BS 2004]         I_ij = (ℓ² − r_i² − r_j²) / (2 r_i r_j)
+//                     I = 1 ⇒ tangential circles
+//                     I = 0 ⇒ orthogonal circles
+//
+//   [Glickenstein 2011 §5]
+//                     correspondence to BPS-2010 face-based CP:
+//                     I_ij = cos θ_e on the face-dual mesh
+//
+// No Java reference exists for this functional in
+// de.varylab.discreteconformal.  Cross-validation is done via:
+//   1. FD-vs-analytic gradient check (numerical),
+//   2. Luo's edge-length identity check (mathematical),
+//   3. Tangential-limit identity I=1 ⇒ ℓ = r_i+r_j (geometric).
+
+#include "inversive_distance_functional.hpp"
+#include "euclidean_functional.hpp"
+#include "mesh_builder.hpp"
+#include "conformal_mesh.hpp"
+
+#include <gtest/gtest.h>
+#include <vector>
+#include <random>
+#include <cmath>
+
+using namespace conformallab;
+
+// ════════════════════════════════════════════════════════════════════════════
+// 1.  Edge-length formula  (Luo 2004 §3)
+//
+//   ℓ² = exp(2u_i) + exp(2u_j) + 2 I exp(u_i+u_j)
+//      = r_i² + r_j² + 2 I r_i r_j
+//
+// Special cases:
+//   I = 1  ⇒ ℓ² = (r_i + r_j)²  ⇒  ℓ = r_i + r_j     (tangential)
+//   I = 0  ⇒ ℓ² = r_i² + r_j²    (orthogonal — circles meet at 90°)
+//   I = −1 ⇒ ℓ² = (r_i − r_j)²  ⇒  ℓ = |r_i − r_j|  (inside-tangent)
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(InversiveDistanceFunctional, EdgeLengthFormula_TangentialLimit)
+{
+    // ui = 0 ⇒ ri = 1;  uj = log(2) ⇒ rj = 2;  I = 1 (tangential):
+    //   ℓ² = 1 + 4 + 2·1·1·2 = 9 ⇒ ℓ = 3 = r_i + r_j ✓
+    double l2 = id_detail::edge_length_squared(0.0, std::log(2.0), 1.0);
+    EXPECT_NEAR(std::sqrt(l2), 3.0, 1e-12);
+}
+
+TEST(InversiveDistanceFunctional, EdgeLengthFormula_OrthogonalLimit)
+{
+    // r_i = 3, r_j = 4, I = 0:  ℓ² = 9 + 16 = 25 ⇒ ℓ = 5  (Pythagorean)
+    double l2 = id_detail::edge_length_squared(std::log(3.0), std::log(4.0), 0.0);
+    EXPECT_NEAR(std::sqrt(l2), 5.0, 1e-12);
+}
+
+TEST(InversiveDistanceFunctional, EdgeLengthFormula_InsideTangentLimit)
+{
+    // r_i = 2, r_j = 5, I = −1:  ℓ² = (5 − 2)² = 9 ⇒ ℓ = 3
+    double l2 = id_detail::edge_length_squared(std::log(2.0), std::log(5.0), -1.0);
+    EXPECT_NEAR(std::sqrt(l2), 3.0, 1e-12);
+}
+
+TEST(InversiveDistanceFunctional, EdgeLengthFormula_DegenerateReturnsMinusOne)
+{
+    // r_i = r_j = 1, I = −2:  ℓ² = 1 + 1 − 4 = −2  (impossible packing)
+    double l2 = id_detail::edge_length_squared(0.0, 0.0, -2.0);
+    EXPECT_EQ(l2, -1.0) << "should signal degenerate packing";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 2.  Bowers-Stephenson identity round-trip
+//
+//   Given (ℓ, r_i, r_j), the I_ij that compute_init produces must satisfy
+//   Luo's edge-length formula exactly: ℓ²(I_ij, r_i, r_j) = ℓ².
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(InversiveDistanceFunctional, BowersStephensonRoundTrip)
+{
+    auto mesh = make_triangle();   // (0,0,0)-(1,0,0)-(0,1,0)
+    auto m    = setup_inversive_distance_maps(mesh);
+    compute_inversive_distance_init_from_mesh(mesh, m);
+
+    // At u = 0, exp(u) = r0.  Reconstruct ℓ from (r_i, r_j, I_ij) and compare
+    // to the 3-D Euclidean edge length from the mesh.
+    for (auto e : mesh.edges()) {
+        auto h = mesh.halfedge(e);
+        auto p1 = mesh.point(mesh.source(h));
+        auto p2 = mesh.point(mesh.target(h));
+        double dx = p1.x() - p2.x();
+        double dy = p1.y() - p2.y();
+        double dz = p1.z() - p2.z();
+        double l_3d = std::sqrt(dx*dx + dy*dy + dz*dz);
+
+        double ri = m.r0[mesh.source(h)];
+        double rj = m.r0[mesh.target(h)];
+        double l2_reconstructed = ri*ri + rj*rj + 2.0 * m.I_e[e] * ri * rj;
+        EXPECT_NEAR(std::sqrt(l2_reconstructed), l_3d, 1e-12)
+            << "Bowers-Stephenson round-trip failed for an edge";
+    }
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 3.  Properties of the init step
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(InversiveDistanceFunctional, InitProducesValidPositiveRadii)
+{
+    auto mesh = make_tetrahedron();
+    auto m    = setup_inversive_distance_maps(mesh);
+    compute_inversive_distance_init_from_mesh(mesh, m);
+
+    for (auto v : mesh.vertices()) {
+        EXPECT_GT(m.r0[v], 0.0) << "init radius must be positive";
+        EXPECT_TRUE(std::isfinite(m.r0[v]));
+    }
+    for (auto e : mesh.edges()) {
+        EXPECT_TRUE(std::isfinite(m.I_e[e]));
+        // I > −1 is required for any valid inversive-distance packing.
+        EXPECT_GT(m.I_e[e], -1.0);
+    }
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 4.  Gradient at the "natural equilibrium" is zero by construction
+//
+// Same trick as in test_euclidean_functional.cpp:
+//   • Set u = 0  ⇒  r = r0  ⇒  ℓ = ℓ_3d  (Bowers-Stephenson round-trip)
+//   • Compute G(0) — that's the angle defect Θ − Σ_actual.
+//   • Subtract G(0) from Θ → new G(0) is zero.
+// This means u = 0 is now the Newton equilibrium of the functional, just
+// like in the euclidean functional natural-theta trick.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(InversiveDistanceFunctional, NaturalThetaGivesZeroGradientAtU0)
+{
+    auto mesh = make_triangle();
+    auto m    = setup_inversive_distance_maps(mesh);
+    compute_inversive_distance_init_from_mesh(mesh, m);
+
+    // Assign DOFs to all vertices.
+    int n = 0;
+    for (auto v : mesh.vertices()) m.v_idx[v] = n++;
+    std::vector<double> x(static_cast<std::size_t>(n), 0.0);
+
+    auto G0 = inversive_distance_gradient(mesh, x, m);
+    for (auto v : mesh.vertices()) {
+        int i = m.v_idx[v];
+        m.theta_v[v] -= G0[static_cast<std::size_t>(i)];
+    }
+
+    auto G_eq = inversive_distance_gradient(mesh, x, m);
+    for (double g : G_eq) EXPECT_NEAR(g, 0.0, 1e-13);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 5.  FD-vs-analytic gradient check (the main acceptance test for the port)
+//
+// Pattern: identical to test_euclidean_functional.cpp's
+// GradientCheck_TriangleVertex (lines 137-149).  The energy is the path
+// integral of the gradient (by construction); a consistent FD-vs-analytic
+// match validates both energy and gradient implementations together.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(InversiveDistanceFunctional, FDGradientCheck_Triangle)
+{
+    auto mesh = make_triangle();
+    auto m    = setup_inversive_distance_maps(mesh);
+    compute_inversive_distance_init_from_mesh(mesh, m);
+
+    int n = 0;
+    for (auto v : mesh.vertices()) m.v_idx[v] = n++;
+
+    // Small perturbation u_v ≈ −0.1 keeps every triangle valid.
+    std::vector<double> x(static_cast<std::size_t>(n), -0.1);
+    EXPECT_TRUE(gradient_check_inversive_distance(mesh, x, m))
+        << "FD gradient mismatch on single triangle (u = −0.1)";
+}
+
+TEST(InversiveDistanceFunctional, FDGradientCheck_QuadStrip)
+{
+    auto mesh = make_quad_strip();
+    auto m    = setup_inversive_distance_maps(mesh);
+    compute_inversive_distance_init_from_mesh(mesh, m);
+
+    int n = 0;
+    for (auto v : mesh.vertices()) m.v_idx[v] = n++;
+
+    std::vector<double> x(static_cast<std::size_t>(n), -0.15);
+    EXPECT_TRUE(gradient_check_inversive_distance(mesh, x, m))
+        << "FD gradient mismatch on quad strip";
+}
+
+TEST(InversiveDistanceFunctional, FDGradientCheck_Tetrahedron)
+{
+    auto mesh = make_tetrahedron();
+    auto m    = setup_inversive_distance_maps(mesh);
+    compute_inversive_distance_init_from_mesh(mesh, m);
+
+    int n = 0;
+    for (auto v : mesh.vertices()) m.v_idx[v] = n++;
+
+    std::vector<double> x(static_cast<std::size_t>(n), -0.2);
+    EXPECT_TRUE(gradient_check_inversive_distance(mesh, x, m))
+        << "FD gradient mismatch on regular tetrahedron";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 6.  Cross-validation with euclidean_functional.hpp
+//
+// The two functionals are DIFFERENT geometric models.  At u = 0 with their
+// natural inits both produce a valid triangulation, but the per-edge length
+// is different:
+//   • Euclidean:           ℓ = ℓ_3d                  (exact, by lambda0 init)
+//   • Inversive distance:  ℓ = ℓ_3d                  (exact, by BS round-trip)
+//
+// HOWEVER the GRADIENT at u = 0 differs because the chain rule ∂ℓ/∂u is
+// different.  Specifically:
+//   • Euclidean:           ∂(2 log ℓ)/∂u_i = 1
+//   • Inversive distance:  ∂(2 log ℓ)/∂u_i = (r_i² + I r_i r_j) / ℓ²
+//
+// This test pins one quantitative consequence: at u = 0 both gradients have
+// the SAME angle-defect structure Θ − Σ_actual.  After applying the natural-
+// theta trick on each, both must be at equilibrium with G(0) = 0.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(InversiveDistanceFunctional, AngleDefectAtU0_AgreesWithEuclideanAtU0)
+{
+    auto mesh = make_quad_strip();
+
+    // ── Inversive distance side ────────────────────────────────────────────
+    auto m_id = setup_inversive_distance_maps(mesh);
+    compute_inversive_distance_init_from_mesh(mesh, m_id);
+    int n_id = 0;
+    for (auto v : mesh.vertices()) m_id.v_idx[v] = n_id++;
+    std::vector<double> x_id(static_cast<std::size_t>(n_id), 0.0);
+    auto G_id = inversive_distance_gradient(mesh, x_id, m_id);
+
+    // ── Euclidean side (same mesh, same DOF order) ─────────────────────────
+    auto m_eu = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, m_eu);
+    int n_eu = 0;
+    for (auto v : mesh.vertices()) m_eu.v_idx[v] = n_eu++;
+    std::vector<double> x_eu(static_cast<std::size_t>(n_eu), 0.0);
+    auto G_eu = euclidean_gradient(const_cast<ConformalMesh&>(mesh), x_eu, m_eu);
+
+    // Both should report the same actual angle sum per vertex at u = 0
+    // (since both reproduce ℓ = ℓ_3d at u = 0).  Therefore  Θ − Σ_actual
+    // is identical for the two functionals (Θ default 2π in both).
+    ASSERT_EQ(G_id.size(), G_eu.size());
+    for (std::size_t i = 0; i < G_id.size(); ++i) {
+        EXPECT_NEAR(G_id[i], G_eu[i], 1e-10)
+            << "angle-defect mismatch at u=0, DOF " << i
+            << ": id=" << G_id[i] << "  eu=" << G_eu[i];
+    }
+}

code/tests/cgal/test_layout.cpp

@@ -0,0 +1,369 @@
+// test_layout.cpp
+//
+// Phase 5 — Layout / embedding tests.
+//
+// Strategy: a conformal map that is the identity (natural equilibrium x*=0)
+// should reproduce the mesh's own edge lengths.  We verify:
+//   1. Euclidean layout preserves edge lengths (to solver tolerance).
+//   2. Spherical layout preserves arc-lengths on S².
+//   3. Layout2D has correct size (one entry per vertex).
+//   4. Serialisation round-trip: save JSON → load → DOF vector unchanged.
+//   5. Serialisation round-trip: save XML  → load → DOF vector unchanged.
+//   6. HyperIdeal layout returns success and finite positions.
+
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include "euclidean_functional.hpp"
+#include "spherical_functional.hpp"
+#include "hyper_ideal_functional.hpp"
+#include "newton_solver.hpp"
+#include "layout.hpp"
+#include "serialization.hpp"
+#include <gtest/gtest.h>
+#include <cmath>
+#include <vector>
+#include <filesystem>
+
+using namespace conformallab;
+
+// ────────────────────────────────────────────────────────────────────────────
+// Helpers
+// ────────────────────────────────────────────────────────────────────────────
+
+// Euclidean edge length from layout (2D)
+static double layout_edge_len(const Layout2D& lay,
+                               ConformalMesh& mesh, Halfedge_index h)
+{
+    auto vi = mesh.source(h).idx();
+    auto vj = mesh.target(h).idx();
+    return (lay.uv[vi] - lay.uv[vj]).norm();
+}
+
+// Spherical arc-length from layout (3D)
+static double layout_arc_len(const Layout3D& lay,
+                              ConformalMesh& mesh, Halfedge_index h)
+{
+    auto& a = lay.pos[mesh.source(h).idx()];
+    auto& b = lay.pos[mesh.target(h).idx()];
+    double c = std::max(-1.0, std::min(1.0, a.dot(b)));
+    return std::acos(c);
+}
+
+// Euclidean edge length from maps + x (the "true" target)
+static double maps_edge_len(const EuclideanMaps& m,
+                             ConformalMesh& mesh,
+                             Halfedge_index h,
+                             const std::vector<double>& x)
+{
+    auto get_u = [&](Vertex_index v) -> double {
+        int iv = m.v_idx[v]; return (iv >= 0) ? x[static_cast<std::size_t>(iv)] : 0.0;
+    };
+    Edge_index e = mesh.edge(h);
+    double lam = m.lambda0[e] + get_u(mesh.source(h)) + get_u(mesh.target(h));
+    return std::exp(lam * 0.5);
+}
+
+// Spherical arc-length from maps + x
+static double maps_arc_len(const SphericalMaps& m,
+                            ConformalMesh& mesh,
+                            Halfedge_index h,
+                            const std::vector<double>& x)
+{
+    auto get_u = [&](Vertex_index v) -> double {
+        int iv = m.v_idx[v]; return (iv >= 0) ? x[static_cast<std::size_t>(iv)] : 0.0;
+    };
+    Edge_index e = mesh.edge(h);
+    double lam = m.lambda0[e] + get_u(mesh.source(h)) + get_u(mesh.target(h));
+    return 2.0 * std::asin(std::min(std::exp(lam * 0.5), 1.0));
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 1 — Euclidean layout preserves edge lengths (identity map)
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(Layout, Euclidean_PreservesEdgeLengths)
+{
+    auto mesh = make_quad_strip();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    // Pin first vertex; natural equilibrium so x* = 0
+    auto vit = mesh.vertices().begin();
+    maps.v_idx[*vit++] = -1;
+    int idx = 0;
+    for (; vit != mesh.vertices().end(); ++vit) maps.v_idx[*vit] = idx++;
+    const int n = idx;
+
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto G0 = euclidean_gradient(mesh, x0, maps);
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv >= 0) maps.theta_v[v] -= G0[static_cast<std::size_t>(iv)];
+    }
+
+    // Solve (identity map)
+    auto res = newton_euclidean(mesh, x0, maps, 1e-10, 100);
+    ASSERT_TRUE(res.converged);
+
+    auto layout = euclidean_layout(mesh, res.x, maps);
+    ASSERT_TRUE(layout.success);
+    EXPECT_EQ(layout.uv.size(), mesh.number_of_vertices());
+
+    // Every edge: |layout_len - expected_len| < 1e-8
+    for (auto e : mesh.edges()) {
+        Halfedge_index h = mesh.halfedge(e, 0);
+        double expected = maps_edge_len(maps, mesh, h, res.x);
+        double actual   = layout_edge_len(layout, mesh, h);
+        EXPECT_NEAR(actual, expected, 1e-7)
+            << "Edge " << e << ": expected " << expected << " got " << actual;
+    }
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 2 — Euclidean layout: correct vertex count
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(Layout, Euclidean_CorrectVertexCount)
+{
+    auto mesh = make_triangle();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    auto vit = mesh.vertices().begin();
+    maps.v_idx[*vit++] = -1;
+    int idx = 0;
+    for (; vit != mesh.vertices().end(); ++vit) maps.v_idx[*vit] = idx++;
+    const int n = idx;
+
+    std::vector<double> x(static_cast<std::size_t>(n), 0.0);
+    auto layout = euclidean_layout(mesh, x, maps);
+
+    EXPECT_TRUE(layout.success);
+    EXPECT_EQ(static_cast<int>(layout.uv.size()),
+              static_cast<int>(mesh.number_of_vertices()));
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 3 — Euclidean triangle layout: root face is a valid triangle
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(Layout, Euclidean_TriangleIsNonDegenerate)
+{
+    auto mesh = make_triangle();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    for (auto v : mesh.vertices()) maps.v_idx[v] = static_cast<int>(v.idx());
+
+    std::vector<double> x(mesh.number_of_vertices(), 0.0);
+    auto layout = euclidean_layout(mesh, x, maps);
+    ASSERT_TRUE(layout.success);
+
+    // Area of layout triangle > 0
+    const auto& A = layout.uv[0];
+    const auto& B = layout.uv[1];
+    const auto& C = layout.uv[2];
+    double area = std::abs((B - A).x() * (C - A).y()
+                         - (C - A).x() * (B - A).y()) * 0.5;
+    EXPECT_GT(area, 1e-10) << "Layout triangle must be non-degenerate";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 4 — Spherical layout: arc-lengths preserved (identity map)
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(Layout, Spherical_PreservesArcLengths)
+{
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    // Solve to identity
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto G0 = spherical_gradient(mesh, x0, maps);
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv >= 0) maps.theta_v[v] -= G0[static_cast<std::size_t>(iv)];
+    }
+    auto res = newton_spherical(mesh, x0, maps, 1e-10, 100);
+    ASSERT_TRUE(res.converged);
+
+    auto layout = spherical_layout(mesh, res.x, maps);
+    ASSERT_TRUE(layout.success);
+    EXPECT_EQ(layout.pos.size(), mesh.number_of_vertices());
+
+    for (auto e : mesh.edges()) {
+        Halfedge_index h = mesh.halfedge(e, 0);
+        double expected = maps_arc_len(maps, mesh, h, res.x);
+        double actual   = layout_arc_len(layout, mesh, h);
+        EXPECT_NEAR(actual, expected, 1e-6)
+            << "Spherical edge " << e << ": expected " << expected << " got " << actual;
+    }
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 5 — Spherical layout: all positions on unit sphere
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(Layout, Spherical_PositionsOnUnitSphere)
+{
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), 0.0);
+    auto layout = spherical_layout(mesh, x, maps);
+    ASSERT_TRUE(layout.success);
+
+    for (auto v : mesh.vertices()) {
+        double r = layout.pos[v.idx()].norm();
+        EXPECT_NEAR(r, 1.0, 1e-12) << "Vertex " << v << " not on unit sphere, |p| = " << r;
+    }
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 6 — HyperIdeal layout returns success and finite positions
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(Layout, HyperIdeal_SuccessAndFinitePositions)
+{
+    auto mesh = make_triangle();
+    auto maps = setup_hyper_ideal_maps(mesh);
+    int n = assign_all_dof_indices(mesh, maps);
+
+    // Natural equilibrium at (b=1, a=0.5)
+    std::vector<double> xbase(static_cast<std::size_t>(n));
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v]; if (iv >= 0) xbase[iv] = 1.0;
+    }
+    for (auto e : mesh.edges()) {
+        int ie = maps.e_idx[e]; if (ie >= 0) xbase[ie] = 0.5;
+    }
+    auto G0 = evaluate_hyper_ideal(mesh, xbase, maps, false).gradient;
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v]; if (iv >= 0) maps.theta_v[v] += G0[iv];
+    }
+    for (auto e : mesh.edges()) {
+        int ie = maps.e_idx[e]; if (ie >= 0) maps.theta_e[e] += G0[ie];
+    }
+
+    auto res = newton_hyper_ideal(mesh, xbase, maps, 1e-9, 100);
+    ASSERT_TRUE(res.converged);
+
+    auto layout = hyper_ideal_layout(mesh, res.x, maps);
+    EXPECT_TRUE(layout.success);
+    ASSERT_EQ(layout.uv.size(), mesh.number_of_vertices());
+
+    for (auto v : mesh.vertices()) {
+        auto& p = layout.uv[v.idx()];
+        EXPECT_FALSE(std::isnan(p.x())) << "NaN in layout x for v" << v;
+        EXPECT_FALSE(std::isnan(p.y())) << "NaN in layout y for v" << v;
+        // Poincaré disk: all points within the unit disk
+        EXPECT_LE(p.norm(), 1.0 + 1e-9) << "Point outside Poincaré disk for v" << v;
+    }
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 7 — JSON serialisation round-trip
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(Serialization, JSON_RoundTrip)
+{
+    auto mesh = make_triangle();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    auto vit = mesh.vertices().begin();
+    maps.v_idx[*vit++] = -1;
+    int idx = 0;
+    for (; vit != mesh.vertices().end(); ++vit) maps.v_idx[*vit] = idx++;
+    const int n = idx;
+
+    std::vector<double> x0(static_cast<std::size_t>(n), -0.05);
+    auto G0 = euclidean_gradient(mesh, std::vector<double>(n, 0.0), maps);
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv >= 0) maps.theta_v[v] -= G0[static_cast<std::size_t>(iv)];
+    }
+    auto res = newton_euclidean(mesh, x0, maps, 1e-10, 100);
+    ASSERT_TRUE(res.converged);
+
+    auto layout = euclidean_layout(mesh, res.x, maps);
+
+    const std::string path = "/tmp/conformallab_test_round_trip.json";
+    ASSERT_NO_THROW(save_result_json(path, res, "euclidean",
+        static_cast<int>(mesh.number_of_vertices()),
+        static_cast<int>(mesh.number_of_faces()),
+        &layout));
+    ASSERT_TRUE(std::filesystem::exists(path));
+
+    NewtonResult res2;
+    std::string geom;
+    Layout2D layout2;
+    auto x2 = load_result_json(path, &res2, &geom, &layout2);
+
+    EXPECT_EQ(geom, "euclidean");
+    EXPECT_EQ(res2.converged,  res.converged);
+    EXPECT_EQ(res2.iterations, res.iterations);
+    ASSERT_EQ(x2.size(), res.x.size());
+    for (std::size_t i = 0; i < x2.size(); ++i)
+        EXPECT_NEAR(x2[i], res.x[i], 1e-14);
+
+    EXPECT_TRUE(layout2.success);
+    ASSERT_EQ(layout2.uv.size(), layout.uv.size());
+    for (std::size_t i = 0; i < layout2.uv.size(); ++i) {
+        EXPECT_NEAR(layout2.uv[i].x(), layout.uv[i].x(), 1e-12);
+        EXPECT_NEAR(layout2.uv[i].y(), layout.uv[i].y(), 1e-12);
+    }
+    std::filesystem::remove(path);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 8 — XML serialisation round-trip
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(Serialization, XML_RoundTrip)
+{
+    auto mesh = make_triangle();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    auto vit = mesh.vertices().begin();
+    maps.v_idx[*vit++] = -1;
+    int idx = 0;
+    for (; vit != mesh.vertices().end(); ++vit) maps.v_idx[*vit] = idx++;
+    const int n = idx;
+
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto G0 = euclidean_gradient(mesh, x0, maps);
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv >= 0) maps.theta_v[v] -= G0[static_cast<std::size_t>(iv)];
+    }
+    auto res = newton_euclidean(mesh, x0, maps, 1e-10, 100);
+    ASSERT_TRUE(res.converged);
+
+    auto layout = euclidean_layout(mesh, res.x, maps);
+
+    const std::string path = "/tmp/conformallab_test_round_trip.xml";
+    ASSERT_NO_THROW(save_result_xml(path, res, "euclidean",
+        static_cast<int>(mesh.number_of_vertices()),
+        static_cast<int>(mesh.number_of_faces()),
+        &layout));
+    ASSERT_TRUE(std::filesystem::exists(path));
+
+    NewtonResult res2;
+    std::string geom;
+    Layout2D layout2;
+    auto x2 = load_result_xml(path, &res2, &geom, &layout2);
+
+    EXPECT_EQ(geom, "euclidean");
+    EXPECT_EQ(res2.converged,  res.converged);
+    ASSERT_EQ(x2.size(), res.x.size());
+    for (std::size_t i = 0; i < x2.size(); ++i)
+        EXPECT_NEAR(x2[i], res.x[i], 1e-12);
+
+    EXPECT_TRUE(layout2.success);
+    ASSERT_EQ(layout2.uv.size(), layout.uv.size());
+
+    std::filesystem::remove(path);
+}

code/tests/cgal/test_mesh_io.cpp

@@ -0,0 +1,170 @@
+// test_mesh_io.cpp
+//
+// Phase 4b — CGAL::IO mesh round-trip tests.
+//
+// Tests:
+//   1. OFF write + read round-trip: vertex count, face count, edge count preserved.
+//   2. OBJ write + read round-trip: same topology checks.
+//   3. load_mesh() throws on non-existent file.
+//   4. Round-trip preserves vertex positions (within floating-point precision).
+
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include "mesh_io.hpp"
+#include <gtest/gtest.h>
+#include <cmath>
+#include <cstdio>
+#include <filesystem>
+#include <stdexcept>
+
+using namespace conformallab;
+
+// Helper: temp file path with given extension
+static std::string tmp_path(const char* ext)
+{
+    return std::string("/tmp/conformallab_test.") + ext;
+}
+
+// Helper: delete file if it exists
+static void rm(const std::string& path)
+{
+    std::filesystem::remove(path);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// OFF round-trip: tetrahedron
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(MeshIO, OFF_RoundTrip_Tetrahedron)
+{
+    auto path = tmp_path("off");
+    rm(path);
+
+    auto mesh_out = make_tetrahedron();
+    ASSERT_TRUE(write_mesh(path, mesh_out))
+        << "write_mesh should succeed for tetrahedron";
+
+    ConformalMesh mesh_in;
+    ASSERT_TRUE(read_mesh(path, mesh_in))
+        << "read_mesh should succeed for the written OFF file";
+
+    EXPECT_EQ(mesh_in.number_of_vertices(), mesh_out.number_of_vertices());
+    EXPECT_EQ(mesh_in.number_of_faces(),    mesh_out.number_of_faces());
+    EXPECT_EQ(mesh_in.number_of_edges(),    mesh_out.number_of_edges());
+
+    rm(path);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// OFF round-trip: quad strip
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(MeshIO, OFF_RoundTrip_QuadStrip)
+{
+    auto path = tmp_path("off");
+    rm(path);
+
+    auto mesh_out = make_quad_strip();
+    ASSERT_TRUE(write_mesh(path, mesh_out));
+
+    ConformalMesh mesh_in;
+    ASSERT_TRUE(read_mesh(path, mesh_in));
+
+    EXPECT_EQ(mesh_in.number_of_vertices(), mesh_out.number_of_vertices());
+    EXPECT_EQ(mesh_in.number_of_faces(),    mesh_out.number_of_faces());
+
+    rm(path);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// OBJ round-trip: tetrahedron
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(MeshIO, OBJ_RoundTrip_Tetrahedron)
+{
+    auto path = tmp_path("obj");
+    rm(path);
+
+    auto mesh_out = make_tetrahedron();
+    ASSERT_TRUE(write_mesh(path, mesh_out))
+        << "write_mesh (OBJ) should succeed";
+
+    ConformalMesh mesh_in;
+    ASSERT_TRUE(read_mesh(path, mesh_in))
+        << "read_mesh (OBJ) should succeed";
+
+    EXPECT_EQ(mesh_in.number_of_vertices(), mesh_out.number_of_vertices());
+    EXPECT_EQ(mesh_in.number_of_faces(),    mesh_out.number_of_faces());
+
+    rm(path);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// load_mesh throws on missing file
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(MeshIO, LoadMeshThrowsOnMissingFile)
+{
+    EXPECT_THROW(load_mesh("/tmp/does_not_exist_conflab.off"), std::runtime_error);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Vertex positions survive a round-trip (OFF)
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(MeshIO, OFF_VertexPositionsPreserved)
+{
+    auto path = tmp_path("off");
+    rm(path);
+
+    auto mesh_out = make_tetrahedron();
+    ASSERT_TRUE(write_mesh(path, mesh_out));
+
+    ConformalMesh mesh_in;
+    ASSERT_TRUE(read_mesh(path, mesh_in));
+
+    // Collect and sort vertex positions from both meshes to compare
+    auto collect_sorted = [](const ConformalMesh& m) {
+        std::vector<std::array<double,3>> pts;
+        for (auto v : m.vertices()) {
+            auto p = m.point(v);
+            pts.push_back({CGAL::to_double(p.x()),
+                           CGAL::to_double(p.y()),
+                           CGAL::to_double(p.z())});
+        }
+        std::sort(pts.begin(), pts.end());
+        return pts;
+    };
+
+    auto pts_out = collect_sorted(mesh_out);
+    auto pts_in  = collect_sorted(mesh_in);
+
+    ASSERT_EQ(pts_out.size(), pts_in.size());
+    for (std::size_t i = 0; i < pts_out.size(); ++i) {
+        EXPECT_NEAR(pts_out[i][0], pts_in[i][0], 1e-10);
+        EXPECT_NEAR(pts_out[i][1], pts_in[i][1], 1e-10);
+        EXPECT_NEAR(pts_out[i][2], pts_in[i][2], 1e-10);
+    }
+
+    rm(path);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// save_mesh / load_mesh convenience wrappers
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(MeshIO, SaveLoadConvenienceWrappers)
+{
+    auto path = tmp_path("off");
+    rm(path);
+
+    auto mesh_out = make_quad_strip();
+    EXPECT_NO_THROW(save_mesh(path, mesh_out));
+
+    ConformalMesh mesh_in;
+    EXPECT_NO_THROW(mesh_in = load_mesh(path));
+
+    EXPECT_EQ(mesh_in.number_of_vertices(), mesh_out.number_of_vertices());
+
+    rm(path);
+}

code/tests/cgal/test_newton_phase9a.cpp

@@ -0,0 +1,264 @@
+// test_newton_phase9a.cpp
+//
+// Phase 9a Newton solvers — convergence tests for the two new
+// circle-packing functionals.
+//
+// Validates that:
+//   • newton_cp_euclidean()        — face-based BPS-2010 functional.
+//   • newton_inversive_distance()  — vertex-based Luo-2004 functional.
+// both reach a Newton equilibrium (‖G‖∞ < 1e-8) in < 30 iterations
+// on a range of test meshes, and that the converged solution satisfies
+// the relevant geometric invariants.
+
+#include "newton_solver.hpp"
+#include "cp_euclidean_functional.hpp"
+#include "inversive_distance_functional.hpp"
+#include "mesh_builder.hpp"
+#include "conformal_mesh.hpp"
+
+#include <gtest/gtest.h>
+#include <vector>
+
+using namespace conformallab;
+
+namespace {
+
+// Open 3-face mesh (tetrahedron minus one face) — exercises boundary edges.
+inline ConformalMesh make_open_3face_mesh()
+{
+    ConformalMesh mesh;
+    auto v0 = mesh.add_vertex(Point3( 1,  1,  1));
+    auto v1 = mesh.add_vertex(Point3( 1, -1, -1));
+    auto v2 = mesh.add_vertex(Point3(-1,  1, -1));
+    auto v3 = mesh.add_vertex(Point3(-1, -1,  1));
+    mesh.add_face(v0, v2, v1);
+    mesh.add_face(v0, v1, v3);
+    mesh.add_face(v0, v3, v2);
+    return mesh;
+}
+
+} // anonymous
+
+// ════════════════════════════════════════════════════════════════════════════
+// 1. CP-Euclidean Newton — orthogonal circle packing
+//
+// Setup matches CPEuclideanFunctionalTest.java (Java parity at the
+// solver level): θ_e = π/2 everywhere, φ_f = 2π for all faces.  Use
+// the "natural-phi" trick (analog of natural-theta in Euclidean):
+// adjust φ so that ρ = 0 is the natural equilibrium → Newton must
+// converge in zero iterations.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonPhase9a, CPEuclidean_NaturalPhi_ClosedTetrahedron_ConvergesInZeroIterations)
+{
+    auto mesh = make_tetrahedron();
+    auto m    = setup_cp_euclidean_maps(mesh);
+    const int n = assign_cp_euclidean_face_dof_indices(mesh, m);
+    ASSERT_EQ(n, 3);
+
+    // Natural-phi: shift φ_f so the gradient at ρ = 0 is zero.
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto G0 = cp_euclidean_gradient(mesh, x0, m);
+    for (auto f : mesh.faces()) {
+        int i = m.f_idx[f];
+        if (i < 0) continue;
+        m.phi_f[f] -= G0[static_cast<std::size_t>(i)];
+    }
+
+    auto res = newton_cp_euclidean(mesh, x0, m);
+    EXPECT_TRUE(res.converged);
+    EXPECT_EQ(res.iterations, 0)
+        << "natural-phi pre-shift should make x=0 the equilibrium";
+    EXPECT_LT(res.grad_inf_norm, 1e-10);
+    for (double r : res.x) EXPECT_NEAR(r, 0.0, 1e-12);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 2. CP-Euclidean Newton — perturbed equilibrium converges back to 0
+//
+// Same setup as test 1, but start from a small perturbation.  The
+// strictly-convex BPS-2010 energy means Newton must converge back
+// to the natural-phi equilibrium ρ = 0.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonPhase9a, CPEuclidean_PerturbedStart_ConvergesBackToEquilibrium)
+{
+    auto mesh = make_tetrahedron();
+    auto m    = setup_cp_euclidean_maps(mesh);
+    const int n = assign_cp_euclidean_face_dof_indices(mesh, m);
+
+    // Apply natural-phi (equilibrium at ρ=0).
+    std::vector<double> x0_zero(static_cast<std::size_t>(n), 0.0);
+    auto G0 = cp_euclidean_gradient(mesh, x0_zero, m);
+    for (auto f : mesh.faces()) {
+        int i = m.f_idx[f];
+        if (i < 0) continue;
+        m.phi_f[f] -= G0[static_cast<std::size_t>(i)];
+    }
+
+    // Start from a perturbation.
+    std::vector<double> x0 = {0.1, -0.2, 0.15};
+    auto res = newton_cp_euclidean(mesh, x0, m);
+
+    EXPECT_TRUE(res.converged);
+    EXPECT_LT(res.iterations, 30);
+    EXPECT_LT(res.grad_inf_norm, 1e-8);
+    // Strictly-convex unique minimum → converges back to ρ=0.
+    for (double r : res.x) EXPECT_NEAR(r, 0.0, 1e-6);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 3. CP-Euclidean Newton — open mesh (boundary edges)
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonPhase9a, CPEuclidean_OpenTetrahedron_NaturalPhi_Converges)
+{
+    auto mesh = make_open_3face_mesh();
+    auto m    = setup_cp_euclidean_maps(mesh);
+    const int n = assign_cp_euclidean_face_dof_indices(mesh, m);
+    ASSERT_EQ(n, 2);
+
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto G0 = cp_euclidean_gradient(mesh, x0, m);
+    for (auto f : mesh.faces()) {
+        int i = m.f_idx[f];
+        if (i < 0) continue;
+        m.phi_f[f] -= G0[static_cast<std::size_t>(i)];
+    }
+
+    auto res = newton_cp_euclidean(mesh, x0, m);
+    EXPECT_TRUE(res.converged);
+    EXPECT_LT(res.iterations, 30);
+    EXPECT_LT(res.grad_inf_norm, 1e-8);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 4. Inversive-Distance Newton — natural-theta on triangle
+//
+// At u = 0, Bowers-Stephenson init reproduces the input edge lengths
+// exactly.  Natural-theta then shifts Θ so the gradient is zero, making
+// u = 0 the equilibrium.  Newton must converge in zero iterations.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonPhase9a, InversiveDistance_NaturalTheta_Triangle_ConvergesInZero)
+{
+    auto mesh = make_triangle();
+    auto m    = setup_inversive_distance_maps(mesh);
+    compute_inversive_distance_init_from_mesh(mesh, m);
+
+    int n = 0;
+    for (auto v : mesh.vertices()) m.v_idx[v] = n++;
+
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto G0 = inversive_distance_gradient(mesh, x0, m);
+    for (auto v : mesh.vertices()) {
+        int i = m.v_idx[v];
+        m.theta_v[v] -= G0[static_cast<std::size_t>(i)];
+    }
+
+    auto res = newton_inversive_distance(mesh, x0, m);
+    EXPECT_TRUE(res.converged);
+    EXPECT_EQ(res.iterations, 0);
+    EXPECT_LT(res.grad_inf_norm, 1e-10);
+    for (double u : res.x) EXPECT_NEAR(u, 0.0, 1e-12);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 5. Inversive-Distance Newton — perturbed start on quad strip
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonPhase9a, InversiveDistance_PerturbedQuadStrip_Converges)
+{
+    auto mesh = make_quad_strip();
+    auto m    = setup_inversive_distance_maps(mesh);
+    compute_inversive_distance_init_from_mesh(mesh, m);
+
+    // Pin vertex 0; index the rest.
+    auto vit = mesh.vertices().begin();
+    m.v_idx[*vit++] = -1;
+    int n = 0;
+    for (; vit != mesh.vertices().end(); ++vit) m.v_idx[*vit] = n++;
+
+    // Natural-theta with the pin in place.
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto G0 = inversive_distance_gradient(mesh, x0, m);
+    for (auto v : mesh.vertices()) {
+        int i = m.v_idx[v];
+        if (i >= 0) m.theta_v[v] -= G0[static_cast<std::size_t>(i)];
+    }
+
+    // Perturb away from the equilibrium and watch it return.
+    std::vector<double> x_pert(static_cast<std::size_t>(n), -0.05);
+    auto res = newton_inversive_distance(mesh, x_pert, m);
+
+    EXPECT_TRUE(res.converged);
+    EXPECT_LT(res.iterations, 30);
+    EXPECT_LT(res.grad_inf_norm, 1e-8);
+    // Strictly-convex unique minimum on the open domain → back to 0.
+    for (double u : res.x) EXPECT_NEAR(u, 0.0, 1e-6);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 6. Inversive-Distance Newton — tetrahedron (closed mesh)
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonPhase9a, InversiveDistance_PerturbedTetrahedron_Converges)
+{
+    auto mesh = make_tetrahedron();
+    auto m    = setup_inversive_distance_maps(mesh);
+    compute_inversive_distance_init_from_mesh(mesh, m);
+
+    // Closed mesh — pin one vertex to remove the gauge mode.
+    auto vit = mesh.vertices().begin();
+    m.v_idx[*vit++] = -1;
+    int n = 0;
+    for (; vit != mesh.vertices().end(); ++vit) m.v_idx[*vit] = n++;
+
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto G0 = inversive_distance_gradient(mesh, x0, m);
+    for (auto v : mesh.vertices()) {
+        int i = m.v_idx[v];
+        if (i >= 0) m.theta_v[v] -= G0[static_cast<std::size_t>(i)];
+    }
+
+    std::vector<double> x_pert(static_cast<std::size_t>(n), -0.1);
+    auto res = newton_inversive_distance(mesh, x_pert, m);
+
+    EXPECT_TRUE(res.converged);
+    EXPECT_LT(res.iterations, 30);
+    EXPECT_LT(res.grad_inf_norm, 1e-8);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 7. CP-Euclidean Newton — uses analytic Hessian (NOT FD)
+//
+// Regression guard: verify the solver actually calls cp_euclidean_hessian
+// (the analytic 2×2-per-edge formula) rather than degenerating to a
+// per-iteration FD pass.  If iteration count exceeds a tight upper bound
+// for a tiny mesh, that would suggest a slow inner Hessian computation
+// or a wrong-sign mistake.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonPhase9a, CPEuclidean_UsesAnalyticHessian)
+{
+    auto mesh = make_tetrahedron();
+    auto m    = setup_cp_euclidean_maps(mesh);
+    const int n = assign_cp_euclidean_face_dof_indices(mesh, m);
+
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto G0 = cp_euclidean_gradient(mesh, x0, m);
+    for (auto f : mesh.faces()) {
+        int i = m.f_idx[f];
+        if (i < 0) continue;
+        m.phi_f[f] -= G0[static_cast<std::size_t>(i)];
+    }
+
+    // Strong perturbation — quadratic Newton with analytic Hessian
+    // should still converge in a handful of iterations.
+    std::vector<double> x_pert = {0.5, -0.4, 0.3};
+    auto res = newton_cp_euclidean(mesh, x_pert, m);
+
+    EXPECT_TRUE(res.converged);
+    EXPECT_LE(res.iterations, 10)
+        << "analytic Hessian: expect very fast convergence on a 3-DOF problem";
+}

