c22ea96 | Maxim Kochurov | 30 March 2022, 09:44:36 UTC | Scope separator for netcdf (#5663) * add a failing test * fix a failing test with changes in model.py * better test * change docs in model.py * fix tests * use :: saparator * fix typo | 30 March 2022, 09:44:36 UTC |
ab93967 | Ricardo | 30 March 2022, 06:00:43 UTC | Bump version for b6 release | 30 March 2022, 09:25:46 UTC |
b536f91 | Ricardo | 30 March 2022, 05:51:39 UTC | Remove stale `test_sample_generate_values` sample_posterior_predictive no longer accepts non-integer samples | 30 March 2022, 07:36:59 UTC |
7aafe10 | Ricardo | 30 March 2022, 05:49:16 UTC | Update stale xfail tests | 30 March 2022, 07:36:59 UTC |
042a1c5 | Ricardo | 30 March 2022, 05:43:01 UTC | Add rng update test for size from observed | 30 March 2022, 07:36:59 UTC |
5392468 | Ricardo | 29 March 2022, 09:30:30 UTC | Update "traditional" rng default_update in `change_rv_size` A "traditional" update is `rng.default_update = rv_constructor(rng=rng).owner.outputs[0]` | 29 March 2022, 18:35:56 UTC |
b9cdd5c | Ricardo | 29 March 2022, 09:19:32 UTC | Rename internal variables in `change_rv_size` | 29 March 2022, 18:35:56 UTC |
72f255d | Ricardo | 29 March 2022, 09:07:54 UTC | Do not set default_updates in RandomVariables returned by PyMC distributions | 29 March 2022, 18:35:56 UTC |
4984ef3 | Ricardo | 29 March 2022, 10:43:09 UTC | Do not set rng updates for RandomVariables that are not computed in the graph | 29 March 2022, 18:35:56 UTC |
73c9e3c | Rob Zinkov | 29 March 2022, 11:30:49 UTC | Recognize cast data in InferenceData | 29 March 2022, 11:30:49 UTC |
fa015e3 | Ricardo | 21 March 2022, 11:44:29 UTC | Implement default transform for Mixtures | 28 March 2022, 09:18:27 UTC |
0b9f9cb | Ricardo | 21 March 2022, 12:25:35 UTC | Specify `default_output` of MarginalMixtureRV This is necessary for transforms to work with Aeppl | 28 March 2022, 09:18:27 UTC |
21b289a | Ricardo | 21 March 2022, 10:15:56 UTC | Use dispatching for default transform instead of overriding `__new__` | 28 March 2022, 09:18:27 UTC |
f0cdb1f | Ricardo | 21 March 2022, 10:25:56 UTC | Pass rv to `_default_transform` dispatch function | 28 March 2022, 09:18:27 UTC |
324d900 | Ricardo | 21 March 2022, 10:07:18 UTC | Rename `_get_default_transform` to `_default_transform` | 28 March 2022, 09:18:27 UTC |
d8f7d22 | Ricardo | 22 March 2022, 08:44:18 UTC | Rename `moment_marginal_mixture` function For consistency with other Mixture dispatched functions | 28 March 2022, 09:18:27 UTC |
623fc91 | Ricardo | 21 March 2022, 10:06:07 UTC | Rename `get_moment` to `moment` | 28 March 2022, 09:18:27 UTC |
9c2ba7a | Thomas J. Fan | 25 March 2022, 19:59:26 UTC | Remove pandas dependency (#5633) * CLN Removes pandas import from tests * FIX Adds importorskip to test_data_container * MAINT Removes pandas dependency | 25 March 2022, 19:59:26 UTC |
7aa520c | Ricardo | 23 March 2022, 11:59:31 UTC | Add test for interactions between missing, default and explicit updates in `compile_pymc` | 25 March 2022, 16:49:44 UTC |
e4a07c1 | Michael Osthege | 24 March 2022, 11:17:06 UTC | Upgrade all cache actions to v3 | 24 March 2022, 13:54:00 UTC |
d322b81 | Ricardo | 22 March 2022, 09:01:35 UTC | Refine Interval docstrings and fix typo | 23 March 2022, 12:19:23 UTC |
