Age | Commit message (Collapse) | Author |
|
Remove exception raise. Fix failed tests. Reword documentation
|
|
|
|
Link the doc for subsampling & point file readers, some reorg
|
|
|
|
I copy-pasted what is done in RipsComplex. Before that, sphinx was
grouping nb_points with off_file, as if it couldn't be used with points.
|
|
|
|
|
|
Wdist
|
|
Co-Authored-By: Marc Glisse <marc.glisse@inria.fr>
|
|
Fix #144 : Reorder import and docstrings. Modify test_representations
|
|
|
|
|
|
|
|
|
|
but don't actually use annotations. A quick try with
sphinx-autodoc-typehints failed, I didn't investigate much, it can wait.
|
|
tests files and doc
|
|
Fix #158 - Add numpy ndarray to init a cubical and a periodic one
|
|
|
|
|
|
|
|
|
|
list
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
complex. Add boolean exact version mechanism
|
|
and some minor doc changes along the way.
(why were we documenting a hasse diagram that doesn't exist?)
|
|
Persistence graphical tool fix
|
|
|
|
Minor tweaks to representations
|
|
|
|
Fix #146.
|
|
wording for the case where CGAL is missing
comparing pointers is sufficient to detect if we are doing fit_transform
epsilon is an additive error for bottleneck, so it is unsafe to use 1e-3 or 1e-4 as default
|
|
uniform -> even
matmul -> indexing
min+max -> clip
|
|
|
|
|
|
sklearn-tda
|
|
Apparently some users expect that it will magically "fix" other
simplices so it remains a valid filtration.
|
|
I think this interacts better with sphinx automodule
|
|
|
|
|
|
|
|
|
|
|