Skip to content

Fix multi-source heat-method geodesic distances (Dirichlet-pinned Poisson solve) #73

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
petrasvestartas merged 3 commits into from
Jul 22, 2026
Merged

Fix multi-source heat-method geodesic distances (Dirichlet-pinned Poisson solve) #73

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
8e0a359
fix(geodesics): correct multi-source heat distances via Dirichlet-pin…
jf--- Jul 10, 2026
e6889ed
feat(geodesics): accept compas Mesh or (V,F) via @overload
jf--- Jul 22, 2026
99f3de8
docs(geodesics): regenerate example image on fixed multi-source backend
jf--- Jul 22, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,10 +14,13 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
* Added image reference in `docs/examples/example_isolines.md` and added `Measure: examples/example_measure.md` to the Examples nav in `mkdocs.yml`.
* Added `docs/examples/example_booleans_with_edges.md` and `example_booleans_with_face_source.md` (with their images) covering the corefinement intersection-edge and face-source-tracking variants of the boolean operations; both wired into the Examples nav.
* Added `docs/examples/example_boolean_difference_mesh_meshes.md` (CSG drilled rounded cube) and its Examples nav entry.
* Added `data/flat_star_disk_irregular_rim.off` — a flat synthetic fixture whose irregular boundary sampling exposes multi-source heat-method normalization errors against a dependency-free euclidean truth (on a flat mesh the euclidean distance to the nearest source is exact in the near band and a hard lower bound everywhere), plus the regression test `tests/test_geodesics.py::test_heat_geodesic_multisource_boundary_sources` that fails on the previous backend.
* `heat_geodesic_distances`, `HeatGeodesicSolver`, `geodesic_isolines`, and `geodesic_isolines_split` now accept a `compas.datastructures.Mesh` directly, in addition to the `(vertices, faces)` tuple — both input shapes are declared via `@overload` and coerced through one internal `_as_vertices_faces` seam.

### Fixed

* Fixed `docs/examples/example_projection.py` imports: `project_mesh_on_mesh`, `project_points_on_mesh`, `pull_mesh_on_mesh`, and `pull_points_on_mesh` were incorrectly imported from `compas_cgal.meshing` instead of `compas_cgal.projection`, causing `ImportError` for anyone running the example.
* Fixed multi-source geodesic distances collapsing near the source set: `heat_geodesic_distances`, `HeatGeodesicSolver`, `geodesic_isolines`, and `geodesic_isolines_split` now use an own Eigen implementation of the heat method (Crane et al. 2017) with a Dirichlet-constrained Poisson step instead of CGAL's `Heat_method_3`, whose `value_at_source_set` normalizes the Poisson potential through `min_s |phi(i) - phi(s)|` — a distance between *values* that folds every vertex whose potential lands inside the source-value spread (seeding all 508 boundary vertices of the new fixture read 0.54x the true distance in the near band, with 83% of band vertices below the euclidean lower bound). Distances are now exactly 0 at sources, non-negative, and accurate for arbitrary source sets; single-source results are unchanged within discretization error. Empty source sets now raise `ValueError` and meshes with a connected component containing no source raise `RuntimeError` (both previously returned meaningless values). The docstring claim that the intrinsic-Delaunay variant was used has been dropped (the binding instantiated the class default, `Direct`).

### Changed

Expand Down
Loading
Loading