delta/python-packages/numpy.git/numpy/random, branch main github.com: numpy/numpy.git MNT: compatibility with cython3 2023-05-14T01:33:45+00:00 Thomas A Caswell tcaswell@gmail.com 2023-05-14T01:33:45+00:00 888fd7719965719321f160f79051aa5caf42b9ac This is fallout from https://github.com/cython/cython/pull/5386 
This is fallout from https://github.com/cython/cython/pull/5386
Also annotate the implementation of the kahan_sum() function, not just the declaration. 2023-05-04T07:35:21+00:00 Stefan Behnel stefan_ml@behnel.de 2023-05-04T07:35:21+00:00 6aca5f6552225a097cb23c875f2b18de86ca7f39 






Revert useless changes.
2023-05-04T07:33:20+00:00

Stefan Behnel
stefan_ml@behnel.de

2023-05-04T07:33:20+00:00

4ff6f470cf53a77a04dbec7f333a844ac3474a0c












Merge branch 'main' into cython3_noexcept
2023-05-04T07:29:53+00:00

scoder
stefan_ml@behnel.de

2023-05-04T07:29:53+00:00

442c8f48d3146ec32c7d5387310e171276cf10ac












Add "noexcept" markers to functions that do not raise exceptions.
2023-05-04T07:18:16+00:00

Stefan Behnel
stefan_ml@behnel.de

2023-05-04T07:18:16+00:00

3e4a6cba2da27bbe2a6e12c163238e503c9f6a07












MAINT: random: In a test module, don't use a global non-deterministically Generator.
2023-05-01T17:11:58+00:00

warren
warren.weckesser@gmail.com

2023-05-01T17:11:58+00:00

5d51fd9d38ada80cf35c33388d2753af69ba1538












BUG: random: Don't return negative values from Generator.geometric.
2023-05-01T03:28:44+00:00

warren
warren.weckesser@gmail.com

2023-04-27T15:52:47+00:00

40ae83411b275b68425b5c85a7243c69599edc59

In C, when the integer part of a double exceeds the maximum int64, casting
the double to int64_t is undefined behavior.  On some platforms, the result
is 2**64-1 and on others it is -2**64.  That results in rng.geometric()
returning -2**64 when p was sufficiently small on some platforms.

The fix is to never invoke undefined behavior by ensuring we don't attempt
to cast very large double values to int64_t.





In C, when the integer part of a double exceeds the maximum int64, casting
the double to int64_t is undefined behavior.  On some platforms, the result
is 2**64-1 and on others it is -2**64.  That results in rng.geometric()
returning -2**64 when p was sufficiently small on some platforms.

The fix is to never invoke undefined behavior by ensuring we don't attempt
to cast very large double values to int64_t.







BUG: accept zeros on numpy.random dirichlet function (#23440)
2023-04-11T07:38:03+00:00

Paulo Almeida
pcca24102002@gmail.com

2023-04-11T07:38:03+00:00

fe893743b83173b72648dc79d86eda905230d1e9

Changed alpha value error to pass a null value. This way, dirichlet function (on the generator, not mtrand) won't raise a value exception at 0. Also added test.






Changed alpha value error to pass a null value. This way, dirichlet function (on the generator, not mtrand) won't raise a value exception at 0. Also added test.








DOC: Add `n_children` param to rng.spawn() and bitgen.spawn() docs
2023-03-14T08:11:06+00:00

Sebastian Berg
sebastianb@nvidia.com

2023-03-13T11:34:29+00:00

94a9b8641bf4d1d5e5c4387df79341d4f0817f1b

The original PR forgot to include the Parameters section (and thus
the paraeter itself).





The original PR forgot to include the Parameters section (and thus
the paraeter itself).







Merge pull request #23103 from SimonAltrogge/typing-fix
2023-03-05T19:17:47+00:00

Bas van Beek
43369155+BvB93@users.noreply.github.com

2023-03-05T19:17:47+00:00

e8226132c6f8f8452f412b62858fd9128378c01f

TYP: Fix return type to float on _FloatLike_co arguments




TYP: Fix return type to float on _FloatLike_co arguments