v5.15.1

What's Changed

Bugfixes 🪲

• Allow mutable shape in partially imputed variables by @ricardoV94 in #7314
• Fix bug in Truncated with Deterministic inputs by @ricardoV94 in #7315
• Fix spurious inputs failure when building Truncated variables by @ricardoV94 in #7328

Documentation 📖

• Add docstring to pymc.sampling.jax.sample_jax_nuts by @andrewdipper in #7313
• Updated pymc.step_methods.slicer.Slice docstring by @ksk-17090k1 in #7322
• Fix error in logcdf example by @ricardoV94 in #7336

Maintenance 🔧

• Add blas_cores argument to pm.sample by @aseyboldt in #7318
• Update pymc.sample docstring to fix incorrect property name by @treszkai in #7321
• Implement CustomProgress that does not output empty divs when disabled by @tomicapretto in #7290
• Use more numerically stable function in TruncatedNormal area computation by @velochy in #7305
• Get rid of most intX usages by @ricardoV94 in #7330
• Fix typing in multiple places by @michaelosthege in #7333
• Split convert observed data by @michaelosthege in #7334
• Add test for freeze_rv_and_dims in JAX backend by @ricardoV94 in #7324

New Contributors

• @andrewdipper made their first contribution in #7313
• @ksk-17090k1 made their first contribution in #7322
• @lhelleckes made their first contribution in #7299

Full Changelog: v5.15.0...v5.15.1

v5.15.0

What's Changed

Major Changes 🛠

• Add warning about change of sign in hessian functions by @aseyboldt in #6312

New Features 🎉

• Skewed Student-T distribution by @fonnesbeck in #7252
• Allow customizing style of model_graph nodes by @wd60622 in #7302

Bugfixes 🪲

• Fix bug when freezing_rv_and_dims after a model transformation by @ricardoV94 in #7296
• Fix need for dummy data in sample_posterior_predictive by @ricardoV94 in #7265

Maintenance 🔧

• Suggest var_names when using deprecated API for partial traces by @lancechua in #7289

New Contributors

• @lancechua made their first contribution in #7289

Full Changelog: v5.14.0...v5.14.1

v5.14.0

What's Changed

Major Changes 🛠

• Implement default_transform and transform argument for distributions by @aerubanov in #7207

New Features 🎉

• Allow freezing only subset of data and dims by @ricardoV94 in #7248

Bugfixes 🪲

• Fix gradient bug in models with max operation by @ricardoV94 in #7261
• Convert random variables to value variables so pm.sample(var_names) works correctly by @tomicapretto in #7284
• Fix draws of Weibull when alpha and beta implicitly define size by @tomicapretto in #7288

Documentation 📖

• Add introduction section to Transformations API page to document role of Transforms by @mkusnetsov in #7232
• Fix link to API quick start. Closes #7266. by @twiecki in #7275

Maintenance 🔧

• Fix typo in progressbar message by @zaxtax in #7271
• Add time remaining column to progress bars by @fonnesbeck in #7273

New Contributors

• @mkusnetsov made their first contribution in #7232

Full Changelog: v5.13.1...v5.14.0

v5.13.1

What's Changed

New Features 🎉

• Implement several RandomVariables as SymbolicRandomVariables by @ricardoV94 in #7239

Bugfixes 🪲

• Fix bug in compute_deterministics by @ricardoV94 in #7249
• Fix typo of Data containers in model graph by @tomicapretto in #7250

Maintenance 🔧

• Fix DOI visibility badge in README by @AndreaBlengino in #7245

New Contributors

• @AndreaBlengino made their first contribution in #7245

Full Changelog: v5.13.0...v5.13.1

v5.13.0

What's Changed

Major Changes 🛠

• Make coords and data always mutable by @ricardoV94 in #7047
• Bump PyTensor dependency and drop support for python 3.9 by @ricardoV94 in #7227

New Features 🎉

• Add compute_deterministics helper by @ricardoV94 in #7238
• Allow Truncation of CustomDist by @ricardoV94 in #6947

Bugfixes 🪲

• Fix join logp for multivariate RVs by @ricardoV94 in #7215

Documentation 📖

• Add HSGPPeriodic to Docs by @juanitorduz in #7230
• Add graphic for contributors. by @twiecki in #7229
• Fix link and typo in developer_guide.md by @pipme in #7219
• Fix type-hint typo in CustomDist docs by @hchen19 in #7243

Maintenance 🔧

• Use exponentially scaled modified Bessel function in periodic HSGPs by @dehorsley in #7228
• Replace fastprogress progress bars with rich by @fonnesbeck in #7233
• Make default STEP_METHODS a list that can be modified by @ricardoV94 in #7231

New Contributors

• @pipme made their first contribution in #7219
• @hchen19 made their first contribution in #7243

