Repository navigation
What's Changed
- [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci[bot] in #952
- fix: support negative offsets, rectangular shapes, and negative axes in sparse.diagonal by @rautaditya2606 in #955
- [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci[bot] in #953
- [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci[bot] in #956
- fix: avoid dense column selectors in GCXS basic slicing by @Doribelove in #957
- fix(io): support CSR, CSC, and DOK formats in save_npz and load_npz by @rautaditya2606 in #960
- DOC: name the array parameter x in expand_dims and flip by @hxperl in #958
- [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci[bot] in #961
- fix: preserve zero-dimensional arrays in sparse conversions by @Doribelove in #963
- fix: GCXS matmul with zero output columns by @raashish1601 in #972
- fix: return the identity for all/any over an empty axis by @raashish1601 in #971
- fix: honour dtype in asarray for sparse inputs by @raashish1601 in #975
- fix: sort unique_counts output when the fill value is inserted by @raashish1601 in #974
- fix: keep the product dtype in 1-D dot and matmul by @raashish1601 in #973
- docs: explain GCXS on the front page by @Doribelove in #968
- docs: clarify COO idx_dtype capacity for from_numpy by @Doribelove in #967
- feat: support Fortran order in COO.flatten by @Doribelove in #964
- fix: normalize negative axis in argmax and argmin by @Rodrigo-Palma in #966
- fix: accept GCXS input in triu, tril and diagonal by @raashish1601 in #978
- fix: sort NaNs to the end in sparse.sort by @raashish1601 in #979
- fix: allow any and all on arrays with a non-boolean fill value by @raashish1601 in #980
- fix: return the first NaN in argmax and argmin by @raashish1601 in #976
- fix: support ufuncs with several outputs such as divmod, modf and frexp by @raashish1601 in #982
- fix: broadcast array bounds in clip by @raashish1601 in #981
- fix: skip complex NaNs in nansum and nanprod by @CAOShurong in #988
- docs: explain fill-value propagation in where by @Doribelove in #987
- fix: index GCXS arrays pointwise with several index arrays by @raashish1601 in #985
- fix: move the advanced-index axis to the front like NumPy when advanced indices are separated by @raashish1601 in #984
- fix: put new axes in the right place when indexing GCXS after an integer index by @raashish1601 in #983
- Add validation and clean up
compressed_axes=Noneambiguity inGCXS. by @hameerabbasi in #1015
New Contributors
- @rautaditya2606 made their first contribution in #955
- @Doribelove made their first contribution in #957
- @hxperl made their first contribution in #958
- @raashish1601 made their first contribution in #972
- @Rodrigo-Palma made their first contribution in #966
- @CAOShurong made their first contribution in #988
Full Changelog: 0.19.2...0.19.3