0fd439f | Michael Osthege | 12 February 2021, 13:31:25 UTC | Bump patch version for 3.11.1 release | 12 February 2021, 16:35:56 UTC |
1d37d31 | ricardoV94 | 12 February 2021, 13:27:24 UTC | Fix Triangular bounds (#4470) * Small fix Triangular logp and logcdf methods * Add tests for invalid parameters Uniform, Triangular, DiscreteUniform | 12 February 2021, 13:27:24 UTC |
e46f490 | Martin Ingram | 12 February 2021, 13:21:04 UTC | Transform jax samples (#4427) * Transform samples from sample_numpyro_nuts * Add `pymc3.sampling_jax._transform_samples` function which transforms draws * Modify `pymc3.sampling_jax.sample_numpyro_nuts` function to use this function to return transformed samples * Add release note * Update pymc3/sampling_jax.py Co-authored-by: Junpeng Lao <junpenglao@gmail.com> * Added a small test * Split jax tests into their own workflow Co-authored-by: Michael Osthege <michael.osthege@outlook.com> Co-authored-by: Junpeng Lao <junpenglao@gmail.com> | 12 February 2021, 13:21:04 UTC |
41a25d5 | Thomas Wiecki | 08 February 2021, 13:31:16 UTC | ⬆️ UPGRADE: Autoupdate pre-commit config (#4466) * ⬆️ UPGRADE: Autoupdate pre-commit config * ⬆️ UPGRADE: Autoupdate pre-commit config Co-authored-by: twiecki <twiecki@users.noreply.github.com> | 08 February 2021, 13:31:16 UTC |
e40304b | ricardoV94 | 08 February 2021, 09:26:01 UTC | Unseed (and slightly refactor) TestMatchesScipy (#4461) * Refactor and unseed TestMatchesScipy * Show arguments when check_logcdf fails * Reduce 64bit precision test_skew_normal * Mark xfail 32-bit for Gamma logcdf * Reduce 64bit precision test_inverse_gamma_alt_params | 08 February 2021, 09:26:01 UTC |
0c21de4 | Sayam Kumar | 06 February 2021, 18:33:46 UTC | Fix Dirichlet.logp (#4454) * Fix Dirichlet.logp by checking number of categories > 1 only at event dims * Update test_distributions.py * Removed the shape validation check to even work for last dimensional shape as 1. Modified the `test_dirichlet` function to check for the same. * Added a test to check Dirichlet.logp with different batch shapes. * Tested exact Dirichlet.logp values againt scipy implementation Given a mention in RELEASE-NOTES.md | 06 February 2021, 18:33:46 UTC |
b6660f9 | Chandan Kumar Roy | 05 February 2021, 13:41:32 UTC | minor correction in sampling.py and starting.py (#4458) Make deepcopy of start dicts in pm.sample and `pm.find_MAP` to prevent inplace modification of user variables closes #4456 | 05 February 2021, 13:41:32 UTC |
e467bb9 | Amarjit | 01 February 2021, 17:30:50 UTC | Removed unused imports using autoflake | 04 February 2021, 15:34:53 UTC |
2a5f65e | Amarjit | 01 February 2021, 17:06:53 UTC | Updated .pre-commit-config.yaml | 04 February 2021, 15:34:53 UTC |
60e0dae | ricardoV94 | 03 February 2021, 17:39:12 UTC | Reduce number of test points in slow logcdf methods (#4459) | 03 February 2021, 17:39:12 UTC |
3cd9eb3 | Michael Osthege | 02 February 2021, 15:07:13 UTC | Update Style guide to reflect the current workflow (#4457) | 02 February 2021, 15:07:13 UTC |
b77e2d1 | Ashish | 31 January 2021, 20:14:40 UTC | Updating docstring of a class according to PEP 257 in a file (#4449) * one class docs updated * Updated with docs with PEP 257 in one class * Updated with docs with PEP 257 in one class * updated * update * Update * update * Update pymc3/step_methods/metropolis.py Co-authored-by: Thomas Wiecki <thomas.wiecki@gmail.com> | 31 January 2021, 20:14:40 UTC |
07679ec | Kaustubh | 31 January 2021, 10:29:30 UTC | Avoid unclear TypeError when using theano.shared variables as input to distribution parameters (#4445) * Added default testvalue support for theano.shared Co-authored-by: Ricardo <ricardo.vieira1994@gmail.com> | 31 January 2021, 10:29:30 UTC |
2d3ec8f | ricardoV94 | 29 January 2021, 16:23:27 UTC | Add tests for tt.switch related bugs (#4448) * Add tests for edge cases * Add release-note | 29 January 2021, 16:23:27 UTC |
03d7af5 | Michael Osthege | 28 January 2021, 10:25:51 UTC | Upgrade to Theano-PyMC 1.1.2 | 28 January 2021, 15:12:11 UTC |
43e7572 | Michael Osthege | 28 January 2021, 10:03:40 UTC | Support imputations with ndarray data (#4439) * Support imputations with ndarray data (closes #4437) | 28 January 2021, 10:03:40 UTC |
823906a | ricardoV94 | 22 January 2021, 19:44:54 UTC | Evaluate only desired branch in log1mexp_numpy (#4428) * Evaluate only desired function in log1mexp_numpy * Do not change input inplace * Add release note | 22 January 2021, 19:44:54 UTC |