0246701 | twiecki | 21 March 2022, 07:04:35 UTC | ⬆️ UPGRADE: Autoupdate pre-commit config | 22 March 2022, 16:30:59 UTC |
40937a7 | Purna Chandra Mansingh | 22 March 2022, 10:53:09 UTC | Update docstring of SkewNormal, Triangular, Gumbel, Logistic, LogitNormal (#5596) * fixed doctring of SkewNormal * fixed doctring of Triangular * fixed doctring of pymc.Gumbel * fixed doctring of pymc.Logistic * fixed doctring of pymc.LogitNormal * fixed optional parameters | 22 March 2022, 10:53:09 UTC |
c20cda2 | Purna Chandra Mansingh | 22 March 2022, 10:51:57 UTC | Update docstring in Interpolated, Rice, Moyal, AsymmetricLaplace and PolyaGamma distribution (#5600) | 22 March 2022, 10:51:57 UTC |
ff694fb | Purna Chandra Mansingh | 22 March 2022, 10:47:43 UTC | Update docstring in ChiSquared, HalfNormal, Wald, Pareto, ExGaussion and VonMiese distribution (#5595) * Fixed ChiSquared, HalfNormal, Wald, Pareto, ExGaussion and VonMiese docstring * fixed optional parameter * fixed TensorVariable | 22 March 2022, 10:47:43 UTC |
a281885 | Ricardo | 22 March 2022, 08:53:01 UTC | Bump version for b5 release | 22 March 2022, 09:50:41 UTC |
7a6bd02 | Michael Osthege | 20 March 2022, 14:54:29 UTC | Ignore flaky sampling_jax.py mypy results | 21 March 2022, 20:55:10 UTC |
80b5e86 | Michael Osthege | 20 March 2022, 14:26:32 UTC | Upgrade dependencies in pre-commit environment | 21 March 2022, 20:55:10 UTC |
4e21ae5 | Michael Osthege | 20 March 2022, 14:16:30 UTC | Remove `BaseStochasticGradient` step method Closes #5138 | 21 March 2022, 20:55:10 UTC |
52682eb | Purna Chandra Mansingh | 21 March 2022, 20:40:32 UTC | Parametrize Binomial and Categorical distributions via logit_p (#5637) | 21 March 2022, 20:40:32 UTC |
5b31ec7 | Osvaldo Martin | 21 March 2022, 19:24:30 UTC | Fix bug with default init_nuts (#5622) | 21 March 2022, 19:24:30 UTC |
436f94a | Michael Osthege | 20 March 2022, 15:05:59 UTC | Parametrize some more tests by floatX | 21 March 2022, 16:36:04 UTC |
0fb7999 | Michael Osthege | 20 March 2022, 13:44:21 UTC | Run only selected tests under float32 Closes #5626 | 21 March 2022, 16:36:04 UTC |
e2cfd11 | Michael Osthege | 20 March 2022, 13:25:52 UTC | Remove XFAILed exact sample comparison test | 21 March 2022, 16:36:04 UTC |
e77e238 | Ricardo | 18 March 2022, 13:46:55 UTC | Raise ValueError if random variables are present in the logp graph Aeppl allows for graphs containing random variables. PyMC models do not generally allow for this, with the current exception of models that include SimulatorRVs. | 21 March 2022, 11:48:28 UTC |
b895e40 | Michael Osthege | 20 March 2022, 00:46:39 UTC | Fix Windows jobs & redistribute to reduce runtime Closes #5475 | 20 March 2022, 11:31:41 UTC |
e0a0325 | Michael Osthege | 19 March 2022, 17:28:28 UTC | Drop latest ArviZ tests, upgrade to Ubuntu 20.04 and add MacOS jobs | 20 March 2022, 11:31:41 UTC |
8087861 | Michael Osthege | 19 March 2022, 14:40:20 UTC | Combine test workflows The steps within each job were not changed in this commit. | 20 March 2022, 11:31:41 UTC |
0a9e99e | Michael Osthege | 19 March 2022, 14:56:49 UTC | Remove old, unused build script | 20 March 2022, 11:31:41 UTC |
a8e5b53 | Michael Osthege | 19 March 2022, 21:21:55 UTC | Update contributing guide with etiquette section (#5611) * Update outdated parts of contributing guide and PR template * Add section on contribution etiquette Co-authored-by: Ravin Kumar <ravinsdrive@gmail.com> Co-authored-by: Oriol Abril <oriol.abril.pla@gmail.com> | 19 March 2022, 21:21:55 UTC |
849a64d | Michael Osthege | 18 March 2022, 17:29:00 UTC | Refine trigger filter for both PRs and pushes Previously only one of the trigger conditions had the filter. This is a follow-up to #5364. | 19 March 2022, 14:39:27 UTC |
8b063f9 | Ricardo | 18 March 2022, 14:15:39 UTC | Do not set RNG updates inplace in compile_pymc | 18 March 2022, 18:44:18 UTC |
80f8195 | Ricardo | 18 March 2022, 11:52:44 UTC | Rename `logp_transform` to `_get_default_transform` and move it to transforms.py | 18 March 2022, 14:34:16 UTC |
02860d3 | Ricardo | 18 March 2022, 12:40:09 UTC | Do not make transforms module accessible at root level | 18 March 2022, 14:24:54 UTC |
4e8f96b | Ricardo | 13 January 2022, 12:55:02 UTC | Create helper wrapper around Aeppl IntervalTransform | 18 March 2022, 14:24:54 UTC |
f0f4252 | Ricardo | 13 January 2022, 12:54:34 UTC | Add docstrings to transforms.simplex | 18 March 2022, 14:24:54 UTC |
144b0ba | Maxim Kochurov | 18 March 2022, 12:44:52 UTC | WIP Improve scoped models (#5607) * use slashes to separate context * add fix for nested prefix * fix test * refactor prefix * revert precommit change * fix windows issue * fis smc test * Update pymc/tests/test_model.py Co-authored-by: Ricardo Vieira <28983449+ricardoV94@users.noreply.github.com> * Update pymc/tests/test_model.py Co-authored-by: Ricardo Vieira <28983449+ricardoV94@users.noreply.github.com> * update tests Co-authored-by: Ricardo Vieira <28983449+ricardoV94@users.noreply.github.com> | 18 March 2022, 12:44:52 UTC |
e8c07ef | Purna Chandra Mansingh | 18 March 2022, 11:01:53 UTC | Remove sd optional kwarg from distributions (#5583) | 18 March 2022, 11:01:53 UTC |
a0cff37 | markvrma | 15 February 2022, 15:34:38 UTC | Generalize Multinomial moment to arbitrary dimensions | 18 March 2022, 10:44:34 UTC |
5a44793 | Ricardo | 18 March 2022, 09:27:20 UTC | Make rv_size_is_none more robust | 18 March 2022, 10:44:34 UTC |
2db28f0 | Michael Osthege | 17 March 2022, 09:41:30 UTC | Bump for b4 release | 17 March 2022, 11:19:10 UTC |
07f5847 | Michael Osthege | 16 March 2022, 17:33:02 UTC | Repair pip install step in arviz workflow The non-SSL cloning is no longer supported by GitHub. | 17 March 2022, 01:02:16 UTC |
416d49f | S | 16 March 2022, 07:42:37 UTC | Updated convert_shape and convert_size Closes #5394 | 17 March 2022, 01:02:16 UTC |
09bddde | theorashid | 15 March 2022, 14:23:05 UTC | Allow n=0 in Binomial logp and logcdf | 15 March 2022, 14:23:05 UTC |
08b6b01 | Michael Osthege | 12 March 2022, 15:37:14 UTC | Fix more typing problems | 15 March 2022, 10:14:43 UTC |
97f0f79 | Michael Osthege | 12 March 2022, 15:25:58 UTC | Make `joint_logp` always return lists when `sum=False` | 15 March 2022, 10:14:43 UTC |
7cc570a | Ricardo | 14 March 2022, 13:28:15 UTC | Rename ZeroInflatedPoisson theta parameter to mu For consistency with the base Poisson distribution | 14 March 2022, 21:01:07 UTC |
e89b594 | Ricardo | 14 March 2022, 13:23:30 UTC | Replace ZeroInflated distributions with Mixtures | 14 March 2022, 21:01:07 UTC |