code/tests/cgal/test_newton_solver.cpp

@@ -0,0 +1,500 @@
+// test_newton_solver.cpp
+//
+// Phase 4 — Newton solver tests.
+//
+// Design principle:
+//   We test convergence to a KNOWN equilibrium.  For the spherical tetrahedron
+//   x* = 0 is built-in (G(0) ≈ 0 by construction).  For Euclidean meshes we
+//   use "natural theta": set theta_v[v] = actual angle sum at x=0, which makes
+//   x* = 0 the exact equilibrium by definition.
+//   For HyperIdeal we use the same "natural target" trick at a valid base point
+//   (b=1.0, a=0.5), since x=0 is degenerate for the HyperIdeal functional.
+//
+// Tests:
+//   Spherical:
+//     1. Converges from x=[-0.2,...] to x*=0 (spherical tetrahedron).
+//     2. Converges in few iterations (quadratic convergence near equilibrium).
+//     3. Converges from a large perturbation x=[-0.5,...].
+//     4. Result fields (x.size, grad_inf_norm) are self-consistent.
+//
+//   Euclidean:
+//     5. Converges (triangle, 1 pinned vertex, natural theta).
+//     6. Converges (quad strip, 1 pinned vertex, natural theta).
+//     7. Converges with explicitly chosen mixed pinned/variable layout.
+//
+//   HyperIdeal:
+//     8.  Converges on triangle (all variable, natural targets).
+//     9.  Result fields self-consistent.
+//    10.  Converges on tetrahedron (10 DOFs, larger mesh).
+//    11.  SparseQR: result consistent (valid starting region).
+//
+//   SparseQR fallback (direct unit tests):
+//    12.  solve_linear_system recovers correct solution on rank-deficient matrix.
+//    13.  solve_linear_system sets fallback_used=true on a singular matrix.
+//    14.  Euclidean Newton on closed tetrahedron (no pinned vertex) converges
+//         via SparseQR gauge-mode handling.
+
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include "euclidean_functional.hpp"
+#include "spherical_functional.hpp"
+#include "hyper_ideal_functional.hpp"
+#include "newton_solver.hpp"
+#include <Eigen/Dense>
+#include <gtest/gtest.h>
+#include <cmath>
+#include <vector>
+
+using namespace conformallab;
+
+// ────────────────────────────────────────────────────────────────────────────
+// Helper: set theta_v[v] = actual angle sum at x=0 for each variable vertex.
+//
+// Requires that DOF indices have already been assigned (v_idx populated).
+// Uses euclidean_gradient directly so the formula is exact and consistent.
+// ────────────────────────────────────────────────────────────────────────────
+static void set_natural_euclidean_theta(ConformalMesh& mesh, EuclideanMaps& maps, int n)
+{
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    // G[iv] = theta_v[v] - sum_alpha(x=0)
+    // so  sum_alpha(x=0) = theta_v[v] - G[iv]
+    auto G = euclidean_gradient(mesh, x0, maps);
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv < 0) continue;
+        maps.theta_v[v] -= G[static_cast<std::size_t>(iv)];
+    }
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Spherical 1 — Converges from moderate perturbation
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonSolver, Spherical_ConvergesFromPerturbation)
+{
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x0(static_cast<std::size_t>(n), -0.2);
+    auto res = newton_spherical(mesh, x0, maps, /*tol=*/1e-8, /*max_iter=*/50);
+
+    EXPECT_TRUE(res.converged)
+        << "Newton (spherical) should converge; grad_inf_norm = " << res.grad_inf_norm;
+    EXPECT_LT(res.grad_inf_norm, 1e-8);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Spherical 2 — Quadratic convergence: few iterations suffice
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonSolver, Spherical_FewIterations)
+{
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x0(static_cast<std::size_t>(n), -0.2);
+    auto res = newton_spherical(mesh, x0, maps, /*tol=*/1e-8, /*max_iter=*/50);
+
+    EXPECT_LE(res.iterations, 20)
+        << "Newton should converge in ≤ 20 iterations; took " << res.iterations;
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Spherical 3 — Large perturbation: global convergence via line search
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonSolver, Spherical_ConvergesFromLargePerturbation)
+{
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x0(static_cast<std::size_t>(n), -0.5);
+    auto res = newton_spherical(mesh, x0, maps, /*tol=*/1e-8, /*max_iter=*/100);
+
+    EXPECT_TRUE(res.converged)
+        << "Newton (spherical, large perturbation) should converge; "
+           "grad_inf_norm = " << res.grad_inf_norm;
+    EXPECT_LT(res.grad_inf_norm, 1e-8);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Spherical 4 — Result fields are self-consistent
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonSolver, Spherical_ResultFieldsConsistent)
+{
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x0(static_cast<std::size_t>(n), -0.1);
+    auto res = newton_spherical(mesh, x0, maps, /*tol=*/1e-8);
+
+    EXPECT_EQ(static_cast<int>(res.x.size()), n);
+
+    // Reported grad_inf_norm must match re-computed gradient at res.x
+    auto G = spherical_gradient(mesh, res.x, maps);
+    double actual_inf = 0.0;
+    for (double v : G) actual_inf = std::max(actual_inf, std::abs(v));
+    EXPECT_NEAR(actual_inf, res.grad_inf_norm, 1e-10);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Euclidean 1 — Triangle with 1 pinned vertex + natural theta
+//
+// make_triangle(): 3 vertices.  Pin v0 → 2 free DOFs.
+// With natural theta: x* = 0 (G(0) = 0 by construction).
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonSolver, Euclidean_ConvergesTrianglePinned)
+{
+    auto mesh = make_triangle();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    // Pin first vertex, assign sequential DOFs to the other two
+    auto vit = mesh.vertices().begin();
+    Vertex_index v0 = *vit++;
+    maps.v_idx[v0] = -1;     // pinned
+    int idx = 0;
+    for (; vit != mesh.vertices().end(); ++vit)
+        maps.v_idx[*vit] = idx++;
+    int n = idx;             // = 2
+
+    set_natural_euclidean_theta(mesh, maps, n);
+
+    std::vector<double> x0(static_cast<std::size_t>(n), -0.1);
+    auto res = newton_euclidean(mesh, x0, maps, /*tol=*/1e-8, /*max_iter=*/50);
+
+    EXPECT_TRUE(res.converged)
+        << "Newton (Euclidean, triangle, pinned) should converge; "
+           "grad_inf_norm = " << res.grad_inf_norm;
+    EXPECT_LT(res.grad_inf_norm, 1e-8);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Euclidean 2 — Quad strip with 1 pinned vertex + natural theta
+//
+// make_quad_strip(): 4 vertices, 2 faces.  Pin v0 → 3 free DOFs.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonSolver, Euclidean_ConvergesQuadStripPinned)
+{
+    auto mesh = make_quad_strip();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    // Pin first vertex
+    auto vit = mesh.vertices().begin();
+    Vertex_index v0 = *vit++;
+    maps.v_idx[v0] = -1;
+    int idx = 0;
+    for (; vit != mesh.vertices().end(); ++vit)
+        maps.v_idx[*vit] = idx++;
+    int n = idx;             // = 3
+
+    set_natural_euclidean_theta(mesh, maps, n);
+
+    std::vector<double> x0(static_cast<std::size_t>(n), -0.15);
+    auto res = newton_euclidean(mesh, x0, maps, /*tol=*/1e-8, /*max_iter=*/50);
+
+    EXPECT_TRUE(res.converged)
+        << "Newton (Euclidean, quad strip, pinned) should converge; "
+           "grad_inf_norm = " << res.grad_inf_norm;
+    EXPECT_LT(res.grad_inf_norm, 1e-8);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Euclidean 3 — Mixed pinned layout: explicit vertex assignment
+//
+// Quad strip: v0 pinned, v1/v2/v3 free.
+// Natural theta set AFTER DOF assignment so that x* = 0 is the equilibrium
+// for the free vertices (with v0 fixed at u0=0).
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonSolver, Euclidean_ConvergesMixedPinned)
+{
+    auto mesh = make_quad_strip();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    // Explicitly assign DOF indices
+    auto vit = mesh.vertices().begin();
+    Vertex_index v0 = *vit++;
+    Vertex_index v1 = *vit++;
+    Vertex_index v2 = *vit++;
+    Vertex_index v3 = *vit;
+
+    maps.v_idx[v0] = -1;   // pinned at u0 = 0
+    maps.v_idx[v1] =  0;
+    maps.v_idx[v2] =  1;
+    maps.v_idx[v3] =  2;
+    const int n = 3;
+
+    // Set natural theta AFTER pinning so that x* = [0,0,0] is the equilibrium
+    set_natural_euclidean_theta(mesh, maps, n);
+
+    std::vector<double> x0 = {-0.1, -0.15, -0.05};
+    auto res = newton_euclidean(mesh, x0, maps, /*tol=*/1e-8, /*max_iter=*/50);
+
+    EXPECT_TRUE(res.converged)
+        << "Newton (Euclidean, mixed pinned) should converge; "
+           "grad_inf_norm = " << res.grad_inf_norm;
+    EXPECT_LT(res.grad_inf_norm, 1e-8);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Helper: set HyperIdeal target angles to actual sums at a non-degenerate
+// base point (b_base, a_base), making that point the equilibrium x*.
+//
+// Note: x = 0 is degenerate for the HyperIdeal functional (log-space; the
+// functional requires b_i > 0 / a_e > 0).  We therefore choose a valid base
+// point, evaluate G there, and absorb G into the targets so that G(xbase) = 0.
+// Newton tests then start from a perturbation of xbase.
+//
+// Returns xbase so callers can construct a perturbed starting point.
+// ════════════════════════════════════════════════════════════════════════════
+static std::vector<double> set_natural_hyper_ideal_targets(
+    ConformalMesh& mesh, HyperIdealMaps& maps, int n,
+    double b_base = 1.0, double a_base = 0.5)
+{
+    const auto sz = static_cast<std::size_t>(n);
+    std::vector<double> xbase(sz, 0.0);
+
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv >= 0) xbase[static_cast<std::size_t>(iv)] = b_base;
+    }
+    for (auto e : mesh.edges()) {
+        int ie = maps.e_idx[e];
+        if (ie >= 0) xbase[static_cast<std::size_t>(ie)] = a_base;
+    }
+
+    // G = Σβ - theta_target (initial target = 0 → G = Σβ = "actual" angles)
+    auto G = evaluate_hyper_ideal(mesh, xbase, maps, /*energy=*/false).gradient;
+
+    // Set target := actual so that G(xbase) = actual - target = 0
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv < 0) continue;
+        maps.theta_v[v] += G[static_cast<std::size_t>(iv)];
+    }
+    for (auto e : mesh.edges()) {
+        int ie = maps.e_idx[e];
+        if (ie < 0) continue;
+        maps.theta_e[e] += G[static_cast<std::size_t>(ie)];
+    }
+
+    return xbase;
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// HyperIdeal 1 — Triangle, all DOFs variable, converges from perturbation
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonSolver, HyperIdeal_ConvergesTriangleAllVariable)
+{
+    auto mesh = make_triangle();
+    auto maps = setup_hyper_ideal_maps(mesh);
+    int n = assign_all_dof_indices(mesh, maps);
+
+    // xbase = (b=1.0, a=0.5) is the equilibrium after natural-target setup.
+    auto xbase = set_natural_hyper_ideal_targets(mesh, maps, n);
+
+    // Perturb by +0.2 uniformly
+    std::vector<double> x0 = xbase;
+    for (auto& v : x0) v += 0.2;
+
+    auto res = newton_hyper_ideal(mesh, x0, maps, /*tol=*/1e-7, /*max_iter=*/100);
+
+    EXPECT_TRUE(res.converged)
+        << "Newton (HyperIdeal, triangle) should converge; "
+           "grad_inf_norm = " << res.grad_inf_norm;
+    EXPECT_LT(res.grad_inf_norm, 1e-7);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// HyperIdeal 2 — Result fields self-consistent
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonSolver, HyperIdeal_ResultFieldsConsistent)
+{
+    auto mesh = make_triangle();
+    auto maps = setup_hyper_ideal_maps(mesh);
+    int n = assign_all_dof_indices(mesh, maps);
+
+    auto xbase = set_natural_hyper_ideal_targets(mesh, maps, n);
+
+    std::vector<double> x0 = xbase;
+    for (auto& v : x0) v += 0.1;
+
+    auto res = newton_hyper_ideal(mesh, x0, maps, /*tol=*/1e-7, /*max_iter=*/100);
+
+    EXPECT_EQ(static_cast<int>(res.x.size()), n);
+
+    // Reported grad_inf_norm must match re-computed gradient at res.x
+    auto G = evaluate_hyper_ideal(mesh, res.x, maps, false).gradient;
+    double actual_inf = 0.0;
+    for (double v : G) actual_inf = std::max(actual_inf, std::abs(v));
+    EXPECT_NEAR(actual_inf, res.grad_inf_norm, 1e-9);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// HyperIdeal 3 — Tetrahedron (10 DOFs): 4 vertex b-vals + 6 edge a-vals
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonSolver, HyperIdeal_ConvergesTetrahedron)
+{
+    auto mesh = make_tetrahedron();
+    auto maps = setup_hyper_ideal_maps(mesh);
+    int n = assign_all_dof_indices(mesh, maps);
+
+    auto xbase = set_natural_hyper_ideal_targets(mesh, maps, n);
+
+    // Perturb by +0.15
+    std::vector<double> x0 = xbase;
+    for (auto& v : x0) v += 0.15;
+
+    auto res = newton_hyper_ideal(mesh, x0, maps, /*tol=*/1e-7, /*max_iter=*/200);
+
+    EXPECT_TRUE(res.converged)
+        << "Newton (HyperIdeal, tetrahedron) should converge; "
+           "grad_inf_norm = " << res.grad_inf_norm;
+    EXPECT_LT(res.grad_inf_norm, 1e-7);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// HyperIdeal 4 — SparseQR fallback: solver returns a result (no crash)
+//
+// With all targets = 0 the equilibrium is not at x=0 but the solver should
+// at minimum not crash and return a consistent result struct.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NewtonSolver, HyperIdeal_SparseQRFallbackNoCrash)
+{
+    auto mesh = make_triangle();
+    auto maps = setup_hyper_ideal_maps(mesh);
+    int n = assign_all_dof_indices(mesh, maps);
+
+    // Leave targets at their default (0): solver tries to solve but the
+    // "equilibrium" is at some unknown x*.  With valid starting point the
+    // Hessian is positive-definite and the solver should not crash.
+    // We don't assert convergence — just that the result struct is consistent.
+    std::vector<double> x0(static_cast<std::size_t>(n), 1.0);
+    // Mix vertex / edge DOFs: b=1.0, a=0.5 (valid region of the functional)
+    for (auto e : mesh.edges()) {
+        int ie = maps.e_idx[e];
+        if (ie >= 0) x0[static_cast<std::size_t>(ie)] = 0.5;
+    }
+    auto res = newton_hyper_ideal(mesh, x0, maps, /*tol=*/1e-7, /*max_iter=*/50);
+
+    // Struct fields must always be populated
+    EXPECT_EQ(static_cast<int>(res.x.size()), n);
+    EXPECT_GE(res.iterations, 0);
+    EXPECT_FALSE(std::isnan(res.grad_inf_norm));
+    EXPECT_FALSE(std::isinf(res.grad_inf_norm));
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// SparseQR fallback — Test 12: solve_linear_system recovers correct solution
+//
+// The public API solve_linear_system(A, rhs) must return the correct answer
+// for a well-conditioned full-rank system (LDLT path taken).
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SparseQRFallback, FullRankSystem_CorrectSolution)
+{
+    // Build a simple 3×3 diagonal PD matrix: A = diag(1, 2, 3)
+    Eigen::SparseMatrix<double> A(3, 3);
+    A.insert(0, 0) = 1.0;
+    A.insert(1, 1) = 2.0;
+    A.insert(2, 2) = 3.0;
+    A.makeCompressed();
+
+    Eigen::VectorXd rhs(3);
+    rhs << 1.0, 4.0, 9.0;   // solution = [1, 2, 3]
+
+    bool fallback = true;     // expect it to be set to false (LDLT succeeds)
+    Eigen::VectorXd x = conformallab::solve_linear_system(A, rhs, &fallback);
+
+    EXPECT_FALSE(fallback) << "Full-rank system: LDLT should succeed (no SparseQR needed)";
+    EXPECT_NEAR(x[0], 1.0, 1e-12);
+    EXPECT_NEAR(x[1], 2.0, 1e-12);
+    EXPECT_NEAR(x[2], 3.0, 1e-12);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// SparseQR fallback — Test 13: fallback_used=true on a singular matrix
+//
+// Construct a symmetric 3×3 matrix of rank 1 where LDLT fails (the (2,2)
+// pivot is zero).  SparseQR finds the minimum-norm least-squares solution.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SparseQRFallback, SingularMatrix_FallbackActivated)
+{
+    // A = [[2, 0, 0],
+    //      [0, 0, 0],   ← zero pivot → LDLT failure
+    //      [0, 0, 3]]
+    // rhs compatible with the row space: [2, 0, 3] → solution [1, 0, 1]
+    Eigen::SparseMatrix<double> A(3, 3);
+    A.insert(0, 0) = 2.0;
+    // row/col 1 deliberately all-zero
+    A.insert(2, 2) = 3.0;
+    A.makeCompressed();
+
+    Eigen::VectorXd rhs(3);
+    rhs << 2.0, 0.0, 3.0;
+
+    bool fallback = false;
+    Eigen::VectorXd x = conformallab::solve_linear_system(A, rhs, &fallback);
+
+    EXPECT_TRUE(fallback) << "Singular matrix: SparseQR fallback must be triggered";
+    // SparseQR min-norm solution: x[0]=1, x[1]=0, x[2]=1
+    EXPECT_NEAR(x[0], 1.0, 1e-10);
+    EXPECT_NEAR(x[1], 0.0, 1e-10);
+    EXPECT_NEAR(x[2], 1.0, 1e-10);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// SparseQR fallback — Test 14: Euclidean Newton on a closed mesh, no pinning
+//
+// make_tetrahedron() is a closed surface (4 vertices, 4 faces).  Without a
+// pinned vertex the Euclidean Hessian has a 1-D null space (uniform scale
+// gauge mode): H·1 = 0.  SimplicialLDLT fails on this rank-deficient H;
+// SparseQR finds the min-norm Newton step orthogonal to the null space.
+//
+// The gradient always lives in the row space of H (Σ G_v = 0 by angle-sum
+// invariance), so the SparseQR step is also the Newton step and the solver
+// converges to the natural equilibrium.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SparseQRFallback, Euclidean_ClosedMeshNoPinConverges)
+{
+    auto mesh = make_tetrahedron();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    // Assign all 4 vertices as free DOFs (no pinning).
+    int idx = 0;
+    for (auto v : mesh.vertices())
+        maps.v_idx[v] = idx++;
+    const int n = idx;   // = 4
+
+    // Natural theta: equilibrium at x* = 0.
+    set_natural_euclidean_theta(mesh, maps, n);
+
+    std::vector<double> x0(static_cast<std::size_t>(n), -0.1);
+    auto res = newton_euclidean(mesh, x0, maps, /*tol=*/1e-8, /*max_iter=*/100);
+
+    EXPECT_TRUE(res.converged)
+        << "Euclidean Newton on closed tetrahedron (no pin) must converge via SparseQR; "
+           "grad_inf_norm = " << res.grad_inf_norm;
+    EXPECT_LT(res.grad_inf_norm, 1e-8);
+}

code/tests/cgal/test_phase6.cpp

@@ -0,0 +1,406 @@
+// test_phase6.cpp
+//
+// Phase 6 — Tests for:
+//   - gauss_bonnet.hpp  : euler_characteristic, genus, GB sum/rhs, check, enforce
+//   - cut_graph.hpp     : compute_cut_graph (tree-cotree algorithm)
+//   - layout.hpp Phase6 : exact hyperbolic trilateration math
+//                         normalise_euclidean (centroid + PCA)
+//                         normalise_hyperbolic (Möbius centering)
+
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include "euclidean_functional.hpp"
+#include "spherical_functional.hpp"
+#include "hyper_ideal_functional.hpp"
+#include "newton_solver.hpp"
+#include "gauss_bonnet.hpp"
+#include "cut_graph.hpp"
+#include "layout.hpp"
+#include <gtest/gtest.h>
+#include <cmath>
+#include <complex>
+#include <vector>
+
+using namespace conformallab;
+
+// mesh_builder.hpp provides make_triangle(), make_tetrahedron(), make_quad_strip().
+
+// ════════════════════════════════════════════════════════════════════════════
+// GaussBonnet — topology helpers
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(GaussBonnet, EulerCharacteristic_Triangle)
+{
+    // V=3 E=3 F=1  →  χ = 1
+    auto m = make_triangle();
+    EXPECT_EQ(euler_characteristic(m), 1);
+}
+
+TEST(GaussBonnet, EulerCharacteristic_QuadStrip)
+{
+    // V=4 E=5 F=2  →  χ = 1
+    auto m = make_quad_strip();
+    EXPECT_EQ(euler_characteristic(m), 1);
+}
+
+TEST(GaussBonnet, EulerCharacteristic_Tetrahedron)
+{
+    // V=4 E=6 F=4  →  χ = 2
+    auto m = make_tetrahedron();
+    EXPECT_EQ(euler_characteristic(m), 2);
+}
+
+TEST(GaussBonnet, Genus_Tetrahedron_IsZero)
+{
+    auto m = make_tetrahedron();
+    EXPECT_EQ(genus(m), 0);
+}
+
+TEST(GaussBonnet, RHS_Triangle)
+{
+    // χ=1  →  2π·χ = 2π
+    auto m = make_triangle();
+    EXPECT_NEAR(gauss_bonnet_rhs(m), 2.0 * M_PI, 1e-12);
+}
+
+TEST(GaussBonnet, RHS_Tetrahedron)
+{
+    // χ=2  →  2π·χ = 4π
+    auto m = make_tetrahedron();
+    EXPECT_NEAR(gauss_bonnet_rhs(m), 4.0 * M_PI, 1e-12);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// GaussBonnet — sum / deficit / check / enforce
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(GaussBonnet, DeficitIsNonZeroWithDefaultTheta)
+{
+    // Default theta_v = 2π at all V=4 vertices of quad-strip  →
+    //   Σ(2π−2π) = 0,  rhs = 2π·1 = 2π  →  deficit = −2π ≠ 0.
+    auto m = make_quad_strip();
+    auto maps = setup_euclidean_maps(m);
+    double def = gauss_bonnet_deficit(m, maps);
+    EXPECT_GT(std::abs(def), 1e-6);
+}
+
+TEST(GaussBonnet, EnforceAdjustsTheta_Deficit_BecomesZero)
+{
+    auto m = make_quad_strip();
+    auto maps = setup_euclidean_maps(m);
+    // Set clearly wrong theta
+    for (auto v : m.vertices()) maps.theta_v[v] = M_PI;
+    EXPECT_GT(std::abs(gauss_bonnet_deficit(m, maps)), 1e-6);
+
+    enforce_gauss_bonnet(m, maps);
+    EXPECT_NEAR(gauss_bonnet_deficit(m, maps), 0.0, 1e-10);
+}
+
+TEST(GaussBonnet, CheckThrowsWhenViolated)
+{
+    auto m = make_triangle();
+    auto maps = setup_euclidean_maps(m);
+    // theta_v = 2π everywhere  →  sum = 0, rhs = 2π  →  |deficit| = 2π
+    EXPECT_THROW(check_gauss_bonnet(m, maps), std::runtime_error);
+}
+
+TEST(GaussBonnet, CheckPassesAfterEnforce)
+{
+    auto m = make_triangle();
+    auto maps = setup_euclidean_maps(m);
+    enforce_gauss_bonnet(m, maps);
+    EXPECT_NO_THROW(check_gauss_bonnet(m, maps));
+}
+
+TEST(GaussBonnet, SumAfterEnforce_EqualsRHS)
+{
+    auto m = make_tetrahedron();
+    auto maps = setup_euclidean_maps(m);
+    // theta_v starts at 2π everywhere; sum = 0, rhs = 4π
+    enforce_gauss_bonnet(m, maps);
+    double lhs = gauss_bonnet_sum(m, maps);
+    double rhs = gauss_bonnet_rhs(m);
+    EXPECT_NEAR(lhs, rhs, 1e-10);
+}
+
+TEST(GaussBonnet, ManuallySetAnalyticalTheta_PassesCheck)
+{
+    // Single triangle (χ=1): need Σ(2π−Θ_v) = 2π.
+    // Set each of the 3 boundary vertices to Θ_v = 4π/3.
+    // Then Σ(2π − 4π/3) = 3·(2π/3) = 2π. ✓
+    auto m = make_triangle();
+    auto maps = setup_euclidean_maps(m);
+    for (auto v : m.vertices()) maps.theta_v[v] = 4.0 * M_PI / 3.0;
+    EXPECT_NO_THROW(check_gauss_bonnet(m, maps));
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// CutGraph — tree-cotree algorithm
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(CutGraph, Triangle_ZeroCutEdges)
+{
+    // Open disk → genus 0 → 0 cut edges.
+    auto m = make_triangle();
+    auto cg = compute_cut_graph(m);
+    EXPECT_EQ(cg.cut_edge_indices.size(), 0u);
+}
+
+TEST(CutGraph, QuadStrip_ZeroCutEdges)
+{
+    // Open disk → genus 0 → 0 cut edges.
+    auto m = make_quad_strip();
+    auto cg = compute_cut_graph(m);
+    EXPECT_EQ(cg.cut_edge_indices.size(), 0u);
+}
+
+TEST(CutGraph, Tetrahedron_ZeroCutEdges_ClosedSphere)
+{
+    // Closed genus-0 surface → 2g = 0 cut edges.
+    auto m = make_tetrahedron();
+    auto cg = compute_cut_graph(m);
+    EXPECT_EQ(cg.cut_edge_indices.size(), 0u);
+    EXPECT_EQ(cg.genus, 0);
+}
+
+TEST(CutGraph, FlagsVsIndicesConsistent)
+{
+    // Every index in cut_edge_indices has flag=true;
+    // count of true flags == number of indices.
+    auto m = make_tetrahedron();
+    auto cg = compute_cut_graph(m);
+    for (std::size_t idx : cg.cut_edge_indices)
+        EXPECT_TRUE(cg.cut_edge_flags[idx]);
+
+    std::size_t count = 0;
+    for (bool b : cg.cut_edge_flags) count += b ? 1u : 0u;
+    EXPECT_EQ(count, cg.cut_edge_indices.size());
+}
+
+TEST(CutGraph, IsCutMethodMatchesFlags)
+{
+    auto m = make_tetrahedron();
+    auto cg = compute_cut_graph(m);
+    for (auto e : m.edges()) {
+        bool flag = cg.cut_edge_flags[static_cast<std::size_t>(e.idx())];
+        EXPECT_EQ(cg.is_cut(e), flag);
+    }
+}
+
+TEST(CutGraph, FlagsVectorHasCorrectSize)
+{
+    auto m = make_quad_strip();
+    auto cg = compute_cut_graph(m);
+    EXPECT_EQ(cg.cut_edge_flags.size(), m.number_of_edges());
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Exact hyperbolic trilateration — Möbius + law of cosines
+// ════════════════════════════════════════════════════════════════════════════
+
+namespace {
+
+/// Poincaré-disk hyperbolic distance.
+double hyp_dist_disk(Eigen::Vector2d a, Eigen::Vector2d b)
+{
+    double num = (a.x()-b.x())*(a.x()-b.x()) + (a.y()-b.y())*(a.y()-b.y());
+    double da  = 1.0 - a.x()*a.x() - a.y()*a.y();
+    double db_  = 1.0 - b.x()*b.x() - b.y()*b.y();
+    double arg = 1.0 + 2.0*num/(da*db_);
+    return std::acosh(std::max(1.0, arg));
+}
+
+/// Reproduce the exact trilateration formula from layout.hpp detail namespace.
+Eigen::Vector2d trilaterate_hyp(Eigen::Vector2d pa, Eigen::Vector2d pb,
+                                  double D, double da, double db)
+{
+    if (D < 1e-12 || da < 1e-12) return pa;
+    if (std::sinh(da) * std::sinh(D) < 1e-14) return pa;
+    using C = std::complex<double>;
+    auto mobius_fwd = [](C z, C center) -> C {
+        return (z - center) / (C(1.0) - std::conj(center) * z);
+    };
+    auto mobius_inv = [](C w, C center) -> C {
+        return (w + center) / (C(1.0) + std::conj(center) * w);
+    };
+    C a(pa.x(), pa.y()), b(pb.x(), pb.y());
+    C b_mapped = mobius_fwd(b, a);
+    double theta_b = std::arg(b_mapped);
+    double cos_alpha = (std::cosh(da) * std::cosh(D) - std::cosh(db))
+                     / (std::sinh(da) * std::sinh(D));
+    cos_alpha = std::max(-1.0, std::min(1.0, cos_alpha));
+    double alpha = std::acos(cos_alpha);
+    C pc_origin = std::tanh(da * 0.5) * std::exp(C(0.0, theta_b + alpha));
+    C pc_c = mobius_inv(pc_origin, a);
+    return Eigen::Vector2d(pc_c.real(), pc_c.imag());
+}
+
+} // namespace
+
+TEST(HyperbolicTrilateration, RecoveredDistances_Small)
+{
+    // pa at origin, pb at tanh(D/2) on x-axis.
+    double D = 0.8, da = 0.5, db = 0.6;
+    Eigen::Vector2d pa(0.0, 0.0);
+    Eigen::Vector2d pb(std::tanh(D * 0.5), 0.0);
+
+    Eigen::Vector2d pc = trilaterate_hyp(pa, pb, D, da, db);
+
+    EXPECT_NEAR(hyp_dist_disk(pa, pb), D,  1e-10);
+    EXPECT_NEAR(hyp_dist_disk(pa, pc), da, 1e-9);
+    EXPECT_NEAR(hyp_dist_disk(pb, pc), db, 1e-9);
+}
+
+TEST(HyperbolicTrilateration, RecoveredDistances_Large)
+{
+    double D = 2.0, da = 1.5, db = 1.0;
+    Eigen::Vector2d pa(0.0, 0.0);
+    Eigen::Vector2d pb(std::tanh(D * 0.5), 0.0);
+
+    Eigen::Vector2d pc = trilaterate_hyp(pa, pb, D, da, db);
+
+    EXPECT_NEAR(hyp_dist_disk(pa, pc), da, 1e-8);
+    EXPECT_NEAR(hyp_dist_disk(pb, pc), db, 1e-8);
+}
+
+TEST(HyperbolicTrilateration, ResultInsidePoincareDisk)
+{
+    double D = 1.2, da = 0.9, db = 0.7;
+    Eigen::Vector2d pa(0.0, 0.0);
+    Eigen::Vector2d pb(std::tanh(D * 0.5), 0.0);
+
+    Eigen::Vector2d pc = trilaterate_hyp(pa, pb, D, da, db);
+    EXPECT_LT(pc.norm(), 1.0);
+}
+
+TEST(HyperbolicTrilateration, OffOrigin_StillCorrect)
+{
+    // Place pa somewhere in the disk (not at origin) to test Möbius map.
+    double D = 0.6, da = 0.4, db = 0.5;
+    // pa at (0.3, 0.0) in Poincaré coords.
+    Eigen::Vector2d pa(0.3, 0.0);
+    // pb at distance D from pa — compute pb in Poincaré disk:
+    // Möbius inverse: tanh(D/2) maps back from origin frame.
+    using C = std::complex<double>;
+    C a_c(pa.x(), pa.y());
+    C pb_c = (std::tanh(D*0.5) + a_c) / (C(1.0) + std::conj(a_c) * std::tanh(D*0.5));
+    Eigen::Vector2d pb(pb_c.real(), pb_c.imag());
+
+    Eigen::Vector2d pc = trilaterate_hyp(pa, pb, D, da, db);
+    EXPECT_LT(pc.norm(), 1.0);
+    EXPECT_NEAR(hyp_dist_disk(pa, pc), da, 1e-8);
+    EXPECT_NEAR(hyp_dist_disk(pb, pc), db, 1e-8);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Normalisation — euclidean_layout (centroid) + hyper_ideal_layout (Möbius)
+// ════════════════════════════════════════════════════════════════════════════
+
+/// Build a solved euclidean layout for the quad-strip at the natural equilibrium.
+static Layout2D make_euclidean_layout_normalised(bool normalise)
+{
+    auto mesh = make_quad_strip();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    // Pin first vertex; assign free DOF indices to all others.
+    auto vit = mesh.vertices().begin();
+    maps.v_idx[*vit++] = -1;
+    int idx = 0;
+    for (; vit != mesh.vertices().end(); ++vit) maps.v_idx[*vit] = idx++;
+    const int n = idx;
+
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    // Adjust theta so G(x=0) = 0 (natural equilibrium).
+    auto G0 = euclidean_gradient(mesh, x0, maps);
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv >= 0) maps.theta_v[v] -= G0[static_cast<std::size_t>(iv)];
+    }
+
+    auto res = newton_euclidean(mesh, x0, maps, 1e-10, 100);
+    return euclidean_layout(mesh, res.x, maps, nullptr, nullptr, normalise);
+}
+
+TEST(Normalisation, Euclidean_CentroidAtOrigin)
+{
+    auto L = make_euclidean_layout_normalised(true);
+    ASSERT_GT(L.uv.size(), 0u);
+
+    double cx = 0.0, cy = 0.0;
+    for (auto& p : L.uv) { cx += p.x(); cy += p.y(); }
+    int n = static_cast<int>(L.uv.size());
+    EXPECT_NEAR(cx / n, 0.0, 1e-9);
+    EXPECT_NEAR(cy / n, 0.0, 1e-9);
+}
+
+TEST(Normalisation, Euclidean_NormalisePreservesEdgeLengthRatios)
+{
+    auto L_no  = make_euclidean_layout_normalised(false);
+    auto L_yes = make_euclidean_layout_normalised(true);
+
+    // Ratio of the lengths of the first two edges must be preserved.
+    ASSERT_GE(L_no.uv.size(), 4u);
+    // edges 0→1 and 0→2 (the two edges from vertex 0 in the quad-strip)
+    double len0_no  = (L_no.uv[1]  - L_no.uv[0]).norm();
+    double len1_no  = (L_no.uv[2]  - L_no.uv[0]).norm();
+    double len0_yes = (L_yes.uv[1] - L_yes.uv[0]).norm();
+    double len1_yes = (L_yes.uv[2] - L_yes.uv[0]).norm();
+
+    if (len1_no > 1e-12 && len1_yes > 1e-12) {
+        double ratio_no  = len0_no  / len1_no;
+        double ratio_yes = len0_yes / len1_yes;
+        EXPECT_NEAR(ratio_no, ratio_yes, 1e-9);
+    }
+}
+
+/// Build a solved hyper-ideal layout for a triangle at natural equilibrium.
+static Layout2D make_hyper_ideal_layout_normalised(bool normalise)
+{
+    auto mesh = make_triangle();
+    auto maps = setup_hyper_ideal_maps(mesh);
+    int n = assign_all_dof_indices(mesh, maps);
+
+    std::vector<double> xbase(static_cast<std::size_t>(n));
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v]; if (iv >= 0) xbase[iv] = 1.0;
+    }
+    for (auto e : mesh.edges()) {
+        int ie = maps.e_idx[e]; if (ie >= 0) xbase[ie] = 0.5;
+    }
+    auto G0 = evaluate_hyper_ideal(mesh, xbase, maps, false).gradient;
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v]; if (iv >= 0) maps.theta_v[v] += G0[iv];
+    }
+    for (auto e : mesh.edges()) {
+        int ie = maps.e_idx[e]; if (ie >= 0) maps.theta_e[e] += G0[ie];
+    }
+    auto res = newton_hyper_ideal(mesh, xbase, maps, 1e-9, 100);
+    return hyper_ideal_layout(mesh, res.x, maps, nullptr, nullptr, normalise);
+}
+
+TEST(Normalisation, HyperIdeal_PositionsInsidePoincareDisk)
+{
+    auto L = make_hyper_ideal_layout_normalised(true);
+    ASSERT_GT(L.uv.size(), 0u);
+    for (auto& p : L.uv) {
+        EXPECT_LT(p.norm(), 1.0 + 1e-9)
+            << "Vertex outside Poincaré disk: r=" << p.norm();
+    }
+}
+
+TEST(Normalisation, HyperIdeal_CenteringReducesAverageRadius)
+{
+    // After Möbius centering the average Euclidean radius inside the disk
+    // must be ≤ the unconstrained average.
+    auto L_no  = make_hyper_ideal_layout_normalised(false);
+    auto L_yes = make_hyper_ideal_layout_normalised(true);
+
+    int n = static_cast<int>(L_no.uv.size());
+    double r_no = 0.0, r_yes = 0.0;
+    for (int i = 0; i < n; ++i) {
+        r_no  += L_no.uv[i].norm();
+        r_yes += L_yes.uv[i].norm();
+    }
+    EXPECT_LE(r_yes / n, r_no / n + 1e-9);
+}

