You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
scipy 1.9 introduces some problems in our codebase.
It looks to be related to np.vectorize in combination with scipy.linalg.expm.
An example coming from test running (with github actions):
tests/tests_geomstats/test_backends2.py:62: in <module>
class TestBackends(TestCase, metaclass=Parametrizer):
tests/conftest.py:[22](https://github.com/geomstats/geomstats/runs/7686708451?check_suite_focus=true#step:6:23)0: in __new__
test_data = _get_test_data(
tests/conftest.py:305: in _get_test_data
test_data_ = getattr(testing_data, data_fn_str)()
tests/data/backends_data.py:91: in unary_op_like_np_test_data
dict(func_name="linalg.cholesky", a=SPDMatrices(3).random_point()),
geomstats/geometry/spd_matrices.py:101: in random_point
spd_mat = GeneralLinear.exp(Matrices.to_symmetric(mat))
geomstats/geometry/lie_group.py:226: in exp
return expm(tangent_vec)
geomstats/_backend/numpy/linalg.py:27: in expm
return _np.vectorize(_scipy.linalg.expm, signature="(n,m)->(n,m)")(x)
/opt/hostedtoolcache/Python/3.8.13/x64/lib/python3.8/site-packages/numpy/lib/function_base.py:[23](https://github.com/geomstats/geomstats/runs/7686708451?check_suite_focus=true#step:6:24)28: in __call__
return self._vectorize_call(func=func, args=vargs)
/opt/hostedtoolcache/Python/3.8.13/x64/lib/python3.8/site-packages/numpy/lib/function_base.py:[24](https://github.com/geomstats/geomstats/runs/7686708451?check_suite_focus=true#step:6:25)02: in _vectorize_call
res = self._vectorize_call_with_signature(func, args)
/opt/hostedtoolcache/Python/3.8.13/x64/lib/python3.8/site-packages/numpy/lib/function_base.py:2442: in _vectorize_call_with_signature
results = func(*(arg[index] for arg in args))
/opt/hostedtoolcache/Python/3.8.13/x64/lib/python3.8/site-packages/scipy/linalg/_matfuncs.py:[33](https://github.com/geomstats/geomstats/runs/7686708451?check_suite_focus=true#step:6:34)5: in expm
lu = bandwidth(aw)
_cythonized_array_utils.pyx:104: in scipy.linalg._cythonized_array_utils.bandwidth
???
_cythonized_array_utils.pyx:114: in scipy.linalg._cythonized_array_utils.bandwidth_c
???
stringsource:660: in View.MemoryView.memoryview_cwrapper
???
stringsource:[35](https://github.com/geomstats/geomstats/runs/7686708451?check_suite_focus=true#step:6:36)0: in View.MemoryView.memoryview.__cinit__
???
scipy 1.9 introduces some problems in our codebase.
It looks to be related to
np.vectorize
in combination withscipy.linalg.expm
.An example coming from test running (with github actions):
see also #1619.
The text was updated successfully, but these errors were encountered: