Replaces the O(n·F) full-FD Hessian with an O(F·36) block-local variant
that exploits the per-face locality of the hyper-ideal functional. Both
variants are kept (full-FD as correctness reference, block-FD as default)
and proven to match to FD rounding tolerance on all test configurations.
Java parity note
────────────────
HyperIdealFunctional.java line 295-298 declares:
public boolean hasHessian() { return false; }
i.e. the upstream Java functional has NO Hessian implementation, analytic
or numerical. Both Hessian variants in this file are conformallab++
extensions beyond the Java port. Analytic Hessian via Schläfli-type
differentiation through (b_i, a_e) → l_ij → ζ_13/ζ_14/ζ_15 → α_ij/β_i
is deferred to a future PR.
Implementation
──────────────
* code/include/hyper_ideal_functional.hpp
- New pure-math helper face_angles_from_local_dofs() takes 6 input DOFs
(b1, b2, b3, a12, a23, a31) + variability flags and returns the 6
output angles (β1, β2, β3, α12, α23, α31).
- Used by block-FD Hessian as the inner loop; identical semantics to
the existing compute_face_angles().
* code/include/hyper_ideal_hessian.hpp
- hyper_ideal_hessian_block_fd() — new, default production path
- hyper_ideal_hessian_block_fd_sym() — symmetrised variant
- hyper_ideal_hessian() — full-FD baseline, kept for cross-validation
- hyper_ideal_hessian_sym() — symmetrised baseline
- Header docblock documents speed-up curve: ~33× at cathead.obj scale,
~1166× at brezel.obj scale.
Tests (7 new in test_hyper_ideal_hessian.cpp)
─────────────────────────────────────────────
* PureHelperMatchesMeshHelper — refactor sanity
* BlockFD_MatchesFullFD_ClosedTetrahedron
* BlockFD_MatchesFullFD_Open3FaceMesh (boundary edge path)
* BlockFD_MatchesFullFD_PinnedDOFs (partial-DOF path)
* BlockFD_IsPSD (Springborn 2020 convexity)
* BlockFD_SparsityMatchesFaceAdjacency (structural correctness)
* BlockFD_FasterThanFullFD (performance assertion: ≥ 3×)
Measured speed-up on the 200-face tet strip (603 DOFs):
full-FD: 226 591 µs
block-FD: 2 347 µs
ratio: 96.5×
The assertion uses ≥ 3× to leave wide CI-hardware tolerance.
Test count
──────────
CGAL suite: 184 → 191 (+7). Zero skips.
Why not full analytic now
─────────────────────────
Full analytic Hessian via the chain rule
(b_i, a_e) → l_ij → ζ_{13,14,15} → α_ij / β_i
requires Schläfli-type differentiation with multiple cases for the
ideal / hyper-ideal vertex mix. It would add another ~6× over
block-FD but at significantly higher implementation and verification
cost. Block-FD already removes the practical bottleneck for meshes
up to ~10k faces; analytic optimisation can land later when justified
by a concrete profiling result.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Removes the only architectural lock-in spotted in the MVP audit before
Phase 9a starts: the wrapper hard-coded Simple_cartesian<double> as the
kernel inside discrete_conformal_map_euclidean. This would have broken
any Surface_mesh<P> where P came from a different kernel (e.g. EPIC).
Change
──────
* CGAL::Kernel_traits<typename TriangleMesh::Point>::Kernel is now used
to deduce the kernel from the mesh's point type.
* The full Default_traits<...> instantiation is wrapped in
internal_np::Lookup_named_param_def so a future `geom_traits(...)`
named parameter can override the entire traits class without changes
to the wrapper body (CGAL idiom, used by every CGAL package).
* New test `KernelIsDeducedFromMeshPointType` pins the contract
explicitly with static_asserts.
Why now
───────
Phase 9a (Inversive-Distance) will copy this same template pattern.
Fixing the kernel deduction once here keeps the design free for any
user kernel; doing it after 9a would mean two parallel hard-coded
kernel sites to refactor.
Tests
─────
CGAL suite: 184/184 passed, 0 skipped (was 183).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
First step of the Phase 8 Hybrid MVP. Adds a thin CGAL-conformant public
API layer over the existing implementation, validated by 7 acceptance
tests. Total CGAL test count: 183 (was 176), 0 skipped.
New public headers
──────────────────
* code/include/CGAL/Conformal_map_traits.h
- ConformalMapTraits concept documentation
- Default_conformal_map_traits<Surface_mesh<P>, K> specialisation
- Static property-map accessors: vertex_points, theta_map,
vertex_index_map, lambda0_map
* code/include/CGAL/Discrete_conformal_map.h
- User-facing entry: discrete_conformal_map_euclidean(mesh, np)
- Conformal_map_result<FT> struct (u, iter, ‖G‖, converged flags)
- Natural-theta default: x = 0 is the equilibrium when no Θ supplied
- Honours user-provided Θ via vertex_curvature_map named parameter
* code/include/CGAL/Conformal_map/internal/parameters.h
- 4 named-parameter tags in CGAL::Conformal_map::internal_np:
vertex_curvature_map, gradient_tolerance,
max_iterations, fixed_vertex_map
- User-facing helpers in CGAL::parameters::*
Tests (test_cgal_traits_mvp.cpp, 7 cases)
─────────────────────────────────────────
* DefaultTraitsTypes: compile-time type sanity (static_assert)
* AccessorsReuseExistingMaps: traits accessors return identical pmaps
* SingleTriangleConverges,
QuadStripConverges: end-to-end Euclidean wrapper passes
* MaxIterationsTakesEffect: named parameter is read
* GradientToleranceTakesEffect: tolerance override changes Newton end-state
* WrapperMatchesLegacyAPI: cross-API result equality at 1e-10
Architecture
────────────
3-layer wrapper as designed (doc/api/cgal-package.md):
Layer 1: code/include/*.hpp (existing algorithms, unchanged)
Layer 2: CGAL/Conformal_map/internal/ (adapter, parameter tags)
Layer 3: CGAL/Conformal_map_traits.h, CGAL/Discrete_conformal_map.h
(user-facing)
No algorithm duplication. Existing 176 + 36 tests untouched.