code/tests/cgal/test_phase7.cpp

@@ -0,0 +1,485 @@
+// test_phase7.cpp
+//
+// Phase 7 — Tests for Java-parity layout features:
+//   - MobiusMap            : identity, inverse, compose, from_three, is_identity
+//   - best_root_face       : selects a valid face; interior bonus
+//   - halfedge_uv          : size, non-seam consistency, seam divergence
+//   - Priority BFS         : vertex ordering / depth correctness
+//   - normalise_euclidean  : halfedge_uv centroid at origin
+//   - period_matrix.hpp    : τ in upper half-plane, SL(2,ℤ) reduction
+//   - fundamental_domain.hpp: parallelogram CCW, generators, tiling_copy
+
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include "euclidean_functional.hpp"
+#include "hyper_ideal_functional.hpp"
+#include "newton_solver.hpp"
+#include "layout.hpp"
+#include "period_matrix.hpp"
+#include "fundamental_domain.hpp"
+#include <gtest/gtest.h>
+#include <cmath>
+#include <complex>
+#include <vector>
+#include <limits>
+
+using namespace conformallab;
+using C = std::complex<double>;
+
+// ════════════════════════════════════════════════════════════════════════════
+// MobiusMap
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(MobiusMap, Identity_AppliesAsIdentity)
+{
+    MobiusMap id = MobiusMap::identity();
+    C z(0.3, 0.7);
+    C w = id.apply(z);
+    EXPECT_NEAR(w.real(), z.real(), 1e-12);
+    EXPECT_NEAR(w.imag(), z.imag(), 1e-12);
+}
+
+TEST(MobiusMap, Identity_IsIdentity)
+{
+    EXPECT_TRUE(MobiusMap::identity().is_identity());
+}
+
+TEST(MobiusMap, NonIdentity_IsNotIdentity)
+{
+    // T(z) = z + 1 — translation, clearly not identity
+    MobiusMap T{ C(1), C(1), C(0), C(1) };
+    EXPECT_FALSE(T.is_identity());
+}
+
+TEST(MobiusMap, Inverse_ComposeIsIdentity)
+{
+    // T(z) = (2z + 1) / (z + 3)
+    MobiusMap T{ C(2), C(1), C(1), C(3) };
+    MobiusMap TinvT = T.inverse().compose(T);
+    EXPECT_TRUE(TinvT.is_identity(1e-9));
+}
+
+TEST(MobiusMap, Compose_OrderCorrect)
+{
+    // S: z ↦ z + 1,   T: z ↦ 2z
+    // S.compose(T) means S applied after T: z ↦ 2z + 1
+    MobiusMap S{ C(1), C(1), C(0), C(1) };   // z + 1
+    MobiusMap T{ C(2), C(0), C(0), C(1) };   // 2z
+    MobiusMap ST = S.compose(T);
+    C z(1.0, 0.0);
+    // S(T(z)) = S(2) = 3
+    EXPECT_NEAR(ST.apply(z).real(), 3.0, 1e-12);
+    EXPECT_NEAR(ST.apply(z).imag(), 0.0, 1e-12);
+}
+
+TEST(MobiusMap, FromThree_RecoversMap)
+{
+    // Known map T(z) = (z + i) / (1 + 0·z) — translation by i
+    C w1 = C(0, 1) + C(0, 1);   // T(i) = 2i
+    C w2 = C(1, 0) + C(0, 1);   // T(1) = 1 + i
+    C w3 = C(-1, 0) + C(0, 1);  // T(-1) = -1 + i
+    MobiusMap T = MobiusMap::from_three(C(0, 1), w1, C(1, 0), w2, C(-1, 0), w3);
+    // Verify T maps a fourth point correctly: T(0) = i
+    C result = T.apply(C(0, 0));
+    EXPECT_NEAR(result.real(), 0.0, 1e-9);
+    EXPECT_NEAR(result.imag(), 1.0, 1e-9);
+}
+
+TEST(MobiusMap, FromThree_DegenerateReturnsIdentity)
+{
+    // Three coincident points → singular system → identity fallback
+    C z(0.5, 0.5);
+    MobiusMap T = MobiusMap::from_three(z, z, z, z, z, z);
+    // Should not crash; returns identity (or at least a valid map)
+    // We just check the result is finite
+    C w = T.apply(C(0.1, 0.2));
+    EXPECT_FALSE(std::isnan(w.real()));
+    EXPECT_FALSE(std::isnan(w.imag()));
+}
+
+TEST(MobiusMap, Apply_Vector2d)
+{
+    MobiusMap id = MobiusMap::identity();
+    Eigen::Vector2d p(0.4, 0.6);
+    Eigen::Vector2d q = id.apply(p);
+    EXPECT_NEAR(q.x(), p.x(), 1e-12);
+    EXPECT_NEAR(q.y(), p.y(), 1e-12);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// best_root_face
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(BestRootFace, ReturnsValidFace_Triangle)
+{
+    auto mesh = make_triangle();
+    Face_index f = detail::best_root_face(mesh);
+    EXPECT_NE(f, Face_index());
+    EXPECT_GE(f.idx(), 0);
+}
+
+TEST(BestRootFace, ReturnsValidFace_Tetrahedron)
+{
+    auto mesh = make_tetrahedron();
+    Face_index f = detail::best_root_face(mesh);
+    EXPECT_NE(f, Face_index());
+    // Tetrahedron has 4 faces — best is one of them
+    EXPECT_LT(static_cast<std::size_t>(f.idx()), mesh.number_of_faces());
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// halfedge_uv — size and non-seam consistency
+// ════════════════════════════════════════════════════════════════════════════
+
+// Helper: build equilibrium Euclidean layout for a given mesh.
+// Uses x = 0 (identity scale factor) which is the equilibrium for natural edge lengths.
+static Layout2D make_euclidean_layout(ConformalMesh& mesh)
+{
+    EuclideanMaps maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    // Pin first vertex (DOF = -1); assign sequential indices to the rest.
+    auto vit = mesh.vertices().begin();
+    maps.v_idx[*vit++] = -1;
+    int idx = 0;
+    for (; vit != mesh.vertices().end(); ++vit) maps.v_idx[*vit] = idx++;
+    std::vector<double> x(static_cast<std::size_t>(idx), 0.0);
+    return euclidean_layout(mesh, x, maps);
+}
+
+TEST(HalfedgeUV, Size_EqualsNumberOfHalfedges_Triangle)
+{
+    auto mesh = make_triangle();
+    auto lay   = make_euclidean_layout(mesh);
+    EXPECT_EQ(lay.halfedge_uv.size(), mesh.number_of_halfedges());
+}
+
+TEST(HalfedgeUV, Size_EqualsNumberOfHalfedges_QuadStrip)
+{
+    auto mesh = make_quad_strip();
+    auto lay   = make_euclidean_layout(mesh);
+    EXPECT_EQ(lay.halfedge_uv.size(), mesh.number_of_halfedges());
+}
+
+TEST(HalfedgeUV, NonBorderHalfedges_MatchUV)
+{
+    // For an open mesh with no cut graph the layout has no seams.
+    // Every non-border halfedge h must satisfy:
+    //   halfedge_uv[h] == uv[source(h)]
+    auto mesh = make_quad_strip();
+    auto lay   = make_euclidean_layout(mesh);
+    for (auto h : mesh.halfedges()) {
+        if (mesh.is_border(h)) continue;
+        std::size_t hi  = static_cast<std::size_t>(h.idx());
+        std::size_t vi  = static_cast<std::size_t>(mesh.source(h).idx());
+        EXPECT_NEAR(lay.halfedge_uv[hi].x(), lay.uv[vi].x(), 1e-10)
+            << "halfedge " << hi << " source vertex " << vi;
+        EXPECT_NEAR(lay.halfedge_uv[hi].y(), lay.uv[vi].y(), 1e-10)
+            << "halfedge " << hi << " source vertex " << vi;
+    }
+}
+
+TEST(HalfedgeUV, BorderHalfedges_AreZero)
+{
+    auto mesh = make_triangle();
+    auto lay   = make_euclidean_layout(mesh);
+    bool found_border = false;
+    for (auto h : mesh.halfedges()) {
+        if (!mesh.is_border(h)) continue;
+        std::size_t hi = static_cast<std::size_t>(h.idx());
+        EXPECT_NEAR(lay.halfedge_uv[hi].x(), 0.0, 1e-12);
+        EXPECT_NEAR(lay.halfedge_uv[hi].y(), 0.0, 1e-12);
+        found_border = true;
+    }
+    EXPECT_TRUE(found_border);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Priority BFS — depth ordering
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(PriorityBFS, Layout_SucceedsOnOpenMesh)
+{
+    auto mesh = make_quad_strip();
+    auto lay   = make_euclidean_layout(mesh);
+    EXPECT_TRUE(lay.success);
+}
+
+TEST(PriorityBFS, Layout_NoSeamOnOpenMesh)
+{
+    auto mesh = make_quad_strip();
+    auto lay   = make_euclidean_layout(mesh);
+    EXPECT_FALSE(lay.has_seam);
+}
+
+TEST(PriorityBFS, AllVerticesPlaced)
+{
+    auto mesh = make_tetrahedron();
+    // Tetrahedron is closed; layout without cut graph will have a seam
+    EuclideanMaps maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    auto vit = mesh.vertices().begin();
+    maps.v_idx[*vit++] = -1;
+    int idx = 0;
+    for (; vit != mesh.vertices().end(); ++vit) maps.v_idx[*vit] = idx++;
+    std::vector<double> x(static_cast<std::size_t>(idx), 0.0);
+    auto lay = euclidean_layout(mesh, x, maps);
+    EXPECT_TRUE(lay.success);
+    // All UVs must be finite
+    for (auto& p : lay.uv) {
+        EXPECT_FALSE(std::isnan(p.x()));
+        EXPECT_FALSE(std::isnan(p.y()));
+    }
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// normalise_euclidean — centroid + PCA applied to both uv and halfedge_uv
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(NormaliseEuclidean, UVCentroidAtOrigin)
+{
+    auto mesh = make_quad_strip();
+    auto lay   = make_euclidean_layout(mesh);
+    normalise_euclidean(lay);
+
+    Eigen::Vector2d mean = Eigen::Vector2d::Zero();
+    for (auto& p : lay.uv) mean += p;
+    mean /= static_cast<double>(lay.uv.size());
+    EXPECT_NEAR(mean.x(), 0.0, 1e-10);
+    EXPECT_NEAR(mean.y(), 0.0, 1e-10);
+}
+
+TEST(NormaliseEuclidean, HalfedgeUVCentroidAlsoShifted)
+{
+    // After normalisation: the non-border halfedge_uv entries should also be
+    // centred (since they are shifted by the same mean as uv).
+    // We verify that the mean of non-border halfedge_uv is near (0,0).
+    auto mesh = make_quad_strip();
+    auto lay   = make_euclidean_layout(mesh);
+    normalise_euclidean(lay);
+
+    Eigen::Vector2d mean = Eigen::Vector2d::Zero();
+    int count = 0;
+    for (auto h : mesh.halfedges()) {
+        if (mesh.is_border(h)) continue;
+        mean += lay.halfedge_uv[static_cast<std::size_t>(h.idx())];
+        ++count;
+    }
+    if (count > 0) mean /= static_cast<double>(count);
+    EXPECT_NEAR(mean.x(), 0.0, 1e-9);
+    EXPECT_NEAR(mean.y(), 0.0, 1e-9);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// PeriodMatrix — reduce_to_fundamental_domain
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(PeriodMatrix, ReduceToFD_AlreadyInFD)
+{
+    // τ = i is in F  (|i|=1, Re(i)=0, Im(i)=1>0)
+    C tau(0.0, 1.0);
+    C reduced = reduce_to_fundamental_domain(tau);
+    EXPECT_TRUE(is_in_fundamental_domain(reduced));
+    EXPECT_NEAR(reduced.real(), 0.0, 1e-10);
+    EXPECT_NEAR(reduced.imag(), 1.0, 1e-10);
+}
+
+TEST(PeriodMatrix, ReduceToFD_ShiftsRealPart)
+{
+    // τ = 2 + 3i  →  T step: τ -= 2  →  3i  (|3i|=3≥1, Re=0)
+    C tau(2.0, 3.0);
+    C reduced = reduce_to_fundamental_domain(tau);
+    EXPECT_TRUE(is_in_fundamental_domain(reduced, 1e-9));
+    EXPECT_NEAR(reduced.real(), 0.0, 1e-10);
+    EXPECT_NEAR(reduced.imag(), 3.0, 1e-10);
+}
+
+TEST(PeriodMatrix, ReduceToFD_InvertsSmallTau)
+{
+    // τ = 0.5i  →  |0.5i|=0.5<1  →  S: τ↦-1/(0.5i) = 2i
+    C tau(0.0, 0.5);
+    C reduced = reduce_to_fundamental_domain(tau);
+    EXPECT_TRUE(is_in_fundamental_domain(reduced, 1e-9));
+    EXPECT_NEAR(reduced.real(), 0.0, 1e-10);
+    EXPECT_NEAR(reduced.imag(), 2.0, 1e-10);
+}
+
+TEST(PeriodMatrix, ReduceToFD_ThrowsForNonUpperHalfPlane)
+{
+    C tau(0.5, -1.0);  // Im < 0 → not in upper half-plane
+    EXPECT_THROW(reduce_to_fundamental_domain(tau), std::domain_error);
+}
+
+TEST(PeriodMatrix, IsInFundamentalDomain_Square)
+{
+    EXPECT_TRUE(is_in_fundamental_domain(C(0.0, 1.0)));        // i
+    EXPECT_TRUE(is_in_fundamental_domain(C(0.3, 1.5)));        // inside
+    EXPECT_FALSE(is_in_fundamental_domain(C(0.6, 1.5)));       // Re > 1/2
+    EXPECT_FALSE(is_in_fundamental_domain(C(0.0, 0.5)));       // |τ| < 1
+}
+
+TEST(PeriodMatrix, ComputePeriodMatrix_UnitSquare)
+{
+    // ω_1 = (1, 0), ω_2 = (0, 1)  →  τ = i
+    HolonomyData hol;
+    hol.translations = { Eigen::Vector2d(1.0, 0.0), Eigen::Vector2d(0.0, 1.0) };
+    PeriodData pd = compute_period_matrix(hol, /*reduce=*/false);
+    EXPECT_EQ(pd.genus(), 1);
+    EXPECT_GT(pd.tau.imag(), 0.0);
+    EXPECT_NEAR(pd.tau.real(), 0.0, 1e-10);
+    EXPECT_NEAR(pd.tau.imag(), 1.0, 1e-10);
+}
+
+TEST(PeriodMatrix, ComputePeriodMatrix_ReducedTau_InFD)
+{
+    // ω_1 = (1, 0), ω_2 = (0.5, 0.25)  →  τ = 0.5 + 0.25i
+    // |τ| = sqrt(0.25 + 0.0625) ≈ 0.559 < 1 → needs S step
+    HolonomyData hol;
+    hol.translations = { Eigen::Vector2d(1.0, 0.0), Eigen::Vector2d(0.5, 0.25) };
+    PeriodData pd = compute_period_matrix(hol, /*reduce=*/true);
+    EXPECT_TRUE(pd.in_fundamental_domain);
+    EXPECT_TRUE(is_in_fundamental_domain(pd.tau, 1e-9));
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// FundamentalDomain — genus-1 parallelogram
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(FundamentalDomain, Genus1_HasFourVertices)
+{
+    HolonomyData hol;
+    hol.translations = { Eigen::Vector2d(1.0, 0.0), Eigen::Vector2d(0.0, 1.0) };
+    FundamentalDomain fd = compute_fundamental_domain_genus1(hol);
+    EXPECT_EQ(fd.vertices.size(), 4u);
+    EXPECT_TRUE(fd.is_valid());
+}
+
+TEST(FundamentalDomain, Genus1_VerticesMatchGenerators_UnitSquare)
+{
+    Eigen::Vector2d w1(1.0, 0.0), w2(0.0, 1.0);
+    HolonomyData hol;
+    hol.translations = { w1, w2 };
+    FundamentalDomain fd = compute_fundamental_domain_genus1(hol);
+    // Expected (CCW): origin, w1, w1+w2, w2
+    EXPECT_NEAR(fd.vertices[0].x(), 0.0, 1e-12);
+    EXPECT_NEAR(fd.vertices[0].y(), 0.0, 1e-12);
+    EXPECT_NEAR(fd.vertices[1].x(), w1.x(), 1e-12);
+    EXPECT_NEAR(fd.vertices[1].y(), w1.y(), 1e-12);
+    EXPECT_NEAR(fd.vertices[2].x(), (w1 + w2).x(), 1e-12);
+    EXPECT_NEAR(fd.vertices[2].y(), (w1 + w2).y(), 1e-12);
+    EXPECT_NEAR(fd.vertices[3].x(), w2.x(), 1e-12);
+    EXPECT_NEAR(fd.vertices[3].y(), w2.y(), 1e-12);
+}
+
+TEST(FundamentalDomain, Genus1_CCWOrientation)
+{
+    // After possible swap, the signed area = cross(v1-v0, v3-v0) > 0 (CCW)
+    HolonomyData hol;
+    hol.translations = { Eigen::Vector2d(1.0, 0.0), Eigen::Vector2d(0.0, 1.0) };
+    FundamentalDomain fd = compute_fundamental_domain_genus1(hol);
+    Eigen::Vector2d v0 = fd.vertices[0], v1 = fd.vertices[1], v3 = fd.vertices[3];
+    double cross = (v1 - v0).x() * (v3 - v0).y() - (v1 - v0).y() * (v3 - v0).x();
+    EXPECT_GT(cross, 0.0);
+}
+
+TEST(FundamentalDomain, Genus1_CCWEnforced_WhenInputIsCW)
+{
+    // If we give CW generators (w2 × w1 < 0), the polygon must still be CCW.
+    // w1 = (0,1), w2 = (1,0):  cross w1×w2 = 0*0 - 1*1 = -1 < 0  →  should swap
+    HolonomyData hol;
+    hol.translations = { Eigen::Vector2d(0.0, 1.0), Eigen::Vector2d(1.0, 0.0) };
+    FundamentalDomain fd = compute_fundamental_domain_genus1(hol);
+    Eigen::Vector2d v0 = fd.vertices[0], v1 = fd.vertices[1], v3 = fd.vertices[3];
+    double cross = (v1 - v0).x() * (v3 - v0).y() - (v1 - v0).y() * (v3 - v0).x();
+    EXPECT_GT(cross, 0.0);
+}
+
+TEST(FundamentalDomain, Genus1_EdgeIdentifications)
+{
+    HolonomyData hol;
+    hol.translations = { Eigen::Vector2d(1.0, 0.0), Eigen::Vector2d(0.0, 1.0) };
+    FundamentalDomain fd = compute_fundamental_domain_genus1(hol);
+    EXPECT_EQ(fd.edge_identifications.size(), 2u);
+    // bottom ≡ top: (0,2)
+    EXPECT_EQ(fd.edge_identifications[0].first,  0);
+    EXPECT_EQ(fd.edge_identifications[0].second, 2);
+    // right ≡ left: (1,3)
+    EXPECT_EQ(fd.edge_identifications[1].first,  1);
+    EXPECT_EQ(fd.edge_identifications[1].second, 3);
+}
+
+TEST(FundamentalDomain, Genus1_GeneratorsStored)
+{
+    Eigen::Vector2d w1(2.0, 1.0), w2(-1.0, 3.0);
+    HolonomyData hol;
+    hol.translations = { w1, w2 };
+    FundamentalDomain fd = compute_fundamental_domain_genus1(hol);
+    EXPECT_EQ(fd.generators.size(), 2u);
+    // Generators are w1 and w2 (possibly swapped to ensure CCW)
+    // Their sum of norms matches the originals
+    double norm_gen = fd.generators[0].norm() + fd.generators[1].norm();
+    double norm_in  = w1.norm() + w2.norm();
+    EXPECT_NEAR(norm_gen, norm_in, 1e-10);
+}
+
+TEST(FundamentalDomain, HigherGenus_ReturnsEmpty)
+{
+    HolonomyData hol;
+    hol.translations = {
+        Eigen::Vector2d(1, 0), Eigen::Vector2d(0, 1),
+        Eigen::Vector2d(2, 0), Eigen::Vector2d(0, 2)   // g=2, 4 generators
+    };
+    FundamentalDomain fd = compute_fundamental_domain(hol);
+    // g > 1 returns empty (TODO Phase 8)
+    EXPECT_FALSE(fd.is_valid());
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// tiling_copy / tiling_neighbourhood
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(TilingCopy, ShiftAppliedToAllUV)
+{
+    auto mesh = make_quad_strip();
+    auto lay   = make_euclidean_layout(mesh);
+
+    Eigen::Vector2d w1(3.0, 0.0), w2(0.0, 2.0);
+    // m=1, n=2 → expected shift = w1 + 2*w2 = (3, 4)
+    Layout2D copy = tiling_copy(lay, w1, w2, 1, 2);
+    Eigen::Vector2d expected_shift(3.0, 4.0);
+    for (std::size_t i = 0; i < lay.uv.size(); ++i) {
+        EXPECT_NEAR(copy.uv[i].x(), lay.uv[i].x() + expected_shift.x(), 1e-12);
+        EXPECT_NEAR(copy.uv[i].y(), lay.uv[i].y() + expected_shift.y(), 1e-12);
+    }
+}
+
+TEST(TilingCopy, ZeroShift_IsSameAsCopy)
+{
+    auto mesh = make_quad_strip();
+    auto lay   = make_euclidean_layout(mesh);
+    Eigen::Vector2d w1(1, 0), w2(0, 1);
+    Layout2D copy = tiling_copy(lay, w1, w2, 0, 0);
+    for (std::size_t i = 0; i < lay.uv.size(); ++i) {
+        EXPECT_NEAR(copy.uv[i].x(), lay.uv[i].x(), 1e-12);
+        EXPECT_NEAR(copy.uv[i].y(), lay.uv[i].y(), 1e-12);
+    }
+}
+
+TEST(TilingNeighbourhood, CorrectCount)
+{
+    auto mesh = make_quad_strip();
+    auto lay   = make_euclidean_layout(mesh);
+    HolonomyData hol;
+    hol.translations = { Eigen::Vector2d(1, 0), Eigen::Vector2d(0, 1) };
+    // m_max=1, n_max=1 → (2*1+1) * (2*1+1) = 9 tiles
+    auto tiles = tiling_neighbourhood(lay, hol, 1, 1);
+    EXPECT_EQ(tiles.size(), 9u);
+}
+
+TEST(TilingNeighbourhood, EmptyHolonomy_ReturnsSingleTile)
+{
+    auto mesh = make_quad_strip();
+    auto lay   = make_euclidean_layout(mesh);
+    HolonomyData hol;  // no translations
+    auto tiles = tiling_neighbourhood(lay, hol);
+    EXPECT_EQ(tiles.size(), 1u);
+}

code/tests/cgal/test_pipeline.cpp

@@ -0,0 +1,292 @@
+// test_pipeline.cpp
+//
+// Phase 4c — End-to-end pipeline tests and library-user examples.
+//
+// These tests exercise the full conformallab++ pipeline as a user would:
+//
+//   1. Build (or load) a mesh
+//   2. Set up maps and assign DOFs
+//   3. Configure target angles / targets
+//   4. Solve with Newton
+//   5. Inspect / export the result
+//
+// Each test mirrors a realistic usage scenario documented in the README.
+//
+// Tests:
+//   1. Pipeline_Euclidean_TriangleToEquilibrium
+//        Read mesh → setup Euclidean maps → solve → verify convergence
+//   2. Pipeline_Spherical_TetrahedronToEquilibrium
+//        Setup spherical tetrahedron → solve → verify angles sum to 4π
+//   3. Pipeline_HyperIdeal_TriangleRoundTrip
+//        Build triangle → setup HyperIdeal → solve → verify G ≈ 0
+//   4. Pipeline_MeshIO_SolveAndExport
+//        Build mesh → solve → write OFF → reload → verify vertex count intact
+//   5. Pipeline_AllThreeGeometries_SameTopology
+//        Same quad-strip mesh solved under all three geometries: all converge
+
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include "mesh_io.hpp"
+#include "euclidean_functional.hpp"
+#include "spherical_functional.hpp"
+#include "hyper_ideal_functional.hpp"
+#include "newton_solver.hpp"
+#include <gtest/gtest.h>
+#include <cmath>
+#include <vector>
+#include <filesystem>
+
+using namespace conformallab;
+
+// ────────────────────────────────────────────────────────────────────────────
+// Shared helpers
+// ────────────────────────────────────────────────────────────────────────────
+
+static void pin_first_vertex_euclidean(ConformalMesh& mesh, EuclideanMaps& maps, int& n)
+{
+    auto vit = mesh.vertices().begin();
+    Vertex_index v0 = *vit++;
+    maps.v_idx[v0] = -1;
+    int idx = 0;
+    for (; vit != mesh.vertices().end(); ++vit)
+        maps.v_idx[*vit] = idx++;
+    n = idx;
+}
+
+static void set_natural_euclidean_theta(ConformalMesh& mesh, EuclideanMaps& maps, int n)
+{
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto G = euclidean_gradient(mesh, x0, maps);
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv < 0) continue;
+        maps.theta_v[v] -= G[static_cast<std::size_t>(iv)];
+    }
+}
+
+static std::vector<double> set_natural_hyper_ideal_targets(
+    ConformalMesh& mesh, HyperIdealMaps& maps, int n,
+    double b_base = 1.0, double a_base = 0.5)
+{
+    const auto sz = static_cast<std::size_t>(n);
+    std::vector<double> xbase(sz, 0.0);
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv >= 0) xbase[static_cast<std::size_t>(iv)] = b_base;
+    }
+    for (auto e : mesh.edges()) {
+        int ie = maps.e_idx[e];
+        if (ie >= 0) xbase[static_cast<std::size_t>(ie)] = a_base;
+    }
+    auto G = evaluate_hyper_ideal(mesh, xbase, maps, false).gradient;
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv < 0) continue;
+        maps.theta_v[v] += G[static_cast<std::size_t>(iv)];
+    }
+    for (auto e : mesh.edges()) {
+        int ie = maps.e_idx[e];
+        if (ie < 0) continue;
+        maps.theta_e[e] += G[static_cast<std::size_t>(ie)];
+    }
+    return xbase;
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 1 — Euclidean full pipeline: triangle → equilibrium
+//
+// Simulates a user doing:
+//   auto mesh = make_triangle();
+//   auto maps = setup_euclidean_maps(mesh);
+//   compute_euclidean_lambda0_from_mesh(mesh, maps);
+//   // … set target angles and DOF indices …
+//   auto result = newton_euclidean(mesh, x0, maps);
+//   assert(result.converged);
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(Pipeline, Euclidean_TriangleToEquilibrium)
+{
+    // ── Step 1: build mesh ────────────────────────────────────────────────
+    auto mesh = make_triangle();
+
+    // ── Step 2: set up maps ───────────────────────────────────────────────
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    // ── Step 3: assign DOFs (pin v0) ──────────────────────────────────────
+    int n = 0;
+    pin_first_vertex_euclidean(mesh, maps, n);
+    ASSERT_EQ(n, 2);
+
+    // ── Step 4: choose natural target angles → x* = 0 ────────────────────
+    set_natural_euclidean_theta(mesh, maps, n);
+
+    // ── Step 5: solve ─────────────────────────────────────────────────────
+    std::vector<double> x0(static_cast<std::size_t>(n), -0.1);
+    auto result = newton_euclidean(mesh, x0, maps);
+
+    // ── Step 6: verify ────────────────────────────────────────────────────
+    EXPECT_TRUE(result.converged)
+        << "Euclidean pipeline: triangle should converge; "
+           "grad_inf_norm = " << result.grad_inf_norm;
+    EXPECT_LT(result.grad_inf_norm, 1e-8);
+    EXPECT_EQ(static_cast<int>(result.x.size()), n);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 2 — Spherical full pipeline: tetrahedron → equilibrium
+//
+// The spherical tetrahedron equilibrium x* = 0 is built into the maps.
+// After solving, the total angle defect Σ(Θ_v − Σα_v) should be ≈ 0.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(Pipeline, Spherical_TetrahedronToEquilibrium)
+{
+    // ── Steps 1–3 ─────────────────────────────────────────────────────────
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    // ── Step 4: solve ─────────────────────────────────────────────────────
+    std::vector<double> x0(static_cast<std::size_t>(n), -0.2);
+    auto result = newton_spherical(mesh, x0, maps);
+
+    // ── Step 5: verify convergence ────────────────────────────────────────
+    EXPECT_TRUE(result.converged)
+        << "Spherical pipeline: tetrahedron should converge; "
+           "grad_inf_norm = " << result.grad_inf_norm;
+    EXPECT_LT(result.grad_inf_norm, 1e-8);
+
+    // ── Step 6: verify geometric invariant — total angle defect ≈ 0 ──────
+    auto G_final = spherical_gradient(mesh, result.x, maps);
+    double total_defect = 0.0;
+    for (double gv : G_final) total_defect += gv;
+    EXPECT_NEAR(total_defect, 0.0, 1e-7)
+        << "Spherical: total angle defect should vanish at equilibrium";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 3 — HyperIdeal full pipeline: triangle → equilibrium
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(Pipeline, HyperIdeal_TriangleRoundTrip)
+{
+    // ── Steps 1–3 ─────────────────────────────────────────────────────────
+    auto mesh = make_triangle();
+    auto maps = setup_hyper_ideal_maps(mesh);
+    int n = assign_all_dof_indices(mesh, maps);
+
+    // ── Step 4: natural targets (equilibrium at b=1.0, a=0.5) ────────────
+    auto xbase = set_natural_hyper_ideal_targets(mesh, maps, n);
+
+    // Verify: gradient at xbase must be ≈ 0 before solving
+    auto G_at_base = evaluate_hyper_ideal(mesh, xbase, maps, false).gradient;
+    double max_g = 0.0;
+    for (double v : G_at_base) max_g = std::max(max_g, std::abs(v));
+    ASSERT_LT(max_g, 1e-10) << "Natural target setup: gradient at base should be ~0";
+
+    // ── Step 5: perturb and solve ─────────────────────────────────────────
+    std::vector<double> x0 = xbase;
+    for (auto& v : x0) v += 0.3;
+
+    auto result = newton_hyper_ideal(mesh, x0, maps);
+
+    // ── Step 6: verify ────────────────────────────────────────────────────
+    EXPECT_TRUE(result.converged)
+        << "HyperIdeal pipeline: triangle should converge; "
+           "grad_inf_norm = " << result.grad_inf_norm;
+    EXPECT_LT(result.grad_inf_norm, 1e-8);
+
+    // Solution should be close to xbase (same equilibrium)
+    for (int i = 0; i < n; ++i) {
+        EXPECT_NEAR(result.x[static_cast<std::size_t>(i)],
+                    xbase[static_cast<std::size_t>(i)], 1e-6)
+            << "DOF " << i << " should recover the equilibrium value";
+    }
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 4 — Mesh I/O in the pipeline: solve → write → reload → check
+//
+// Demonstrates: compute a conformal factor on a mesh, write it to OFF, reload
+// and check that the mesh topology is preserved.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(Pipeline, MeshIO_SolveAndExport)
+{
+    // ── Build and solve ───────────────────────────────────────────────────
+    auto mesh = make_quad_strip();
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    int n = 0;
+    pin_first_vertex_euclidean(mesh, maps, n);
+    set_natural_euclidean_theta(mesh, maps, n);
+
+    std::vector<double> x0(static_cast<std::size_t>(n), -0.1);
+    auto result = newton_euclidean(mesh, x0, maps);
+    ASSERT_TRUE(result.converged) << "Solver must converge before export test";
+
+    // ── Write mesh ────────────────────────────────────────────────────────
+    const std::string tmp_path = "/tmp/conformallab_pipeline_test.off";
+    ASSERT_NO_THROW(save_mesh(tmp_path, mesh));
+    ASSERT_TRUE(std::filesystem::exists(tmp_path));
+
+    // ── Reload and verify topology ────────────────────────────────────────
+    ConformalMesh mesh2;
+    ASSERT_NO_THROW(mesh2 = load_mesh(tmp_path));
+
+    EXPECT_EQ(mesh2.number_of_vertices(), mesh.number_of_vertices())
+        << "Vertex count must survive OFF round-trip";
+    EXPECT_EQ(mesh2.number_of_faces(), mesh.number_of_faces())
+        << "Face count must survive OFF round-trip";
+
+    std::filesystem::remove(tmp_path);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Test 5 — All three geometries, same quad-strip topology
+//
+// Validates that the solver infrastructure works uniformly: the same mesh
+// topology is solvable under Euclidean, Spherical, and HyperIdeal geometries.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(Pipeline, AllThreeGeometries_QuadStrip)
+{
+    // ── Euclidean ──────────────────────────────────────────────────────────
+    {
+        auto mesh = make_quad_strip();
+        auto maps = setup_euclidean_maps(mesh);
+        compute_euclidean_lambda0_from_mesh(mesh, maps);
+        int n = 0;
+        pin_first_vertex_euclidean(mesh, maps, n);
+        set_natural_euclidean_theta(mesh, maps, n);
+        std::vector<double> x0(static_cast<std::size_t>(n), -0.1);
+        auto res = newton_euclidean(mesh, x0, maps);
+        EXPECT_TRUE(res.converged) << "Euclidean: quad strip should converge";
+    }
+
+    // ── HyperIdeal ────────────────────────────────────────────────────────
+    {
+        auto mesh = make_quad_strip();
+        auto maps = setup_hyper_ideal_maps(mesh);
+        int n = assign_all_dof_indices(mesh, maps);
+        auto xbase = set_natural_hyper_ideal_targets(mesh, maps, n);
+        std::vector<double> x0 = xbase;
+        for (auto& v : x0) v += 0.1;
+        auto res = newton_hyper_ideal(mesh, x0, maps);
+        EXPECT_TRUE(res.converged) << "HyperIdeal: quad strip should converge";
+    }
+
+    // ── Spherical (tetrahedron: smallest closed mesh with all vertices free) ─
+    {
+        auto mesh = make_spherical_tetrahedron();
+        auto maps = setup_spherical_maps(mesh);
+        compute_lambda0_from_mesh(mesh, maps);
+        int n = assign_vertex_dof_indices(mesh, maps);
+        std::vector<double> x0(static_cast<std::size_t>(n), -0.1);
+        auto res = newton_spherical(mesh, x0, maps);
+        EXPECT_TRUE(res.converged) << "Spherical: tetrahedron should converge";
+    }
+}