7d4006f | Ricardo | 13 March 2022, 14:49:21 UTC | Round moment of discrete mixtures | 14 March 2022, 21:01:07 UTC |
7ec9a24 | Ricardo | 11 March 2022, 16:52:50 UTC | Allow broadcasting of mixture components | 14 March 2022, 21:01:07 UTC |
5410885 | Ricardo | 11 March 2022, 15:43:54 UTC | Implement Mixture logcdf | 14 March 2022, 21:01:07 UTC |
0190aa5 | Ricardo | 11 March 2022, 17:45:18 UTC | Test Mixture dtype | 14 March 2022, 21:01:07 UTC |
21a2ba1 | Ricardo | 11 March 2022, 15:54:46 UTC | Implement Constant logcdf | 14 March 2022, 21:01:07 UTC |
edb832c | Ricardo | 11 March 2022, 17:25:22 UTC | Allow Constant to be integer or float | 14 March 2022, 21:01:07 UTC |
8cb71a5 | Rob Zinkov | 13 March 2022, 11:56:02 UTC | Adding release notes for recent changes to `sampling_jax` This fixes #5587 | 14 March 2022, 10:00:14 UTC |
cf73e9c | twiecki | 14 March 2022, 07:05:19 UTC | ⬆️ UPGRADE: Autoupdate pre-commit config | 14 March 2022, 09:00:42 UTC |
0a1993f | theorashid | 13 March 2022, 14:35:45 UTC | Use checkout action in nightly workflow (#5591) | 13 March 2022, 14:35:45 UTC |
86f8d98 | Purna Chandra Mansingh | 12 March 2022, 08:22:17 UTC | Update pymc.HalfStudentT docsting (#5555) * fixed pymc.HalfStudentT docsting * added :context: close-figs and removed sd from dist * undo the sd change | 12 March 2022, 08:22:17 UTC |
a14e4f9 | Purna Chandra Mansingh | 12 March 2022, 08:22:08 UTC | Update pymc.HalfCauchy docsting (#5554) * fixed pymc.HalfCauchy docsting * added :context: close-figs | 12 March 2022, 08:22:08 UTC |
7c3900c | Purna Chandra Mansingh | 12 March 2022, 08:21:59 UTC | Update pymc.Cauchy docsting (#5553) * fixed pymc.Cauchy docsting * added :context: close-figs | 12 March 2022, 08:21:59 UTC |
32380ec | Purna Chandra Mansingh | 12 March 2022, 08:21:23 UTC | Update pymc.StudentT docsting (#5552) * fixed pymc.StudentT docsting * removed sd from dist * undo the sd change | 12 March 2022, 08:21:23 UTC |
b9d4bb8 | Purna Chandra Mansingh | 12 March 2022, 08:21:04 UTC | Update pymc.Laplace docsting (#5551) * fixed pymc.Laplace docsting * fixed pymc.Laplace docsting | 12 March 2022, 08:21:04 UTC |
7f434bb | Purna Chandra Mansingh | 12 March 2022, 08:20:49 UTC | Update pymc.TruncatedNormal docstring (#5546) * fixed TruncateNormal docsting * fixed TruncateNormal docsting * removed sd from dist * undo the sd change | 12 March 2022, 08:20:49 UTC |
54c9dfa | Christian Ritter | 12 March 2022, 08:20:33 UTC | update Exponential numpydoc (#5491) Co-authored-by: Oriol Abril-Pla <oriol.abril.pla@gmail.com> | 12 March 2022, 08:20:33 UTC |
5eaa516 | Ricardo | 11 March 2022, 15:24:23 UTC | Update aesara and aeppl dependencies | 11 March 2022, 20:19:58 UTC |
b799547 | Rob Zinkov | 11 March 2022, 20:18:35 UTC | Adding NUTS sampler from blackjax to sampling_jax (#5477) * Adding NUTS sampler from blackjax to sampling_jax * Lint fixes * Refactor get_jaxified_logp * Install blackjax in workflows * Fix url * Simplify function * Add documentation * Add library versions * Move to more appropriate section * Fix docstrings * Add jax to dev environment * Jax doesn't work on windows, remove it * Exclude jax * Fix merge * Remove progress bar functionality for now * Pre-commit fix | 11 March 2022, 20:18:35 UTC |