da76320 | Michael Osthege | 22 January 2021, 17:42:40 UTC | Mention removal of ArviZ-delegated plotting and stats in release notes (#4432) * Mention removal of ArviZ-delegated plotting and stats * Mention most important breaking changes first | 22 January 2021, 17:42:40 UTC |
aedc8e9 | Eelke Spaak | 21 January 2021, 07:48:56 UTC | link to Theano-PyMC 1.1.0 release notes, mention BART.predict, update release date | 21 January 2021, 08:20:27 UTC |
4bd23c0 | Eelke Spaak | 19 January 2021, 12:37:39 UTC | reformulate release notes opening sentence | 21 January 2021, 08:20:27 UTC |
457dabc | Eelke Spaak | 19 January 2021, 12:31:45 UTC | update release notes for 3.11.0 | 21 January 2021, 08:20:27 UTC |
4656cd5 | Eelke Spaak | 19 January 2021, 12:24:23 UTC | increment version to 3.11.0 in __init__.py | 21 January 2021, 08:20:27 UTC |
89916ad | ricardoV94 | 20 January 2021, 18:05:50 UTC | Increase logcdf coverage for invalid parameter values (#4421) * Test logcdf methods outside of finite edges * Fix failing logcdf methods * Disable failing tests temporarily * Revert removed xfail * Update docstring * Add more informative comment | 20 January 2021, 18:05:50 UTC |
a3c2060 | Osvaldo Martin | 19 January 2021, 16:13:01 UTC | Bart: Refactor splitting variables and predictions (#4310) * refactor split variables, add prior split variables, add predict function * remove dirichlet * remove unused import * use to_numpy to convert a series/dataframe to an array. * use already defined variable | 19 January 2021, 16:13:01 UTC |
7ef2de4 | Bas Nijholt | 19 January 2021, 15:03:58 UTC | make sure weights always sum up to 1, fix #4369 (#4370) * make sure weights always sum up to 1, fix #4369 * always normalize self.weights in update_weights_beta * rephrase | 19 January 2021, 15:03:58 UTC |
84c88c2 | Michael Osthege | 15 January 2021, 21:58:06 UTC | Mention SMC/named models related PR #4365 | 19 January 2021, 12:11:16 UTC |
54d288d | Bas Nijholt | 22 December 2020, 14:57:35 UTC | Add tests related to named models A test was added to asserts that strace.name is a string (this was not the case before #4365). Non-empty model names are actually not supported (again, see #4365) so attempting to SMC-sample a named model will now raise a NotImplementedError. | 19 January 2021, 12:11:16 UTC |
4f079d4 | Bas Nijholt | 21 December 2020, 16:57:10 UTC | pass a string to NDArray(name=...) Currently, `name` is a `Model`. | 19 January 2021, 12:11:16 UTC |
e03c227 | ricardoV94 | 19 January 2021, 11:43:38 UTC | find_MAP informative error (#4423), Closes #3143 * Raise ValueError when there are no continuous variables Remove deprecated block of code * pre-commit * Remove unused imports | 19 January 2021, 11:43:38 UTC |
37ca5ea | Raul-ing Average | 19 January 2021, 11:35:58 UTC | Remove Plots & Diagnostics (#4397), Closes #4371 * :tada: Start removing Diagnostics & Plots in PyMC3 development :fire: Remove arviz plots * Remove directly imported arviz plots used in pymc3 plots :fire: Remove all plots from PyMC3 Plots module :fire: Remove PyMC3 plots references in Docs :art: Mention Plotting & Diagnostics in API page and remove plots reference in __init__.py :rewind: Revert posterior_plot function, test, and docs :art: Add deprecation warning to posterior_plot function :art: Add context on plot import and import back into __init__.py :pencil2: Add warning and details of posterior_plot added Update docs/source/api/plots.rst Co-authored-by: Alexandre ANDORRA <andorra.alexandre@gmail.com> Update docs/source/api/plots.rst Co-authored-by: Alexandre ANDORRA <andorra.alexandre@gmail.com> Update pymc3/plots/__init__.py Co-authored-by: Alexandre ANDORRA <andorra.alexandre@gmail.com> Update pymc3/plots/__init__.py Co-authored-by: Alexandre ANDORRA <andorra.alexandre@gmail.com> :pencil2: Update docs to add stats.rst details :pencil2: Minor docs notation for posterioplot function(s) :pencil: Add breakline before docstring title * :pencil2: small typo and remove summary in plots.rst | 19 January 2021, 11:35:58 UTC |
32b5c94 | Kaustubh | 19 January 2021, 11:32:27 UTC | Fix logp of bounded variables when value lies near the boundary (#4418) Closes #3938 | 19 January 2021, 11:32:27 UTC |
acb3261 | Michael Osthege | 19 January 2021, 09:07:24 UTC | Repair change_flags reference and remove pytest.mark.run decorators We do not have pytest-ordering in the dependencies and tests should be unordered anyway. | 19 January 2021, 11:26:17 UTC |