code/tests/cgal/test_scalability_smoke.cpp

@@ -0,0 +1,201 @@
+// test_scalability_smoke.cpp
+//
+// Scalability smoke tests — convergence on large real-world meshes.
+//
+// PURPOSE
+//   These tests verify that the Newton solver converges correctly on meshes
+//   significantly larger than the unit tests (which use tiny synthetic meshes).
+//   They do NOT assert on wall-clock time — timing is printed for information
+//   only, so the tests remain stable on slow CI hardware (Raspberry Pi ARM64).
+//
+//   If Newton fails to converge here, it is a correctness regression, not a
+//   performance regression.  See doc/math/complexity.md for timing context.
+//
+// MESHES USED
+//   cathead.obj   V=131,  F=248,  genus=0, open  — small, sanity check
+//   brezel.obj    V=6910, F=13824, genus=2, closed — large genus-2 mesh (χ=−2)
+//   brezel2.obj   V=2622, F=5248,  genus=2, closed — smaller genus-2 mesh (χ=−2)
+//
+// NOTE: both brezel meshes are genus-2. The naming follows the Java original
+// where "brezel2" is a different triangulation, not a different genus.
+//
+// EXPECTED RESULTS
+//   Newton converges in < 30 iterations for all Euclidean meshes (strictly
+//   convex energy, quadratic convergence from u=0).
+//   Cut graph produces 2g seam edges: 2 for brezel, 4 for brezel2.
+//
+// Tests:
+//   1. SmokeEuclidean.CatHead_SmallOpen
+//   2. SmokeEuclidean.Brezel_LargeGenus2
+//   3. SmokeEuclidean.Brezel2_Genus2_CutGraph
+
+#include "conformal_mesh.hpp"
+#include "mesh_io.hpp"
+#include "euclidean_functional.hpp"
+#include "gauss_bonnet.hpp"
+#include "newton_solver.hpp"
+#include "cut_graph.hpp"
+#include <gtest/gtest.h>
+#include <chrono>
+#include <iostream>
+#include <vector>
+#include <cmath>
+#include <string>
+
+using namespace conformallab;
+using Clock = std::chrono::steady_clock;
+using Ms    = std::chrono::milliseconds;
+
+// ── Helpers ──────────────────────────────────────────────────────────────────
+
+static int setup_open_mesh_dofs(ConformalMesh& mesh, EuclideanMaps& maps)
+{
+    int idx = 0;
+    for (auto v : mesh.vertices())
+        maps.v_idx[v] = mesh.is_border(v) ? -1 : idx++;
+    return idx;
+}
+
+static int setup_closed_mesh_dofs(ConformalMesh& mesh, EuclideanMaps& maps)
+{
+    auto vit = mesh.vertices().begin();
+    maps.v_idx[*vit++] = -1;
+    int idx = 0;
+    for (; vit != mesh.vertices().end(); ++vit)
+        maps.v_idx[*vit] = idx++;
+    return idx;
+}
+
+static void apply_natural_theta(ConformalMesh& mesh, EuclideanMaps& maps, int n)
+{
+    std::vector<double> x0(static_cast<std::size_t>(n), 0.0);
+    auto G0 = euclidean_gradient(mesh, x0, maps);
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv >= 0) maps.theta_v[v] -= G0[static_cast<std::size_t>(iv)];
+    }
+}
+
+// ── Test 1 — cathead.obj (V=131, F=248, open) ────────────────────────────────
+
+TEST(SmokeEuclidean, CatHead_SmallOpen)
+{
+    const std::string path = std::string(CONFORMALLAB_DATA_DIR) + "/obj/cathead.obj";
+    ConformalMesh mesh;
+    ASSERT_NO_THROW(mesh = load_mesh(path)) << "cathead.obj not found: " << path;
+
+    EXPECT_EQ(131u, mesh.number_of_vertices());
+    EXPECT_EQ(248u, mesh.number_of_faces());
+
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+    const int n = setup_open_mesh_dofs(mesh, maps);
+    apply_natural_theta(mesh, maps, n);
+
+    // Start from a small perturbation so Newton actually iterates.
+    std::vector<double> x0(static_cast<std::size_t>(n), -0.05);
+
+    auto t0 = Clock::now();
+    auto res = newton_euclidean(mesh, x0, maps, 1e-9, 200);
+    auto dt  = std::chrono::duration_cast<Ms>(Clock::now() - t0).count();
+
+    std::cout << "[SmokeEuclidean.CatHead] V=" << mesh.number_of_vertices()
+              << " F=" << mesh.number_of_faces()
+              << " iter=" << res.iterations
+              << " ||G||=" << res.grad_inf_norm
+              << " time=" << dt << "ms\n";
+
+    EXPECT_TRUE(res.converged)   << "Newton did not converge on cathead.obj";
+    EXPECT_LT(res.iterations, 30) << "Newton took ≥ 30 iterations — unexpected";
+    EXPECT_LT(res.grad_inf_norm, 1e-8);
+}
+
+// ── Test 2 — brezel.obj (V=6910, F=13824, genus=2) ───────────────────────────
+// Primary scalability target: largest mesh in the test suite.
+// Newton is started from a small perturbation (x0 = −0.05) so it must
+// actually iterate rather than exit immediately from the trivial equilibrium.
+
+TEST(SmokeEuclidean, Brezel_LargeGenus2)
+{
+    const std::string path = std::string(CONFORMALLAB_DATA_DIR) + "/obj/brezel.obj";
+    ConformalMesh mesh;
+    ASSERT_NO_THROW(mesh = load_mesh(path)) << "brezel.obj not found: " << path;
+
+    EXPECT_EQ(6910u,  mesh.number_of_vertices());
+    EXPECT_EQ(13824u, mesh.number_of_faces());
+
+    // Euler characteristic: V - E + F = −2 for genus-2 closed surface
+    const int chi = static_cast<int>(mesh.number_of_vertices())
+                  - static_cast<int>(mesh.number_of_edges())
+                  + static_cast<int>(mesh.number_of_faces());
+    EXPECT_EQ(-2, chi) << "brezel.obj must be genus-2 (χ=−2)";
+
+    auto maps = setup_euclidean_maps(mesh);
+    compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+    const int n = setup_closed_mesh_dofs(mesh, maps);
+    enforce_gauss_bonnet(mesh, maps);
+    apply_natural_theta(mesh, maps, n);
+
+    // Start from a small perturbation so Newton actually iterates.
+    std::vector<double> x0(static_cast<std::size_t>(n), -0.05);
+
+    // Newton solve
+    auto t0  = Clock::now();
+    auto res = newton_euclidean(mesh, x0, maps, 1e-9, 200);
+    auto dt_newton = std::chrono::duration_cast<Ms>(Clock::now() - t0).count();
+
+    // Cut graph
+    auto t1  = Clock::now();
+    CutGraph cg = compute_cut_graph(mesh);
+    auto dt_cut = std::chrono::duration_cast<Ms>(Clock::now() - t1).count();
+
+    std::cout << "[SmokeEuclidean.Brezel] V=" << mesh.number_of_vertices()
+              << " F=" << mesh.number_of_faces()
+              << " iter=" << res.iterations
+              << " ||G||=" << res.grad_inf_norm
+              << " newton=" << dt_newton << "ms"
+              << " cut=" << dt_cut << "ms\n";
+
+    EXPECT_TRUE(res.converged)    << "Newton did not converge on brezel.obj";
+    EXPECT_LT(res.iterations, 30) << "Newton took ≥ 30 iterations";
+    EXPECT_LT(res.grad_inf_norm, 1e-8);
+
+    // Genus-2: 2g = 4 seam edges
+    EXPECT_EQ(4u, cg.cut_edge_indices.size())
+        << "brezel.obj (genus 2) must yield 2g=4 cut edges";
+    EXPECT_EQ(2, cg.genus);
+}
+
+// ── Test 3 — brezel2.obj (V=2622, F=5248, genus=2) ───────────────────────────
+
+TEST(SmokeEuclidean, Brezel2_Genus2_CutGraph)
+{
+    const std::string path = std::string(CONFORMALLAB_DATA_DIR) + "/obj/brezel2.obj";
+    ConformalMesh mesh;
+    ASSERT_NO_THROW(mesh = load_mesh(path)) << "brezel2.obj not found: " << path;
+
+    EXPECT_EQ(2622u, mesh.number_of_vertices());
+    EXPECT_EQ(5248u, mesh.number_of_faces());
+
+    const int chi = static_cast<int>(mesh.number_of_vertices())
+                  - static_cast<int>(mesh.number_of_edges())
+                  + static_cast<int>(mesh.number_of_faces());
+    EXPECT_EQ(-2, chi) << "brezel2.obj must be genus-2 (χ=−2)";
+
+    // Cut graph only — Newton on genus-2 requires full DOF setup
+    // (tested separately in test_geometry_utils.cpp HomologyGenerators suite)
+    auto t0  = Clock::now();
+    CutGraph cg = compute_cut_graph(mesh);
+    auto dt_cut = std::chrono::duration_cast<Ms>(Clock::now() - t0).count();
+
+    std::cout << "[SmokeEuclidean.Brezel2] V=" << mesh.number_of_vertices()
+              << " F=" << mesh.number_of_faces()
+              << " cut=" << dt_cut << "ms"
+              << " seams=" << cg.cut_edge_indices.size() << "\n";
+
+    // Genus-2: 2g = 4 seam edges
+    EXPECT_EQ(4u, cg.cut_edge_indices.size())
+        << "brezel2.obj (genus 2) must yield 2g=4 cut edges";
+    EXPECT_EQ(2, cg.genus);
+}

code/tests/cgal/test_spherical_functional.cpp

@@ -0,0 +1,358 @@
+// test_spherical_functional.cpp  (Phase 3c + 3e)
+//
+// Phase 3c — SphericalFunctional ported to ConformalMesh.
+//
+// Corresponds to de.varylab.discreteconformal.functional.SphericalFunctionalTest.
+//
+// Test map (Java → C++)
+// ──────────────────────
+//   testHessian (Ignored)         → GradientCheck_Hessian     (ported)
+//   testGradientWithHyperIdeal…   → GradientCheck_OctaFaceVertex   (ported)
+//   testGradientInExtendedDomain  → GradientCheck_SpherTetVertex    (ported)
+//   testGradientWithHyperelliptic → GradientCheck_SpherTetAllDofs   (ported)
+//   testFunctionalAtNaNValue      → AnglesFiniteAtKnownPoint         (ported)
+//
+// Energy model
+// ────────────
+//   The energy is computed as the Schläfli path integral E(x) = ∫₀¹⟨G(tx),x⟩dt
+//   using 10-point Gauss-Legendre quadrature. The gradient check therefore
+//   verifies that G is curl-free (the integrability / exactness condition of
+//   the spherical discrete conformal functional). This is equivalent to the
+//   Java FunctionalTest gradient check.
+
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include "spherical_functional.hpp"
+#include "spherical_hessian.hpp"
+#include <gtest/gtest.h>
+#include <cmath>
+#include <vector>
+
+using namespace conformallab;
+
+// ════════════════════════════════════════════════════════════════════════════
+// Cross-module Hessian check: spherical_gradient() ↔ spherical_hessian()
+//
+// Java @Ignore reason: "no Hessian implemented" — the Java functional test
+// was written before the Hessian existed.  In C++ the analytic spherical
+// Hessian (spherical_hessian.hpp, Phase 3f) is complete.
+//
+// This test verifies cross-module consistency between the functional and
+// the Hessian module.  The spherical Hessian is NSD (negative semi-definite)
+// because the spherical energy is concave — hessian_check_spherical() uses
+// the sign-corrected FD check appropriate for the spherical case.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalFunctional, GradientCheck_Hessian)
+{
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), -0.2);
+
+    EXPECT_TRUE(hessian_check_spherical(mesh, x, maps))
+        << "Cross-module: spherical_gradient() and spherical_hessian() are inconsistent";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Angle formula: octahedron-face triangle has all angles = π/2
+//
+// The triangle (1,0,0)–(0,1,0)–(0,0,1) has l_ij = π/2 for all edges.
+// Half-angle formula: s = 3π/4, s_ij = π/4 for all three.
+// All angles = π/2 (right-angled spherical triangle).
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalFunctional, OctaFaceAnglesAreRightAngles)
+{
+    // l_ij = π/2 for all edges (octahedron face on unit sphere)
+    const double l = PI_SPHER / 2.0;
+    auto fa = spherical_angles(l, l, l);
+
+    ASSERT_TRUE(fa.valid) << "Equilateral spherical triangle must be valid";
+    EXPECT_NEAR(PI_SPHER / 2.0, fa.alpha1, 1e-12);
+    EXPECT_NEAR(PI_SPHER / 2.0, fa.alpha2, 1e-12);
+    EXPECT_NEAR(PI_SPHER / 2.0, fa.alpha3, 1e-12);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Angle sum of a spherical triangle exceeds π (positive curvature)
+//
+// For the spherical tetrahedron face (arccos(−1/3) ≈ 1.9106 per edge):
+// The dihedral angle = arccos(1/3) ≈ 70.53°; by symmetry the face angles
+// (vertex angles of the spherical triangle) are all equal.
+// Angle sum must be > π and equal 3·arccos(1/3) ≈ 3·1.2310 ≈ 3.693 rad.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalFunctional, SpherTetAngleSumExceedsPi)
+{
+    // Edge length of spherical tetrahedron face: arccos(−1/3)
+    const double l = std::acos(-1.0 / 3.0);
+    auto fa = spherical_angles(l, l, l);
+
+    ASSERT_TRUE(fa.valid);
+    EXPECT_GT(fa.alpha1 + fa.alpha2 + fa.alpha3, PI_SPHER)
+        << "Angle sum of spherical triangle must exceed π";
+
+    // By symmetry all three angles must be equal
+    EXPECT_NEAR(fa.alpha1, fa.alpha2, 1e-12);
+    EXPECT_NEAR(fa.alpha2, fa.alpha3, 1e-12);
+
+    // For a regular spherical tetrahedron with edge arccos(−1/3):
+    // half-angle: tan(α/2) = √(sin(l/2)/sin(3l/2)) = √3 → α/2 = π/3 → α = 2π/3.
+    // (arccos(1/3) ≈ 1.231 is the 3D dihedral angle of a Euclidean tetrahedron, not this.)
+    double expected = 2.0 * PI_SPHER / 3.0;   // 120°
+    EXPECT_NEAR(fa.alpha1, expected, 1e-10);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Gradient check: octahedron-face triangle, vertex DOFs only
+//
+// Sets λ° from mesh geometry (unit sphere), all u_i = −0.3 (slightly smaller).
+// Mirrors Java testGradientWithHyperIdeal… on a single-triangle mesh.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalFunctional, GradientCheck_OctaFaceVertex)
+{
+    auto mesh = make_octahedron_face();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    // Small uniform conformal factor: shrink the triangle slightly.
+    std::vector<double> x(static_cast<std::size_t>(n), -0.3);
+
+    EXPECT_TRUE(gradient_check_spherical(mesh, x, maps))
+        << "Gradient check failed on octahedron-face triangle (vertex DOFs)";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Gradient check: spherical tetrahedron (4 faces), vertex DOFs only
+//
+// Closed surface; exercises accumulation over multiple faces per vertex.
+// Mirrors Java testGradientInTheExtendedDomain.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalFunctional, GradientCheck_SpherTetVertex)
+{
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), -0.2);
+
+    EXPECT_TRUE(gradient_check_spherical(mesh, x, maps))
+        << "Gradient check failed on spherical tetrahedron (vertex DOFs)";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Gradient check: spherical tetrahedron, all DOFs (vertex + edge)
+//
+// Exercises the edge-gradient branch: G_e = α_opp⁺ + α_opp⁻ − π.
+// Mirrors Java testGradientWithHyperellipticCurve.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalFunctional, GradientCheck_SpherTetAllDofs)
+{
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_all_spherical_dof_indices(mesh, maps);
+
+    // Small but non-zero values; vertex DOFs negative, edge DOFs zero.
+    // Edge DOF adjusts the effective log-length Λ_ij = λ°_ij + u_i + u_j + λ_e.
+    std::vector<double> x(static_cast<std::size_t>(n), 0.0);
+    // Set vertex DOFs (indices 0..3) to -0.2 to keep triangle well-formed.
+    for (int i = 0; i < 4; ++i) x[static_cast<std::size_t>(i)] = -0.2;
+
+    EXPECT_TRUE(gradient_check_spherical(mesh, x, maps))
+        << "Gradient check failed on spherical tetrahedron (all DOFs)";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Angles are finite at a known interior point
+//
+// Mirrors Java testFunctionalAtNaNValue: choose DOFs that could hit
+// a degenerate branch (l_ij → 0 or triangle inequality fails) and check
+// the gradient vector is free of NaN/Inf.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalFunctional, AnglesFiniteAtKnownPoint)
+{
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    // u_i = -1.5: contracts the triangle heavily but stays non-degenerate.
+    std::vector<double> x(static_cast<std::size_t>(n), -1.5);
+    auto G = spherical_gradient(mesh, x, maps);
+
+    for (std::size_t i = 0; i < G.size(); ++i) {
+        EXPECT_FALSE(std::isnan(G[i])) << "Gradient component " << i << " is NaN";
+        EXPECT_FALSE(std::isinf(G[i])) << "Gradient component " << i << " is Inf";
+    }
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Gradient check: fan-4 mesh on unit sphere, vertex DOFs only
+//
+// Make a fan of 4 triangles around the north pole (0,0,1);
+// rim vertices projected onto the equator.
+// Exercises high-valence vertex gradient accumulation.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalFunctional, GradientCheck_SpherFan4Vertex)
+{
+    // Build a fan with 4 spherical triangles manually (can't use make_fan
+    // directly because those vertices are not on the unit sphere).
+    ConformalMesh mesh;
+    auto center = mesh.add_vertex(Point3(0, 0, 1));  // north pole
+    const int n_rim = 4;
+    std::vector<Vertex_index> rim(n_rim);
+    const double dtheta = 2.0 * PI_SPHER / n_rim;
+    const double phi    = PI_SPHER / 4.0;  // 45° colatitude
+    for (int i = 0; i < n_rim; ++i) {
+        double theta = i * dtheta;
+        rim[i] = mesh.add_vertex(Point3(
+            std::sin(phi) * std::cos(theta),
+            std::sin(phi) * std::sin(theta),
+            std::cos(phi)));
+    }
+    for (int i = 0; i < n_rim; ++i)
+        mesh.add_face(center, rim[i], rim[(i + 1) % n_rim]);
+
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int ndof = assign_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(ndof), -0.3);
+
+    EXPECT_TRUE(gradient_check_spherical(mesh, x, maps))
+        << "Gradient check failed on spherical fan-4 mesh";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Gradient check: mixed pinned/variable vertices
+//
+// One vertex pinned (u_v = 0 fixed), others variable.
+// Verifies that the gradient accumulation skips pinned vertices correctly.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalFunctional, GradientCheck_MixedPinnedVertices)
+{
+    auto mesh = make_octahedron_face();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+
+    // Pin v0, make v1 and v2 variable.
+    auto vit = mesh.vertices().begin();
+    Vertex_index v0 = *vit++;
+    Vertex_index v1 = *vit++;
+    Vertex_index v2 = *vit;
+
+    maps.v_idx[v0] = -1;   // pinned
+    maps.v_idx[v1] = 0;
+    maps.v_idx[v2] = 1;
+
+    std::vector<double> x = {-0.2, -0.4};
+
+    EXPECT_TRUE(gradient_check_spherical(mesh, x, maps))
+        << "Gradient check failed for mixed pinned/variable vertices";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Phase 3e — Gauge-fix for closed spherical surfaces
+//
+// On a closed spherical surface, the functional has a gauge mode:
+//   E(u + t·1)  is maximised at some t*.
+//   At t*, the sum of all vertex gradients equals zero: Σ G_v = 0.
+//
+// Test: start from a point with non-zero ΣG_v, apply the gauge shift,
+// and verify ΣG_v(x + t*·1) ≈ 0.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalFunctional, GaugeFix_SpherTetVertexZerosSumGv)
+{
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    // Off-gauge starting point: all u_i = -0.5
+    std::vector<double> x(static_cast<std::size_t>(n), -0.5);
+
+    // Compute ΣG_v before gauge shift.
+    {
+        auto G = spherical_gradient(mesh, x, maps);
+        double sum = 0.0;
+        for (auto v : mesh.vertices()) {
+            int iv = maps.v_idx[v];
+            if (iv >= 0) sum += G[static_cast<std::size_t>(iv)];
+        }
+        // At -0.5 the surface is compressed; ΣG_v should be non-zero.
+        EXPECT_NE(sum, 0.0) << "Pre-gauge ΣG_v should be non-zero";
+    }
+
+    // Compute gauge shift and apply.
+    double t = spherical_gauge_shift(mesh, x, maps);
+    std::vector<double> x_fixed = x;
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv >= 0)
+            x_fixed[static_cast<std::size_t>(iv)] += t;
+    }
+
+    // Verify ΣG_v ≈ 0 at the gauge-fixed point.
+    {
+        auto G = spherical_gradient(mesh, x_fixed, maps);
+        double sum = 0.0;
+        for (auto v : mesh.vertices()) {
+            int iv = maps.v_idx[v];
+            if (iv >= 0) sum += G[static_cast<std::size_t>(iv)];
+        }
+        EXPECT_NEAR(sum, 0.0, 1e-6)
+            << "After gauge fix, Σ G_v should vanish; t* = " << t;
+    }
+}
+
+TEST(SphericalFunctional, GaugeFix_ApplyInPlace)
+{
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    // x = -0.3: compressed but inside the valid spherical domain.
+    std::vector<double> x(static_cast<std::size_t>(n), -0.3);
+
+    apply_spherical_gauge(mesh, x, maps);
+
+    // After in-place gauge fix, ΣG_v must be near 0.
+    auto G = spherical_gradient(mesh, x, maps);
+    double sum = 0.0;
+    for (auto v : mesh.vertices()) {
+        int iv = maps.v_idx[v];
+        if (iv >= 0) sum += G[static_cast<std::size_t>(iv)];
+    }
+    EXPECT_NEAR(sum, 0.0, 1e-6)
+        << "apply_spherical_gauge must drive Σ G_v to zero";
+}
+
+TEST(SphericalFunctional, GaugeFix_AlreadyAtGaugeReturnsTNearZero)
+{
+    // A symmetric, equilateral spherical tetrahedron at x=0 is already
+    // at the gauge maximum (by symmetry, ΣG_v = 0).
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), 0.0);
+    double t = spherical_gauge_shift(mesh, x, maps);
+    // Symmetric starting point → t* should be very close to 0.
+    EXPECT_NEAR(t, 0.0, 1e-5)
+        << "Gauge shift from the symmetric point should be ~0; got " << t;
+}

code/tests/cgal/test_spherical_hessian.cpp

@@ -0,0 +1,205 @@
+// test_spherical_hessian.cpp
+//
+// Phase 3f — Spherical cotangent-Laplace Hessian.
+//
+// The Hessian of the spherical discrete conformal energy is the "spherical
+// cotangent Laplacian" with edge weight
+//   w_k = cot(β_k),  β_k = (π − αi − αj + αk) / 2
+// for the edge (vi, vj) with opposite vertex vk and angles αi, αj, αk.
+//
+// In the flat limit (αi+αj+αk → π) this reduces to the Euclidean cotangent
+// Laplacian (β_k → αk, w_k → cot(αk)).
+//
+// Tests:
+//   1. Spherical cot weights match Euclidean weights in the near-flat limit.
+//   2. Hessian is symmetric.
+//   3. Hessian has H·1 ≈ 0 on the spherical tetrahedron (null-space property).
+//   4. Finite-difference check (the primary correctness criterion).
+
+#include "conformal_mesh.hpp"
+#include "mesh_builder.hpp"
+#include "spherical_hessian.hpp"
+#include "euclidean_hessian.hpp"  // for Euclidean comparison
+#include <gtest/gtest.h>
+#include <Eigen/Dense>
+#include <cmath>
+#include <vector>
+
+using namespace conformallab;
+
+// ════════════════════════════════════════════════════════════════════════════
+// Spherical cot weights reduce to Euclidean cot weights in the flat limit
+//
+// For a nearly-flat equilateral spherical triangle (l → 0, α → 60°):
+//   β_k = (π − 60° − 60° + 60°)/2 = 60°  →  cot(60°) = 1/√3  ✓
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalHessian, CotWeights_NearFlatEquilateral)
+{
+    // Use a very small equilateral spherical triangle: α1=α2=α3=60°
+    const double alpha = PI / 3.0;
+    auto sw = spherical_cot_weights(alpha, alpha, alpha);
+
+    ASSERT_TRUE(sw.valid);
+
+    const double expected = 1.0 / std::sqrt(3.0);   // = cot(60°)
+    EXPECT_NEAR(sw.w12, expected, 1e-12);
+    EXPECT_NEAR(sw.w23, expected, 1e-12);
+    EXPECT_NEAR(sw.w31, expected, 1e-12);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Spherical cot weights are positive for the regular spherical tetrahedron
+//
+// Each face has α_k = 2π/3 (120°), angle sum = 2π.
+// β_k = (π − 2π/3 − 2π/3 + 2π/3)/2 = (π − 2π/3)/2 = π/6
+// w_k = cot(π/6) = √3
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalHessian, CotWeights_RegularSphericalTetrahedronFace)
+{
+    const double alpha = 2.0 * PI / 3.0;   // 120°
+    auto sw = spherical_cot_weights(alpha, alpha, alpha);
+
+    ASSERT_TRUE(sw.valid);
+
+    const double expected = std::sqrt(3.0);   // cot(π/6)
+    EXPECT_NEAR(sw.w12, expected, 1e-10);
+    EXPECT_NEAR(sw.w23, expected, 1e-10);
+    EXPECT_NEAR(sw.w31, expected, 1e-10);
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Hessian is symmetric: H[i,j] == H[j,i]
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalHessian, HessianIsSymmetric)
+{
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), -0.2);
+    auto H = spherical_hessian(mesh, x, maps);
+
+    Eigen::MatrixXd Hd = Eigen::MatrixXd(H);
+    EXPECT_NEAR((Hd - Hd.transpose()).norm(), 0.0, 1e-12)
+        << "Spherical Hessian must be symmetric";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// H·1 is NOT zero for the spherical Hessian (unlike the Euclidean case).
+//
+// In the Euclidean case, a uniform shift u_i → u_i + c scales all edge
+// lengths by e^c, leaving angles unchanged → H·1 = 0 exactly.
+//
+// In the spherical case, l_ij = 2·asin(exp(λ_ij/2)) is NOT a linear
+// function of the DOFs, so a uniform shift DOES change the angles
+// → H·1 ≠ 0 in general.  This test verifies that property.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalHessian, ConstantVectorNotInNullSpace)
+{
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), -0.2);
+    auto H = spherical_hessian(mesh, x, maps);
+
+    Eigen::VectorXd ones = Eigen::VectorXd::Ones(n);
+    Eigen::VectorXd Hones = H * ones;
+
+    // H·1 should be non-trivial (norm well above zero)
+    EXPECT_GT(Hones.norm(), 1e-3)
+        << "Spherical H·1 should be non-zero; norm = " << Hones.norm();
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Hessian is negative semi-definite at the equilibrium point (x = 0)
+//
+// The spherical discrete conformal energy is concave in the vertex DOFs
+// (unlike the Euclidean case which is convex).  At the equilibrium x = 0,
+// the Hessian is NSD: all eigenvalues ≤ 0, with a multi-dimensional null
+// space corresponding to degenerate directions.
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalHessian, HessianIsNegativeSemiDefiniteAtEquilibrium)
+{
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    // x = 0 is the equilibrium for the regular spherical tetrahedron.
+    std::vector<double> x(static_cast<std::size_t>(n), 0.0);
+    auto H = spherical_hessian(mesh, x, maps);
+
+    Eigen::MatrixXd Hd = Eigen::MatrixXd(H);
+    Eigen::SelfAdjointEigenSolver<Eigen::MatrixXd> es(Hd);
+    double max_ev = es.eigenvalues().maxCoeff();
+
+    EXPECT_LE(max_ev, 1e-9)
+        << "Largest eigenvalue of spherical Hessian at equilibrium must be ≤ 0; got " << max_ev;
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Finite-difference Hessian check: octahedron-face triangle (vertex DOFs)
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalHessian, FDCheck_OctaFaceVertex)
+{
+    auto mesh = make_octahedron_face();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), -0.3);
+
+    EXPECT_TRUE(hessian_check_spherical(mesh, x, maps))
+        << "FD Hessian check failed on octahedron-face triangle";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Finite-difference Hessian check: spherical tetrahedron (vertex DOFs)
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalHessian, FDCheck_SpherTetVertex)
+{
+    auto mesh = make_spherical_tetrahedron();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+    int n = assign_vertex_dof_indices(mesh, maps);
+
+    std::vector<double> x(static_cast<std::size_t>(n), -0.2);
+
+    EXPECT_TRUE(hessian_check_spherical(mesh, x, maps))
+        << "FD Hessian check failed on spherical tetrahedron";
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Finite-difference Hessian check: mixed pinned/variable vertices
+// ════════════════════════════════════════════════════════════════════════════
+
+TEST(SphericalHessian, FDCheck_MixedPinnedVertices)
+{
+    auto mesh = make_octahedron_face();
+    auto maps = setup_spherical_maps(mesh);
+    compute_lambda0_from_mesh(mesh, maps);
+
+    auto vit = mesh.vertices().begin();
+    Vertex_index v0 = *vit++;
+    Vertex_index v1 = *vit++;
+    Vertex_index v2 = *vit;
+
+    maps.v_idx[v0] = -1;   // pinned
+    maps.v_idx[v1] = 0;
+    maps.v_idx[v2] = 1;
+
+    std::vector<double> x = {-0.2, -0.3};
+
+    EXPECT_TRUE(hessian_check_spherical(mesh, x, maps))
+        << "FD Hessian check failed for mixed pinned/variable vertices";
+}

code/tests/test_discrete_elliptic_utility.cpp

@@ -0,0 +1,39 @@
+// Port of de.varylab.discreteconformal.util.DiscreteEllipticUtilityTest (Java/JUnit).
+// Tests the normalizeModulus function that moves a complex number tau into the
+// fundamental domain of the modular group SL(2,Z).
+
+#include "discrete_elliptic_utility.hpp"
+#include <gtest/gtest.h>
+#include <complex>
+#include <cmath>
+
+using namespace conformallab;
+
+// Corresponds to Java testNormalizeModulus()
+TEST(DiscreteEllipticUtilityTest, NormalizeModulus) {
+    // tau already in fundamental domain → should be returned unchanged
+    std::complex<double> tau(0.45, 1.1);
+    auto tauNorm = normalizeModulus(tau);
+    EXPECT_NEAR(0.45, tauNorm.real(), 1E-12);
+    EXPECT_NEAR(1.1,  tauNorm.imag(), 1E-12);
+
+    // tau = i/3  (|tau| < 1) → inversion gives 3i
+    tau = std::complex<double>(0.0, 1.0 / 3.0);
+    tauNorm = normalizeModulus(tau);
+    EXPECT_NEAR(3.0, tauNorm.imag(), 1E-12);
+    EXPECT_NEAR(0.0, tauNorm.real(), 1E-12);
+}
+
+// Corresponds to Java testNormalizeModulusPeriodShift()
+// Two tau values that differ by a T-shift (integer shift of Re) must normalize
+// to the same point in the fundamental domain.
+TEST(DiscreteEllipticUtilityTest, NormalizeModulusPeriodShift) {
+    std::complex<double> tau1(0.3,  1.0);
+    std::complex<double> tau2(-0.7, 1.0);   // tau2 = tau1 - 1
+
+    auto n1 = normalizeModulus(tau1);
+    auto n2 = normalizeModulus(tau2);
+
+    EXPECT_NEAR(n1.real(), n2.real(), 1E-12) << "real parts should be equal";
+    EXPECT_NEAR(n1.imag(), n2.imag(), 1E-12) << "imag parts should be equal";
+}

code/tests/test_hyper_ideal_visualization_utility.cpp

@@ -0,0 +1,50 @@
+// Port of de.varylab.discreteconformal.plugin.HyperIdealVisualizationPluginTest (Java/JUnit).
+//
+// Tests the conversion from a hyperbolic circle (hyperboloid model)
+// to its Euclidean representation (Poincaré disk model).
+//
+// Java test: static method HyperIdealVisualizationPlugin
+//              .getEuclideanCircleFromHyperbolic(double[] center, double radius)
+// C++ port:  conformallab::getEuclideanCircleFromHyperbolic(Vector4d, double)
+//              in hyper_ideal_visualization_utility.hpp
+
+#include "hyper_ideal_visualization_utility.hpp"
+#include <gtest/gtest.h>
+#include <cmath>
+
+using namespace conformallab;
+
+// Corresponds to Java testGetEuclideanCircleFromHyperbolic_Centered()
+//
+// A hyperbolic circle centered at the origin (0,0,0,1) with radius 1.
+// In the Poincaré disk this maps to a Euclidean circle centered at (0,0)
+// with radius  sinh(1) / (cosh(1) + 1).
+TEST(HyperIdealVisualizationUtilityTest, EuclideanCircleFromHyperbolic_Centered)
+{
+    Eigen::Vector4d center(0.0, 0.0, 0.0, 1.0);
+    const double radius = 1.0;
+
+    auto result = getEuclideanCircleFromHyperbolic(center, radius);
+
+    const double expected_r = std::sinh(1.0) / (std::cosh(1.0) + 1.0);
+    EXPECT_NEAR(0.0,        result[0], 1E-12) << "Euclidean cx should be 0";
+    EXPECT_NEAR(0.0,        result[1], 1E-12) << "Euclidean cy should be 0";
+    EXPECT_NEAR(expected_r, result[2], 1E-12) << "Euclidean radius mismatch";
+}
+
+// Corresponds to Java testGetEuclideanCircleFromHyperbolic_OffCenter()
+//
+// A hyperbolic circle centered at (sinh(1),0,0,cosh(1)) with radius 1.
+// By symmetry (center is on the x-axis, circle is symmetric about it):
+//   • the Euclidean center lies on the x-axis  → cy = 0
+//   • the Euclidean center equals the radius   → cx = r  (the circle passes through the Poincaré origin)
+TEST(HyperIdealVisualizationUtilityTest, EuclideanCircleFromHyperbolic_OffCenter)
+{
+    Eigen::Vector4d center(std::sinh(1.0), 0.0, 0.0, std::cosh(1.0));
+    const double radius = 1.0;
+
+    auto result = getEuclideanCircleFromHyperbolic(center, radius);
+
+    EXPECT_NEAR(result[0], result[2], 1E-12) << "cx should equal Euclidean radius";
+    EXPECT_NEAR(0.0, result[1],       1E-12) << "cy should be 0 (x-axis symmetry)";
+}

code/tests/test_p2_utility.cpp

@@ -0,0 +1,87 @@
+// Port of de.varylab.discreteconformal.math.P2BigTest (Java/JUnit).
+// Tests 2-D projective geometry utilities: perpendicular bisectors,
+// point-from-lines, and direct isometries in the Euclidean plane.
+//
+// The Java test compared double precision (P2) against BigDecimal precision
+// (P2Big) to 1E-10.  Here we compare double against long double to the
+// same tolerance.
+
+#include "p2_utility.hpp"
+#include <gtest/gtest.h>
+#include <Eigen/Dense>
+#include <cmath>
+
+using namespace conformallab;
+
+// Corresponds to Java P2BigTest.testMakeDirectIsometryFromFramesEuclidean()
+//
+// Computes the Euclidean isometry mapping frame (s1,s2) to frame (t1,t2)
+// with both double and long-double precision, and checks:
+//   1. The two precisions agree to 1E-10 (precision stability).
+//   2. The matrix actually maps s1→t1 and s2→t2.
+TEST(P2UtilityTest, MakeDirectIsometryFromFramesEuclidean) {
+    using V3d  = Eigen::Vector3d;
+    using V3ld = Eigen::Matrix<long double, 3, 1>;
+
+    V3d s1(-1.4142135623730963, 0.0, 1.0);
+    V3d s2( 1.4142135623730951, 0.0, 1.0);
+    V3d t1(-2.828427124746189,  2.4494897427831805, 1.0);
+    V3d t2( 0.0,               2.4494897427831783, 1.0);
+
+    // double precision
+    auto T = makeDirectIsometryFromFramesEuclidean<double>(s1, s2, t1, t2);
+
+    // long double precision (analogous to Java's BigDecimal P2Big)
+    V3ld s1l = s1.cast<long double>();
+    V3ld s2l = s2.cast<long double>();
+    V3ld t1l = t1.cast<long double>();
+    V3ld t2l = t2.cast<long double>();
+    auto Tl = makeDirectIsometryFromFramesEuclidean<long double>(s1l, s2l, t1l, t2l);
+
+    // 1. double vs long double must agree to 1E-10
+    for (int i = 0; i < 3; ++i)
+        for (int j = 0; j < 3; ++j)
+            EXPECT_NEAR((double)Tl(i,j), T(i,j), 1E-10)
+                << "element (" << i << "," << j << ") differs between precisions";
+
+    // 2. T must map s1 → t1 and s2 → t2  (verify isometry correctness)
+    auto map_s1 = T * s1;
+    auto map_s2 = T * s2;
+    EXPECT_NEAR(euclideanDistanceP2(map_s1, t1), 0.0, 1E-9) << "T*s1 should equal t1";
+    EXPECT_NEAR(euclideanDistanceP2(map_s2, t2), 0.0, 1E-9) << "T*s2 should equal t2";
+}
+
+// Corresponds to Java P2BigTest.testPerpendicularBisector()
+TEST(P2UtilityTest, PerpendicularBisector) {
+    Eigen::Vector3d p1(0.5, 0.0, 1.0);
+    Eigen::Vector3d q1(0.0, 0.5, 1.0);
+
+    auto bisector = perpendicularBisectorEuclidean(p1, q1);
+
+    EXPECT_NEAR( 0.5, bisector(0), 1E-10);
+    EXPECT_NEAR(-0.5, bisector(1), 1E-10);
+    EXPECT_NEAR( 0.0, bisector(2), 1E-10);
+}
+
+// Corresponds to Java P2BigTest.testPerpendicularBisectorIntersection()
+//
+// The intersection of the perpendicular bisectors of two edges must be
+// equidistant from the endpoints of each edge (circumcenter property).
+TEST(P2UtilityTest, PerpendicularBisectorIntersection) {
+    Eigen::Vector3d p1(0.5, 0.0, 1.0);
+    Eigen::Vector3d q1(0.0, 1.0, 1.0);
+    Eigen::Vector3d p2(1.0, 0.0, 1.0);
+    Eigen::Vector3d q2(0.0, 1.5, 1.0);
+
+    auto l1 = perpendicularBisectorEuclidean(p1, q1);
+    auto l2 = perpendicularBisectorEuclidean(p2, q2);
+    auto o  = pointFromLines(l1, l2);    // circumcenter
+
+    // o must be equidistant from p1 and q1
+    EXPECT_NEAR(euclideanDistanceP2(p1, o),
+                euclideanDistanceP2(q1, o), 1E-10);
+
+    // o must be equidistant from p2 and q2
+    EXPECT_NEAR(euclideanDistanceP2(p2, o),
+                euclideanDistanceP2(q2, o), 1E-10);
+}

