Age | Commit message (Collapse) | Author | |
---|---|---|---|
2020-05-07 | doc | Marc Glisse | |
2020-05-06 | Reimplement the bottleneck python wrapper with pybind11 | Marc Glisse | |
2020-05-06 | Refactor the numpy -> C++ range conversion | Marc Glisse | |
If we want to reuse it for bottleneck... | |||
2020-05-06 | Avoid reading outside of allocated region | Marc Glisse | |
The result was unused, but better be safe. | |||
2020-05-03 | Fewer copies and no GIL for hera | Marc Glisse | |
Now the input arrays are not copied as long as they use a float64 data type, even if they are not contiguous. That's not important here, but I wanted an example of how to do it. More importantly, no need to hold the GIL. I was too lazy to benchmark to see if that changed anything... | |||
2020-04-30 | Merge pull request #285 from mglisse/wass-autodiff | Marc Glisse | |
Automatic differentiation for Wasserstein distance | |||
2020-04-29 | Merge remote-tracking branch 'origin/master' into gen2 | Marc Glisse | |
2020-04-25 | Merge pull request #292 from VincentRouvreau/one_biblio_for_all_sphinx | Vincent Rouvreau | |
Fix bibliography for sphinx | |||
2020-04-24 | Code review: plain instead of unsrt for biblio - concatenate biblio files - ↵ | ROUVREAU Vincent | |
undo cgal citation removal | |||
2020-04-24 | Fix bibliography for sphinx | ROUVREAU Vincent | |
2020-04-23 | Merge remote-tracking branch 'origin/master' into compute_persistence | Marc Glisse | |
2020-04-23 | Fix pasto in the doc | Marc Glisse | |
2020-04-22 | Reduce the probability of failure of test_dtm | Marc Glisse | |
It is expected that hnsw sometimes misses one neighbor, which has an impact on the DTM, especially if the number of neighbors considered is low. | |||
2020-04-22 | Document incompatibility of matching=True and enable_autodiff | Marc Glisse | |
2020-04-22 | Merge remote-tracking branch 'origin/master' into wass-autodiff | Marc Glisse | |
2020-04-22 | Clarify the doc of enable_autodiff | Marc Glisse | |
2020-04-22 | Test value of computed gradient | Marc Glisse | |
2020-04-22 | Rename local variables | Marc Glisse | |
2020-04-22 | Document several optional dependencies of knn | Marc Glisse | |
2020-04-20 | Fix #279 | ROUVREAU Vincent | |
2020-04-20 | Drop one comment | Marc Glisse | |
2020-04-20 | Check that dependencies are present before testing | Marc Glisse | |
2020-04-20 | Merge remote-tracking branch 'origin/master' into wass-autodiff | Marc Glisse | |
2020-04-20 | Merge pull request #265 from mglisse/dtm | Marc Glisse | |
DTM | |||
2020-04-20 | Explicitly pass sort_results=True on some tests | Marc Glisse | |
2020-04-20 | Reintroduce _proj_on_diag, with a unit test | Marc Glisse | |
2020-04-20 | Add __license__ | Marc Glisse | |
2020-04-19 | Test comparison with persistence_pairs() | Marc Glisse | |
2020-04-19 | Drop redundant test | Marc Glisse | |
torch.isnan(None) raises an exception anyway | |||
2020-04-19 | Test gradient | Marc Glisse | |
2020-04-19 | Unwrap the result | Marc Glisse | |
2020-04-18 | enable_autodiff for POT wasserstein_distance | Marc Glisse | |
2020-04-17 | Simplify distance-to-diagonal in Wasserstein | Marc Glisse | |
2020-04-17 | Merge remote-tracking branch 'origin/master' into filtration | Marc Glisse | |
2020-04-14 | Check that the gradient is not NaN | Marc Glisse | |
This can easily happen with pytorch, and there is special code to avoid it. | |||
2020-04-14 | Merge remote-tracking branch 'origin/master' into dtm | Marc Glisse | |
2020-04-14 | Doc improvements | Marc Glisse | |
2020-04-14 | EagerPy in cmake | Marc Glisse | |
2020-04-14 | More testing | Marc Glisse | |
2020-04-13 | Tweak to detect fit_transform | Marc Glisse | |
2020-04-13 | Remove left-over code | Marc Glisse | |
eagerpy is only used with enable_autodiff | |||
2020-04-13 | Generalize enable_autodiff to more implementations | Marc Glisse | |
Still limited to L^p | |||
2020-04-13 | Fix NaN gradient with pytorch | Marc Glisse | |
2020-04-13 | Small autodiff tweaks | Marc Glisse | |
2020-04-13 | enable_autodiff with keops | Marc Glisse | |
This doesn't seem like the best way to handle it, we may want to handle it like a wrapper that gets the indices from knn (whatever backend) and then computes the distances. | |||
2020-04-12 | Parallelize the "precomputed" case of knn | Marc Glisse | |
It is supposed to be possible to compile numpy with openmp, but it looks like it isn't done in any of the usual packages. It may be possible to refactor that code so there is less redundancy. | |||
2020-04-11 | Use longer names | Marc Glisse | |
2020-04-07 | Document dependencies for building the doc | Marc Glisse | |
2020-04-06 | Remove trailing whitespace | Marc Glisse | |
2020-04-06 | Merge remote-tracking branch 'upstream/master' into wbary-theo | Marc Glisse | |