Full Changelog: v5.12.0...v5.13.0

v5.12.0

What's Changed

Major Changes 🛠

• Bump PyTensor dependency and support Python 3.12 by @ricardoV94 in #7203

New Features 🎉

• Add _print_name to Truncated and CustomDists by @ricardoV94 in #7205
• Add icdf functions for Beta, Gamma, Chisquared and StudentT distributions by @amyoshino in #6845
• Add var_names argument to sample by @fonnesbeck in #7206

Bugfixes 🪲

• add _moment function for bacward backward compatibility by @aerubanov in #7216

Documentation 📖

• Fix expression for the variance in ZINB docstring by @aloctavodia in #7200
• List more math function in API docs by @brandonhorsley in #7211
• Improve example on expanded sample_posterior_predictive by @ricardoV94 in #7222

Maintenance 🔧

• Removing unnecessary comp_shape from class NormalMixture by @mohammed052 in #7098

New Contributors

• @mohammed052 made their first contribution in #7098
• @brandonhorsley made their first contribution in #7211

Full Changelog: v5.11.0...v5.12.0

v5.11.0

What's Changed

Major Changes 🛠

• Rename moment to support_point by @aerubanov in #7166
• Remove intX and floatX calls from distributions by @aerubanov in #7114
• Remove deprecated Bound distribution by @ricardoV94 in #7176
• Remove deprecated sample_posterior_predictive_w by @aloctavodia in #7179

New Features 🎉

• Add option to save model graph to an image by @aloctavodia in #7158
• Run convergence checks when using JAX samplers by @ricardoV94 in #7165

Bugfixes 🪲

• Fix compute_log_prior in models with Deterministics by @OriolAbril in #7168
• Fix error in warn_treedepth when using multiple NUTS samplers by @ricardoV94 in #7182
• Fix error in dataset_to_point_list when chain, draw are not the leading dims by @OriolAbril in #7180
• Refactor get_tau_sigma and support lists of variables by @tvwenger in #7185

Documentation 📖

• Update link of Bayesian Analysis with Python book to third edition by @aloctavodia in #7161
• Fix typo in developer guide by @erik-werner in #7170
• Add examples explaining advanced applications of sample_posterior_predictive by @ricardoV94 in #7014
• Updated docstrings in pymc.model.core.Model by @apalermo01 in #7118
• Explain how to obtain the model graphviz in a non-Ipython environment by @Armavica in #7181
• Improve HSGP and ZeroInflated / Hurdle distributions docs by @AlexAndorra in #7189

Maintenance 🔧

• Reduce number of minimum draw in warning message by @aloctavodia in #7163
• Add sitemap to docs by @OriolAbril in #7171
• Add gufunc signature to SymbolicRandomVariables by @ricardoV94 in #7159

New Contributors

• @apalermo01 made their first contribution in #7118

Full Changelog: v5.10.4...v5.11.0

v5.10.4

What's Changed

New Features 🎉

• Add a flag to LKJCorr to return the unpacked correlation matrix by @velochy in #7100
• Allow jitter boolean to be set through nuts_sampler_kwargs by @VMBoehm in #7083
• Logprob derivation of Min for Discrete IID distributions by @Dhruvanshu-Joshi in #6968
• Implement compute_log_prior utility by @aloctavodia in #7149

Bugfixes 🪲

• GP do not fail when xdims cannot be constant folded by @bwengals in #7111
• Fix copying of shared variables in fgraph_from_model by @ricardoV94 in #7153

Documentation 📖

• Suppressed pip install bambi output in GLM core notebook by @jaharvey8 in #7090
• Gitpod instructions: use latest version of pymc by @reshamas in #7106
• Moved 'Implementing a Random Variable' to the last position in 'How To' section by @OmGhadge in #7108
• Add Google Scholar link for newest article by @aloctavodia in #7156

Maintenance 🔧

• Extend dataset_to_point_dict to accept both dataset and dict of dataarray by @OriolAbril in #7097
• Fix progressbar bug for parallel SMC sampling by @jucor in #7079
• Rectify return type hints in logprob module rewrites by @AryanNanda17 in #7125
• Deprecate block_diag from math module in favor of PyTensor by @AryanNanda17 in #7132

New Contributors

• @jucor made their first contribution in #7079
• @pre-commit-ci made their first contribution in #7110
• @velochy made their first contribution in #7100
• @VMBoehm made their first contribution in #7083
• @OmGhadge made their first contribution in #7108
• @AryanNanda17 made their first contribution in #7125

Full Changelog: v5.10.3...v5.10.4

v5.10.3

What's Changed

New Features 🎉

• Implement logcdf for TruncatedNormal by @LukeLB in #7034

Bugfixes 🪲

• Fix issue with sampling of PartialObservedRVs by @ricardoV94 in #7071

Full Changelog: v5.10.2...v5.10.3

v5.10.2

What's Changed

New Features 🎉

• Allow batched scalar sigma in ZeroSumNormal by @ricardoV94 in #7063

Bugfixes 🪲

• Fix failing default transform for LKJCorr by @juanitorduz in #7065

Full Changelog: v5.10.1...v5.10.2