5481a12 | Thomas Wiecki | 18 January 2021, 11:40:54 UTC | Bump to most recent ArviZ | 19 January 2021, 11:26:17 UTC |
dff8fb9 | Eelke Spaak | 12 January 2021, 16:09:55 UTC | updating test_pandas_to_array for Theano 1.1.0 compatibility | 19 January 2021, 11:26:17 UTC |
6ad83f6 | Eelke Spaak | 11 January 2021, 15:09:03 UTC | update test to accommodate int type casting | 19 January 2021, 11:26:17 UTC |
59cfef0 | Michael Osthege | 11 January 2021, 14:36:39 UTC | Don't test float128 on Windows (unsupported) | 19 January 2021, 11:26:17 UTC |
405fd10 | Michael Osthege | 11 January 2021, 00:09:48 UTC | Upgrade to Theano-PyMC 1.1.0 Many gof imports were changed to explicit imports. This commit also changes the usage of change_flags because of pending deprecation. | 19 January 2021, 11:26:17 UTC |
e46b1b4 | Brandon T. Willard | 04 January 2021, 17:09:31 UTC | Update as_op, theano.gof imports and stack_search usage | 19 January 2021, 11:26:17 UTC |
a607709 | Kaustubh | 18 January 2021, 12:00:42 UTC | Fixed Dirichlet.random returning output in wrong shapes (#4416) | 18 January 2021, 12:00:42 UTC |
4e2c099 | ricardoV94 | 16 January 2021, 15:35:51 UTC | UserWarning if doing predictive sampling with models containing Potentials (#4419) * Raise warning when sampling with Potentials * Add warning to fast_sample_ppc and unittests * Add release note * Avoid sampling in unittests | 16 January 2021, 15:35:51 UTC |
2a3d9a3 | ricardoV94 | 16 January 2021, 10:56:51 UTC | Dirichlet multinomial (continued) (#4373) * Add implementation of DM distribution. * Fix class name mistake. * Add DM dist to exported multivariate distributions. * Export DirichletMultinomial in pymc3.distributions As suggested in https://github.com/pymc-devs/pymc3/pull/3639#issuecomment-559822430 Also see: https://github.com/pymc-devs/pymc3/pull/3639#issuecomment-559840453 but this seems to be part of a broader discussion. * Attempt at matching Multinomial initialization. * Add some simple tests for DM. * Correctly deal with 1d n and 2d alpha. * Fix typo in DM random. * Fix faulty tests for DM. * Drop redundant initialization test for DM. * Add test that DM is normalized for n=1 case. * Add DM test case based on BetaBinomial. * Update pymc3/distributions/multivariate.py * - Infer shape by default (copied code from Dirichlet Distribution) - Add default shape in `test_distributions_random.py` * - Use size information in random method - Change random unittests * - Restore merge accidental deletions * - Underscore missing * - More merge cleaning * Bring DirichletMultinomial initialization into alignment with Multinomial. * Align all DM tests with Multinomial. * Align DirichletMultinomial random implementation with Multinomial. * Match DM random method to Multinomial implementation. * Change alpha -> a Remove _repr_latex_ * Run pre-commit * Keep standard order of methods random and logp * Update docstrings for valid input types. Progress on batch test. * Add new test to ensure DM matches BetaBinom * Change DM alpha -> a in docstrings. * Test two additional parameterization shapes in `test_dirichlet_multinomial_random`. * Revert debugging comments. * Revert unrelated changes. * Fix minor Black inconsistency. * Drop no-longer-functional reshaping code. * Assert shape of random samples is as expected. * Explicitly test random sample shapes, including batch dimensions. * Sort imports. * Simplify _random It should be okay to not explicitly change the input dtype as in the multinomial, because the input to the np.random.dirichlet should be safe (it's fine to have float32 to float64 overflow from 1.00 to 1.01..., underflow from 0.01, to 0.0 would still be problematic, but we don't know if this is an issue yet...). The output of the numpy.random.dirichlet to numpy.random.multinomial should be safe since it is already in float64 by then. We still need to convert to the previous dtype, since numpy changes it by default. size_ argument was no longer being used. * Reorder tests more logically * Refactor tests Merged mode tests since shape must be given explicitly anyway Moved test_dirichlet_multinomial_random to test_distributions_random.py and renamed it to test_dirichlet_multinomial_shapes * Require shape argument Also allow more forgiveness if user passes lists instead of arrays (WIP/suggestion only) * Remove unused