doc/api/cgal-package.md

@@ -0,0 +1,403 @@
+# Phase 8 — CGAL Package Design
+
+> **Status: design frozen, implementation planned.**
+> This document captures the strategic decisions taken before the first
+> line of Phase 8 code is written.  The decisions were taken on 2026-05-19
+> after the docstring/architecture audit at the end of Phase 7.
+>
+> The design is informed by the CGAL package submission guidelines at
+> https://www.cgal.org/developers.html and by reading the existing
+> `Polygon_mesh_processing` and `Surface_mesh_parameterization` packages.
+
+---
+
+## Strategic position
+
+| Question | Decision | Rationale |
+|---|---|---|
+| Submission to CGAL? | **Pre-submission-ready, not submission-bound.** 12+ months horizon, optional. | Keep design freedom, no editor-review pressure. Structure is valuable on its own. |
+| License | **MIT preserved.** | CGAL submission would require LGPL — deferred. Current users (academic + industrial) profit from MIT. |
+| Mesh-type flexibility | **Generic `FaceGraph + HalfedgeGraph`.** | Maximum CGAL value: works with `Surface_mesh`, `Polyhedron_3`, OpenMesh-adapter, pmp. |
+| Parameter style | **Named Parameters** (`CGAL::parameters::vertex_curvature_map(...).max_iterations(50)`). | CGAL standard; identical UX to `PMP::triangulate_*`. |
+| Default kernel | **`CGAL::Simple_cartesian<double>`.** | Status quo. Conformal geometry does not require exact predicates. |
+| Backward compatibility | **Dual-layer wrapper.** `code/include/*.hpp` stays as implementation; `include/CGAL/*.h` is thin wrapper. | Existing 176 + 36 tests unchanged. New API gets new tests. |
+| Algorithm code | **No duplication.** New CGAL headers delegate to existing code via property-map adapters. | Single source of truth; no parallel maintenance. |
+
+---
+
+## Architecture
+
+### Three-layer model
+
+```
+┌──────────────────────────────────────────────────────────────────┐
+│  Layer 3: Public CGAL API           include/CGAL/*.h              │
+│  ─────────────────────────                                         │
+│  • Conformal_map_traits.h           ← concept + default model     │
+│  • Discrete_conformal_map.h         ← user-facing entry           │
+│  • Conformal_layout.h, ...                                         │
+│  Named parameters, generic over FaceGraph, Doxygen-documented.     │
+└──────────────────────────────────────────────────────────────────┘
+                              ▲
+                              │ thin wrapper, no algorithm code
+                              │
+┌──────────────────────────────────────────────────────────────────┐
+│  Layer 2: Adapter / Traits          include/CGAL/Conformal_map/   │
+│  ─────────────────────────                                         │
+│  • Default_traits.h                 ← maps generic FaceGraph to   │
+│                                       Surface_mesh property maps  │
+│  • Property_map_adapter.h           ← read/write u, θ, α via      │
+│                                       boost::property_map traits  │
+└──────────────────────────────────────────────────────────────────┘
+                              ▲
+                              │ uses existing algorithms as-is
+                              │
+┌──────────────────────────────────────────────────────────────────┐
+│  Layer 1: Implementation            code/include/*.hpp            │
+│  ─────────────────────────                                         │
+│  euclidean_functional.hpp, layout.hpp, newton_solver.hpp, ...     │
+│  Hardcoded to Surface_mesh + Simple_cartesian — unchanged.        │
+└──────────────────────────────────────────────────────────────────┘
+```
+
+---
+
+## 8a — Traits class & concepts
+
+### `ConformalMapTraits` concept
+
+The concept lists the types and operations every Traits model must provide.
+
+```cpp
+namespace CGAL {
+
+// Concept (documentation only; no code):
+struct ConformalMapTraits {
+    // Types
+    using Triangle_mesh = ...;            // model of FaceGraph + HalfedgeGraph
+    using FT            = ...;            // typically double
+    using Vertex_descriptor = boost::graph_traits<Triangle_mesh>::vertex_descriptor;
+    using Halfedge_descriptor = ...;
+    using Face_descriptor    = ...;
+
+    // Read access (input geometry)
+    using Vertex_point_map = ...;         // model of ReadablePropertyMap
+                                           //   key:   Vertex_descriptor
+                                           //   value: K::Point_3
+
+    // Read/write access (conformal data)
+    using Lambda_pmap       = ...;        // u_v   (scale factor)         RW
+    using Theta_pmap        = ...;        // Θ_v   (target curvature)     R
+    using Vertex_index_pmap = ...;        // DOF index (−1 = pinned)      RW
+    using Edge_alpha_pmap   = ...;        // α_e   (hyperbolic only)      RW
+    using Face_type_pmap    = ...;        // geometry tag per face        R
+
+    // Optional output
+    using UV_pmap           = ...;        // halfedge → (u, v) ∈ ℝ²       W
+    using Holonomy_pmap     = ...;        // seam edge → ω ∈ ℂ            W
+};
+
+}
+```
+
+### `Default_conformal_map_traits<TM>`
+
+The default model wraps `Surface_mesh` property maps so existing code keeps
+working through the new public API.
+
+```cpp
+template <class TriangleMesh,
+          class K = CGAL::Simple_cartesian<double>>
+struct Default_conformal_map_traits;
+
+// Specialisation for Surface_mesh:
+template <class K>
+struct Default_conformal_map_traits<CGAL::Surface_mesh<typename K::Point_3>, K> {
+    using Triangle_mesh    = CGAL::Surface_mesh<typename K::Point_3>;
+    using FT               = typename K::FT;
+    using Vertex_point_map = typename Triangle_mesh::Point_property_map;
+    using Lambda_pmap      = typename Triangle_mesh::template Property_map<vertex_descriptor, FT>;
+    // ... etc, using "conformal:lambda" property names
+};
+
+// Generic specialisation for other FaceGraph models will be added in 8a.2.
+```
+
+### Concept-checks
+
+```cpp
+// include/CGAL/Conformal_map_concept_checks.h
+template <class Traits>
+struct Conformal_map_traits_check {
+    static_assert(boost::is_same<...>::value, "Traits::FT must be a floating-point type");
+    static_assert(is_face_graph<Traits::Triangle_mesh>::value);
+    // ...
+};
+```
+
+---
+
+## 8b — Public CGAL header hierarchy
+
+### User-facing entry
+
+```cpp
+// include/CGAL/Discrete_conformal_map.h
+namespace CGAL {
+
+template <class TriangleMesh, class NamedParameters = parameters::Default_named_parameters>
+bool discrete_conformal_map_euclidean(TriangleMesh& mesh,
+                                       const NamedParameters& np = parameters::default_values());
+
+template <class TriangleMesh, class NamedParameters = ...>
+bool discrete_conformal_map_spherical(TriangleMesh& mesh,
+                                       const NamedParameters& np = ...);
+
+template <class TriangleMesh, class NamedParameters = ...>
+bool discrete_conformal_map_hyperbolic(TriangleMesh& mesh,
+                                        const NamedParameters& np = ...);
+
+} // namespace CGAL
+```
+
+### Named parameter vocabulary
+
+| Parameter | Type | Default | Meaning |
+|---|---|---|---|
+| `vertex_curvature_map(pmap)` | ReadablePropertyMap | `2π` at interior, `π` at boundary | Θᵥ values |
+| `fixed_vertex_pmap(pmap)` | ReadablePropertyMap<bool> | First vertex pinned | Which vertices are pinned (gauge) |
+| `max_iterations(n)` | int | 200 | Newton iteration limit |
+| `gradient_tolerance(eps)` | FT | 1e-10 | `‖G‖∞` threshold |
+| `vertex_index_map(pmap)` | LvaluePropertyMap | DOF auto-assigned | Allows user to override DOF assignment |
+| `output_uv_map(pmap)` | WritablePropertyMap | none | If set, writes UV layout into pmap |
+| `cut_graph(cg)` | `Conformal_cut_graph` | auto-computed | Pre-computed seam edges (mandatory for closed surfaces) |
+| `geom_traits(t)` | model of ConformalMapTraits | `Default_*` | Custom traits |
+
+### Modular headers
+
+```
+include/CGAL/
+├── Discrete_conformal_map.h          ← user-facing entry (1 include for casual use)
+├── Conformal_map_traits.h            ← concept + Default_conformal_map_traits
+├── Conformal_map_concept_checks.h
+├── Conformal_newton_solver.h         ← standalone Newton (advanced users)
+├── Conformal_layout.h                ← layout + holonomy
+├── Conformal_cut_graph.h             ← orthogonal algorithm
+├── Conformal_period_matrix.h         ← genus-1 τ (conformallab++ unique)
+├── Conformal_holonomy.h              ← Möbius holonomy (conformallab++ unique)
+└── Conformal_map/                    ← CGAL convention: implementation details
+    ├── Default_traits.h
+    ├── Property_map_adapter.h
+    ├── Newton_iteration.h
+    └── Internal_helpers.h
+```
+
+---
+
+## 8c — CGAL-style documentation
+
+```
+doc/Conformal_map/
+├── PackageDescription.txt            ← CGAL Doxygen package file
+├── Conformal_map.txt                 ← Doxygen User_manual
+├── examples.txt                      ← linkable example code
+├── dependencies                       ← textual list
+└── fig/                              ← pipeline diagrams, math figures
+```
+
+All public functions, concepts, and types require Doxygen.  See **Phase 7.5** (below).
+
+---
+
+## 8d — CGAL test format
+
+```
+test/Conformal_map/
+├── CMakeLists.txt                    ← CGAL-format, uses find_package(CGAL)
+├── test_euclidean_traits.cpp         ← traits concept checks
+├── test_polyhedron_3_backend.cpp     ← tests with Polyhedron_3 as mesh
+├── test_named_parameters.cpp
+└── data/                             ← test meshes
+```
+
+The existing GTest suite at `code/tests/cgal/` remains; CGAL-format tests are
+added alongside as a separate target.  CI runs both.
+
+---
+
+## 8e — Declarative YAML pipeline
+
+A lightweight YAML format for reproducible experiments.  CLI accepts
+`--pipeline experiment.yml`; the validator checks `require`/`provide` tokens
+before execution.
+
+**Full spec:** [doc/concepts/declarative-pipeline.md](../concepts/declarative-pipeline.md)
+— token vocabulary, validation algorithm, 5 complete examples.
+
+```yaml
+pipeline:
+  name: flat_torus_period
+  geometry: euclidean
+  input:  { source: data/torus.off }
+  steps:
+    - { id: setup,  unit: setup_euclidean_maps,        provide: [maps_initialised] }
+    - { id: gb,     unit: enforce_gauss_bonnet,        require: [maps_initialised], provide: [gauss_bonnet_satisfied] }
+    - { id: solve,  unit: newton_euclidean,            require: [gauss_bonnet_satisfied], provide: [x_converged] }
+    - { id: cut,    unit: compute_cut_graph,           require: [mesh_closed],            provide: [cut_graph] }
+    - { id: layout, unit: euclidean_layout,            require: [x_converged, cut_graph], provide: [layout_uv, holonomy] }
+    - { id: period, unit: compute_period_matrix,       require: [holonomy],               provide: [tau] }
+  output:
+    layout: out/torus_layout.off
+    json:   out/torus_result.json
+```
+
+---
+
+## Phase 7.5 — Doxygen infrastructure (prerequisite)
+
+Before any Phase 8 code, the existing API surface must be extractable.
+This is the prerequisite that bridges Phase 7 → Phase 8.
+
+```
+Phase 7.5 — Doxygen infrastructure
+──────────────────────────────────
+  • Doxyfile (CGAL-conform: INPUT=code/include + include/CGAL,
+    EXCLUDE_PATTERNS="* 2.hpp")
+  • doxygen-awesome-css as theme (matches CGAL house style)
+  • CMake target:  cmake --build build --target doc
+  • CI job:  doc-build → publishes to Codeberg Pages or gitea-pages
+  • Extract baseline once → snapshot what is actually exported today
+  • Top-5 central headers (3 functional + conformal_mesh + layout)
+    upgraded to Doxygen comments; the rest follows during Phase 8 implementation
+```
+
+The baseline snapshot doubles as the API-design review tool: before designing
+the public CGAL wrapper, we see exactly which functions, classes and free
+operators exist and need to be wrapped or hidden.
+
+---
+
+## Validation criteria
+
+Phase 8a is "done" when:
+
+1. `cgal.ConformalTraits.Polyhedron_3_works` passes.
+2. `cgal.ConformalTraits.Surface_mesh_default_works` passes — identical results to the legacy API.
+3. The Inversive-Distance functional (Phase 9a) is implementable as the *first* new client of the traits API without architectural changes — no breaking changes to the trait concept.
+4. A user can write `#include <CGAL/Discrete_conformal_map.h>` and call `discrete_conformal_map_euclidean(mesh, parameters::vertex_curvature_map(theta))` against a `Polyhedron_3` and get a valid layout.
+
+If any of these fail, the design is iterated before continuing.
+
+---
+
+## Implementation strategy — "Hybrid MVP" (decided 2026-05-19)
+
+After cost/benefit re-evaluation, the plan is **not** to build Phase 8 in full
+before resuming the port.  Instead:
+
+```
+┌────────────────────────────────────────────────────────────────────┐
+│  PHASE 8 MVP (3–5 days)                                            │
+│  ─────────────────────                                              │
+│  Just enough CGAL-style architecture for Phase 9a to validate it.  │
+│                                                                     │
+│  • Conformal_map_traits.h        concept + Default<Surface_mesh,K> │
+│  • Discrete_conformal_map.h      ONE entry: _euclidean()           │
+│  • 4 named parameters            Θ-map, max_iter, tol, pin         │
+│  • Concept-check header                                             │
+│  • Doxygen on these 3 headers                                       │
+└────────────────────────────────────────────────────────────────────┘
+                                ▼
+┌────────────────────────────────────────────────────────────────────┐
+│  PHASE 9a — Inversive-Distance (3–5 days)                          │
+│  ──────────────────────────────────                                 │
+│  Built directly against the new traits API.  This is the           │
+│  acceptance test for the MVP.                                       │
+│                                                                     │
+│  If painless  → MVP design is sound, continue with Phase 9b/9c     │
+│  If painful   → iterate the traits design before going further     │
+└────────────────────────────────────────────────────────────────────┘
+                                ▼
+┌────────────────────────────────────────────────────────────────────┐
+│  PHASE 9b — Analytic HyperIdeal Hessian (1 week)                   │
+│  PHASE 9c — 4g-polygon fundamental domain (1 week)                  │
+│  ────────────────────────────────────────                           │
+│  Port really finished. v0.9.0 release possible.                     │
+└────────────────────────────────────────────────────────────────────┘
+                                ▼
+┌────────────────────────────────────────────────────────────────────┐
+│  PHASE 8 EXTENSIONS — only on demand                                │
+│  ─────────────────────────────────                                   │
+│  • 8a.2 generic FaceGraph specialisation     when Polyhedron_3 user │
+│  • 8b extend to spherical + hyperbolic       when 9a pattern proven │
+│  • 8c full User_manual + Reference_manual    when submission planned│
+│  • 8d CGAL-format test directory             when submission planned│
+│  • 8e YAML pipeline + CLI flag               orthogonal, any time   │
+│                                                                     │
+│  Each extension only when there is a concrete trigger.  No          │
+│  speculative architecture for a hypothetical CGAL submission.       │
+└────────────────────────────────────────────────────────────────────┘
+```
+
+**Why this order?**
+
+- Port-completion (Phase 9) is the higher-confidence value: well-defined,
+  ~3 weeks of work, finishes Goal A.
+- Full Phase 8 (3–4 weeks) speculative — only pays off if CGAL submission
+  actually happens, which is uncertain.
+- Phase 8 MVP captures the architectural insight (traits + named params)
+  without the long tail.  If the rest of Phase 8 is ever wanted, it's
+  additive — nothing built in the MVP needs to be thrown away.
+
+**Total committed budget: 2 weeks (MVP + 9a) + 2 weeks (9b + 9c) = ~4 weeks
+net work, 6–8 weeks calendar.**  After that, the port is finished.
+
+## Phase 8 MVP scope — what is and isn't in the first cut
+
+| Item | MVP | Later | Reason |
+|---|:---:|:---:|---|
+| `Conformal_map_traits.h` concept | ✅ | — | Core abstraction |
+| `Default_conformal_map_traits<Surface_mesh, K>` | ✅ | — | Status-quo wrapper |
+| Generic `FaceGraph` specialisation | — | ✅ 8a.2 | Speculative until asked |
+| `Discrete_conformal_map.h` — `_euclidean()` | ✅ | — | First entry function |
+| `Discrete_conformal_map.h` — `_spherical()`, `_hyperbolic()` | — | ✅ 8b.2 | Pattern-replicates once 9a works |
+| Named parameters: `vertex_curvature_map`, `max_iterations`, `gradient_tolerance`, `fixed_vertex_pmap` | ✅ | — | Essential 4 |
+| Named parameters: rest (`output_uv_map`, `cut_graph`, …) | — | ✅ 8b.2 | Additive |
+| Doxygen on MVP headers | ✅ | — | Same time anyway |
+| Doxygen on legacy `code/include/*` | partial | ✅ 8c | Bulk later |
+| `PackageDescription.txt` | — | ✅ 8c | Only if submitting |
+| User_manual.md | — | ✅ 8c | Only if submitting |
+| `test/Conformal_map/` CGAL-style | — | ✅ 8d | Only if submitting |
+| YAML pipeline + CLI flag | — | ✅ 8e | Orthogonal, any time |
+
+---
+
+## TODO checklist
+
+### MVP track (committed work, ~4 weeks)
+- [x] Phase 7.5: Doxyfile + CMake doc target + duplicate cleanup
+- [ ] **Phase 8 MVP — Traits + one wrapper**
+  - [ ] `Conformal_map_traits.h` — concept documentation
+  - [ ] `Default_conformal_map_traits<Surface_mesh, K>`
+  - [ ] `Conformal_map_concept_checks.h`
+  - [ ] `Discrete_conformal_map.h` with `_euclidean()` only
+  - [ ] 4 named parameters: Θ-map, max_iter, tol, pin
+  - [ ] Test: `cgal.ConformalTraits.Surface_mesh_default_works`
+- [ ] **Phase 9a — Inversive-Distance (acceptance test for MVP)**
+  - [ ] `inversive_distance_functional.hpp` against new traits
+  - [ ] Gradient check + Newton convergence tests
+  - [ ] Doxygen on new headers
+- [ ] **Phase 9b — Analytic HyperIdeal Hessian**
+  - [ ] Replace FD in `hyper_ideal_hessian.hpp`
+  - [ ] Symmetry + PSD checks unchanged
+- [ ] **Phase 9c — 4g-polygon for genus g > 1**
+  - [ ] Extend `compute_fundamental_domain()` beyond genus 1
+
+### On-demand track (only with concrete trigger)
+- [ ] 8a.2: Generic `FaceGraph` specialisation (trigger: Polyhedron_3 user)
+- [ ] 8b.2: `_spherical()` + `_hyperbolic()` entry functions (trigger: pattern proven)
+- [ ] 8b.2: `Conformal_layout.h`, `Conformal_cut_graph.h` wrappers
+- [ ] 8c: `doc/Conformal_map/PackageDescription.txt` (trigger: submission planned)
+- [ ] 8c: User_manual + Reference_manual (trigger: submission planned)
+- [ ] 8d: `test/Conformal_map/` CGAL-style tests (trigger: submission planned)
+- [ ] 8e: YAML validator + CLI `--pipeline` flag (orthogonal, any time)

doc/api/contracts.md

@@ -0,0 +1,69 @@
+# Processing Unit Contracts
+
+Each pipeline unit has explicit preconditions and guarantees.
+A pipeline is valid if every unit's preconditions are satisfied by the outputs
+of all preceding units.
+
+---
+
+## Contract table
+
+| Unit | Preconditions | Provides |
+|---|---|---|
+| `load_mesh()` | Valid file path, supported format (OFF/OBJ/PLY) | Manifold, oriented, triangulated `ConformalMesh` |
+| `setup_*_maps()` | Triangulated mesh | Initialised property maps; `lambda0` zeroed |
+| `compute_*_lambda0_from_mesh()` | `setup_*_maps()` called | `lambda0[e]` set from 3-D edge lengths |
+| `check_gauss_bonnet()` | `theta_v[v]` set for all vertices | Throws `std::runtime_error` if Σ(2π−Θᵥ) ≠ 2π·χ(M) |
+| `enforce_gauss_bonnet()` | `theta_v[v]` set | Σ(2π−Θᵥ) = 2π·χ(M) guaranteed; `theta_v` modified |
+| `newton_euclidean()` | GB satisfied · DOFs assigned · `lambda0` initialised | `NewtonResult.x` — converged scale factors; `.converged`, `.iterations`, `.grad_inf_norm` |
+| `newton_spherical()` | GB satisfied · DOFs assigned · gauge vertex pinned | Same as above |
+| `newton_hyper_ideal()` | `assign_all_dof_indices()` called · `lambda0` initialised | Same as above |
+| `compute_cut_graph()` | Closed, orientable, triangulated mesh | `CutGraph.cut_edge_flags` — 2g seam edges; `.genus` |
+| `euclidean_layout()` | `newton_euclidean()` converged | `Layout2D.uv[v]` · `.halfedge_uv[h]` · `HolonomyData.translations` |
+| `spherical_layout()` | `newton_spherical()` converged | `Layout3D.xyz[v]` |
+| `hyper_ideal_layout()` | `newton_hyper_ideal()` converged | `Layout2D.uv[v]` · `.halfedge_uv[h]` · `HolonomyData.mobius_maps` |
+| `normalise_euclidean()` | `layout.success == true` | Centroid at origin, major axis = x-axis |
+| `normalise_hyperbolic()` | `layout.success == true` | Weighted centroid at Poincaré disk origin |
+| `normalise_spherical()` | `layout.success == true` | Area centroid at north pole |
+| `compute_period_matrix()` | `hol.translations.size() >= 2` | `PeriodData.tau ∈ ℍ`; optionally SL(2,ℤ)-reduced |
+| `compute_fundamental_domain()` | `HolonomyData` (genus 1) | CCW parallelogram `{0, ω₁, ω₁+ω₂, ω₂}` + edge identifications |
+| `tiling_neighbourhood()` | `Layout2D` + `HolonomyData` (genus 1) | `(2m+1)·(2n+1)` translated layout copies |
+| `save_layout_off()` | `Layout2D` + mesh | OFF file with UV coordinates |
+| `save_result_json/xml()` | `NewtonResult` + optional `Layout2D` | JSON/XML serialised result |
+
+---
+
+## DOF assignment conventions
+
+```cpp
+// Euclidean / Spherical: pin one vertex, assign sequential indices
+maps.v_idx[first_vertex] = -1;          // pinned: u_v = 0
+int idx = 0;
+for (auto v : remaining_vertices)
+    maps.v_idx[v] = idx++;
+
+// HyperIdeal: all vertices and edges are free DOFs
+int n_dofs = assign_all_dof_indices(mesh, maps);
+// maps.v_idx[v] ∈ [0, n_v)
+// maps.e_idx[e] ∈ [n_v, n_v + n_e)
+```
+
+`-1` in `v_idx` or `e_idx` means the DOF is pinned (fixed at its `lambda0` value).
+The solver never writes to pinned DOFs. All indexing is 0-based and contiguous.
+
+---
+
+## Gauss–Bonnet — the most common source of failure
+
+Prescribing angles that violate Gauss–Bonnet means no conformal factor can realise the
+target metric — the Newton solver will iterate indefinitely without converging.
+
+```cpp
+// Option A: verify before solving (throws on violation)
+check_gauss_bonnet(mesh, maps);
+
+// Option B: auto-correct (redistributes defect uniformly across all vertices)
+enforce_gauss_bonnet(mesh, maps);
+```
+
+The target violation is `|Σ(2π−Θᵥ) − 2π·χ(M)| > tol`. Default tolerance: `1e-10`.

doc/api/extending.md

@@ -0,0 +1,141 @@
+# Extending conformallab++
+
+## Adding a new functional
+
+All three existing functionals (`euclidean_functional.hpp`, `spherical_functional.hpp`,
+`hyper_ideal_functional.hpp`) follow the same pattern. Copy the structure of the
+simplest one (`euclidean_functional.hpp`) as a template.
+
+### Step 1 — Maps struct
+
+```cpp
+// my_functional.hpp
+struct MyMaps {
+    ConformalMesh::Property_map<Vertex_index, double> lambda0;  // initial log-lengths
+    ConformalMesh::Property_map<Vertex_index, double> theta_v;  // target angles
+    ConformalMesh::Property_map<Vertex_index, int>    v_idx;    // DOF index, -1 = pinned
+    // add edge DOFs if needed:
+    ConformalMesh::Property_map<Edge_index,   int>    e_idx;
+};
+
+MyMaps setup_my_maps(ConformalMesh& mesh);
+void compute_my_lambda0_from_mesh(ConformalMesh& mesh, MyMaps& maps);
+```
+
+### Step 2 — Gradient
+
+The gradient must satisfy: `G_v = Σ(angle contributions at v) − theta_v[v]`.
+
+```cpp
+std::vector<double> my_gradient(
+    const ConformalMesh& mesh,
+    const std::vector<double>& x,
+    const MyMaps& maps)
+{
+    std::vector<double> G(n_dofs, 0.0);
+    for (auto f : mesh.faces()) {
+        // compute angles in face f given current x
+        // accumulate into G[maps.v_idx[v]] for each vertex v of f
+    }
+    // subtract target angles
+    for (auto v : mesh.vertices())
+        if (maps.v_idx[v] >= 0)
+            G[maps.v_idx[v]] -= maps.theta_v[v];
+    return G;
+}
+```
+
+### Step 3 — Gradient check test
+
+Before claiming correctness, verify with finite differences. Copy any `GradientCheck_*`
+test suite from `tests/cgal/test_*_functional.cpp`:
+
+```cpp
+TEST(MyFunctional, GradientCheck_Triangle) {
+    ConformalMesh mesh = make_single_triangle();
+    MyMaps maps = setup_my_maps(mesh);
+    // ... assign DOFs, set natural theta ...
+
+    const double eps = 1e-5;
+    auto G = my_gradient(mesh, x0, maps);
+    for (int i = 0; i < n; ++i) {
+        std::vector<double> xp = x0, xm = x0;
+        xp[i] += eps;  xm[i] -= eps;
+        double fd = (my_energy(mesh, xp, maps) - my_energy(mesh, xm, maps)) / (2*eps);
+        EXPECT_NEAR(G[i], fd, 1e-7) << "DOF " << i;
+    }
+}
+```
+
+### Step 4 — Hessian and Newton wrapper
+
+```cpp
+// Reuse solve_linear_system from newton_solver.hpp
+Eigen::SparseMatrix<double> H = my_hessian(mesh, x, maps);
+bool used_fallback;
+auto dx = conformallab::solve_linear_system(H, -G, &used_fallback);
+```
+
+Or write a thin `newton_my()` wrapper following the structure of `newton_euclidean()`.
+
+---
+
+## Adding a new geometry mode
+
+To add a new space (e.g. de Sitter, flat 3-torus):
+
+1. **Trilateration function** — implement `trilaterate_my(p1, p2, l12, l13, l23)` that
+   places a third point given two placed points and three edge lengths.
+   This is the only geometry-specific part of the layout.
+
+2. **BFS structure** — reuse the priority-BFS loop from `euclidean_layout()` in `layout.hpp`.
+   The loop itself is geometry-agnostic; swap in your trilateration function.
+
+3. **Holonomy** — if the space has a holonomy group, record the transition maps at seam edges
+   the same way `HolonomyData` does for translations (Euclidean) or Möbius maps (hyperbolic).
+
+---
+
+## Adding a new processing unit
+
+1. Declare preconditions and outputs explicitly (see [contracts.md](contracts.md)).
+
+2. Write a header-only implementation in `code/include/my_unit.hpp`.
+
+3. Add tests in `code/tests/cgal/test_my_unit.cpp` and register in
+   `code/tests/cgal/CMakeLists.txt`.
+
+4. Pipeline validation is currently manual (documented contracts). A compile-time or
+   runtime check via the declarative YAML pipeline (Phase 8e) is the planned upgrade —
+   see [cgal-package.md](cgal-package.md).
+
+---
+
+## Porting from Java
+
+When porting a class from the Java library:
+
+1. Locate the original at [github.com/varylab/conformallab](https://github.com/varylab/conformallab)
+   under `src/main/java/de/varylab/discreteconformal/`.
+
+2. The Java library uses `CoHDS` (half-edge data structure) with intrusive vertex/edge data.
+   Map these to CGAL property maps:
+   ```
+   Java: vertex.getLambda()     → C++: maps.lambda0[v]
+   Java: edge.getAlpha()        → C++: maps.e_alpha[e]
+   Java: vertex.getTheta()      → C++: maps.theta_v[v]
+   Java: vertex.getSolverIndex() → C++: maps.v_idx[v]
+   ```
+
+3. Java uses `HalfedgeInterface` adapters with named accessors like `getOppositeVertex()`.
+   C++ equivalent:
+   ```cpp
+   // Java: h.getOppositeVertex()
+   Vertex_index v_opp = mesh.target(mesh.next(h));
+
+   // Java: h.getNextHalfedge().getOppositeVertex()
+   Vertex_index v_opp2 = mesh.target(mesh.next(mesh.next(h)));
+   ```
+
+4. Port the test cases from the Java `@Test` methods. The "natural theta" trick
+   (`theta_v[v] = actual_angle_sum`) works the same way in both languages.

doc/api/headers.md

@@ -0,0 +1,74 @@
+# Public Headers (`code/include/`)
+
+All algorithms are header-only. Include the headers you need directly —
+there is no compiled library to link against (only GTest for tests and
+CGAL/Eigen for the CGAL-dependent headers).
+
+## Core mesh type
+
+| Header | Description |
+|---|---|
+| `conformal_mesh.hpp` | `ConformalMesh` = `CGAL::Surface_mesh<Point3>`. Property-map naming convention. Index type aliases. `GeometryType` enum. |
+| `constants.hpp` | `conformallab::PI`, `TWO_PI` |
+| `mesh_builder.hpp` | `make_triangle()` / `make_tetrahedron()` / `make_quad_strip()` / `make_fan()` / `make_open_cylinder()` — test mesh factories |
+| `mesh_io.hpp` | `load_mesh()` / `save_mesh()` via `CGAL::IO` (OFF / OBJ / PLY) |
+| `mesh_utils.hpp` | `cgal_to_eigen()` — convert `ConformalMesh` vertex positions to `Eigen::MatrixXd` |
+
+## Special functions
+
+| Header | Description |
+|---|---|
+| `clausen.hpp` | `clausen_cl2(θ)` (Clausen Cl₂), `lobachevsky(θ)` (Л), `im_li2(θ)` (ImLi₂ = imaginary part of dilogarithm) |
+
+## HyperIdeal geometry (H²)
+
+| Header | Description |
+|---|---|
+| `hyper_ideal_geometry.hpp` | `ζ₁₃`, `ζ₁₄`, `ζ₁₅` (Springborn 2020), `l_from_zeta()`, `alpha_ij()`, `beta_i()`, `sigma_i()`, `sigma_ij()` |
+| `hyper_ideal_utility.hpp` | Tetrahedron volumes (Meyerhoff formula, Kolpakov–Mednykh) |
+| `hyper_ideal_visualization_utility.hpp` | Poincaré disk projection, circumcircle helpers, Lorentz boost |
+| `hyper_ideal_functional.hpp` | `HyperIdealMaps`, `setup_hyper_ideal_maps()`, `compute_hyper_ideal_lambda0_from_mesh()`, `assign_all_dof_indices()`, `hyper_ideal_gradient()`, `hyper_ideal_energy()` |
+| `hyper_ideal_hessian.hpp` | `hyper_ideal_hessian()` — symmetric FD Hessian (Phase 9b: analytic) |
+
+## Spherical geometry (S²)
+
+| Header | Description |
+|---|---|
+| `spherical_geometry.hpp` | Spherical arc-length, half-angle formula, spherical law of cosines |
+| `spherical_functional.hpp` | `SphericalMaps`, `setup_spherical_maps()`, `compute_spherical_lambda0_from_mesh()`, `spherical_gradient()`, `spherical_energy()` |
+| `spherical_hessian.hpp` | `spherical_hessian()` — analytic Hessian via ∂α/∂u from law of cosines |
+
+## Euclidean geometry (ℝ²)
+
+| Header | Description |
+|---|---|
+| `euclidean_geometry.hpp` | Euclidean corner angle (t-value / atan2), edge lengths from DOF vector |
+| `euclidean_functional.hpp` | `EuclideanMaps`, `setup_euclidean_maps()`, `compute_euclidean_lambda0_from_mesh()`, `euclidean_gradient()`, `euclidean_energy()` |
+| `euclidean_hessian.hpp` | `euclidean_hessian()` — cotangent Laplacian (Pinkall–Polthier 1993) |
+
+## Solver
+
+| Header | Description |
+|---|---|
+| `newton_solver.hpp` | `newton_euclidean()`, `newton_spherical()`, `newton_hyper_ideal()`, `solve_linear_system()` (SimplicialLDLT + SparseQR fallback), `NewtonResult` struct |
+
+## Preprocessing
+
+| Header | Description |
+|---|---|
+| `gauss_bonnet.hpp` | `euler_characteristic()`, `genus()`, `gauss_bonnet_sum()`, `gauss_bonnet_rhs()`, `gauss_bonnet_deficit()`, `check_gauss_bonnet()`, `enforce_gauss_bonnet()` |
+
+## Layout and holonomy
+
+| Header | Description |
+|---|---|
+| `cut_graph.hpp` | `CutGraph` struct, `compute_cut_graph()` — tree-cotree algorithm (Erickson–Whittlesey 2005), produces 2g seam edges |
+| `layout.hpp` | `euclidean_layout()`, `spherical_layout()`, `hyper_ideal_layout()`, `normalise_{euclidean,hyperbolic,spherical}()`. Structs: `Layout2D`, `Layout3D`, `HolonomyData`. `MobiusMap` (T(z)=(az+b)/(cz+d), `from_three`, `compose`, `inverse`, `apply`). Priority-BFS, `halfedge_uv`. |
+
+## Post-processing
+
+| Header | Description |
+|---|---|
+| `period_matrix.hpp` | `PeriodData`, `compute_period_matrix()`, `reduce_to_fundamental_domain()`, `is_in_fundamental_domain()` — period ratio τ = ω₂/ω₁ ∈ ℍ, SL(2,ℤ) reduction |
+| `fundamental_domain.hpp` | `FundamentalDomain`, `compute_fundamental_domain()` (genus 1: CCW parallelogram; genus > 1: empty, TODO Phase 9c), `tiling_copy()`, `tiling_neighbourhood()` |
+| `serialization.hpp` | `save_result_json()`, `load_result_json()`, `save_result_xml()`, `load_result_xml()`, `save_layout_off()` |