65dcb49 | Sagar | 11 March 2022, 15:26:23 UTC | Return an empty string for empty model representation (#5577) | 11 March 2022, 15:26:23 UTC |
4626712 | aloctavodia | 10 March 2022, 10:25:03 UTC | fix extract_Q_estimate function | 11 March 2022, 09:36:53 UTC |
b903e57 | aloctavodia | 10 March 2022, 08:06:39 UTC | stack array stats | 11 March 2022, 09:36:53 UTC |
c3c76c5 | Michael Osthege | 11 March 2022, 02:23:09 UTC | Run mypy in pre-commit for continuous improvement of type hints (#5549) * Remove numpy Tester instance from public API It was making mypy follow into the tests module regardles of its config. See https://stackoverflow.com/a/70367929 * Move all mypy config to mypy.init Because typing_copilot can't deal with the other formats. * Fix typing problems and add pre-commit mypy step | 11 March 2022, 02:23:09 UTC |
b919153 | Osvaldo Martin | 10 March 2022, 22:10:43 UTC | update release note (#5573) | 10 March 2022, 22:10:43 UTC |
d150d55 | mirko-m | 10 March 2022, 19:09:27 UTC | Add tests for the output type of ode_func (#5414) Co-authored-by: Mirko Moeller <mirko.moeller@coanda.ca> Co-authored-by: Michael Osthege <michael.osthege@outlook.com> | 10 March 2022, 19:09:27 UTC |
069de73 | Purna Chandra Mansingh | 10 March 2022, 18:42:23 UTC | Update pymc.LogNormal docsting (#5556) * fixed pymc.LogNormal docsting * fixed pymc.LogNormal docsting | 10 March 2022, 18:42:23 UTC |
44c5495 | theorashid | 09 March 2022, 23:03:22 UTC | use grep from __init__.py to get version | 10 March 2022, 09:13:44 UTC |
2dd4c8c | aloctavodia | 08 March 2022, 15:29:26 UTC | allow external step method | 10 March 2022, 08:22:09 UTC |
50ceb2b | aloctavodia | 08 March 2022, 09:54:00 UTC | remove bart | 10 March 2022, 08:22:09 UTC |
620b11d | Ricardo Vieira | 09 March 2022, 02:56:09 UTC | Refactor Mixture distribution for V4 (#5438) * Set `expand=True` when calling `change_size` in `SymbolicDistribution` * Move NormalMixture tests to their own class * Move mixture random tests from test_distributions_random to test_mixture * Use specific imports in test_mixture * Reenable Mixture tests in pytest workflow * Refactor Mixture distribution Mixtures now use an `OpFromGraph` that encapsulates the Aesara random method. This is used so that logp can be easily dispatched to the distribution without requiring involved pattern matching. The Mixture random and logp methods now fully respect the support dimensionality of its components, whereas previously only the logp method did, leading to inconsistencies between the two methods. In the case where the weights (or size) indicate the need for more draws than what is given by the component distributions, the latter are resized to ensure there are no repeated draws. This refactoring forces Mixture components to be basic RandomVariables, meaning that nested Mixtures or Mixtures of Symbolic distributions (like Censored) are not currently possible. Co-authored-by: Larry Dong <larry.dong@mail.utoronto.ca> * Add warning when using iterable with single Mixture component * Update Mixture docstrings * Emphasize equivalency between iterable of components and single batched component * Add example with mixture of two distinct distributions * Add example with multivariate components * Refactor NormalMixture * Refactor TestMixtureVsLatent The two tests relied on implicit behavior of