import `to_tuple` * Simplify logic to handle list as input for `a` * Raise ShapeError in random() * Finish batch and repr unittests * Add note about mode * Tiny rewording * Change mode to _defaultval * Revert comment for Multinomial mode * Update shape check logic * Add DM to release notes. * Minor docstring revisions as suggested by @AlexAndorra. * Revise the revision. * Add comment clarifying bounds checking in logp() * Address review suggestions * Update `matches_beta_binomial` to take into consideration float precision * Add DM to multivariate distributions docs. Co-authored-by: Byron Smith <me@byronjsmith.com> Co-authored-by: Colin <ColCarroll@users.noreply.github.com> | 16 January 2021, 10:56:51 UTC |
1769258 | Chris Fonnesbeck | 10 January 2021, 16:49:06 UTC | [WIP] Replace installation instructions on README with wiki links (#4413) * Updated Mauna Loa notebook * Replaced installation instructions with links to platform-specific guides on wiki * Fixed rst link syntax | 10 January 2021, 16:49:06 UTC |
c81e0c9 | Moritz Schubert | 08 January 2021, 22:14:56 UTC | documented the default value of BinaryGibbsMetropolis' transit_p parameter (#4411) Co-authored-by: Moritz Schubert <moritz.schubert@uni-marburg.de> | 08 January 2021, 22:14:56 UTC |
9c54d91 | VasanthManiVasi | 08 January 2021, 15:58:14 UTC | Update PULL_REQUEST_TEMPLATE.md | 08 January 2021, 17:31:01 UTC |
935ae4d | Marco Gorelli | 07 January 2021, 17:14:22 UTC | :label: fix some typing in _TraceDict | 08 January 2021, 13:32:44 UTC |
8b1f64c | ricardoV94 | 05 January 2021, 17:33:40 UTC | Improve ExGaussian logcdf and refactor logp (#4407) * - Improve ExGaussian logcdf and refactor logp * - Remove unused std_cdf * - Add release note * - Remove unused import `theano` | 05 January 2021, 17:33:40 UTC |
044c407 | Chandan Kumar Roy | 05 January 2021, 15:44:29 UTC | Asymmetric Laplace distribution added (#4392) * asymmetric laplace distribution added * unit test for random function added * continuous.py updated * suggested changes added * updated * RELEASE-NOTES.md and continuous.rst updated | 05 January 2021, 15:44:29 UTC |
34b05d8 | Marco Gorelli | 05 January 2021, 14:43:27 UTC | Test on oldest supported versions during end-to-end CI (#4374) * test on oldest supported versions during end-to-end CI * less strict pin * try 1.15 pin * try pinning pandas and scipy too * only test betabinom if scipy modern enough * revert sort_conda changes (will do in separate PR) * generate requirements-dev from py37 env | 05 January 2021, 14:43:27 UTC |
aabec02 | Raul-ing Average | 05 January 2021, 09:44:08 UTC | Docs - Update PyMC3 Installation focus via Conda Forge (#4401) * :pencil2: Update PyMC3 Docs installation by Conda Forge :pencil2: Update README.rst Add note about pip installation Co-authored-by: Thomas Wiecki <thomas.wiecki@gmail.com> Update README.rst Add context to pip installation and dependencies Co-authored-by: Thomas Wiecki <thomas.wiecki@gmail.com> Apply suggestions from code review Co-authored-by: Thomas Wiecki <thomas.wiecki@gmail.com> :pencil2: README installation update * Update README.rst * :pencil2: Focus on Theano-PyMC3 instead of Theano for installation * Update README.rst * Update README.rst Co-authored-by: Thomas Wiecki <thomas.wiecki@gmail.com> | 05 January 2021, 09:44:08 UTC |
982bb4d | ricardoV94 | 04 January 2021, 20:56:27 UTC | Remove redundant log1pexp reimplementation in Logistic CDF (#4406) Also: Reorder logp and random to be in line with rest of the library | 04 January 2021, 20:56:27 UTC |
e783106 | Eelke Spaak | 04 January 2021, 11:39:24 UTC | Better handling of variable names in e.g. GraphViz graphs (#4403) * handle changed API of theano.gof.graph.stack_search * convert n and eta to tensors, explicitly list parameters for repr * improve robustness of get_repr_for_variable * Revert "handle changed API of theano.gof.graph.stack_search" This reverts commit 6238bff6c38f5b0ff3b822e908780efb14401f50. | 04 January 2021, 11:39:24 UTC |
240c372 | Thomas Wiecki | 04 January 2021, 09:12:57 UTC | ⬆️ UPGRADE: Autoupdate pre-commit config (#4402) Co-authored-by: twiecki <twiecki@users.noreply.github.com> | 04 January 2021, 09:12:57 UTC |