doc/api/pipeline.md

@@ -0,0 +1,214 @@
+# Pipeline API Reference
+
+The full pipeline from mesh loading to layout and serialisation.
+See [doc/architecture/overall_pipeline.md](../architecture/overall_pipeline.md) for the
+Mermaid diagram and stage descriptions.
+
+---
+
+## Minimal Euclidean pipeline
+
+```cpp
+#include "conformal_mesh.hpp"
+#include "mesh_io.hpp"
+#include "euclidean_functional.hpp"
+#include "gauss_bonnet.hpp"
+#include "newton_solver.hpp"
+#include "layout.hpp"
+
+using namespace conformallab;
+
+ConformalMesh mesh = load_mesh("input.off");
+
+// 1. Set up property maps and initialise log-edge-lengths from 3-D positions
+EuclideanMaps maps = setup_euclidean_maps(mesh);
+compute_euclidean_lambda0_from_mesh(mesh, maps);
+
+// 2. Assign DOF indices — pin first vertex (gauge fix for open meshes)
+auto vit = mesh.vertices().begin();
+maps.v_idx[*vit++] = -1;           // pinned: u_v = 0
+int idx = 0;
+for (; vit != mesh.vertices().end(); ++vit)
+    maps.v_idx[*vit] = idx++;
+
+// 3. Set target angles — "natural equilibrium" makes x* = 0
+std::vector<double> x0(idx, 0.0);
+auto G0 = euclidean_gradient(mesh, x0, maps);
+for (auto v : mesh.vertices())
+    if (maps.v_idx[v] >= 0)
+        maps.theta_v[v] -= G0[maps.v_idx[v]];
+
+// 4. Check Gauss–Bonnet (mandatory before solving)
+check_gauss_bonnet(mesh, maps);     // throws if Σ(2π−Θᵥ) ≠ 2π·χ(M)
+
+// 5. Solve
+NewtonResult res = newton_euclidean(mesh, x0, maps);
+// res.converged, res.x, res.iterations, res.grad_inf_norm
+
+// 6. Layout
+Layout2D layout = euclidean_layout(mesh, res.x, maps);
+// layout.uv[v.idx()], layout.halfedge_uv[h.idx()]
+```
+
+---
+
+## Layout with holonomy (closed surfaces)
+
+```cpp
+#include "cut_graph.hpp"
+#include "period_matrix.hpp"
+#include "fundamental_domain.hpp"
+
+// Tree-cotree cut graph — 2g seam edges
+CutGraph cg = compute_cut_graph(mesh);
+// cg.cut_edge_flags[e.idx()] — true if seam edge
+// cg.genus                   — topological genus g
+
+// Layout with holonomy tracking
+HolonomyData hol;
+Layout2D layout = euclidean_layout(mesh, res.x, maps, &cg, &hol, /*normalise=*/true);
+// hol.translations[i] — lattice generator ωᵢ ∈ ℂ (Euclidean)
+
+// Period matrix (genus 1 flat torus)
+PeriodData pd = compute_period_matrix(hol);
+// pd.tau               — complex period ratio τ = ω₂/ω₁ ∈ ℍ
+// pd.omega             — {ω₁, ω₂} as complex<double>
+// pd.in_fundamental_domain — true if SL(2,ℤ)-reduced to |τ|≥1, |Re(τ)|<½
+
+// Fundamental domain parallelogram
+FundamentalDomain fd = compute_fundamental_domain(hol);
+// fd.vertices[0..3]    — CCW corners: {0, ω₁, ω₁+ω₂, ω₂}
+// fd.generators[0..1]  — ω₁, ω₂
+
+// Tiling copies for universal cover visualisation
+auto tiles = tiling_neighbourhood(layout, hol, /*m_max=*/2, /*n_max=*/2);
+// returns (2·m_max+1)·(2·n_max+1) translated layout copies
+```
+
+---
+
+## HyperIdeal pipeline
+
+```cpp
+#include "hyper_ideal_functional.hpp"
+#include "newton_solver.hpp"
+#include "layout.hpp"
+#include "cut_graph.hpp"
+
+HyperIdealMaps maps = setup_hyper_ideal_maps(mesh);
+compute_hyper_ideal_lambda0_from_mesh(mesh, maps);
+
+// HyperIdeal has both vertex and edge DOFs — assign all automatically
+int n_dofs = assign_all_dof_indices(mesh, maps);
+// No vertex needs to be pinned — strictly convex functional
+
+std::vector<double> x0(n_dofs, 0.0);
+NewtonResult res = newton_hyper_ideal(mesh, x0, maps);
+
+// Layout with Möbius holonomy
+CutGraph cg = compute_cut_graph(mesh);
+HolonomyData hol;
+Layout2D layout = hyper_ideal_layout(mesh, res.x, maps, &cg, &hol);
+// hol.mobius_maps[i] — Möbius isometry Tᵢ ∈ SU(1,1) per cut edge
+```
+
+---
+
+## Spherical pipeline
+
+```cpp
+#include "spherical_functional.hpp"
+#include "newton_solver.hpp"
+#include "layout.hpp"
+
+SphericalMaps maps = setup_spherical_maps(mesh);
+compute_spherical_lambda0_from_mesh(mesh, maps);
+
+// Pin one vertex (gauge fix) + assign DOFs
+auto vit = mesh.vertices().begin();
+maps.v_idx[*vit++] = -1;
+int idx = 0;
+for (; vit != mesh.vertices().end(); ++vit)
+    maps.v_idx[*vit] = idx++;
+
+std::vector<double> x0(idx, 0.0);
+NewtonResult res = newton_spherical(mesh, x0, maps);
+
+// 3-D layout on the unit sphere
+Layout3D slayout = spherical_layout(mesh, res.x, maps);
+// slayout.xyz[v.idx()] — 3-D position on S²
+```
+
+---
+
+## SparseQR fallback — direct use
+
+```cpp
+#include "newton_solver.hpp"
+
+bool used_fallback = false;
+Eigen::VectorXd dx = conformallab::solve_linear_system(H, rhs, &used_fallback);
+if (used_fallback)
+    std::cerr << "Warning: H is rank-deficient, used SparseQR\n";
+```
+
+The fallback is automatically invoked by all three `newton_*` functions when
+`SimplicialLDLT` fails. It finds the minimum-norm Newton step orthogonal to the null space.
+
+---
+
+## Serialisation
+
+```cpp
+#include "serialization.hpp"
+#include "mesh_io.hpp"
+
+// Save layout as OFF mesh file
+save_layout_off("layout.off", mesh, layout);
+
+// Full result: DOF vector + metadata + layout UVs
+save_result_json("result.json", res, "euclidean", mesh, &layout);
+save_result_xml ("result.xml",  res, "euclidean", mesh, &layout);
+
+// Round-trip load
+NewtonResult res2;
+std::string geom;
+Layout2D uv2;
+load_result_json("result.json", &res2, &geom, &uv2);
+```
+
+---
+
+## Attaching custom property maps
+
+```cpp
+// Add a custom per-vertex map
+auto [curv, created] = mesh.add_property_map<Vertex_index, double>("v:my_curv", 0.0);
+curv[v] = 3.14;
+
+// Retrieve an existing map
+auto [curv2, found] = mesh.property_map<Vertex_index, double>("v:my_curv");
+```
+
+---
+
+## Halfedge traversal conventions
+
+```cpp
+for (auto f : mesh.faces()) {
+    auto h0 = mesh.halfedge(f);   // canonical halfedge of face f
+    auto h1 = mesh.next(h0);
+    auto h2 = mesh.next(h1);
+
+    Vertex_index v1 = mesh.source(h0);  // = mesh.target(h2)
+    Vertex_index v2 = mesh.source(h1);
+    Vertex_index v3 = mesh.source(h2);
+
+    // Angle at v3 is opposite to halfedge h0 (edge v1–v2)
+    // h_alpha[h0] = α₃,  h_alpha[h1] = α₁,  h_alpha[h2] = α₂
+
+    Edge_index e0 = mesh.edge(h0);
+    bool is_seam = cg.cut_edge_flags[e0.idx()];
+    bool is_boundary = mesh.is_border(mesh.opposite(h0));
+}
+```

doc/api/tests.md

@@ -0,0 +1,80 @@
+# Test Suites
+
+## `conformallab_tests` — always built (no CGAL)
+
+Pure-math tests, only Eigen required. Covers Java utilities ported in Phase 1–2.
+
+| File | What it tests |
+|---|---|
+| `test_clausen.cpp` | Clausen Cl₂, Lobachevsky Л, ImLi₂ — values at known points |
+| `test_hyper_ideal_utility.cpp` | Tetrahedron volumes (Meyerhoff / Kolpakov–Mednykh) |
+| `test_matrix_utility.cpp` | Matrix helpers |
+| `test_surface_curve_utility.cpp` | Surface curve utilities |
+| `test_discrete_elliptic_utility.cpp` | Discrete elliptic functions |
+| `test_p2_utility.cpp` | P2 projective utilities |
+| `test_hyper_ideal_visualization_utility.cpp` | Poincaré disk projection, circumcircle |
+
+**Total: 23 tests, 0 skipped.**
+
+---
+
+## `conformallab_cgal_tests` — built with `-DWITH_CGAL_TESTS=ON` or `-DWITH_CGAL=ON`
+
+All tests have CTest prefix `cgal.` (set via `TEST_PREFIX "cgal."` in CMakeLists).
+
+| Suite | File | Tests | What is verified |
+|---|---|---:|---|
+| `ConformalMeshTopology` | `test_conformal_mesh.cpp` | 4 | Euler characteristic, vertex/edge/face counts |
+| `ConformalMeshTraversal` | `test_conformal_mesh.cpp` | 4 | Halfedge iteration, valence, opposite |
+| `ConformalMeshProperties` | `test_conformal_mesh.cpp` | 5 | Property maps (λ, θ, idx, α, geometry type) |
+| `ConformalMeshValidity` | `test_conformal_mesh.cpp` | 1 | CGAL validity for all factory meshes |
+| `HyperIdealFunctional` | `test_hyper_ideal_functional.cpp` | 7 | FD gradient checks + Hessian symmetry |
+| `SphericalFunctional` | `test_spherical_functional.cpp` | 12 | Angle formula + gradient + gauge-fix + cross-module Hessian check |
+| `EuclideanFunctional` | `test_euclidean_functional.cpp` | 12 | Angle formula + gradient + cross-module Hessian check |
+| `EuclideanHessian` | `test_euclidean_hessian.cpp` | 9 | Cotangent Laplacian structure, FD agreement, PSD, null space |
+| `SphericalHessian` | `test_spherical_hessian.cpp` | 8 | Derivative correctness, NSD at equilibrium |
+| `NewtonSolver` | `test_newton_solver.cpp` | 11 | Convergence: Euclidean ×3, Spherical ×4, HyperIdeal ×4 |
+| `SparseQRFallback` | `test_newton_solver.cpp` | 3 | Full-rank LDLT · singular matrix → QR · closed mesh gauge mode |
+| `MeshIO` | `test_mesh_io.cpp` | 6 | OFF/OBJ round-trips, error handling |
+| `Pipeline` | `test_pipeline.cpp` | 5 | End-to-end: build → setup → solve → export → reload |
+| `Layout` | `test_layout.cpp` | 6 | Edge-length preservation (Eucl./Spher.), Poincaré disk layout |
+| `Serialization` | `test_layout.cpp` | 2 | JSON and XML round-trips (DOF vector + layout UVs) |
+| `GaussBonnet` | `test_phase6.cpp` | 12 | χ, genus, sum/RHS, deficit, check, enforce |
+| `CutGraph` | `test_phase6.cpp` | 6 | Tree-cotree, open/closed meshes, flag–index consistency |
+| `HyperbolicTrilateration` | `test_phase6.cpp` | 4 | Möbius + law of cosines: exact distances, disk interior, off-origin |
+| `Normalisation` | `test_phase6.cpp` | 4 | Euclidean centroid, length ratios, Möbius centring |
+| `MobiusMap` | `test_phase7.cpp` | 8 | Identity, inverse, compose, `from_three`, `apply(Vector2d)` |
+| `BestRootFace` | `test_phase7.cpp` | 2 | Valid root face selection, interior bonus |
+| `HalfedgeUV` | `test_phase7.cpp` | 4 | Size = #halfedges, seam consistency, boundary halfedges = 0 |
+| `PriorityBFS` | `test_phase7.cpp` | 3 | Success, no seam on open meshes, all vertices placed |
+| `NormaliseEuclidean` | `test_phase7.cpp` | 2 | UV centroid = 0, halfedge_uv centroid = 0 |
+| `PeriodMatrix` | `test_phase7.cpp` | 7 | τ ∈ ℍ, SL(2,ℤ) reduction, exception outside ℍ |
+| `FundamentalDomain` | `test_phase7.cpp` | 7 | Genus-1 parallelogram CCW, generators, g > 1 empty |
+| `TilingCopy/Neighbourhood` | `test_phase7.cpp` | 4 | Translation correct, tile count |
+| `CuttingUtility` | `test_geometry_utils.cpp` | 3 | `point_in_triangle_2d`: false, true, unit triangle (Java CuttingUtilityTest) |
+| `UnwrapUtility` | `test_geometry_utils.cpp` | 2 | Corner angle: collinear → π, equilateral → π/3 (Java UnwrapUtilityTest) |
+| `ConvergenceUtility` | `test_geometry_utils.cpp` | 6 | Circumradius + scale-invariant R_f/√A (Java ConvergenceUtilityTests) |
+| `EuclideanLayout` | `test_geometry_utils.cpp` | 2 | Euclidean layout round-trip edge lengths |
+| `SphericalLayout` | `test_geometry_utils.cpp` | 1 | Spherical layout on unit sphere |
+| `HomologyGenerators` | `test_geometry_utils.cpp` | 1 | Genus-2 cut graph: χ = −2, 4 cut edges (`brezel2.obj`) |
+| `SmokeEuclidean` | `test_scalability_smoke.cpp` | 3 | Smoke tests on real meshes: CatHead (open), Brezel genus-1, Brezel2 genus-2 |
+
+**Total: 227 tests, 0 skipped.**
+
+---
+
+## Running tests
+
+```bash
+# All CGAL tests
+ctest --test-dir build -R "^cgal\." --output-on-failure
+
+# One suite
+./build/conformallab_cgal_tests --gtest_filter="PeriodMatrix*"
+
+# One specific test
+./build/conformallab_cgal_tests --gtest_filter="PeriodMatrix.TauInUpperHalfPlane"
+
+# Verbose output with timing
+./build/conformallab_cgal_tests --gtest_filter="NewtonSolver*" --gtest_print_time=1
+```

doc/architecture/design-decisions.md

@@ -0,0 +1,124 @@
+# Key Design Decisions
+
+Rationale for the architectural choices that distinguish conformallab++ from the
+Java original and from generic geometry-processing frameworks.
+
+---
+
+## CGAL `Surface_mesh` as the halfedge data structure
+
+The Java library uses `CoHDS` — a custom intrusive halfedge data structure with
+`CoVertex`, `CoEdge`, `CoFace` types that carry domain-specific data directly as fields.
+
+conformallab++ replaces this with `CGAL::Surface_mesh<Point3>` and attaches data via
+**named property maps**:
+
+```cpp
+// Java: vertex.getLambda()           → C++: maps.lambda0[v]
+// Java: edge.getAlpha()              → C++: maps.e_alpha[e]
+// Java: vertex.getSolverIndex()      → C++: maps.v_idx[v]
+
+auto [lambda0, ok] = mesh.add_property_map<Edge_index, double>("e:lambda0", 0.0);
+lambda0[e] = 1.234;
+```
+
+This decouples the mesh topology from the algorithm data, makes it straightforward
+to attach multiple independent data sets to the same mesh, and will enable the Phase 8
+traits-class design to work with any CGAL-conforming mesh type.
+
+---
+
+## DOF vector convention
+
+All three functionals use the same indexing scheme: a flat `std::vector<double> x`
+indexed by `v_idx[v]` (vertices) and `e_idx[e]` (edges, HyperIdeal only).
+Index value `-1` means "pinned" — the DOF is fixed at zero and excluded from the
+Newton system.
+
+```
+x[maps.v_idx[v]]  = uᵥ   (conformal scale factor, Euclidean/Spherical)
+x[maps.e_idx[e]]  = λₑ   (edge log-length variable, HyperIdeal only)
+-1                        pinned — u_v = 0 / λ_e = 0
+```
+
+This is consistent across all three geometry modes, enabling the same Newton solver
+and linear system infrastructure to serve all three without branching.
+
+---
+
+## Priority-BFS layout
+
+A naive BFS layout places faces in arbitrary order; trilateration errors accumulate
+along the BFS frontier. conformallab++ uses a **priority min-heap on BFS depth**:
+
+```
+depth(face) = max(depth[v_src], depth[v_tgt]) + 1   for each new face
+```
+
+Faces with smaller depth (closer to the root) are placed first. This means each
+face's trilateration uses the two most accurately-placed adjacent vertices, minimising
+error propagation across the mesh.
+
+Root face selection: largest 3-D area face, with an additional 1.5× bonus for
+interior faces over boundary faces. This heuristic places the root where metric
+distortion is lowest.
+
+---
+
+## `halfedge_uv` semantics
+
+`layout.uv[v.idx()]` gives the *primary* UV coordinate of vertex `v` — the position
+from the shallowest BFS visit. At seam edges this is insufficient for GPU rendering:
+two faces sharing a seam vertex need *different* UV values for that vertex.
+
+`layout.halfedge_uv[h.idx()]` stores the UV of `source(h)` **as seen from `face(h)`**:
+
+```
+halfedge h  →  face(h)  →  source(h) has UV = halfedge_uv[h.idx()]
+opposite(h) →  face(h') →  source(h) has UV = halfedge_uv[opposite(h).idx()]
+                            (different value at a seam)
+```
+
+At seam halfedges the two opposite halfedges carry different UV values — each face
+gets its own copy of the seam vertex. This enables a proper GPU texture atlas
+without vertex duplication in the index buffer.
+
+---
+
+## Spherical Hessian sign convention
+
+The spherical energy functional is **concave** (negative semidefinite Hessian).
+Standard Newton would require solving `H·Δx = −G` with NSD `H`, which Cholesky
+cannot handle.
+
+`newton_spherical()` solves `(−H)·Δx = G` instead — algebraically identical,
+but `−H` is PSD and `SimplicialLDLT` works correctly. This sign flip is handled
+transparently inside `newton_spherical()`; callers need not be aware of it.
+
+The gradient sign in spherical mode is also flipped vs. Euclidean:
+- Euclidean: `G_v = actual_sum − Θᵥ`
+- Spherical: `G_v = Θᵥ − actual_sum`
+
+Both conventions drive the same equilibrium condition `G = 0`.
+
+---
+
+## HyperIdeal Hessian via finite differences
+
+The analytic HyperIdeal Hessian requires differentiating through the chain
+`(bᵢ, aₑ) → lᵢⱼ → ζ₁₃/ζ₁₄/ζ₁₅ → αᵢⱼ/βᵢ` with four vertex-type combinations
+per edge — substantial implementation complexity.
+
+conformallab++ uses a **symmetric finite-difference Hessian** instead:
+
+```
+H[i,j] = (G(x + ε·eⱼ)[i] − G(x − ε·eⱼ)[i]) / (2ε),   ε = 1e-5
+```
+
+Properties:
+- O(ε²) accuracy — relative error ≈ 10⁻¹⁰ at ε = 10⁻⁵
+- PSD guaranteed by strict convexity of the HyperIdeal energy (Springborn 2020)
+- Symmetrised automatically: `H = (H + Hᵀ) / 2`
+- Cost: n extra gradient evaluations per Newton step (acceptable for < 500 DOFs)
+
+The analytic Hessian is deferred to Phase 9b. See [roadmap/java-parity.md](../roadmap/java-parity.md).

doc/architecture/geometry-central-comparison.md

@@ -0,0 +1,281 @@
+# conformallab++ vs. geometry-central — Detailed Comparison
+
+> **Purpose of this document.**  
+> conformallab++ and geometry-central (CMU, Keenan Crane's group) both implement
+> discrete conformal equivalence of triangulated surfaces.  They share the same
+> mathematical core but diverge in algorithmic strategy, scope, and target audience.
+> This document maps the overlap precisely, identifies what cannot and should not be
+> adopted, and explains where a side-by-side study creates scientific added value.
+
+---
+
+## 1 — Shared mathematical foundation
+
+Both libraries implement the following chain:
+
+```
+Input mesh M  →  edge lengths ℓᵢⱼ  →  solve for u ∈ ℝᵛ
+  such that  ℓ̃ᵢⱼ = e^{(uᵢ+uⱼ)/2} · ℓᵢⱼ  satisfies  Σα_v(u) = Θᵥ  ∀v
+```
+
+The variational framework that makes this a well-posed optimisation problem goes
+back to **Bobenko & Springborn (2004)**.  The hyperbolic (HyperIdeal) geometry is
+from **Springborn (2020)**.  The geometry-central implementation (Gillespie,
+Springborn & Crane, SIGGRAPH 2021) is an explicit extension of Springborn 2020
+to intrinsic triangulations.
+
+**Key consequence:** the *mathematical problem* is identical.  Any difference in
+output is either a normalization convention or a bug in one of the two libraries —
+making cross-validation directly meaningful.
+
+---
+
+## 2 — Algorithmic comparison
+
+| Dimension | conformallab++ | geometry-central (Gillespie 2021) |
+|---|---|---|
+| **Solver** | Newton–Raphson, analytical Hessian | Newton or Yamabe gradient flow (user choice) |
+| **Convergence** | Quadratic (8–20 iterations on typical meshes) | Newton: quadratic; Yamabe: linear (~hundreds of steps) |
+| **Triangulation** | Fixed throughout — operates on original `Surface_mesh` | Ptolemaic flips applied before/during solve to reach intrinsic Delaunay |
+| **Hessian** | SimplicialLDLT + SparseQR fallback; analytical for Euclidean/Spherical, FD for HyperIdeal | Assembled on the current (possibly flipped) triangulation |
+| **Mesh backend** | CGAL `Surface_mesh<Point_3>` | geometry-central `ManifoldSurfaceMesh` |
+| **Geometry modes** | Euclidean ✓ · Spherical ✓ · HyperIdeal ✓ | Euclidean ✓ · Hyperbolic ✓ · Spherical ✗ |
+| **Open meshes** | ✓ (boundary DOFs pinned) | ✓ |
+
+### Why Newton on a fixed triangulation works well
+
+The Euclidean and HyperIdeal energies are strictly convex after gauge-fixing.
+Newton therefore converges from u=0 in 8–20 iterations for any reasonable mesh.
+The Hessian is the cotangent Laplacian (Euclidean) or its hyperbolic analog —
+well-conditioned on Delaunay meshes, but can degrade on strongly non-Delaunay inputs.
+
+### What Ptolemaic flips add
+
+A Ptolemaic flip replaces diagonal AC with BD in a quadrilateral under the constraint
+that the Ptolemy relation
+
+```
+AC · BD = AB · CD + AD · BC
+```
+
+holds.  This is *conformal-class-preserving* — the new λ₀ values represent the same
+discrete conformal structure.  The gain: the flipped triangulation is intrinsic
+Delaunay, which bounds the off-diagonal Hessian entries and prevents ill-conditioning
+on pathological inputs.
+
+**This is the only genuine algorithmic advantage geometry-central has for the shared
+sub-problem.** It costs nothing mathematically and buys robustness on bad meshes.
+
+---
+
+## 3 — Feature matrix: what exists where
+
+| Feature | conformallab++ | geometry-central | Notes |
+|---|---|---|---|
+| Discrete conformal equivalence (Euclidean) | ✓ | ✓ | Shared core |
+| Discrete conformal equivalence (Hyperbolic/HyperIdeal) | ✓ (Springborn 2020 formulation) | ✓ (Gillespie 2021 extension) | Mathematically equivalent |
+| Discrete conformal equivalence (Spherical) | ✓ | ✗ | Unique to conformallab++ |
+| Analytical Hessian (Euclidean & Spherical) | ✓ | ✓ | |
+| Analytical Hessian (HyperIdeal) | FD (Phase 9b: analytical planned) | ✓ | gc has analytical version |
+| Gauss–Bonnet check & enforce | ✓ | implicit in solver | |
+| Tree-cotree cut graph (2g seam edges) | ✓ | ✗ | Required for period matrix |
+| Priority-BFS layout in ℝ²/S²/Poincaré disk | ✓ | partial (conformal param only) | |
+| Möbius holonomy SU(1,1) | ✓ | ✗ | Unique to conformallab++ |
+| Period matrix τ ∈ ℍ + SL(2,ℤ) reduction | ✓ | ✗ | Unique to conformallab++ |
+| Fundamental domain + tiling | ✓ | ✗ | Unique to conformallab++ |
+| Intrinsic Delaunay triangulation | ✗ | ✓ | gc has via SignpostIntrinsicTriangulation |
+| Ptolemaic flips | ✗ | ✓ | gc's robustness mechanism |
+| Heat method (geodesic distances) | ✗ | ✓ | Auxiliary tool in gc |
+| YAML/declarative pipeline | ✓ (Phase 8e spec) | ✗ | |
+| CGAL-package submission | ✓ (Phase 8 target) | ✗ | |
+| JSON/XML serialisation | ✓ | ✗ | |
+| CLI app | ✓ | ✗ | |
+| Inversive distance functional (Luo 2004) | ✗ (Phase 9a) | ✗ | Neither has it yet |
+| Siegel period matrix Ω (genus g≥2) | ✗ (Phase 10b) | ✗ | |
+
+---
+
+## 4 — What should be adopted — and what should not
+
+### Adopt: Ptolemaic pre-conditioning (GC-2, after Phase 8)
+
+**What:** a single preprocessing pass that Delaunay-izes the input triangulation
+via Ptolemaic flips, updates λ₀ accordingly, then hands off to the existing
+Newton pipeline unchanged.
+
+**Why it fits:**
+- Preserves the conformal class — mathematically sound
+- Drop-in before `compute_euclidean_lambda0_from_mesh()`, no interface change
+- Does not touch cut graph, holonomy, period matrix
+- ~200 lines of code, one new test suite
+
+**Interface sketch:**
+```cpp
+// include/preprocessing_delaunay.hpp  (Phase GC-2)
+void ptolemy_delaunay(ConformalMesh& mesh, EuclideanMaps& maps);
+// Flips edges until all faces satisfy the Delaunay condition.
+// Updates maps.lambda0[e] via the Ptolemy relation after each flip.
+// Prerequisite: compute_euclidean_lambda0_from_mesh() already called.
+// Postcondition: mesh is an intrinsic Delaunay triangulation of the same surface.
+```
+
+### Adopt partially: analytical HyperIdeal Hessian
+
+geometry-central has a closed-form Hessian for the hyperbolic energy.
+conformallab++ currently uses finite differences (Phase 9b plans analytical).
+The geometry-central implementation can serve as a reference for Phase 9b —
+not a code copy, but a mathematical cross-check.
+
+### Do not adopt: full intrinsic triangulations as architecture
+
+**SignpostIntrinsicTriangulation** is geometry-central's core data structure.
+It tracks vertex positions as (face, barycentric coordinates) rather than 3D points.
+Replacing `CGAL::Surface_mesh` with this would require:
+
+1. Rewriting the cut graph algorithm (which works on halfedges of the *original* mesh
+   and must survive across flips — non-trivial bookkeeping)
+2. Tracking seam edges through flip events for holonomy computation
+3. Abandoning the CGAL package target (Phase 8) — CGAL's mesh concepts are extrinsic
+4. Losing the 3D layout output (Poincaré disk, sphere) which clients depend on
+
+**Verdict:** the architecture incompatibility is fundamental, not incidental.
+The period matrix pipeline requires a stable topological cut that does not survive
+arbitrary flip sequences. This is not a solvable engineering problem within the
+current project scope — it would be a different project.
+
+### Do not adopt: Yamabe flow
+
+Newton converges in 8–20 iterations; Yamabe flow needs hundreds.
+The only reason to use Yamabe flow is when the Hessian is indefinite (which happens
+in the Spherical case — and conformallab++ already handles this with the correct
+sign flip in the energy). There is no mesh type where Yamabe flow beats Newton
+on metrics that conformallab++ targets.
+
+---
+
+## 5 — Where cross-comparison creates scientific added value
+
+### 5.1 — Independent cross-validation of the shared core
+
+The discrete conformal equivalence problem for Euclidean and HyperIdeal geometry
+is implemented independently in two codebases, by different groups, with different
+algorithms.  Agreement on:
+
+- the u-vector (after normalization)
+- UV coordinates (up to Möbius transformation)
+- the residual ‖G(u*)‖ at convergence
+
+would constitute **mutual validation without ground truth**.  This is the same
+methodology used in numerical PDE literature to validate independent solvers.
+
+**Concrete protocol:**
+```
+For each test mesh (cathead.obj, brezel.obj, torus_4x4.off, torus_hex_6x6.off):
+  1. Load into both libraries with identical vertex ordering
+  2. Run conformallab++ Newton solver  →  u_clab, UV_clab
+  3. Run geometry-central solver        →  u_gc,   UV_gc
+  4. Normalize both (subtract mean, divide by scale)
+  5. Report  max|u_clab[v] - u_gc[v]|  and  mean conformal distortion difference
+```
+
+Expected: agreement to ≤ 1e-8 on well-conditioned meshes.  Discrepancy would
+indicate a bug or a normalization mismatch worth investigating.
+
+### 5.2 — Convergence study: Newton with vs. without Ptolemaic pre-conditioning
+
+Hypothesis: on non-Delaunay meshes (e.g. a torus refined by subdivision without
+re-meshing), Ptolemaic Delaunay pre-conditioning reduces Newton iteration count.
+
+**Measurable quantities:**
+- Newton iterations to ‖G‖ < 1e-8
+- Hessian condition number κ(H) at u = 0
+- Wall-clock time
+
+This comparison requires only GC-2 to be implemented in conformallab++ and would
+answer the question: *how bad does a mesh have to be before Ptolemaic pre-conditioning
+pays off?*
+
+**Publication potential:** a short note or conference contribution comparing the
+two approaches on a systematic mesh quality benchmark would be self-contained and
+novel — neither library has published this comparison.
+
+### 5.3 — Period matrix and holonomy as differentiating contribution
+
+geometry-central deliberately stops at the conformal parameterization.  The period
+matrix τ and Möbius holonomy computation in conformallab++ extend the pipeline into
+Teichmüller theory territory that geometry-central does not address.
+
+This is the strongest scientific differentiator: conformallab++ can compute
+
+```
+τ = ω_b / ω_a  ∈ ℍ,  SL(2,ℤ)-reduced
+```
+
+for any closed genus-1 surface, and (in Phase 10) the Siegel matrix Ω for genus g≥2.
+No other open-source C++ library does this.  Cross-comparison with geometry-central
+makes this gap explicit and positions conformallab++ as the more complete tool for
+Teichmüller-theoretic applications.
+
+### 5.4 — Spherical geometry as unique contribution
+
+The Spherical geometry mode (angle sums → 4π on a sphere, NSD Hessian with sign
+flip) has no counterpart in geometry-central.  A mathematician interested in
+conformal maps on surfaces of positive curvature (constant curvature +1) has no
+alternative in open-source C++.
+
+### 5.5 — Validation of Springborn 2020 in two independent implementations
+
+Springborn 2020 ("Ideal Hyperbolic Polyhedra and Discrete Uniformization") is the
+shared theoretical reference for both the HyperIdeal geometry mode in conformallab++
+(Phase 2/3) and the hyperbolic component of Gillespie 2021 in geometry-central.
+Cross-checking the ζ-function values (ζ₁₃, ζ₁₄, ζ₁₅) and the resulting angle sums
+on the same meshes would validate both implementations of the paper — a service to
+the discrete geometry community.
+
+---
+
+## 6 — Demarcation: where the comparison ends
+
+| Topic | conformallab++ | geometry-central | Comparable? |
+|---|---|---|---|
+| u-vector at convergence | ✓ | ✓ | ✓ after normalization |
+| UV parameterization | ✓ | ✓ | ✓ up to Möbius |
+| Convergence speed (iterations) | ✓ | ✓ (Newton mode) | ✓ direct |
+| HyperIdeal angle sums | ✓ | ✓ | ✓ |
+| Spherical angle sums | ✓ | ✗ | ✗ |
+| Period matrix τ | ✓ | ✗ | ✗ |
+| Holonomy T_a, T_b | ✓ | ✗ | ✗ |
+| Fundamental domain | ✓ | ✗ | ✗ |
+| Intrinsic Delaunay quality | not tracked | ✓ | — |
+| Mesh topology handling | CGAL halfedge | gc manifold mesh | not comparable |
+| Scalability (large meshes) | not benchmarked yet | benchmarked in paper | comparable if same mesh |
+
+The comparison is meaningful and complete for the **shared conformal core**.
+It ends where conformallab++ continues into Teichmüller theory (holonomy, τ,
+fundamental domain) — that region has no counterpart in geometry-central and
+must be validated by analytic invariants alone (→ `doc/math/validation.md`).
+
+---
+
+## 7 — Practical roadmap for the comparison
+
+| Step | When | What | Effort |
+|---|---|---|---|
+| **GC-1a** | Now | Manual UV comparison on cathead.obj — run both, diff u-vectors | 1 day |
+| **GC-1b** | Now | Add normalization utility to conformallab++ (`normalize_u_vector()`) | 2h |
+| **GC-1c** | After Phase 8 | Automated comparison script (Python or small C++ binary) | 2 days |
+| **GC-2** | After Phase 8 | `ptolemy_delaunay()` preprocessing pass | 1 week |
+| **GC-bench** | After GC-2 | Convergence study: Newton ± Ptolemaic pre-conditioning on 10 meshes | 1 week |
+| **GC-paper** | Phase 10 | Short note on the comparison — period matrix as differentiator | — |
+
+---
+
+## 8 — References
+
+| Reference | Role in this comparison |
+|---|---|
+| **Bobenko, Springborn** — *Variational Principles for Circle Patterns*, Trans. AMS (2004) | Shared variational foundation for all three geometry modes |
+| **Springborn** — *Ideal Hyperbolic Polyhedra and Discrete Uniformization*, DCG (2020) | Mathematical basis for HyperIdeal in conformallab++ AND for Gillespie 2021 |
+| **Gillespie, Springborn, Crane** — *Discrete Conformal Equivalence of Polyhedral Surfaces*, SIGGRAPH (2021) | geometry-central implementation; introduces Ptolemaic flips |
+| **Sharp, Soliman, Crane** — *Navigating Intrinsic Triangulations*, SIGGRAPH (2019) | geometry-central `SignpostIntrinsicTriangulation` — basis for GC-2 |
+| **Sechelmann** — doctoral thesis, TU Berlin (2016) | conformallab++ primary source; covers period matrix, holonomy, all three modes |

doc/architecture/overall_pipeline.md

@@ -1,251 +1,405 @@
-# Draft High-level Architecture
+# conformallab++ — Architecture & Pipeline
 
+## Origin
 
-## Positioning and relation to existing libraries (Nice To Have but maybe to much)
+conformallab++ is a C++ reimplementation of
-ConformalLab++ is not intended to replace established geometry processing libraries such as CGAL, libigl, or Geometry Central. Instead, it acts as an experiment-friendly pipeline layer on top of existing data structures and algorithms provided by these libraries.
+[ConformalLab](https://github.com/varylab/conformallab),
+the Java research library for discrete conformal geometry by
+**Stefan Sechelmann** (TU Berlin, Institut für Mathematik,
+SFB/Transregio 109 *Discretization in Geometry and Dynamics*).
 
-The project revives the ideas and algorithms of the original ConformalLab by Stefan Sechelmann in a modern C++ setting, making them easier to combine with contemporary mesh libraries (e.g. Geometry Central for intrinsic quantities and operators) and to reuse in new experimental setups.
+The algorithmic foundation is his doctoral dissertation:
 
-In practice, ConformalLab++ focuses on:
+> Stefan Sechelmann —  
+> **Variational Methods for Discrete Surface Parameterization: Applications and Implementation**  
+> Doctoral thesis, Technische Universität Berlin, 2016.  
+> DOI: [10.14279/depositonce-5415](https://depositonce.tu-berlin.de/items/8e2988b2-d991-45b5-aad5-9fb7988f3b2f) · CC BY-SA 4.0
 
-+ a clear, declarative pipeline description for conformal and Möbius-based mesh experiments,
+The dissertation develops the variational framework for discrete conformal equivalence:
+discrete uniformization of Riemann surfaces, cone metrics, period matrices, and
+holonomy — all of which are directly implemented in this library.
 
-+ a unified halfedge-based UniversalMesh representation plus optional alternative representations,
+Further links:  
+**Java original:** [github.com/varylab/conformallab](https://github.com/varylab/conformallab) ·
+**Website:** [sechel.de](https://sechel.de/) ·
+**LinkedIn:** [linkedin.com/in/sechel](https://www.linkedin.com/in/sechel/)
 
-adapters and plugin units that wrap robust external implementations where appropriate (e.g. CGAL-based repair or reconstruction, Geometry Central–style discrete geometry operators, or libigl functionality) instead of reimplementing them.
+---
 
+## Positioning
 
+conformallab++ is a **specialised research library for discrete conformal geometry** on
+triangulated surfaces. It is not a general geometry-processing framework.
 
-## High-level geometry pipeline
+The library revives the algorithms of the original ConformalLab by Stefan Sechelmann
-The following diagram outlines the planned high-level pipeline for ConformalLab++. It shows the rough division into preprocessing, processing, and postprocessing. In preprocessing, various input types are first converted into a universal representation (half-edge mesh) via adapter layers and any necessary preprocessing steps. From there, they are processed by the various core processors units and then either exported or visualized in postprocessing, with minor optimizations as necessary.
+in a modern C++ setting. Its core concern is one precise mathematical question:
+
+> Given a triangulated surface, find a conformally equivalent metric that satisfies
+> prescribed curvature (angle-sum) constraints at each vertex.
+
+Everything in the library serves this goal:
+
+| What it is | What it is not |
+|------------|----------------|
+| Discrete conformal maps (Euclidean, spherical, hyperbolic) | General mesh processing |
+| Newton solver for angle-sum energy functionals | Remeshing / boolean / smoothing |
+| Priority-BFS layout into ℝ², S², Poincaré disk | Point-cloud or implicit-field processing |
+| Holonomy, period matrices, fundamental domains | NURBS / parametric modelling |
+| Research platform — experiment-first, CLI-ready | Production rendering engine |
+
+**Relation to existing libraries.**  
+conformallab++ sits *on top of* CGAL, not beside it.
+`CGAL::Surface_mesh` is the internal mesh representation; there is no additional
+abstraction layer. Eigen handles all linear algebra. libigl provides the optional
+interactive viewer. The library adds the conformal-geometry layer that none of these
+provide.
+
+---
+
+## The conformal geometry pipeline
+
+The full pipeline runs in three stages. All stages operate on the same
+`ConformalMesh` (= `CGAL::Surface_mesh<Point3>` with attached property maps).
 
 ```mermaid
 graph TD
-    A1["Explicit<br/>(OBJ, PLY, STL)"]
+    IN["Input\n(OFF / OBJ / PLY / builder)"]
-    A2["Implicit<br/>(SDF, Formulas)"]
+    ADAPTER["Input Adapter\nload_mesh() · make_triangle() · make_quad_strip() …"]
-    A3["Parametric<br/>(NURBS, Curves)"]
+
-    A4["Procedural<br/>(Noise, L-Sys)"]
+    subgraph PRE["① PREPROCESSING"]
-    ADAPTER1["Input Adapter <br/>(Convert to Universal Represenation)"]
+        MAPS["Maps Setup\nsetup_*_maps() + compute_lambda0_from_mesh()"]
-    PREPROCESSING["Preprocessor <br/>(refined triangulation etc.)"]
+        DOF["DOF Assignment\nv_idx[v] · e_idx[e] · pin / free"]
-    EXT["EXTERNAL LIBRARIES<br/>(Plugin System)<br/>- CGAL"]
+        THETA["Target Angles\ntheta_v[v] — cone metric or natural equilibrium"]
-    UNIVERSAL["UNIVERSAL REPRESENTATION FORM<br/>(HalfEdge Mesh Interface)"]
+        GB["Gauss–Bonnet Check\ncheck_gauss_bonnet() / enforce_gauss_bonnet()"]
-    EXPORT_ADAPTER["Export Adapter Layer"]
+        MAPS --> DOF --> THETA --> GB
-    PROCESSING["PROCESSING CORE LAYER"]
+    end
-    OPT["Optimization<br/>- Decimation<br/>- Denoising<br/>- Remeshing"]
+
-    ALGO["Algorithmic<br/>- Boolean Ops<br/>- Smoothing<br/>- Hole Filling"]
+    subgraph CORE["② PROCESSING CORE"]
-    OTHER["Other<br/>- Custom Algos<br/>- Transforms<br/>- Filters"]
+        NEWTON["Newton Solver\nnewton_euclidean/spherical/hyper_ideal()\n→ NewtonResult  x* ∈ ℝⁿ"]
-    POSTPROCESSING["POSTPROCESSING <br/>(refined triangulation etc.)"]
+        CG["Cut Graph  [closed surfaces]\ncompute_cut_graph()\n→ CutGraph  (2g seam edges)"]
-    EXPORT["Export Formats<br/>- OBJ<br/>- PLY<br/>- STL<br/>- Custom"]
+        LAYOUT["Layout / Embedding\neuclidean/spherical/hyper_ideal_layout()\n→ Layout2D / Layout3D\n  · uv[v]  · halfedge_uv[h]\n  · HolonomyData"]
-    VIZ["Visualization<br/>- OpenGL<br/>- QT<br/>- Screenshot"]
+        NORM["Normalisation\nnormalise_euclidean / _hyperbolic / _spherical()"]
-subgraph PRE[PREPROCESSING]
+        PERIOD["Period Matrix  [genus 1]\ncompute_period_matrix()\n→ PeriodData  τ ∈ ℍ"]
-    A1 --> ADAPTER1
+        FD["Fundamental Domain\ncompute_fundamental_domain()\n→ FundamentalDomain + tiling"]
-    A2 --> ADAPTER1
+        NEWTON --> CG --> LAYOUT --> NORM --> PERIOD --> FD
-    A3 --> ADAPTER1
+    end
-    A4 --> ADAPTER1   
+
-    PREPROCESSING <-->ADAPTER1
+    subgraph POST["③ POSTPROCESSING"]
-end
+        SERIAL["Serialisation\nsave_result_json/xml()\nsave_layout_off()"]
-    EXT <--> UNIVERSAL
+        VIZ["Visualisation\nexample_viewer  (libigl / GLFW)"]
-    UNIVERSAL --> PROCESSING
+        CLI["CLI App\nconformallab_core  -i -g -o -j -x -s"]
-    UNIVERSAL --> EXPORT_ADAPTER
+    end
-    ADAPTER1 <--> EXT
+
-    ADAPTER1 <--> UNIVERSAL
+    IN --> ADAPTER --> PRE
-    OPT --> UNIVERSAL
+    PRE --> CORE
-    ALGO --> UNIVERSAL
+    CORE --> POST
-    OTHER --> UNIVERSAL
+
-subgraph CORE[PROCESSING ]
+    style PRE  fill:#dbeafe,stroke:#3b82f6,stroke-width:2px,color:#000
-    PROCESSING --> OPT
+    style CORE fill:#dcfce7,stroke:#16a34a,stroke-width:2px,color:#000
-    PROCESSING --> ALGO
+    style POST fill:#fef9c3,stroke:#ca8a04,stroke-width:2px,color:#000
-    PROCESSING --> OTHER
-end
-subgraph POST[POSTPROCESSING]
-    EXPORT_ADAPTER <--> POSTPROCESSING
-    EXPORT_ADAPTER --> EXPORT
-    EXPORT_ADAPTER --> VIZ
-end
-    style UNIVERSAL fill:#90EE90,stroke:#000,stroke-width:3px,color:#000
-    style PREPROCESSING  fill:#87CEEB,stroke:#000,stroke-width:2px,color:#000
-    style POSTPROCESSING fill:#87CEEB,stroke:#000,stroke-width:2px,color:#000
-    style PROCESSING fill:#87CEEB,stroke:#000,stroke-width:2px,color:#000
-    style EXT fill:#DDA0DD,stroke:#000,stroke-width:2px,color:#000
-    style ADAPTER1 fill:#FFB347,stroke:#000,stroke-width:2px,color:#000
-    style EXPORT_ADAPTER fill:#FFB347,stroke:#000,stroke-width:2px,color:#000
 ```
-All external inputs (file-based, implicit, parametric, procedural) are first converted by an spezfic input adapter into a single universal  interface, which is the canonical internal representation used by the processing stages. The processing core units operates on this universal representation and can use external libraries via a plugin-like extension, while results are passed through an export adapter to file formats or visualization frontends.
 