Next: Phase 9a (Inversive-Distance) as the second client of this API —
the real acceptance test for the trait design.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Unified the codebase language to English throughout. German text appeared
in code comments, test file headers, CI step names, and several markdown
documents. All natural-language text is now English; proper nouns
(Institut für Mathematik, Technische Universität Berlin) are unchanged.
Files changed:
- .gitea/workflows/cpp-tests.yml — CI step names and job comments
- code/include/mesh_utils.hpp — inline comment
- code/tests/cgal/CMakeLists.txt — section comment block
- code/tests/cgal/test_geometry_utils.cpp — full file header + all test comments
- doc/math/references.md — geometry-central section
- doc/math/validation.md — Section 9 (geometry-central cross-validation)
- doc/roadmap/phases.md — Optional geometry-central track (GC-1/2/3)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replaces the Java CoHDS with CGAL::Surface_mesh<Point3> (Simple_cartesian
kernel). Adds domain-specific property maps for lambda/theta/idx/alpha and
face geometry type — the direct C++ equivalent of CoVertex/CoEdge adapters.
New files:
include/conformal_mesh.hpp — ConformalMesh type + property-map helpers
include/mesh_builder.hpp — mesh factories (triangle, tetrahedron,
quad-strip, fan) for tests and examples
tests/cgal/ — second test executable (conformallab_cgal_tests)
built only with -DWITH_CGAL=ON
Test results (local, -DWITH_CGAL=ON):
conformallab_tests: 36 registered | 23 passed | 13 skipped | 0 failed
conformallab_cgal_tests: 14 registered | 14 passed | 0 skipped | 0 failed
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Implements getEuclideanCircleFromHyperbolic() in C++ (hyperboloid model
→ Poincaré disk via Lorentz boost + circumcircle). Ports the 2 Java tests
that were the only remaining candidates not requiring HDS or a solver.
All other unported tests (DataTypesTest, SchottkyIOTest,
UniformizationDataTest, BranchedCoverTorusTest) are blocked by XML
serialization or HDS – stubs remain for Phase 4.
Test totals: 36 registered | 23 passed | 13 skipped | 0 failed.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Fully ported (pure math, no HDS required):
test_discrete_elliptic_utility.cpp – 2 tests
normalizeModulus: move tau into SL(2,Z) fundamental domain
test_p2_utility.cpp – 3 tests
P2 projective geometry (perpendicularBisector, pointFromLines,
makeDirectIsometryFromFrames double vs long double precision)
New headers:
include/discrete_elliptic_utility.hpp – normalizeModulus
include/p2_utility.hpp – P2 Euclidean geometry (templated
on scalar type so double and long double share one implementation)
Stubs (GTEST_SKIP, blocked until HDS port – Phase 4):
test_hyper_ideal_functional.cpp – 5 tests (1 @Ignore in Java)
test_hyper_ideal_hyperelliptic_utility.cpp – 3 tests
test_spherical_functional.cpp – 5 tests
All use CoHDS + HalfEdgeUtils which are not yet ported to C++.
Result: 34 tests total | 21 passed | 13 skipped | 0 failed
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- include/matrix_utility.hpp: 4x4 mapping matrix R·from=to (Eigen)
- include/projective_math.hpp: dehomogenize, hyperbolicDistance,
isOnSegment (collinearity + betweenness via 3D cross/dot),
getPointOnCorrespondingSegment (parameter by arc-length ratio)
- test_matrix_utility.cpp: port of MatrixUtilityTest (1 test)
- test_surface_curve_utility.cpp: port of SurfaceCurveUtilityTest
testIsBetween and testGetPointOnSegment_SegmentEdge (2 tests)
- tolerance adjusted to 1e-12 for matrix inversion (2.7e-15 rounding
from Eigen vs jReality's LU; both well within meaningful accuracy)
Total: 16/16 tests pass
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add Google Test via CMake FetchContent (v1.14.0)
- Add clausen.hpp: Clausen integral, Lobachevsky function, Im(Li2)
- Add hyper_ideal_utility.hpp: generalized and ideal-vertex hyperbolic
tetrahedron volume formulas using Eigen for the 4x4 Gram determinant
- Port ClausenTest (5 tests) and HyperIdealUtilityTest (8 tests) from
Java/JUnit — all 13 pass with same tolerances as the Java originals
- Fix pre-existing VIEWER/viewer case mismatch in CMakeLists.txt
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