2824027 | Andrew Wray | 03 January 2021, 06:47:51 UTC | Add .random method to MvGaussianRandomWalk (#4388) * Implement random for MvGaussianRandomWalk Implements the random method for MvGaussianRandomWalk, partially fixing #4337. * Update docstring for GaussianRandomWalk._random() * Add example to random docstring * Modify MvGaussianRandomWalk.random Improves the implementation by using MvNormal.random as suggested by @Sayam753. Also updated its docstring to add more examples. * Add TestMvGaussianRandomWalk * Pass entire shape to MvNormal This commit rewrites the random method to pass the entire shape into MvNormal. MvGaussianRandomWalk also now supports an integer shape that must match the dimensionality of mu. In this case it is assumed that there are no time steps, and a random sample from the multivariate normal distribution will be returned. * Fix rebase issue * Update pymc3/distributions/timeseries.py Remove equality since the shape parameter cannot have more than 2 dimensions. Co-authored-by: Sayam Kumar <sayamkumar753@yahoo.in> * Update comment pymc3/distributions/timeseries.py Co-authored-by: Sayam Kumar <sayamkumar753@yahoo.in> * Update pymc3/distributions/timeseries.py Adds a more explicit conditional on the time_axis statement. Co-authored-by: Sayam Kumar <sayamkumar753@yahoo.in> * Fix syntax error * Add tests for MvGaussianRandomWalk with RV params Added tests for both chol and cov to be random variables. * Improve comment clarity in MvNormal._random Moves the brittle comment in MvRandom._random's docstring to the actual location in the code for clarity. * Update docstring formatting * Update RELEASE-NOTES.md Co-authored-by: Sayam Kumar <sayamkumar753@yahoo.in> | 03 January 2021, 06:47:51 UTC |
acb8da0 | ricardoV94 | 02 January 2021, 22:50:38 UTC | Increase unittest `check_logcdf` coverage and fix issues with some distribution methods (#4393) * Change `check_logcdf` to test that values below or above domain are properly evaluated. Fix issues with `Uniform`, `HalfNormal`, `Gamma`, and `InverseGamma` distributions. * Add multiple value test for logcdf. Add more informative comment for Gamma and InverseGamma hack. Update Release note. * Update release note * Update docstrings with valid value types * Update RELEASE-NOTES.md * Add more informative comments and remove TODO * TypeError: format -> f-strings * Ignore finite upper limit in Nat domains. Move new checks to `check_logcdf`. * Use `tt.switch` in `DiscreteUniform` for hard boundary (addresses previously failing test in 32bit OS) | 02 January 2021, 22:50:38 UTC |
a21fafa | ricardoV94 | 02 January 2021, 20:55:49 UTC | Update `log1mexp` and remove redundant local reimplementations in the library (#4394) * Fix cutoff value in `log1mexp` and redundant reimplementation in `Exponential.logcdf()` * Get more digits :) * Remove redundant reimplementation in Weibull * Tiny rewrite to use `|` instead of `tt.or_`. Remove TODO comments. * retrigger checks | 02 January 2021, 20:55:49 UTC |
e2ce815 | Raul-ing Average | 31 December 2020, 18:50:24 UTC | Update legibility of formula to proper Latex Format (#4390) * :pencil2: Normal Dist. Function updated to latex, proper * Update function to latex * Add hyperlink * :pencil2: Normal Dist PDF slight update on where conditional * :pencil2: Change evaluated at to 3rd operand and minor update | 31 December 2020, 18:50:24 UTC |
d871b80 | ricardoV94 | 31 December 2020, 11:49:46 UTC | Implement logcdf method for discrete distributions (#4387) * Implement logcdf method for discrete distributions * Add release note * Small reformatting * Add more comprehensive bound check for impossible parameters * Fix bounds for logcdf of DiscreteUniform * Add safe values/ parameters for `BetaBinomial`, `Poisson`, and `HyperGeometric` to avoid errors when evaluating negative logcdfs. * Change `check_selfconsistency_discrete_logcdf` to work with negative domains, such as the DiscreteUniform. * Raise TypeError in logcdf methods that only accept scalar values. Fix docstring formatting. * Small change to logcdf methods of `DiscreteWeibull` and `ZeroInflated` making use of `tt.log1p` and `logaddexp`. More informative comment on workaround for `Poisson.logcdf`. | 31 December 2020, 11:49:46 UTC |
02095c2 | Kyle Beauchamp | 31 December 2020, 08:37:17 UTC | Fix pymc3 to work with latest theano-pymc master (#4382) * Update Theano-PyMC version * Update JAX imports * Update RandomStream imports and arguments * Update use of theano.config.change_flags in pymc3.tests.test_variational_inference Co-authored-by: Brandon T. Willard <brandonwillard@users.noreply.github.com> | 31 December 2020, 08:37:17 UTC |
