MAINT: stats.zmap: restore support for complex data #22405
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
mdhaber
added
scipy.stats
maintenance
j-bowhay
reviewed
Jan 26, 2025
mdhaber
commented
Jan 26, 2025
mdhaber
added
the
backport-candidate
tylerjereddy
pushed a commit
to tylerjereddy/scipy
that referenced
this pull request
Feb 5, 2025
* MAINT: stats.zmap: restore support for complex data * MAINT: stats.zmap: fix masked array support * Apply suggestions from code review
7 tasks
tylerjereddy
added a commit
that referenced
this pull request
Feb 16, 2025
* BUG: immortalize uarray global strings in order to prevent negative refcounts (#22038) #### Reference issue Fixes #21214 Closes #21218 #### What does this implement/fix? This PR addresses the segfaults caused by the occurrence of negative refcounts when uarray static strings were being released when the interpreter didn't exist. --------- Co-authored-by: peterbell10 <peterbell10@live.co.uk> * BUG: special: Fix unchecked memory allocations in `specfun.h` (#22080) * WIP: Fix unchecked memory allocation in aswfa() Summary of the changes: * Add another SF error code, "SF_ERROR_NOMEM", that is used by ufuncs that require internally allocated memory. Just like the other possible errors, how this error is handled can be queried and set using `geterr()` and `seterr()`. Unlike the other error conditions, the default behavior is to raise an exception if a memory allocation fails. * Modify the function `aswfa()` in `specfun.h` to return a status code that indicates when a mempory allocation failed. * Modify the functions `prolate_aswfa_nocv()`, `oblate_aswfa_nocv()`, `prolate_aswfa()` and `oblate_aswfa()` in `sphd_wave.h` to check the return value of `aswfa()`, and set the error state and return values appropriately if `aswfa()` returns `NoMemory`. * MAINT: Rename 'nomem' -> 'memory'. * Fix all uses of unchecked allocations in specfun.h * Fix aswfa() in specfunc.h * Fix refguide check. * Add comment about the return value when nonconvergence is detected in rmn2l. * Change (c/m)alloc calls to use 'new' wrapped in a unique_ptr. * Fix a few more names passed to set_error() * Set outputs to nan in mtu12 when alloc fails Also update docstring-like comments of some C++ functions. * BUG: cluster: `cophenet` intercept invalid linkage matrix count (#22187) * BUG: catch invalid linkage count * Fixes gh-22183. * The input data from gh-22183 causes an out of bounds memory access in a 1-D `memoryview` in the Cython `cophenetic_distances` function. Prevent this by enforcing a check for an allowable upper bound on the cluster membership (4th column) of the linkage matrix `Z` received by `is_valid_linkage`. Co-authored-by: Warren Weckesser <warren.weckesser@gmail.com> * TST: PR 22187 revisions * Adjust `test_gh_22183()` such that the linkage matrix is invalid only because of an excessive observation count, rather than also being invalid because the large observation count is not integral. * Adjust `test_gh_22183()` such that is properly flushes through array API backends. * DOC: backticks --------- Co-authored-by: Warren Weckesser <warren.weckesser@gmail.com> Co-authored-by: Jake Bowhay <60778417+j-bowhay@users.noreply.github.com> * MAINT: stats.Mixture: make return type consistent * MAINT: stats.Mixture: fix inverse functions when mean is undefined (#22337) * MAINT: stats.Mixture: fix inverse when mean is undefined * Apply suggestions from code review * Update scipy/stats/_distribution_infrastructure.py * BUG: special: Fix unchecked malloc in stirling2.h (#22339) Use std::unique_ptr and new (std::nothrow) instead of malloc. Closes gh-22336. * BUG: sparse: fix selecting wrong dtype for coo coords (#22353) * Fixes bug with selecting wrong dtype for coo coords * add test for reshape having wrong dtype * test that smaller dtype is maintained even if intermediate values are big --------- Co-authored-by: Dan Schult <dschult@colgate.edu> * TST: interpolate: small tolerance bump to TestAAA.test_basic_functions * MAINT: Stop installing libhighs Co-authored-by: drew-parsons <drew-parsons@users.noreply.github.com> Closes #22349 * BUG: sparse.linalg.norm: fix return type (#22372) cupy reports cupy/cupy#8866 that sparse.linalg.norm returns a python number instead of a 0-dim numpy array * revert NotImplemented return values in dot & others (#22373) * DOC: sparse.linalg: add two recent functions to namespace and fix doctests (#22374) * fix doctests and make two recent functions visible * update refguide [docs only] * fix capitalization [docs only] Co-authored-by: Daniel Schmitz <40656107+dschmitz89@users.noreply.github.com> --------- Co-authored-by: Daniel Schmitz <40656107+dschmitz89@users.noreply.github.com> * MAINT: stats.zmap: restore support for complex data (#22405) * MAINT: stats.zmap: restore support for complex data * MAINT: stats.zmap: fix masked array support * Apply suggestions from code review * BUG: scipy.spatial: Fix inaccurate orthonormalization in `Rotation.from_matrix()` (#22418) * BUG: fix inaccurate orthonormalization in Rotation.from_matrix() * Code review updates --------- Co-authored-by: Scott Shambaugh <scottshambaugh@users.noreply.github.com> * BUG: special: Fix a memory leak in the AMOS function besy(). (#22423) Use a unique_ptr to manage the heap-allocated array. Closes gh-22314. * CI: build Linux aarch64 wheels on GitHub Actions instead of Cirrus CI This unifies the wheels builds, which is a lot easier on maintenance. In the process, did the following: - update to setup-miniconda 3.1.1. for GHA Linux arm64 runner support - add `conda-remove-defaults` to `setup-miniconda` * CI: migrate Linux aarch64 job to GitHub Actions * CI: remove Cirrus CI configuration [wheel build] * BUG: io.loadmat: throw error on file containing all zeros (#22447) [ci skip] * BUG: PR 22471 revisions * Fixes issues related to backport manual merge conflict resolution during the preparation of SciPy `1.15.2`. * Restore an accidentally-removed size-0 array check in `_contains_nan()` function--this allows SciPy `stats` tests to pass again on the maintenance branch. * Remove some accidentally-added tests in `test_mio.py` [skip circle] * BUG: sparse: fix spmatrix indexing with None and implicit padding to match matrix behavior (#22472) * fix index using None when index is padded with : * align None indexing for spmatrix with np.matrix behavior. And tests. * rewrite spmatrix handling to ease reading * MAINT: pearsonr SIMD-related shim * Fixes #22479. * This small patch allows `test_stats.py::TestRegression::test_regressZEROX` to pass on x86_64 Linux instead of failing due to an extra division by zero warning over a fairly narrow range of NumPy versions near `1.25.2` where SIMD implementation details appear to have been slightly different. [skip circle] * DOC: PR 22471 revisions * Update the SciPy `1.15.2` release notes following backport activity. * BUG: wrap median_filter stability (#22402) * An uninitialized data structure in `_rank_filter()` C++ code was causing stochastic algorithm failures on Linux but not MacOS, so force a zero-based initialization. This allows `test_gh_22250` to pass over thousands of `pytest-repeat` incantations. * Add a regression test for gh-22333, which reliably passes on this branch with `pytest-repeat` (and fails on `main` with sufficient repeats). * MAINT: integrate.cumulative_simpson: bump test tolerance * do not check dtype in test_compare_with_GCVSPL * CI: PR 22471 revisions * temporarily disable testing against `meson` `master` branch in CI because of gh-22534. * DOC: PR 22471 revisions * Update the SciPy `1.15.2` release notes following additional backport activity. * CI: PR 22471 wheel builds [wheel build] * This is an empty commit to test the wheel build matrix in PR 22471. [wheel build] --------- Co-authored-by: Edgar Andrés Margffoy Tuay <andfoy@gmail.com> Co-authored-by: peterbell10 <peterbell10@live.co.uk> Co-authored-by: Warren Weckesser <warren.weckesser@gmail.com> Co-authored-by: Jake Bowhay <60778417+j-bowhay@users.noreply.github.com> Co-authored-by: Matt Haberland <mhaberla@calpoly.edu> Co-authored-by: Parth Nobel <parthnobel@berkeley.edu> Co-authored-by: Dan Schult <dschult@colgate.edu> Co-authored-by: Rohit Goswami <rog32@hi.is> Co-authored-by: Daniel Schmitz <40656107+dschmitz89@users.noreply.github.com> Co-authored-by: Scott Shambaugh <14363975+scottshambaugh@users.noreply.github.com> Co-authored-by: Scott Shambaugh <scottshambaugh@users.noreply.github.com> Co-authored-by: Ralf Gommers <ralf.gommers@gmail.com> Co-authored-by: Matthew Brett <matthew.brett@gmail.com> Co-authored-by: Charles Bousseau <cbousseau@anaconda.com>
tylerjereddy
removed
the
backport-candidate
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Reference issue
Closes gh-22404
What does this implement/fix?
gh-22404 notes that
stats.zmaplost support for complex data. This restores it.Additional information
While restoring it is the path of least resistance, this is part of a bigger issue: most stats functions don't work for complex data, but some stats functions have accidentally done the "right" thing for complex input in the past with documenting or testing support. Ideally, we'd document and test the behavior of all functions with complex input.