+---
 
-## Three-stage processing pipeline
+## Stage ① — Preprocessing
-The geometry processing pipeline in ConformalLab++ is structured into three main stages, all operating on the same universal mesh representation.
 
-### Preprocessing
+### Input adapter
-Converts all external inputs (files, analytic models, procedural sources) into the canonical half-edge mesh form.
 
-Performs optional cleaning and normalization (e.g. fixing degeneracies, rescaling, enforcing orientation).
+All mesh input flows through a single entry point:
 
-Guarantees that downstream stages see a well-formed, consistent mesh (or fail early with clear errors).
+```cpp
+ConformalMesh mesh = load_mesh("input.off");   // OFF · OBJ · PLY
+ConformalMesh mesh = make_quad_strip();         // built-in test meshes
+```
 
-### Processing (core processing units)
+**Precondition guarantee:** the adapter ensures the mesh is a valid, orientable,
+triangulated surface with consistent halfedge structure (CGAL validity).
+Downstream stages never check for degeneracies — they trust the adapter.
 
-Runs one or more core processing units in sequence on the canonical mesh.
+### Maps setup
 
-Each unit takes a mesh of the same type as input and produces a mesh of the same type as output (possibly with enriched attributes).
+Each geometry mode has its own maps struct that attaches property maps to the mesh:
 
-Examples: remeshing, conformal parameterization, smoothing, boolean operations, experiment-specific transforms.
+| Geometry | Setup function | Key property maps |
+|----------|---------------|-------------------|
+| Euclidean (ℝ²) | `setup_euclidean_maps()` | `lambda0[e]`, `theta_v[v]`, `v_idx[v]` |
+| Spherical (S²) | `setup_spherical_maps()` | `lambda0[e]`, `theta_v[v]`, `v_idx[v]` |
+| Hyper-ideal (H²) | `setup_hyper_ideal_maps()` | `beta_v[v]`, `alpha_e[e]`, `v_idx[v]`, `e_idx[e]` |
 
-Each core processing unit conceptually has the shape UniversalMesh -> UniversalMesh, but units also declare:
+`compute_*_lambda0_from_mesh()` initialises log-edge-lengths from the 3-D vertex
+positions. After this step the solver works entirely in scale-factor space; the
+original vertex positions are no longer needed.
 
-+ Preconditions: requirements on the input mesh (e.g. triangulated, manifold, oriented, specific attributes present).
+### DOF assignment & target angles
 
-+ Capabilities: guarantees on the output mesh (e.g. remains manifold, produces curvature attributes, preserves boundary).
+```cpp
+// Pin first vertex (gauge fix for open meshes)
+maps.v_idx[*mesh.vertices().begin()] = -1;
+int idx = 0;
+for (auto v : rest_of_vertices) maps.v_idx[v] = idx++;
 
-A simple example:
+// Set target curvature (cone metric or natural equilibrium)
+maps.theta_v[v] = 2 * M_PI;          // flat interior vertex
+maps.theta_v[v] = M_PI / 3;          // 60° cone singularity
+```
 
-+ RemeshingUnit
+**Natural equilibrium shortcut:** evaluate the gradient at `x = 0`, subtract it from
+`theta_v` — the solver then converges to `x* = 0` identically (useful for tests).
 
-  + Preconditions: triangulated, manifold, oriented
+### Gauss–Bonnet check
 
-  + Capabilities: triangulated, manifold, vertex positions modified, edge count changed
+Before solving, the prescribed angles must satisfy:
- 
-Pipeline composition is valid if, for every adjacent pair of units, the capabilities of the previous unit satisfy the preconditions of the next unit.
 
+$$\sum_{v} (2\pi - \Theta_v) = 2\pi \cdot \chi(M)$$
 
-### Postprocessing
+```cpp
+check_gauss_bonnet(mesh, maps);         // throws if violated
+enforce_gauss_bonnet(mesh, maps);       // distributes residual uniformly
+```
 
-Adapts the processed mesh for external consumption: export to file formats, visualization, analysis tools.
+**This is the most common source of silent Newton non-convergence.**
+Prescribing angles that violate Gauss–Bonnet means no conformal factor exists —
+the solver will iterate without converging.
 
-May attach or convert attribute data (e.g. colors, scalar fields, experiment results) into a format suitable for rendering or further tools.
+---
 
-Does not change the core topology or semantics of the processed mesh, only its representation for the outside world.
+## Stage ② — Processing core
 
-This structure makes it easy to reason about where data enters, is modified, and leaves the system, and keeps experimental algorithms clearly separated from IO concerns.
+### The three geometry modes
 
-## Role of the universal mesh representation
+conformallab++ implements discrete conformal geometry in three model spaces:
-The universal mesh representation (a half-edge mesh interface) is the shared contract between all pipeline stages.
 
-### Single input/output type
+| Mode | Space | Curvature | Typical surfaces |
-Every core processing unit exposes the same function shape conceptually:
+|------|-------|-----------|-----------------|
-UniversalMesh → UniversalMesh.
+| **Euclidean** | ℝ² | K = 0 | Flat tori, developable surfaces, open patches |
-This allows units to be chained in arbitrary order as long as their preconditions on the mesh are satisfied.
+| **Spherical** | S² | K = +1 | Genus-0 (sphere-like) surfaces |
+| **Hyper-ideal** | H² (Poincaré disk) | K = −1 | Genus-g surfaces (g ≥ 1), hyperbolic structures |
 
-### Local, topology-aware access
+All three share the same algorithmic structure — only the angle formula, the
-The half-edge structure encodes vertices, halfedges, faces, and their adjacency relations, which is essential for typical geometry operations (e.g. local neighborhood queries, traversal, edge flips).
+trilateration geometry, and the Hessian sign differ.
 
-Algorithms do not need to know where the mesh came from (file vs. analytic vs. procedural); they only rely on this uniform interface.
+### Newton solver
 
-### Extensibility via attributes
+```
-The universal mesh may carry extensible per-vertex, per-edge, and per-face attributes (e.g. UVs, curvature, experimental scalar fields).
+NewtonResult newton_euclidean  (mesh, x0, maps [, tol, max_iter])
-Core units can read and write these attributes while still conforming to the same mesh type, enabling complex pipelines without changing the central data structure.
+NewtonResult newton_spherical  (mesh, x0, maps [, tol, max_iter])
+NewtonResult newton_hyper_ideal(mesh, x0, maps [, tol, max_iter, hess_eps])
+```
 
+Each iteration:
+1. Evaluate gradient **G** (angle-sum defect per vertex)
+2. Evaluate Hessian **H** (analytical for Euclidean/Spherical; symmetric FD for HyperIdeal)
+3. Solve **H·Δx = −G** — try `SimplicialLDLT`, fall back to `SparseQR` on rank deficiency
+4. Backtracking line search (up to 20 halvings)
 
+**Preconditions:** mesh triangulated + manifold · Gauss–Bonnet satisfied · DOFs assigned  
+**Provides:** `NewtonResult.x` — converged scale factors; `converged`, `iterations`, `grad_inf_norm`
 
-Because all processing units speak the same “mesh language”, you can build pipelines like:
+**SparseQR fallback** handles gauge modes on closed meshes without a pinned vertex.
+The fallback is public API: `solve_linear_system(H, rhs, &used_fallback)`.
+
+### Cut graph (closed surfaces)
+
+For a closed genus-g surface, cutting along 2g independent homology cycles
+turns it into a topological disk — a prerequisite for globally consistent BFS layout.
+
+```cpp
+CutGraph cg = compute_cut_graph(mesh);
+// cg.cut_edge_flags[e.idx()] — true for seam edges
+// cg.cut_edge_indices         — ordered list of 2g seam edges
+// cg.genus                    — g
+```
+
+**Algorithm:** tree-cotree decomposition (Erickson–Whittlesey 2005).  
+**Preconditions:** closed, orientable, triangulated mesh  
+**Provides:** exactly `2g` seam edges whose removal makes the surface simply connected
+
+### Layout / Embedding
+
+BFS-trilateration unfolds the mesh into the target geometry.
+The root face (largest 3-D area, 1.5× interior bonus) is placed first;
+all other faces are placed in **priority order by BFS depth** (min-heap),
+minimising trilateration error accumulation.
+
+```cpp
+HolonomyData hol;
+Layout2D layout = euclidean_layout(mesh, result.x, maps, &cg, &hol, /*normalise=*/true);
+Layout3D slayout = spherical_layout(mesh, result.x, smaps);
+Layout2D hlayout = hyper_ideal_layout(mesh, result.x, hmaps, &cg, &hol);
+```
+
+**Key outputs:**
+
+| Field | Content |
+|-------|---------|
+| `layout.uv[v.idx()]` | Primary UV — first / shallowest-BFS visit |
+| `layout.halfedge_uv[h.idx()]` | UV of `source(h)` as seen from `face(h)` — seam-aware |
+| `layout.has_seam` | True when a vertex was reached via two different paths |
+| `hol.translations[i]` | Translation ω_i across cut edge i (Euclidean / spherical) |
+| `hol.mobius_maps[i]` | Möbius isometry T_i ∈ SU(1,1) across cut edge i (hyperbolic) |
+
+**`halfedge_uv` — texture atlas semantics:**  
+At a seam edge the two opposite halfedges carry *different* UV values.
+This gives each face its own UV copy of a seam vertex, enabling
+a proper GPU texture atlas without vertex duplication.
+
+**Trilateration — geometry by mode:**
+
+| Mode | Method | Accuracy |
+|------|--------|---------|
+| Euclidean | Analytic formula in ℝ² | Exact |
+| Spherical | Spherical law of cosines on S² | Exact |
+| Hyper-ideal | Möbius + hyperbolic law of cosines in Poincaré disk | Exact |
+
+**Preconditions:** `NewtonResult.converged` · mesh · maps · optional `CutGraph`  
+**Provides:** `Layout2D/3D` with `uv`, `halfedge_uv` · `HolonomyData` with translations / Möbius maps
+
+### Möbius maps
+
+`MobiusMap` (T(z) = (az+b)/(cz+d)) is the central algebraic object for hyperbolic geometry:
+
+```cpp
+MobiusMap T = MobiusMap::from_three(z1,w1, z2,w2, z3,w3); // fit to 3 correspondences
+MobiusMap S = T.inverse().compose(U);                       // group operations
+Eigen::Vector2d p2 = T.apply(p);                           // apply to 2-D point
+```
+
+Used for: hyperbolic trilateration · holonomy tracking · normalisation centering.
+
+### Normalisation
+
+After layout, a canonical post-processing step brings the result into a standard position:
+
+| Mode | Method | Effect |
+|------|--------|--------|
+| Euclidean | PCA — centroid → origin, major axis → x-axis | Translation + rotation |
+| Hyperbolic | Weighted Möbius centering (Fréchet mean, 30 iterations) | Maps centroid to disk origin |
+| Spherical | Rodrigues rotation | Maps centroid to north pole |
+
+Both `uv` and `halfedge_uv` are transformed identically.
+
+### Period matrix (genus 1)
+
+From the two holonomy translations ω₁, ω₂ ∈ ℂ read off from the cut graph,
+the conformal type of a flat torus is the SL(2,ℤ)-orbit of:
+
+$$\tau = \omega_2 / \omega_1 \in \mathbb{H}$$
+
+```cpp
+PeriodData pd = compute_period_matrix(hol);
+// pd.tau                   — complex period ratio
+// pd.omega[i]              — lattice generators as complex numbers
+// pd.in_fundamental_domain — after SL(2,ℤ) reduction
+// pd.genus()               — g = omega.size() / 2
+```
+
+SL(2,ℤ) reduction alternates T: τ↦τ+1 and S: τ↦−1/τ steps until
+τ ∈ F = {|τ| ≥ 1, −½ ≤ Re(τ) < ½}.
+
+**Note:** The Siegel period matrix Ω ∈ H_g for genus g ≥ 2 requires integrating
+holomorphic differentials — deferred to Phase 8.
+
+### Fundamental domain
+
+```cpp
+FundamentalDomain fd = compute_fundamental_domain(hol);
+// genus 1: CCW parallelogram {0, ω₁, ω₁+ω₂, ω₂}
+// genus g > 1: empty — 4g-polygon boundary walk deferred to Phase 8
+
+auto tiles = tiling_neighbourhood(layout, hol, /*m_max=*/2, /*n_max=*/2);
+// returns (2·m_max+1)·(2·n_max+1) translated copies of the layout
+// for visualising the universal cover
+```
+
+---
+
+## Stage ③ — Postprocessing
+
+### Serialisation
+
+```cpp
+// Layout as mesh file
+save_layout_off("layout.off", mesh, layout);
+
+// Full result: DOF vector + metadata + layout UVs
+save_result_json("result.json", result, "euclidean", V, F, &layout);
+save_result_xml ("result.xml",  result, "euclidean", V, F, &layout);
+
+// Round-trip load
+NewtonResult res2; std::string geom; Layout2D uv2;
+load_result_json("result.json", &res2, &geom, &uv2);
+```
+
+### CLI app
 
 ```bash
-Preprocessing
+conformallab_core \
-  -> RemeshingUnit
+  -i input.off        # input mesh
-  -> ConformalMapUnit
+  -g euclidean        # geometry: euclidean | spherical | hyper_ideal
-  -> ExperimentSpecificFilter
+  -o layout.off       # layout output
-  -> Export/Postprocessing
+  -j result.json      # JSON serialisation
+  -x result.xml       # XML serialisation
+  -s                  # show input in viewer
+  -v                  # verbose solver output
 ```
-without changing the basic function signature or the surrounding infrastructure, only by reordering or swapping units.
 
+### Interactive viewer
 
-## Declarative pipeline descriptions (Nice To Have but maybe to much) 
+`example_viewer` (libigl / GLFW) shows the 3-D mesh and the 2-D layout
-ConformalLab++ optionally supports a lightweight YAML-based description format for experiments.
+side-by-side. Built automatically with `-DWITH_CGAL=ON`.
-A pipeline consists of an input adapter, a sequence of steps (core processing units), and an output adapter. Each step can specify params as well as structural constraints via require and provide keys. A pipeline is considered valid if, for every step, all require conditions are satisfied by the accumulated provide and expect constraints of previous steps
 
-```yaml
+---
 
-pipeline:
+## Processing unit contracts
-  name: basic_conformal
-  input:
-    adapter: obj_reader
-    source: data/bunny.obj
 
-  steps:
+Each stage has explicit preconditions and guarantees.
-    - id: clean
+A pipeline is valid if every unit's preconditions are satisfied
-      unit: repair_soft_clean
+by the outputs of all preceding units.
-      params:
-        mode: soft
-      require:
-        representation: UniversalMesh
-        triangulated: true
-      provide:
-        triangulated: true
-        manifold: null
 
-    - id: param
+| Unit | Preconditions | Provides |
-      unit: conformal_parameterization
+|------|--------------|---------|
-      params:
+| `load_mesh` | Valid file path, supported format | Manifold, oriented, triangulated `ConformalMesh` |
-        method: cauchy_riemann
+| `setup_*_maps` | Triangulated mesh | Initialised property maps; `lambda0` from 3-D positions |
-      require:
+| `check_gauss_bonnet` | `theta_v` set | Throws if Σ(2π−Θ_v) ≠ 2π·χ |
-        triangulated: true
+| `enforce_gauss_bonnet` | `theta_v` set | Σ(2π−Θ_v) = 2π·χ guaranteed |
-        manifold: true
+| `newton_*` | GB satisfied · DOFs assigned | `NewtonResult.converged` · `x*` · gradient norm |
-      provide:
+| `compute_cut_graph` | Closed, orientable mesh | `2g` seam edges · `CutGraph.genus` |
-        attributes_add: [uv]
+| `euclidean_layout` | `newton_euclidean` converged | `uv[v]` · `halfedge_uv[h]` · `HolonomyData` |
+| `normalise_euclidean` | `layout.success == true` | Centroid at origin · major axis = x-axis |
+| `compute_period_matrix` | `hol.translations.size() >= 2` | `τ ∈ ℍ` · optionally reduced to F |
+| `compute_fundamental_domain` | `HolonomyData` (genus 1) | CCW parallelogram · edge identifications |
 
-  output:
+---
-    adapter: gltf_writer
-    target: out/bunny.glb
 
-```   
+## The three geometry modes in detail
-Each unit may declare a require section describing what it needs from its input (e.g. representation, triangulated, manifold, required attributes) and a provide section describing what it guarantees on its output (e.g. still triangulated, now manifold, adds a uv attribute). Pipelines are considered valid if for every step, the accumulated provide information of all previous steps satisfies the require constraints of the next step.
 
-### Repair Units (Nice To Have but maybe to much)
+```
+              Euclidean            Spherical            Hyper-ideal
+              ─────────────────────────────────────────────────────
+Space         ℝ²                   S²                   H² (Poincaré disk)
+Curvature     K = 0                K = +1               K = −1
+Genus         any (cone metrics)   0                    ≥ 1
+Angle sum     Σα_v = Θ_v           Σα_v = Θ_v           Σβ_v = Θ_v
+Hessian       PSD (cotangent-Lap.) NSD (sign-flip)      PSD (FD, strict conv.)
+Holonomy      translations ω_i     rotations (2-D)      Möbius maps T_i ∈ SU(1,1)
+Period        τ = ω₂/ω₁ ∈ ℍ        —                    axis of T_i
+Normalise     PCA centring         Rodrigues to N pole  weighted Möbius centring
+```
 
-Before entering the core conformal pipeline, input meshes are passed through an explicit preprocessing stage. This stage is responsible for turning “real-world” polygon data (often noisy, inconsistent, or non‑manifold) into a mesh that satisfies the structural requirements of the subsequent units
+---
-
-ConformalLab++ provides a small set of repair units that can be combined as needed:
-
-+ removal of (almost) degenerate triangles (needles, caps),
-
-+ stitching of compatible boundary cycles to close small gaps,
-
-+ enforcing a consistent face orientation where possible.
-
-These units are conceptually similar to geometric repair routines in modern polygon mesh processing toolkits
-
-We distinguish between two typical modes of operation:
-
-+ Soft cleaning: minimally invasive repairs intended for visualization and quick experiments. Soft cleaning tries to improve mesh quality without significantly altering topology or removing components.
-
-+ Hard cleaning: more aggressive repairs intended for numerically robust experiments. Hard cleaning may merge vertices, remove tiny components, and modify local topology to enforce manifoldness and consistent orientation when possible.
-
-## More internal representations (Nice To Have but maybe to much)
-ConformalLab++ distinguishes between internal representation spaces that experiments may move between:
-
-+ UniversalMesh: a halfedge‑based surface mesh representation used for most combinatorial and conformal algorithms.
-
-+ PointCloud: an unstructured set of sample points, optionally with normals and additional attributes, used for sampling, resampling, and certain reconstruction tasks.
-
-+ ImplicitField: a scalar field $f: \mathbf{R}^n \rightarrow \mathbf{R}$
-represented on a grid or as a procedural function, used for iso‑surface extraction and robust shape transformations.
-
-Each space has its own natural algorithms, and dedicated conversion units allow pipelines to move between these representations when needed.
-
-### Conversion units
-
-Typical conversion units include:
-
-+ mesh_to_pointcloud (sampling vertices and surfaces of a UniversalMesh into a PointCloud),
-
-+ pointcloud_to_mesh (surface reconstruction from point samples),
-
-+ mesh_to_implicit (rasterizing a UniversalMesh into a signed distance or occupancy field),
-
-+ implicit_to_mesh (iso‑surface extraction from an ImplicitField).
-
-These units are modeled as regular processing units in the pipeline but change the underlying representation space instead of just transforming a mesh
-
-## Attribute behaviour under topology changes (Nice To Have but maybe to much)
-
-ConformalLab++ treats attributes as first-class data attached to mesh entities (vertices, edges, faces, halfedges). When topology changes, attributes are updated according to simple, explicit rules:
-
-+ Vertex splits / edge splits: attributes on newly created vertices or edges are initialized by interpolation of the incident elements (e.g. linear interpolation of scalar fields along an edge).
-
-+ Edge collapse: attributes on the surviving vertex are computed from the incident vertices, typically by a convex combination or area‑weighted average for scalar quantities.
-
-+ Face operations (flip, subdivision): face attributes are either propagated (for categorical data) or interpolated (for scalar data) to new faces.
-
-By default, ConformalLab++ uses linear interpolation for scalar attributes and nearest‑neighbour propagation for discrete or categorical attributes; units may override these policies where necessary
 
+## Further documentation
 
+| Topic | Document |
+|---|---|
+| Public headers (all 24, with descriptions) | [api/headers.md](../api/headers.md) |
+| Test suites (26 suites, 158 tests, individual counts) | [api/tests.md](../api/tests.md) |
+| Extending the library (new functionals, geometry modes, Java porting guide) | [api/extending.md](../api/extending.md) |
+| Processing unit contracts (preconditions / provides table) | [api/contracts.md](../api/contracts.md) |
+| Phase 8 CGAL package design + declarative pipeline YAML | [api/cgal-package.md](../api/cgal-package.md) |
+| Key design decisions + rationale | [architecture/design-decisions.md](design-decisions.md) |
+| Project structure (directory tree + build targets) | [architecture/project-structure.md](project-structure.md) |
+| Three geometry modes in detail | [math/geometry-modes.md](../math/geometry-modes.md) |
+| References and papers | [math/references.md](../math/references.md) |
+| Development roadmap (Phases 1–10) | [roadmap/phases.md](../roadmap/phases.md) |
+| Java vs. C++ feature parity table | [roadmap/java-parity.md](../roadmap/java-parity.md) |

doc/architecture/phase-9a-validation.md

@@ -0,0 +1,228 @@
+# Phase 9a Validation Report
+
+This document records the mathematical and code-level validation of the two
+Phase 9a functionals against their reference sources.  It is produced as
+part of the Phase 9a acceptance procedure (decided 2026-05-19).
+
+| Functional | Reference | Java original | Lines |
+|---|---|---|---|
+| 9a.1 `cp_euclidean_functional.hpp` | Bobenko-Pinkall-Springborn 2010 | `CPEuclideanFunctional.java` (260) | 320 |
+| 9a.2 `inversive_distance_functional.hpp` | Luo 2004 + Glickenstein 2011 | none | 290 |
+
+---
+
+## 1.  CP-Euclidean (9a.1) — line-by-line Java mapping
+
+Reference file: `/Users/tarikmoussa/Desktop/conformallab/src/de/varylab/discreteconformal/functional/CPEuclideanFunctional.java`
+
+| Java line(s) | Java construct | C++ counterpart | Status |
+|---:|---|---|:---:|
+| 55–58 | `public class CPEuclideanFunctional<V,E,F>` | `namespace conformallab` (header-only, monomorphic on `ConformalMesh`) | ✓ |
+| 61–69 | `thetaMap, phiMap` ctor args | `CPEuclideanMaps { f_idx, theta_e, phi_f }` struct | ✓ |
+| 95–97 | `getDimension = hds.numFaces()` (incl. face 0) | `cp_euclidean_dimension(mesh, m)` counts only `f_idx ≥ 0` | refined |
+| 103–123 | `getNonZeroPattern` returns face-to-face adjacency | implicit in `cp_euclidean_hessian` triplet construction | ✓ |
+| 135–165 | `evaluateHessian` — interior edge `h_jk = sin θ / (cosh Δρ − cos θ)` | `cp_euclidean_hessian()` lines 224-247 | ✓ |
+| 146–151 | "skip if k == 0 / j == 0" gauge fix | `if (j >= 0)` / `if (k >= 0)` guard in triplet emission | equivalent |
+| 178–193 | per-face term `+φ_f · ρ_f` to energy and gradient | `cp_euclidean_energy/gradient` loop over faces | ✓ |
+| 196–232 | per-(directed)-edge term: boundary or interior branch | `cp_euclidean_energy/gradient` halfedge loop with `mesh.is_border(opposite)` check | ✓ |
+| 224–227 | `E += ½ p Δρ + Λ(θ*+p) − θ* ρ_left` | identical formula via `clausen2()` | ✓ |
+| 230 | `grad[left] -= p + θ*` | identical | ✓ |
+| 243–247 | `p(θ*, Δρ) = 2 atan(tan(θ*/2) tanh(Δρ/2))` | `cp_detail::p_function` | ✓ |
+
+### 1.1  Gauge convention divergence
+
+The Java code uses "face index 0 is pinned" as an implicit convention:
+hard-coded `if (face.getIndex() == 0)` short-circuits scatter throughout.
+The C++ port keeps the gauge user-selectable via
+`assign_cp_euclidean_face_dof_indices(mesh, m, pinned_face)`.  The default
+convenience overload picks the first face in iteration order, matching the
+Java behaviour for any mesh whose face 0 is the first iterated face.
+
+### 1.2  Test parity
+
+| Java test                                              | C++ test                                                 | Reproduces what |
+|---|---|---|
+| `init()` lines 53-86 dodecahedron-minus-face setup     | `make_open_tetrahedron()` (3 faces, 3 bdy edges)         | open-mesh boundary code path |
+| `setXGradient(rho)` (base class FunctionalTest)        | `FDGradientCheck_ClosedTetrahedron_RandomRho`            | FD-vs-analytic gradient |
+| `setXHessian(rho)`  (base class FunctionalTest)        | `FDHessianCheck_ClosedTetrahedron_RandomRho`             | FD-vs-analytic Hessian |
+| (none — implicit from FunctionalTest convexity check) | `HessianIsPSD`                                           | BPS-2010 §6 convexity |
+| (none)                                                 | `TangentialLimitGradientEqualsPhi`                       | θ=0 analytic check |
+| (none)                                                 | `PFunctionKnownValues`, `NaturalPhiMakesZeroTheEquilibrium` | helper-function unit tests |
+
+All Java tests reproduced; C++ adds 4 mathematical-property tests
+(PSD, tangential limit, p-function unit, natural-φ equilibrium).
+
+### 1.3  Numerical equivalence — seed-1 random ρ, FD gradient
+
+Java `FunctionalTest` uses `Random(1)` + uniform `[-0.5, +0.5]`.
+C++ uses `std::mt19937(1)` + `uniform_real_distribution<double>(-0.5, 0.5)`.
+The two RNG streams differ; numerical equivalence is therefore tested via
+the *property* "FD matches analytic to 1e-6 absolute" rather than via
+identical numerical traces.  Both implementations satisfy this property.
+
+---
+
+## 2.  Inversive Distance (9a.2) — line-by-line literature mapping
+
+No Java original exists.  The implementation derives directly from three
+published papers; each formula in the header is annotated with its source
+line in the paper.
+
+### 2.1  Edge-length formula  (Luo 2004 §3 / Glickenstein 2011 eq. 2.1)
+
+```
+ℓ_ij(u)²  =  exp(2 u_i) + exp(2 u_j) + 2 I_ij exp(u_i + u_j)
+          =  r_i² + r_j² + 2 I_ij r_i r_j
+```
+
+Implementation: `id_detail::edge_length_squared` (lines 130-138 of
+`inversive_distance_functional.hpp`).
+
+Tested at three special-case limits in
+`test_inversive_distance_functional.cpp` §1:
+
+| Inversive distance | Geometric meaning           | Closed-form ℓ        | Test |
+|:---:|---|---|---|
+| `I = +1` | tangential circles           | `r_i + r_j`         | `EdgeLengthFormula_TangentialLimit` |
+| `I = 0`  | orthogonal circles           | `√(r_i² + r_j²)`    | `EdgeLengthFormula_OrthogonalLimit` |
+| `I = −1` | inside-tangent / inverted    | `|r_i − r_j|`       | `EdgeLengthFormula_InsideTangentLimit` |
+| `I < −1` | impossible packing (no real ℓ) | `nan`/signal         | `EdgeLengthFormula_DegenerateReturnsMinusOne` |
+
+### 2.2  Bowers-Stephenson identity (Bowers-Stephenson 2004)
+
+```
+I_ij  =  ( ℓ_ij² − r_i² − r_j² )  /  ( 2 r_i r_j )
+```
+
+Implementation: `compute_inversive_distance_init_from_mesh` lines 152-172.
+The round-trip property — given (ℓ_3d, r0_i, r0_j) recover ℓ via the Luo
+formula — is tested in `BowersStephensonRoundTrip`.
+
+### 2.3  Gradient (Luo 2004 Lemma 3.1)
+
+```
+∂E/∂u_v  =  Θ_v  −  Σ_{T ∋ v}  α_v(T)
+```
+
+Implementation: `inversive_distance_gradient` lines 191-241.  This is
+structurally identical to the Euclidean functional (same halfedge
+convention, same `euclidean_angles()` law-of-cosines reused).  The only
+difference is the edge-length input: Luo's `ℓ²(I)` instead of Springborn's
+`exp(λ° + u_i + u_j)`.
+
+### 2.4  Energy — Luo's closed 1-form
+
+Luo (2004) proves the gradient 1-form `ω = Σ_v (Θ_v − Σ α_v) du_v` is
+closed on the open domain where every triangle is valid.  Hence the energy
+exists as a path integral.  No general closed-form expression is known
+(Glickenstein 2011 surveys partial results).  The implementation uses the
+same 10-point Gauss-Legendre quadrature as `euclidean_functional.hpp`
+(lines 243-274).
+
+This shared-quadrature choice keeps both functionals identical in the
+energy-evaluation cost and the FD-gradient validation pattern.
+
+### 2.5  Hessian — finite difference for now
+
+Glickenstein 2011 eq. (4.6) gives an analytic Hessian for the inversive-
+distance variational principle, but is more involved than the Springborn
+cot-Laplacian.  For MVP we rely on FD; the analytic form is a future
+optimisation (joining the Phase 9b roadmap for HyperIdeal as a sibling
+optimisation task).
+
+---
+
+## 3.  Cross-validation between 9a.1 and 9a.2
+
+The two functionals describe geometrically distinct circle packings
+(face-dual vs vertex-based) but agree in the *angle-defect structure*
+they expose to the Newton solver: both report `Θ − Σ α_actual` at every
+free DOF.  This is tested directly:
+
+```cpp
+TEST(InversiveDistanceFunctional, AngleDefectAtU0_AgreesWithEuclideanAtU0)
+{
+    auto mesh = make_quad_strip();
+
+    auto G_id = inversive_distance_gradient(mesh, x_id, m_id);
+    auto G_eu = euclidean_gradient            (mesh, x_eu, m_eu);
+
+    for (i in DOFs) EXPECT_NEAR(G_id[i], G_eu[i], 1e-10);
+}
+```
+
+Why this works: at `u = 0`, both initialisation procedures reconstruct the
+input 3-D edge length exactly (`compute_*_lambda0_from_mesh` for Euclidean,
+`compute_inversive_distance_init_from_mesh` for inversive distance).
+Therefore the actual angle sums per vertex are identical, and both
+gradients reduce to the same `Θ_default − Σ α(ℓ_3d)`.
+
+This is the **operational** consequence of Glickenstein 2011 §5:
+"different parametrisations of the same initial discrete metric produce the
+same Newton-time-zero gradient".
+
+---
+
+## 4.  Comparison with `euclidean_functional.hpp`
+
+| Aspect                     | Euclidean                                    | CP-Euclidean (9a.1)                          | Inversive Distance (9a.2)                                          |
+|---|---|---|---|
+| DOF basis                  | per vertex (`u_v = log scale`)                | per face (`ρ_f = log radius`)                | per vertex (`u_v = log radius`)                                    |
+| Per-edge constant          | `λ°_ij = 2 log ℓ°_ij`                         | `θ_e` (intersection angle)                   | `I_ij` (inversive distance)                                        |
+| Edge-length formula        | `ℓ = exp((λ° + u_i + u_j)/2)`                 | (no per-edge ℓ — face-circle radii)          | `ℓ² = r_i² + r_j² + 2 I r_i r_j`                                   |
+| Angles                     | half-tangent law of cosines                   | (face-internal angles via `p(θ*, Δρ)`)       | half-tangent law of cosines (reuses `euclidean_angles`)            |
+| Gradient                   | `Θ_v − Σ α_v(f)`                              | `φ_f − Σ_{h:face(h)=f} (p+θ*)`               | `Θ_v − Σ α_v(f)`                                                   |
+| Energy form                | path integral (10-pt GL)                      | closed form via `½ p Δρ + Λ(θ*+p) − θ* ρ`   | path integral (10-pt GL)                                           |
+| Hessian                    | analytic (cotangent Laplacian)                | analytic (`sin θ / (cosh Δρ − cos θ)`)       | finite difference (analytic deferred — Glickenstein 2011 eq. 4.6)  |
+| Convexity                  | strictly convex after gauge fix               | strictly convex after gauge fix              | locally convex on triangle-inequality domain (Luo 2004 Thm 1.2)    |
+| Gauge fix                  | pin one vertex (`v_idx = −1`)                 | pin one face   (`f_idx = −1`)                | pin one vertex (`v_idx = −1`)                                      |
+
+### 4.1  Structural reuse
+
+`inversive_distance_functional.hpp` reuses `euclidean_angles()` from
+`euclidean_geometry.hpp` unchanged.  The only difference from
+`euclidean_functional.hpp` is the four lines that map `(u_i, u_j, I_ij)` to
+`ℓ_ij²` before invoking `euclidean_angles()`.
+
+`cp_euclidean_functional.hpp` shares no algorithmic code with the Euclidean
+functional — it operates on a different DOF basis and a different energy
+form.  It does share the Clausen function via `clausen2()` from `clausen.hpp`.
+
+---
+
+## 5.  Test counts and acceptance criteria
+
+| Suite | Tests | Status |
+|---|---:|:---:|
+| `cgal.CPEuclideanFunctional.*`        | 10 | ✅ all pass |
+| `cgal.InversiveDistanceFunctional.*`  | 11 | ✅ all pass |
+| **Full CGAL regression**              | **205** | ✅ all pass |
+| Non-CGAL fast suite                   | 36 | ✅ all pass |
+
+### Acceptance criteria — met
+
+- [x]  FD-vs-analytic gradient check passes on all test meshes (both functionals)
+- [x]  FD-vs-analytic Hessian check passes on closed and open tetrahedron (9a.1)
+- [x]  Hessian PSD-property test passes (9a.1)
+- [x]  Bowers-Stephenson round-trip identity verified (9a.2)
+- [x]  Three special-case limits of Luo's edge formula verified at machine ε (9a.2)
+- [x]  Cross-validation against `euclidean_functional` at `u = 0` (9a.2)
+- [x]  All formulas cite their source paper or Java line number
+
+---
+
+## 6.  References
+
+- Bobenko, A. I., Pinkall, U. & Springborn, B. (2010). *Discrete conformal
+  maps and ideal hyperbolic polyhedra.* Geometry & Topology 14, 379–426.
+- Bowers, P. L. & Stephenson, K. (2004). *Uniformizing dessins and Belyĭ
+  maps via circle packing.* Memoirs of the AMS 170(805).
+- Glickenstein, D. (2011). *Discrete conformal variations and scalar
+  curvature on piecewise flat manifolds.* J. Differential Geometry 87(2),
+  201–238.
+- Luo, F. (2004). *Combinatorial Yamabe Flow on Surfaces.* Communications
+  in Contemporary Mathematics 6(5), 765–780.
+- Java original (CPEuclidean only):
+  `/Users/tarikmoussa/Desktop/conformallab/src/de/varylab/discreteconformal/functional/CPEuclideanFunctional.java`
+- Java test:
+  `/Users/tarikmoussa/Desktop/conformallab/src-test/de/varylab/discreteconformal/functional/CPEuclideanFunctionalTest.java`