91993d8 | Sayam Kumar | 30 December 2020, 15:58:49 UTC | Fix MatrixNormal.random (#4368) * Fix MatrixNormal.random * Provided better context for tests * Worked on suggestions * Added a test to signify need of transpose of cholesky matrix. * Used np.swapaxes to take transpose * Given a mention in release notes | 30 December 2020, 15:58:49 UTC |
3cfee77 | Marco Gorelli | 26 December 2020, 15:19:31 UTC | Use sigma instead of sd, remove deprecationwarning (#4344) * remove sd deprecation warning * sd -> sigma in pymc3/tests/test_models_linear.py::TestGLM * update cls.sd in tests * noop * don't delete self.sd | 26 December 2020, 15:19:31 UTC |
93b68f2 | Raul-ing Average | 25 December 2020, 14:15:50 UTC | [Proposal] Track Submodule pymc-examples Git Fix (#4375) * :pushpin: Track submodule pymc-examples to main branch * Configure .gitmodules to specify branch for submodule * closes #4357 * :pushpin: Track submodule pymc-examples to main branch * Configure .gitmodules to specify branch for submodule * closes #4357 * :arrow_up: Commit latest main branch submodule pymc-examples * Updated .gitattributes file * git submodule sync * switch branch * go to parent module, and commit * :pin: Change git submodule update command for docs/source/pymc-examples | 25 December 2020, 14:15:50 UTC |
e0ee447 | Ravin Kumar | 25 December 2020, 10:37:25 UTC | Add Jax compilation more prominently to README (#4380) * Add Jax compilation more prominently * Update README.rst Co-authored-by: Thomas Wiecki <thomas.wiecki@gmail.com> | 25 December 2020, 10:37:25 UTC |
666f71a | Ravin Kumar | 25 December 2020, 10:36:12 UTC | Add PyMCon Talks link to readme (#4381) * Add PyMCon Talks link * Remove a trailing whitespace * Update README.rst Co-authored-by: Thomas Wiecki <thomas.wiecki@gmail.com> | 25 December 2020, 10:36:12 UTC |
b7daf26 | Thomas Wiecki | 25 December 2020, 00:52:38 UTC | Rename disable_bounds_check to check_bounds (#4378) * Rename disable_bounds_check to bounds_check. * Remove trailing white space. * Rename bounds_check to check_bounds as per @ricardoV94 * Set default check_bounds to True. * bounds_check -> check_bounds. * Wording. | 25 December 2020, 00:52:38 UTC |
37239fc | Thomas Wiecki | 23 December 2020, 23:39:06 UTC | Add flag to disable bounds check for speed-up (#4377) * Add flag to disable bounds check. * Move check for model and flag into single line as per @colcarrolls's suggestion. * modelcontext raises a TypeError if no model is found. Catch that. * Add comment. * Add mention in release-notes. * Add test for when boundaries are disabled. | 23 December 2020, 23:39:06 UTC |
0402aab | ricardoV94 | 22 December 2020, 09:10:09 UTC | Add bound to HyperGeometric logp (resolves #4366) (#4367) * - Add bound to HyperGeometric logp - Pass unit tests when scipy logpmf returns nan * - Add release-note * - Replace tt.max and tt.min with tt.switch | 22 December 2020, 09:10:09 UTC |
0ec65e5 | Thomas Wiecki | 21 December 2020, 10:02:12 UTC | ⬆️ UPGRADE: Autoupdate pre-commit config (#4362) Co-authored-by: twiecki <twiecki@users.noreply.github.com> | 21 December 2020, 10:02:12 UTC |
d9cf14c | Bas Nijholt | 21 December 2020, 10:01:35 UTC | fix typo, "togheter" -> "together" (#4363) | 21 December 2020, 10:01:35 UTC |
4e435ea | Marco Gorelli | 20 December 2020, 16:45:04 UTC | Remove pymc3/examples (#4352) * move test files to tests/data, send request to pymc-examples for other files | 20 December 2020, 16:45:04 UTC |
e1af7c9 | Michael Osthege | 20 December 2020, 16:41:05 UTC | Fixing scalar shape handling: treat (1,) as vectors (#4214) Fixing scalar shape handling: treat (1,) as vectors | 20 December 2020, 16:41:05 UTC |
398eae6 | Sayam Kumar | 20 December 2020, 15:53:07 UTC | Fix MvStudentT.random (#4359) * Fix MvStudentT.random * Given a mention in release notes | 20 December 2020, 15:53:07 UTC |
0e9b9a4 | ricardoV94 | 20 December 2020, 12:52:56 UTC | Improve logsumexp to work with infinite values (#4360) * Make logsumexp work with inifinite values, matching scipy behavior * Run pre-commit * Add note to release_notes | 20 December 2020, 12:52:56 UTC |
34447a7 | Marco Gorelli | 19 December 2020, 10:59:54 UTC | Remove notebooks, replace with pymc-examples submodule (#4348) * :fire: remove notebooks from pymc3, replace with pymc-examples submodule * change notebook output_html * revert href change * :memo: release notes * :truck: move examples to pymc-examples * checkout submodule at main * noop * :fire: remove test_examples * keep examples/data for now, so pm.get_data keeps working * keep examples/__init__.py so get_data still works * revert removing examples * revert removing test_examples * fixup so docs work with latest pymc-examples reorg * :art: | 19 December 2020, 10:59:54 UTC |