V3, where the dimensionality of the weights implied the support dimension of mixture distribution. This, however, led to inconsistent behavior between the random method and the logp, as the latter did not enforce this assumption, and did not distinguish if values were mixed across the implied support dimension. In this refactoring, the support dimensionality of the component variables determines the dimensionality of the mixture distribution, regardless of the weights. This leads to consistent behavior between the random and logp methods as asserted by the new checks. Future work will explore allowing the user to specify an artificial support dimensionality that is higher than the one implied by the component distributions, but this is for now not possible. * Remove MixtureSameFamily Behavior is now implemented in Mixture * Add Mixture moments * Update release notes Co-authored-by: Larry Dong <larry.dong@mail.utoronto.ca> | 09 March 2022, 02:56:09 UTC |
7d4162c | Purna Chandra Mansingh | 08 March 2022, 13:54:40 UTC | fixed pymc.Kumaraswamy docsting (#5550) | 08 March 2022, 13:54:40 UTC |
afe210a | Chris Fonnesbeck | 07 March 2022, 17:43:10 UTC | Initiate v4.0.0b3 release (#5562) * Initiate v4.0.0b3 release * Added Maintenance section to release notes with VI update notice * Clean release notes * Removed Cholesky from non-working section of release notes Co-authored-by: Ricardo <ricardo.vieira1994@gmail.com> | 07 March 2022, 17:43:10 UTC |
f71a2b1 | Ricardo | 03 March 2022, 10:00:10 UTC | Automatically resize `sd_dist` in `_LKJCholeskyCov` | 07 March 2022, 16:02:14 UTC |
5023a70 | aloctavodia | 25 February 2022, 08:27:01 UTC | fix predictions when using 1 chain | 07 March 2022, 15:57:11 UTC |
55463de | theorashid | 07 March 2022, 11:21:52 UTC | use new secret for nightly build | 07 March 2022, 11:38:37 UTC |
a98ab6d | Thomas J. Fan | 07 March 2022, 06:05:25 UTC | CLN Remove pandas from bart.utils.plot_dependence (#5558) | 07 March 2022, 06:05:25 UTC |
620546b | Purna Chandra Mansingh | 06 March 2022, 02:39:53 UTC | Doc change the default value of pymc.draw from 500 to 1. (#5543) * changed the default value of drwas from 500 to 1 * fixed samplers_draw docsting * fixed samplers_draw docsting | 06 March 2022, 02:39:53 UTC |
4223b6e | Larry Dong | 04 March 2022, 01:10:03 UTC | replace assert_op by raise_op | 04 March 2022, 08:45:51 UTC |
6833677 | Osvaldo Martin | 04 March 2022, 06:45:55 UTC | update reported warnings after sampling (#5516) * update report * update per comments | 04 March 2022, 06:45:55 UTC |
999160a | theorashid | 03 March 2022, 18:19:50 UTC | adding nightly builds (#5490) * this isn't aesara anymore * release-job didn't exist * remove get_versions() * more functions, less faff * nicer f string * NIGHTLY is a global variable * testing nightly install with date in workflow | 03 March 2022, 18:19:50 UTC |
a0e43da | Ricardo | 03 March 2022, 13:38:18 UTC | Remove pylint skip-file in sampling_jax.py | 03 March 2022, 15:14:44 UTC |
64e9375 | Ricardo | 03 March 2022, 13:37:12 UTC | Add type hints to several functions in sampling_jax.py | 03 March 2022, 15:14:44 UTC |
dd4b940 | Ricardo | 03 March 2022, 13:25:48 UTC | Fix seeding bug in `sample_numpyro_nuts` when more than one chain was sampled | 03 March 2022, 15:14:44 UTC |