Age | Commit message (Collapse) | Author |
|
|
|
|
|
Add build and tests wo cgal and eigen and wo cgal
|
|
|
|
|
|
and roll back the other solution
|
|
Pass a distance function instead of a kernel to choose_n_farthest_points
|
|
Clean-ups in subsampling
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Bottleneck distance for one point or negative coordinates
|
|
|
|
It would require updating the figure.
|
|
|
|
and remove example installation
|
|
+ black reformatting
|
|
This allows correct handling of infinite values.
|
|
suggested by @raphaeltinarrage
|
|
Tests are without matplotlib, so no need to test if present
|
|
|
|
|
|
|
|
It would be possible to emit the duplicate points instead of stopping,
but the current implementation makes that inconvenient.
|
|
|
|
|
|
Added tests for wasserstein distance with tensorflow
|
|
test_wasserstein_distance.py to test_wasserstein_with_tensors.py
|
|
A proposal for simplex_tree reset_filtration (python & C++)
|
|
|
|
|
|
|
|
targets
|
|
Documents for weighted Rips
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Exception is raised when not found. tested
|