52a0479 | Raul-ing Average | 19 December 2020, 10:51:46 UTC | :art: Update CI badge to GitHub Actions in docs (#4356) * closes #4350 | 19 December 2020, 10:51:46 UTC |
4e5edd5 | Marco Gorelli | 17 December 2020, 15:18:47 UTC | Let plot_posterior_predictive_glm work with inferencedata too (#4234) * adapt for both * add multitrace test * add inferencedata test, cover all lines * update release notes * sort imports (it's not checked yet?) * fixup test * fixup PR number * :label: add type annotations, correct docstring * sort * sort * add copyright note * use todict * :art: * don't cover import error (matplotlib is a dev requirement) * don't cover import typechecking * remove optional mpl import * remove optional mpl import * remove optional mpl import * Update RELEASE-NOTES.md * use Python3.7+ type hints | 17 December 2020, 15:18:47 UTC |
b7b145d | Marco Gorelli | 16 December 2020, 12:39:16 UTC | Remove vars from sample_posterior_predictive, other random refactorings (#4343) Co-authored-by: Michael Osthege <michael.osthege@outlook.com> | 16 December 2020, 12:39:16 UTC |
9dbf9ea | Tomoki Ohtsuki | 16 December 2020, 07:46:12 UTC | Add numerically safe ordered probit distribution. (#4232) * Add numerically safer ordered probit distribution. * Fix styling issue. * Use smaller values for probit & cutpoint testing * fix import sorting * Typo fix * add a line to RELEASE-NOTES.md Co-authored-by: Michael Osthege <michael.osthege@outlook.com> Co-authored-by: Michael Osthege <m.osthege@fz-juelich.de> | 16 December 2020, 07:46:12 UTC |
dbcc49e | Marco Gorelli | 14 December 2020, 16:11:18 UTC | Support Python 3.9, drop Python 3.6 (#4332) * :arrow_up: support py39, drop py36 * :memo: add release note | 14 December 2020, 16:11:18 UTC |
f4552e3 | Marco Gorelli | 14 December 2020, 10:18:48 UTC | add slash workflow (#4330) | 14 December 2020, 10:18:48 UTC |
f778046 | Thomas Wiecki | 14 December 2020, 09:40:54 UTC | ⬆️ UPGRADE: Autoupdate pre-commit config (#4334) Co-authored-by: twiecki <twiecki@users.noreply.github.com> | 14 December 2020, 09:40:54 UTC |
b386d94 | Marco Gorelli | 13 December 2020, 17:59:37 UTC | Fixup mypy errors in sampling.py (#4327) * :label: type sampling * :fire: remove deprecated vars from sample_prior_predictive | 13 December 2020, 17:59:37 UTC |
70fdcf9 | Michael Osthege | 13 December 2020, 12:02:30 UTC | Remove theanof.set_theano_conf and instead use the config context (#4329) * Remove theanof.set_theano_conf and instead use the config context properly | 13 December 2020, 12:02:30 UTC |
6f15cbb | Marco Gorelli | 12 December 2020, 18:26:51 UTC | Fix bug whereby partial traces have fewer draws than would be available (#4318) * add test for _choose_chains * fix bug - choose overall maximum * update release notes * :memo: * :art: * minimise diff * minimise diff * Update pymc3/sampling.py | 12 December 2020, 18:26:51 UTC |
2a38198 | Marco Gorelli | 11 December 2020, 14:26:00 UTC | Fix failing pip-from-conda (#4326) * :ambulance: fix pip-from-conda * Update scripts/generate_pip_deps_from_conda.py | 11 December 2020, 14:26:00 UTC |
eb4be99 | Marco Gorelli | 11 December 2020, 13:55:58 UTC | Only pin theano-pymc in requirements.txt (#4322) * :hammer: add scripts to check theano-pymc pins, to generate requirements-dev from conda, to sort conda deps * :wrench: update conda envs and requirements-dev.txt * :pencil: fix typo * :memo: add docstrings * :memo: update pip-to-conda docstring * noop * :fire: remove pin theano checks, just use requirements.txt * local install in arviz compat * :memo: switch -> checkout, note beginner friendly label * :memo: link to beginner friendly issues * run generate pip deps from conda on all conda env files * :wrench: make sure all conda-envs files generate requirements-dev * :twisted_rightwards_arrows: reorder some tests * :art: remove trailing comma * Update scripts/generate_pip_deps_from_conda.py Co-authored-by: Thomas Wiecki <thomas.wiecki@gmail.com> | 11 December 2020, 13:55:58 UTC |