doc/architecture/project-structure.md

@@ -0,0 +1,115 @@
+# Project Structure
+
+```
+ConformalLabpp/
+├── CLAUDE.md                         # Claude Code context file
+├── README.md                         # Entry point — what/why, quick start, doc index
+├── LICENSE                           # MIT
+│
+├── code/                             # All C++ source
+│   ├── CMakeLists.txt                # Root: three build modes (default / CGAL_TESTS / CGAL)
+│   │
+│   ├── include/                      # Header-only library — all algorithms here
+│   │   ├── conformal_mesh.hpp        # ConformalMesh = CGAL::Surface_mesh<Point3>
+│   │   ├── constants.hpp             # conformallab::PI, TWO_PI
+│   │   ├── clausen.hpp               # Cl₂, Lobachevsky Л, ImLi₂
+│   │   ├── hyper_ideal_geometry.hpp  # ζ₁₃/₁₄/₁₅, lᵢⱼ, αᵢⱼ, σᵢ, σᵢⱼ
+│   │   ├── hyper_ideal_utility.hpp   # Tetrahedron volumes (Meyerhoff / Kolpakov–Mednykh)
+│   │   ├── hyper_ideal_visualization_utility.hpp  # Poincaré disk projection, circumcircle helpers
+│   │   ├── hyper_ideal_functional.hpp # HyperIdeal energy + gradient on ConformalMesh
+│   │   ├── hyper_ideal_hessian.hpp   # HyperIdeal Hessian (symmetric FD, Phase 9b: analytic)
+│   │   ├── spherical_geometry.hpp    # Spherical arc-length, half-angle formula
+│   │   ├── spherical_functional.hpp  # Spherical energy + gradient + gauge-fix
+│   │   ├── spherical_hessian.hpp     # Spherical Hessian (∂α/∂u via law of cosines)
+│   │   ├── euclidean_geometry.hpp    # Euclidean corner angle (t-value / atan2)
+│   │   ├── euclidean_functional.hpp  # Euclidean energy + gradient
+│   │   ├── euclidean_hessian.hpp     # Cotangent Laplacian Hessian (Pinkall–Polthier)
+│   │   ├── newton_solver.hpp         # newton_{euclidean,spherical,hyper_ideal} + solve_linear_system
+│   │   ├── gauss_bonnet.hpp          # euler_characteristic, genus, check/enforce_gauss_bonnet
+│   │   ├── cut_graph.hpp             # CutGraph + compute_cut_graph (tree-cotree)
+│   │   ├── layout.hpp                # Priority-BFS layout, MobiusMap, halfedge_uv, HolonomyData
+│   │   ├── mesh_builder.hpp          # make_triangle / make_tetrahedron / make_quad_strip / make_fan
+│   │   ├── mesh_io.hpp               # load_mesh / save_mesh (OFF/OBJ/PLY via CGAL::IO)
+│   │   ├── mesh_utils.hpp            # CGAL → Eigen conversion (cgal_to_eigen)
+│   │   ├── serialization.hpp         # save/load_result_json + save/load_result_xml
+│   │   ├── period_matrix.hpp         # PeriodData, compute_period_matrix, SL(2,ℤ) reduction
+│   │   └── fundamental_domain.hpp    # FundamentalDomain, tiling_copy, tiling_neighbourhood
+│   │
+│   ├── src/
+│   │   ├── apps/v0/conformallab_cli.cpp   # CLI app (Phase 5): load → solve → layout → export
+│   │   └── viewer/simple_viewer.cpp       # libigl/GLFW viewer wrapper
+│   │
+│   ├── examples/                     # Standalone example programs (require WITH_CGAL=ON)
+│   │   ├── example_euclidean.cpp     # Euclidean pipeline end-to-end
+│   │   ├── example_hyper_ideal.cpp   # HyperIdeal pipeline end-to-end
+│   │   ├── example_layout.cpp        # Full pipeline: solve → layout → OFF/JSON/XML + round-trip
+│   │   └── example_viewer.cpp        # Interactive libigl viewer
+│   │
+│   ├── tests/
+│   │   ├── CMakeLists.txt
+│   │   ├── test_clausen.cpp
+│   │   ├── test_hyper_ideal_utility.cpp
+│   │   ├── test_matrix_utility.cpp
+│   │   ├── test_surface_curve_utility.cpp
+│   │   ├── test_discrete_elliptic_utility.cpp
+│   │   ├── test_p2_utility.cpp
+│   │   ├── test_hyper_ideal_visualization_utility.cpp
+│   │   └── cgal/                     # CGAL test suite (WITH_CGAL_TESTS or WITH_CGAL)
+│   │       ├── test_conformal_mesh.cpp
+│   │       ├── test_hyper_ideal_functional.cpp
+│   │       ├── test_spherical_functional.cpp
+│   │       ├── test_euclidean_functional.cpp
+│   │       ├── test_euclidean_hessian.cpp
+│   │       ├── test_spherical_hessian.cpp
+│   │       ├── test_newton_solver.cpp
+│   │       ├── test_mesh_io.cpp
+│   │       ├── test_pipeline.cpp
+│   │       ├── test_layout.cpp
+│   │       ├── test_phase6.cpp
+│   │       ├── test_phase7.cpp
+│   │       └── test_geometry_utils.cpp
+│   │
+│   └── deps/                         # All dependencies as bundled tarballs
+│       ├── tarballs/
+│       │   ├── eigen-3.4.0.tar.gz
+│       │   ├── CGAL-6.1.1.tar.xz
+│       │   ├── libigl-2.6.0.tar.gz
+│       │   ├── libigl-glad.tar.gz
+│       │   └── glfw-3.4.tar.gz
+│       └── single_includes/          # CLI11.hpp, json.hpp (header-only)
+│
+├── doc/
+│   ├── getting-started.md
+│   ├── contributing.md
+│   ├── api/
+│   │   ├── pipeline.md               # Full pipeline API with code examples
+│   │   ├── extending.md              # New functionals, geometry modes, Java porting
+│   │   ├── contracts.md              # Processing unit preconditions/provides table
+│   │   └── cgal-package.md           # Phase 8 CGAL package design (TODO)
+│   ├── architecture/
+│   │   ├── overall_pipeline.md       # Mermaid diagram + stage descriptions
+│   │   ├── design-decisions.md       # Key architectural choices + rationale
+│   │   └── project-structure.md      # This file
+│   ├── math/
+│   │   ├── geometry-modes.md         # Euclidean / Spherical / HyperIdeal comparison
+│   │   └── references.md             # All papers by module
+│   └── roadmap/
+│       ├── phases.md                 # Phases 1–10 with porting/research boundary
+│       └── java-parity.md            # Java vs C++ feature parity table
+│
+└── .gitea/
+    ├── docker/Dockerfile.ci-cpp      # Ubuntu 22.04 ARM64 CI image
+    └── workflows/cpp-tests.yml       # Two-job CI pipeline
+```
+
+## Build targets
+
+| Target | Mode | Description |
+|---|---|---|
+| `conformallab_tests` | default | 36 non-CGAL pure-math tests |
+| `conformallab_cgal_tests` | `WITH_CGAL_TESTS` or `WITH_CGAL` | 158 CGAL tests |
+| `conformallab_core` | `WITH_CGAL` | CLI application |
+| `example_euclidean` | `WITH_CGAL` | Euclidean example program |
+| `example_hyper_ideal` | `WITH_CGAL` | HyperIdeal example program |
+| `example_layout` | `WITH_CGAL` | Full pipeline example |
+| `example_viewer` | `WITH_CGAL` | Interactive viewer |

doc/concepts/declarative-pipeline.md

@@ -0,0 +1,571 @@
+# Declarative YAML Pipeline — Concept & Design
+
+> **Status: Phase 8e — designed, not yet implemented.**  
+> This document is the authoritative design specification.  
+> Implementation target: `code/include/pipeline.hpp` + CLI flag `--pipeline`.
+
+---
+
+## 1 — Core idea
+
+Every algorithm in conformallab++ is a **Processing Unit**: a function with
+explicit preconditions (*require*) and guarantees (*provide*).
+The contract table in [doc/api/contracts.md](../api/contracts.md) lists them all.
+
+A **declarative pipeline** is a YAML file that:
+1. Lists the units to execute and their parameters.
+2. Annotates each step with the tokens it `require`s and `provide`s.
+3. Is validated **before any code runs** — the validator walks the dependency
+   graph and rejects the file if any `require` token is not yet provided by
+   a preceding step.
+
+The result is a **self-documenting, reproducible experiment** that can be
+version-controlled, shared, and re-run identically.
+
+---
+
+## 2 — Token vocabulary
+
+Tokens are short strings.  Each Processing Unit consumes and produces a fixed
+set of tokens.  The validator treats them as a monotonically growing *provided set*.
+
+### Input tokens (provided by the `input:` block)
+| Token | Meaning |
+|---|---|
+| `mesh` | A loaded, triangulated, oriented `ConformalMesh` |
+| `mesh_closed` | Mesh has no boundary (required for cut graph) |
+| `mesh_open` | Mesh has at least one boundary component |
+
+### Setup tokens
+| Token | Produced by | Required by |
+|---|---|---|
+| `maps_euclidean` | `setup_euclidean_maps` | `lambda0_euclidean`, `gauss_bonnet`, `newton_euclidean` |
+| `maps_spherical` | `setup_spherical_maps` | `lambda0_spherical`, `gauss_bonnet`, `newton_spherical` |
+| `maps_hyper_ideal` | `setup_hyper_ideal_maps` | `lambda0_hyper_ideal`, `gauss_bonnet`, `newton_hyper_ideal` |
+| `lambda0_euclidean` | `compute_euclidean_lambda0_from_mesh` | `gauss_bonnet_euclidean`, `newton_euclidean` |
+| `lambda0_spherical` | `compute_spherical_lambda0_from_mesh` | `gauss_bonnet_spherical`, `newton_spherical` |
+| `lambda0_hyper_ideal` | `compute_hyper_ideal_lambda0_from_mesh` | `gauss_bonnet_hyper_ideal`, `newton_hyper_ideal` |
+| `dof_indices` | DOF assignment step | `newton_*` |
+
+### Gauss–Bonnet tokens
+| Token | Produced by |
+|---|---|
+| `gauss_bonnet_euclidean` | `check_gauss_bonnet` or `enforce_gauss_bonnet` (Euclidean maps) |
+| `gauss_bonnet_spherical` | same, Spherical maps |
+| `gauss_bonnet_hyper_ideal` | same, HyperIdeal maps |
+
+### Solver tokens
+| Token | Produced by |
+|---|---|
+| `x_euclidean` | `newton_euclidean` (converged) |
+| `x_spherical` | `newton_spherical` (converged) |
+| `x_hyper_ideal` | `newton_hyper_ideal` (converged) |
+
+### Topology tokens
+| Token | Produced by | Required by |
+|---|---|---|
+| `cut_graph` | `compute_cut_graph` | `euclidean_layout` (closed mesh), `hyper_ideal_layout` |
+
+### Layout tokens
+| Token | Produced by |
+|---|---|
+| `layout_euclidean` | `euclidean_layout` |
+| `layout_spherical` | `spherical_layout` |
+| `layout_hyper_ideal` | `hyper_ideal_layout` |
+| `holonomy_euclidean` | `euclidean_layout` (with cut graph) |
+| `holonomy_hyper_ideal` | `hyper_ideal_layout` (with cut graph) |
+
+### Period / domain tokens
+| Token | Produced by | Required by |
+|---|---|---|
+| `period_matrix` | `compute_period_matrix` | `fundamental_domain` |
+| `fundamental_domain` | `compute_fundamental_domain` | `tiling` |
+| `tiling` | `tiling_neighbourhood` | output steps |
+
+### Output tokens
+| Token | Produced by |
+|---|---|
+| `saved_layout` | `save_layout_off` |
+| `saved_result` | `save_result_json` / `save_result_xml` |
+
+---
+
+## 3 — YAML schema
+
+```yaml
+pipeline:
+  name:     <string>           # human-readable experiment name
+  geometry: euclidean          # euclidean | spherical | hyper_ideal
+  description: |               # optional multi-line description
+    ...
+
+  input:
+    source: <path>             # mesh file (.off / .obj / .ply)
+    # Optional overrides:
+    theta_v: flat              # flat (2π everywhere) | cone:<file> | custom:<file>
+
+  steps:
+    - id:      <string>        # unique step identifier
+      unit:    <function>      # C++ function name (see token table)
+      require: [<token>, ...]  # tokens that must be in the provided set
+      provide: [<token>, ...]  # tokens added to provided set after this step
+      params:                  # optional parameter overrides
+        <key>: <value>
+
+  output:
+    layout: <path>             # optional: save Layout2D as .off with UVs
+    json:   <path>             # optional: save NewtonResult + Layout as JSON
+    xml:    <path>             # optional: save NewtonResult + Layout as XML
+    tau:    <path>             # optional: write τ (period matrix) as text
+    report: <path>             # optional: write human-readable summary
+```
+
+### Parameter defaults by unit
+
+| Unit | Parameter | Default |
+|---|---|---|
+| `newton_euclidean` | `tol` | `1e-8` |
+| `newton_euclidean` | `max_iter` | `200` |
+| `newton_spherical` | `tol` | `1e-8` |
+| `newton_spherical` | `max_iter` | `200` |
+| `newton_hyper_ideal` | `tol` | `1e-8` |
+| `newton_hyper_ideal` | `max_iter` | `200` |
+| `newton_hyper_ideal` | `hess_eps` | `1e-5` |
+| `euclidean_layout` | `normalise` | `false` |
+| `hyper_ideal_layout` | `normalise` | `false` |
+| `spherical_layout` | `normalise` | `false` |
+| `compute_period_matrix` | `reduce` | `true` (SL(2,ℤ)) |
+| `tiling_neighbourhood` | `m` | `1` |
+| `tiling_neighbourhood` | `n` | `1` |
+
+---
+
+## 4 — Validation algorithm
+
+```
+provided = { "mesh" }              ← always available after input is loaded
+
+if input.source has no boundary:
+    provided ← provided ∪ { "mesh_closed" }
+else:
+    provided ← provided ∪ { "mesh_open" }
+
+for each step in pipeline.steps:
+    for token in step.require:
+        if token ∉ provided:
+            ERROR: "Step '<id>' requires '<token>' which is not yet provided.
+                   Add a step that provides it before step '<id>'."
+    provided ← provided ∪ step.provide
+
+for each output key in pipeline.output:
+    check that its required token is in provided
+    (e.g. 'tau' requires 'period_matrix')
+```
+
+The validator runs **before any C++ code executes**.  If validation passes,
+the steps are executed in order.  There is no parallelism — steps are sequential.
+
+---
+
+## 5 — Complete examples
+
+### Example A — Euclidean uniformization of a flat torus (genus 1)
+
+```yaml
+pipeline:
+  name: flat_torus_euclidean
+  geometry: euclidean
+  description: |
+    Euclidean uniformization of the 4×4 torus of revolution.
+    Computes the period matrix τ and the fundamental domain parallelogram.
+
+  input:
+    source: code/data/off/torus_4x4.off
+
+  steps:
+    - id: setup
+      unit: setup_euclidean_maps
+      require: [mesh]
+      provide: [maps_euclidean]
+
+    - id: lambda0
+      unit: compute_euclidean_lambda0_from_mesh
+      require: [mesh, maps_euclidean]
+      provide: [lambda0_euclidean]
+
+    - id: dofs
+      unit: assign_dof_indices_euclidean
+      require: [maps_euclidean]
+      provide: [dof_indices]
+      params:
+        pin_strategy: first_vertex   # pin v₀, assign sequential to rest
+
+    - id: gauss_bonnet
+      unit: enforce_gauss_bonnet
+      require: [maps_euclidean, lambda0_euclidean]
+      provide: [gauss_bonnet_euclidean]
+
+    - id: solve
+      unit: newton_euclidean
+      require: [gauss_bonnet_euclidean, dof_indices]
+      provide: [x_euclidean]
+      params:
+        tol: 1.0e-10
+        max_iter: 200
+
+    - id: cut
+      unit: compute_cut_graph
+      require: [mesh_closed]
+      provide: [cut_graph]
+
+    - id: layout
+      unit: euclidean_layout
+      require: [x_euclidean, cut_graph]
+      provide: [layout_euclidean, holonomy_euclidean]
+      params:
+        normalise: true
+
+    - id: period
+      unit: compute_period_matrix
+      require: [holonomy_euclidean]
+      provide: [period_matrix]
+      params:
+        reduce: true
+
+    - id: domain
+      unit: compute_fundamental_domain
+      require: [holonomy_euclidean]
+      provide: [fundamental_domain]
+
+  output:
+    layout: out/torus_layout.off
+    json:   out/torus_result.json
+    tau:    out/torus_tau.txt
+```
+
+**Expected output (`torus_tau.txt`):**
+```
+tau = 0.000... + 0.9...i    # Re(τ) ≈ 0 (4-fold symmetry), Im(τ) ≈ 1
+|tau| = 0.9...              # ≥ 1 after SL(2,ℤ) reduction
+```
+
+---
+
+### Example B — Spherical uniformization (genus 0)
+
+```yaml
+pipeline:
+  name: cathead_spherical
+  geometry: spherical
+  description: |
+    Map the open cathead mesh to the sphere.
+
+  input:
+    source: code/data/obj/cathead.obj
+
+  steps:
+    - id: setup
+      unit: setup_spherical_maps
+      require: [mesh]
+      provide: [maps_spherical]
+
+    - id: lambda0
+      unit: compute_spherical_lambda0_from_mesh
+      require: [mesh, maps_spherical]
+      provide: [lambda0_spherical]
+
+    - id: dofs
+      unit: assign_dof_indices_spherical
+      require: [maps_spherical]
+      provide: [dof_indices]
+
+    - id: gauss_bonnet
+      unit: enforce_gauss_bonnet
+      require: [maps_spherical, lambda0_spherical]
+      provide: [gauss_bonnet_spherical]
+
+    - id: solve
+      unit: newton_spherical
+      require: [gauss_bonnet_spherical, dof_indices]
+      provide: [x_spherical]
+
+    - id: layout
+      unit: spherical_layout
+      require: [x_spherical]
+      provide: [layout_spherical]
+      params:
+        normalise: true    # rotate centroid to north pole
+
+  output:
+    json: out/cathead_spherical.json
+```
+
+---
+
+### Example C — Hyperbolic uniformization (genus 1, Poincaré disk)
+
+```yaml
+pipeline:
+  name: torus_hyperbolic
+  geometry: hyper_ideal
+  description: |
+    Hyperbolic (hyper-ideal) uniformization of the 8×8 torus.
+    Lays out the mesh in the Poincaré disk with correct Möbius holonomy.
+
+  input:
+    source: code/data/off/torus_8x8.off
+
+  steps:
+    - id: setup
+      unit: setup_hyper_ideal_maps
+      require: [mesh]
+      provide: [maps_hyper_ideal]
+
+    - id: lambda0
+      unit: compute_hyper_ideal_lambda0_from_mesh
+      require: [mesh, maps_hyper_ideal]
+      provide: [lambda0_hyper_ideal]
+
+    - id: dofs
+      unit: assign_all_dof_indices
+      require: [maps_hyper_ideal]
+      provide: [dof_indices]
+      # HyperIdeal: all vertices AND edges are free DOFs — no vertex pinned.
+
+    - id: gauss_bonnet
+      unit: enforce_gauss_bonnet
+      require: [maps_hyper_ideal, lambda0_hyper_ideal]
+      provide: [gauss_bonnet_hyper_ideal]
+
+    - id: solve
+      unit: newton_hyper_ideal
+      require: [gauss_bonnet_hyper_ideal, dof_indices]
+      provide: [x_hyper_ideal]
+      params:
+        tol: 1.0e-10
+
+    - id: cut
+      unit: compute_cut_graph
+      require: [mesh_closed]
+      provide: [cut_graph]
+
+    - id: layout
+      unit: hyper_ideal_layout
+      require: [x_hyper_ideal, cut_graph]
+      provide: [layout_hyper_ideal, holonomy_hyper_ideal]
+      params:
+        normalise: true    # Möbius-centre to disk origin
+
+  output:
+    layout: out/torus_disk.off
+    json:   out/torus_disk.json
+```
+
+---
+
+### Example D — Full pipeline with period matrix and tiling
+
+```yaml
+pipeline:
+  name: torus_full
+  geometry: euclidean
+
+  input:
+    source: code/data/off/torus_hex_6x6.off
+
+  steps:
+    - id: setup
+      unit: setup_euclidean_maps
+      require: [mesh]
+      provide: [maps_euclidean]
+
+    - id: lambda0
+      unit: compute_euclidean_lambda0_from_mesh
+      require: [mesh, maps_euclidean]
+      provide: [lambda0_euclidean]
+
+    - id: dofs
+      unit: assign_dof_indices_euclidean
+      require: [maps_euclidean]
+      provide: [dof_indices]
+
+    - id: gauss_bonnet
+      unit: enforce_gauss_bonnet
+      require: [maps_euclidean, lambda0_euclidean]
+      provide: [gauss_bonnet_euclidean]
+
+    - id: solve
+      unit: newton_euclidean
+      require: [gauss_bonnet_euclidean, dof_indices]
+      provide: [x_euclidean]
+
+    - id: cut
+      unit: compute_cut_graph
+      require: [mesh_closed]
+      provide: [cut_graph]
+
+    - id: layout
+      unit: euclidean_layout
+      require: [x_euclidean, cut_graph]
+      provide: [layout_euclidean, holonomy_euclidean]
+      params:
+        normalise: true
+
+    - id: period
+      unit: compute_period_matrix
+      require: [holonomy_euclidean]
+      provide: [period_matrix]
+
+    - id: domain
+      unit: compute_fundamental_domain
+      require: [holonomy_euclidean]
+      provide: [fundamental_domain]
+
+    - id: tiling
+      unit: tiling_neighbourhood
+      require: [layout_euclidean, holonomy_euclidean]
+      provide: [tiling]
+      params:
+        m: 2    # 5×5 tile grid around origin
+        n: 2
+
+  output:
+    layout: out/hex_torus_layout.off
+    tau:    out/hex_torus_tau.txt
+    json:   out/hex_torus_full.json
+    report: out/hex_torus_summary.txt
+
+# Expected tau for 6-fold symmetric torus:
+#   Re(τ) ≈ 0.5,  Im(τ) ≈ 0.866  (approaches e^{iπ/3} as mesh is refined)
+```
+
+---
+
+### Example E — Validation error (intentional mistake)
+
+```yaml
+pipeline:
+  name: broken_example
+  geometry: euclidean
+
+  input:
+    source: code/data/off/torus_4x4.off
+
+  steps:
+    - id: solve             # ← WRONG: skipped setup and lambda0
+      unit: newton_euclidean
+      require: [gauss_bonnet_euclidean, dof_indices]
+      provide: [x_euclidean]
+
+    - id: layout
+      unit: euclidean_layout
+      require: [x_euclidean, cut_graph]   # ← WRONG: cut_graph never provided
+      provide: [layout_euclidean]
+```
+
+**Validator output:**
+```
+ERROR [step 'solve']: requires 'gauss_bonnet_euclidean' which is not yet provided.
+  Hint: add setup_euclidean_maps → compute_euclidean_lambda0_from_mesh
+        → enforce_gauss_bonnet before 'solve'.
+
+ERROR [step 'layout']: requires 'cut_graph' which is not yet provided.
+  Hint: add compute_cut_graph (requires: mesh_closed) before 'layout'.
+
+Pipeline rejected. 2 contract violations.
+```
+
+---
+
+## 6 — Mapping to C++ code
+
+Each `unit:` name in the YAML maps directly to a C++ function:
+
+| YAML unit | C++ function | Header |
+|---|---|---|
+| `setup_euclidean_maps` | `conformallab::setup_euclidean_maps()` | `euclidean_functional.hpp` |
+| `compute_euclidean_lambda0_from_mesh` | `conformallab::compute_euclidean_lambda0_from_mesh()` | `euclidean_functional.hpp` |
+| `enforce_gauss_bonnet` | `conformallab::enforce_gauss_bonnet()` | `gauss_bonnet.hpp` |
+| `assign_dof_indices_euclidean` | manual pin + sequential loop | `euclidean_functional.hpp` |
+| `assign_all_dof_indices` | `conformallab::assign_all_dof_indices()` | `hyper_ideal_functional.hpp` |
+| `newton_euclidean` | `conformallab::newton_euclidean()` | `newton_solver.hpp` |
+| `newton_spherical` | `conformallab::newton_spherical()` | `newton_solver.hpp` |
+| `newton_hyper_ideal` | `conformallab::newton_hyper_ideal()` | `newton_solver.hpp` |
+| `compute_cut_graph` | `conformallab::compute_cut_graph()` | `cut_graph.hpp` |
+| `euclidean_layout` | `conformallab::euclidean_layout()` | `layout.hpp` |
+| `spherical_layout` | `conformallab::spherical_layout()` | `layout.hpp` |
+| `hyper_ideal_layout` | `conformallab::hyper_ideal_layout()` | `layout.hpp` |
+| `compute_period_matrix` | `conformallab::compute_period_matrix()` | `period_matrix.hpp` |
+| `compute_fundamental_domain` | `conformallab::compute_fundamental_domain()` | `fundamental_domain.hpp` |
+| `tiling_neighbourhood` | `conformallab::tiling_neighbourhood()` | `fundamental_domain.hpp` |
+| `save_layout_off` | `conformallab::save_layout_off()` | `mesh_io.hpp` |
+| `save_result_json` | `conformallab::save_result_json()` | `serialization.hpp` |
+
+---
+
+## 7 — Implementation plan (Phase 8e)
+
+```
+code/include/pipeline.hpp          ← YAML parser + validator + executor
+code/apps/conformallab_pipeline.cpp ← CLI: conformallab_pipeline --pipeline foo.yml
+
+External YAML dependency (header-only, already bundled):
+  code/deps/single_includes/yaml-cpp/yaml.h  ← or single-include yaml.hpp
+  Alternative: use the bundled single_includes/nlohmann/json.hpp for a
+               JSON-based pipeline format (simpler, no new dependency).
+```
+
+### Validator pseudocode
+
+```cpp
+struct PipelineValidator {
+    std::set<std::string> provided;
+
+    void load_input(const YAML::Node& input) {
+        provided.insert("mesh");
+        auto mesh = load_mesh(input["source"].as<std::string>());
+        if (is_closed(mesh)) provided.insert("mesh_closed");
+        else                 provided.insert("mesh_open");
+    }
+
+    void validate_step(const YAML::Node& step) {
+        for (auto& tok : step["require"])
+            if (!provided.count(tok.as<std::string>()))
+                throw PipelineError("Step '" + step["id"].as<std::string>()
+                    + "' requires '" + tok.as<std::string>() + "' not yet provided.");
+        for (auto& tok : step["provide"])
+            provided.insert(tok.as<std::string>());
+    }
+
+    void validate(const YAML::Node& pipeline) {
+        load_input(pipeline["input"]);
+        for (auto& step : pipeline["steps"])
+            validate_step(step);
+    }
+};
+```
+
+---
+
+## 8 — Design decisions
+
+**Why YAML and not JSON or TOML?**
+YAML supports multi-line strings (for `description:`), comments (`#`), and
+anchors/aliases — useful for parametric experiments.  JSON lacks comments.
+TOML lacks the list syntax needed for `require:` / `provide:`.
+
+**Why explicit require/provide instead of auto-inference?**
+Auto-inference would require the validator to know all function signatures
+at parse time — this couples the validator tightly to the C++ code.
+Explicit tokens make the contract visible in the YAML, making it
+self-documenting and readable without the source code.
+
+**Why sequential steps and not a DAG?**
+The pipeline is a linear sequence for now (Phase 8e target).  A DAG-based
+executor (parallel steps where contracts allow) is a natural Phase 10+ extension
+but adds significant complexity.  Linear execution is correct and debuggable.
+
+**Why one YAML per geometry mode?**
+A single YAML could support multiple geometries with conditional blocks, but
+this adds syntactic complexity.  The `geometry:` field at the top locks the mode
+and keeps the YAML readable.  Cross-geometry experiments can chain two pipelines.

doc/contributing.md

@@ -0,0 +1,86 @@
+# Contributing
+
+## Language
+
+**All code, comments, documentation, commit messages, and test descriptions must be in English.**
+
+The project targets CGAL submission and international collaboration. When editing
+files that still contain German-language comments or documentation, replace them
+with English.
+
+---
+
+## Git workflow
+
+- `main` is protected on `origin` (Gitea). Push to `dev`, then open a pull request.
+- `codeberg/main` can be pushed to directly (public mirror, no CI).
+- Both remotes must stay in sync after every significant change:
+  ```bash
+  git push origin HEAD:dev   # triggers CI
+  git push codeberg main     # updates public mirror
+  ```
+- Branch naming: `feature/<topic>`, `fix/<topic>`, `phase<N>-<topic>`
+
+---
+
+## CI
+
+Two jobs run on push to `dev`/`main` or on pull requests:
+
+| Job | What it tests | Trigger |
+|---|---|---|
+| `test-fast` | 23 non-CGAL tests, no Boost | all branches |
+| `test-cgal` | 227 CGAL tests, 0 skipped | `main`, `dev`, PRs only |
+
+A PR is ready to merge when both jobs pass.
+
+The runner is a self-hosted Raspberry Pi (ARM64). See `.gitea/workflows/cpp-tests.yml`
+and `.gitea/docker/Dockerfile.ci-cpp`.
+
+---
+
+## Test standards
+
+Every new algorithm needs:
+
+1. **Gradient check** — finite-difference verification of `G(x) = ∂E/∂x`.
+   Copy any `GradientCheck_*` test suite from `tests/cgal/test_*_functional.cpp`.
+
+2. **Convergence test** — Newton converges on a small mesh using the "natural theta"
+   trick (see [CLAUDE.md](../CLAUDE.md) and [api/extending.md](api/extending.md)).
+
+3. **Registration** — add the `.cpp` file to `code/tests/cgal/CMakeLists.txt`.
+
+Expected CI result: **23 + 227 tests pass, 0 skipped**.
+
+---
+
+## Code style
+
+- C++17. Header-only (`code/include/*.hpp`). No compiled library.
+- `#pragma once` at the top of every header.
+- Everything in the `conformallab` namespace, internal helpers in `conformallab::detail`.
+- Property map names follow the prefix convention: `"v:"` (vertex), `"e:"` (edge), `"f:"` (face).
+- DOF index `-1` always means "pinned/fixed".
+
+---
+
+## Releases
+
+Release tags follow `vMAJOR.MINOR.PATCH`:
+
+```bash
+# Merge dev → main, then tag
+git checkout main && git merge --no-ff dev
+git tag -a vX.Y.Z -m "vX.Y.Z — <one-line summary>"
+git push origin main && git push origin vX.Y.Z
+git push codeberg main && git push codeberg vX.Y.Z
+```
+
+---
+
+## TODO
+
+- [ ] Define code review checklist
+- [ ] Add `.clang-format` configuration
+- [ ] Document how to request CGAL package review

doc/getting-started.md

@@ -0,0 +1,189 @@
+# Getting Started
+
+## Prerequisites
+
+| Tool | Minimum | Notes |
+|---|---|---|
+| C++ compiler (GCC or Clang) | C++17 | GCC 11+ or Clang 14+ recommended |
+| CMake | 3.20 | |
+| Boost headers | 1.70 | Only for `-DWITH_CGAL_TESTS=ON` or `-DWITH_CGAL=ON` |
+| Wayland/X11 dev headers | — | Only for `-DWITH_CGAL=ON` (viewer) |
+
+All other dependencies (Eigen 3.4, CGAL 6.1.1, libigl 2.6, GLFW 3.4, GTest 1.14)
+are bundled as tarballs in `code/deps/tarballs/` and extracted automatically at CMake
+configure time. No internet access is required at build time (except GTest, fetched via
+`FetchContent` from GitHub).
+
+Install Boost on your system:
+```bash
+# Ubuntu / Debian
+apt install libboost-dev
+
+# macOS
+brew install boost
+```
+
+---
+
+## Clone
+
+```bash
+git clone https://codeberg.org/TMoussa/ConformalLabpp
+cd ConformalLabpp
+```
+
+---
+
+## Build modes
+
+Three modes with increasing dependencies:
+
+### Mode 1 — Fast tests (default, no system dependencies)
+
+Pure-math tests: Clausen functions, hyper-ideal geometry, matrix utilities.
+
+```bash
+cmake -S code -B build
+cmake --build build --target conformallab_tests -j$(nproc)
+ctest --test-dir build --output-on-failure
+```
+
+Expected: **23 tests pass**.
+
+### Mode 2 — CGAL tests, headless (recommended for CI and development)
+
+Full CGAL test suite. Requires Boost headers only — no display, no Wayland, no GLFW.
+
+```bash
+cmake -S code -B build -DWITH_CGAL_TESTS=ON
+cmake --build build --target conformallab_cgal_tests -j$(nproc)
+ctest --test-dir build -R "^cgal\." --output-on-failure
+```
+
+Expected: **227 tests pass, 0 skipped**.
+
+### Mode 3 — Full local build (CLI app + interactive viewer)
+
+Requires Wayland or X11 development packages (`wayland-scanner`, `libx11-dev`, etc.).
+Automatically enables the viewer library (GLFW + libigl).
+
+```bash
+cmake -S code -B build -DWITH_CGAL=ON
+cmake --build build -j$(nproc)
+```
+
+> **Note:** `-DWITH_CGAL=ON` implies `-DWITH_VIEWER=ON`. Do not use this in headless
+> environments — it will fail with `Failed to find wayland-scanner`.
+
+---
+
+## Running a single test
+
+```bash
+# By GTest filter (fastest, full output)
+./build/conformallab_cgal_tests --gtest_filter="NewtonSolver*"
+./build/conformallab_tests       --gtest_filter="Clausen*"
+
+# By CTest regex
+ctest --test-dir build -R "cgal.NewtonSolver" --output-on-failure
+```
+
+All CGAL tests have the prefix `cgal.` in CTest (set in `tests/cgal/CMakeLists.txt`
+via `TEST_PREFIX "cgal."`).
+
+---
+
+## First run — CLI app
+
+After a full build (`-DWITH_CGAL=ON`):
+
+```bash
+# Euclidean conformal layout
+./bin/conformallab_core -i input.off -g euclidean -o layout.off -j result.json
+
+# Spherical layout
+./bin/conformallab_core -i input.off -g spherical -o sphere.off
+
+# Hyperbolic layout (HyperIdeal)
+./bin/conformallab_core -i input.off -g hyper_ideal -o hyperbolic.off
+
+# Show input mesh in interactive viewer
+./bin/conformallab_core -i input.off -s
+
+# All options
+./bin/conformallab_core --help
+```
+
+## Example programs
+
+```bash
+./build/examples/example_layout    [input.off] [layout.off] [result.json]
+./build/examples/example_euclidean [input.off] [output.off]
+./build/examples/example_hyper_ideal [input.off] [output.off]
+./build/examples/example_viewer    [input.off]   # interactive, requires WITH_VIEWER
+```
+
+`example_layout.cpp` is the best starting point — it shows the complete pipeline in ~120 lines.
+
+**Expected output of `example_euclidean` on the built-in quad-strip mesh:**
+```
+[example_euclidean] No input file given — using make_quad_strip().
+[example_euclidean] Mesh: 6 vertices, 4 faces.
+[example_euclidean] DOFs: 5 (1 vertex pinned).
+[example_euclidean] Solving Newton system…
+[example_euclidean] Converged in 1 iterations.  ||G||_inf = 0
+[example_euclidean] Per-vertex conformal factors u_i:
+  v0  u = 0  (pinned)
+  v1  u = -1.38778e-17       ← ≈ 0 (machine epsilon)
+  ...
+[example_euclidean] Mesh saved to: /tmp/conformallab_euclidean_out.off
+```
+Convergence in 1 iteration is expected: the "natural equilibrium" construction
+sets x* = 0, so a small perturbation (-0.05) needs only one Newton step.
+
+**Quick automated start (no interactive build needed):**
+```bash
+bash scripts/try_it.sh
+```
+This clones nothing (run from inside the repo), builds the CGAL test suite, runs
+all 227+23 tests, and prints a summary.  See `scripts/try_it.sh` for details.
+
+---
+
+## Known issues
+
+### macOS Finder duplicates
+
+macOS Finder sometimes creates duplicate files named `foo 2.hpp` when copying
+the repository.  These cause confusing compile errors ("redefinition of …").
+
+**Fix (run once from the repo root):**
+```bash
+find code/include -name "* 2.*" -delete
+find code/include -name "*\ 2.*" -delete
+```
+
+Files without a ` 2` suffix are always canonical — the duplicates are safe to delete.
+
+### First build is slow
+
+The first CMake configure extracts four tarballs (Eigen 3.4, CGAL 6.1.1,
+libigl 2.6, GLFW 3.4) and downloads GTest via `FetchContent`.
+Allow **30–90 seconds** for the first configure.  Subsequent builds are fast
+(< 10 s incremental).
+
+---
+
+## Rebuilding the CI Docker image
+
+The CI runner is a self-hosted Raspberry Pi (ARM64). After changes to
+`.gitea/docker/Dockerfile.ci-cpp`:
+
+```bash
+docker buildx build \
+  --platform linux/arm64 \
+  -f .gitea/docker/Dockerfile.ci-cpp \
+  -t git.eulernest.eu/conformallab/ci-cpp:latest \
+  --push \
+  .gitea/docker/
+```