b953c40 | Sayam Kumar | 11 December 2020, 06:17:15 UTC | Made sample_shape same across all contexts in draw_values (#4305) * Made sample_shape same across all contexts, thereby resolves #3758 * Pass the failing test * Worked on suggestions * Used to_tuple for size * Given a mention in release notes * Update RELEASE-NOTES.md Co-authored-by: Thomas Wiecki <thomas.wiecki@gmail.com> | 11 December 2020, 06:17:15 UTC |
4fd56fd | semohr | 10 December 2020, 08:08:03 UTC | Small compatibility fix for master of Theano-PyMc (#4293) * Update theano-pymc version * Update infer_shape signatures * Updates for new double-underscore config settings Co-authored-by: Brandon T. Willard <brandonwillard@users.noreply.github.com> | 10 December 2020, 08:08:03 UTC |
0431292 | Marco Gorelli | 08 December 2020, 19:31:02 UTC | rel to abs again (#4292) | 08 December 2020, 19:31:02 UTC |
d595d29 | Marco Gorelli | 08 December 2020, 19:29:26 UTC | fix warnings (#4300) | 08 December 2020, 19:29:26 UTC |
648617d | Marco Gorelli | 08 December 2020, 13:55:11 UTC | Add GitHubAction to auto-update hooks, and other cleanups (#4314) * :arrow_up: update pre-commit hooks * try restarting (flaky test?) | 08 December 2020, 13:55:11 UTC |
6a299a3 | mikkelbue | 08 December 2020, 13:47:51 UTC | Added random_seed argument to the sampler in some MLDA tests (#4313) | 08 December 2020, 13:47:51 UTC |
a9806db | Eelke Spaak | 07 December 2020, 15:40:49 UTC | bumping version to 3.10.0 (#4309) * bumping version to 3.10.0 * adding PR 4211 and 4260 to release notes * typo | 07 December 2020, 15:40:49 UTC |
218dd4a | Chandan Kumar Roy | 07 December 2020, 12:39:00 UTC | CLN run isort on source code files (#4239) | 07 December 2020, 12:39:00 UTC |
a5d1697 | Marco Gorelli | 07 December 2020, 11:23:29 UTC | uncomment test from #4297 (#4302) * uncomment test * :art: | 07 December 2020, 11:23:29 UTC |
86dc132 | rpgoldman | 06 December 2020, 17:09:01 UTC | Update README.rst (#4303) Update python version and theano information. Add some links. | 06 December 2020, 17:09:01 UTC |
de47253 | Marco Gorelli | 06 December 2020, 13:41:46 UTC | test on windows and mac (#4269) * run arviz compat on mac and windows * run test_distributions_random.py too on mac * update env file in windows, activate correct env * rename dataset_to_points_dict with DeprecationWarning * fix underlying bug behind DensityDist stackoverflow tl:dr: The hashable helper function did not appropiately deal with tuples (and the test case did not actually test the memoization). In the process of prior-predictive sampling a model involving a DensityDist, the _compile_theano_function function was called with arguments (sd__log_, []). The _compile_theano_function has a pm.memo.memoize-decorator, which relies on the pm.memo.hashable for hashing of typically unhashable objects. The "hashable" function incorrectly handled tuples, eventually causing a stackoverflow error on Windows. * rewrite dict as literal, use list comprehension * punctuation, naming Co-authored-by: Michael Osthege <michael.osthege@outlook.com> | 06 December 2020, 13:41:46 UTC |
580a32a | ricardoV94 | 05 December 2020, 22:30:33 UTC | Implement more robust jitter init (resolves #4107) (#4298) * Implement robust jitter for init_nuts * Update Release notes * Small documentation formatting * Update pymc3/sampling.py Co-authored-by: Thomas Wiecki <thomas.wiecki@gmail.com> | 05 December 2020, 22:30:33 UTC |
3fa3d1f | Eelke Spaak | 05 December 2020, 11:35:36 UTC | fix regression #4273 (#4297) * informative warnings on bound method logp in DensityDist * black * run test on single core only to avoid dill error on windows/macos * adding tests for DensityDist serialize recursion handling * forgot a test | 05 December 2020, 11:35:36 UTC |
df3ae60 | Marco Gorelli | 05 December 2020, 09:13:54 UTC | Add Normal mixture example, fixup check_toc script (#4294) * add gaussian mixture example * update check-toc script | 05 December 2020, 09:13:54 UTC |
9311899 | ricardoV94 | 05 December 2020, 08:27:48 UTC | - Fix regression caused by #4211 (#4285) * - Fix regression caused by #4211 * - Add test to make sure jitter is being applied to chains starting points by default * - Import appropriate empty context for python < 3.7 * - Apply black formatting * - Change the second check_start_vals to explicitly run on the newly assigned start variable. * - Improve test documentation and add a new condition * Use monkeypatch for more robust test * - Black formatting, once again... | 05 December 2020, 08:27:48 UTC |