v5.26.1

What's Changed

Maintenance 🔧

• Update python_requires to version 3.11 by @maresb in #7924

Full Changelog: v5.26.0...v5.26.1

v5.26.0

What's Changed

Major Changes 🛠

• Bump PyTensor dependency and drop support for NumPy <2.0, and Python 3.10 by @ricardoV94 in #7910
  • Functions that try to infer inputs of the graph fail with more than one input. This includes Model.compile_fn, gradient, jacobian and hessian_diag.
  • Model.compile_logp now expects all model variables as input, even when only a subset of the logp terms is requested.
  • Many pytensor functions from moved from pytensor.graph.basic to pytensor.graph.traversal, including ancestors, graph_inputs, and toposort related functions.
• Remove deprecated noise parameter for GPs by @williambdean in #7886

New Features 🎉

• Implement logcdf for CensoredRV by @asifzubair in #7884
• Derive logprob for Split operation by @ricardoV94 in #7875

Bugfixes 🪲

• Fix bug in mixture logprob inference with None indices by @asifzubair in #7877

Documentation 📖

• Add model_to_mermaid to docs by @williambdean in #7868
• Use rst code-block over code:: by @williambdean in #7882

Maintenance 🔧

• Show more digits of step size in progress_bar by @ricardoV94 in #7870
• Allow for specification of 'var_names' in 'mock_sample' by @tomicapretto in #7906

New Contributors

• @asifzubair made their first contribution in #7871

Full Changelog: v5.25.1...v5.26.0

v5.25.1

What's Changed

Documentation 📖

• Fix release_notes_to_discourse script by @ricardoV94 in #7862
• fix ICAR docstring latex by @williambdean in #7865
• Use latex for LKJCorr docstrings by @williambdean in #7866

Full Changelog: v5.25.0...v5.25.1

v5.25.0

What's Changed

New Features 🎉

• Add experimental dims module with objects that follow dim-based semantics (like xarray without coordinates) by @ricardoV94 in #7820

Documentation 📖

• Small tweaks to Matern Covariance Documentation by @elc45 in #7859
• Add workflow to publish release notes directly to discourse by @ricardoV94 in #7860

New Contributors

• @elc45 made their first contribution in #7859

Full Changelog: v5.24.1...v5.25.0

v5.24.1

What's Changed

Bugfixes 🪲

• Do not fail with zero-sized arrays in dataset_to_point_list by @ricardoV94 in #7856
• Fix bug with pickling PointFunc by @jessegrabowski in #7858

Full Changelog: v5.24.0...v5.24.1

v5.24.0

What's Changed

Major Changes 🛠

• Remove deprecated functionality by @ricardoV94 in #7822
  • Among others, MutableData, ConstantData, coords_mutable.

New Features 🎉

• Implement specialized Hurdle distribution by @ricardoV94 in #7810
• Model to mermaid diagram with model_to_mermaid by @williambdean in #7826
• Add vectorize_over_posterior to pymc.sampling.forward by @lucianopaz in #7841
• Allow censoring Categorical distributions by @ricardoV94 in #7662

Bugfixes 🪲

• Only add dim_length shared variable when adding a new coordinate by @lucianopaz in #7809
• Fix issues with model graph by @ricardoV94 in #7844
• Fix progressbar with nested compound step samplers by @ricardoV94 in #7776

Documentation 📖

• Mention pymc-marketing in README by @williambdean in #7853

Maintenance 🔧

• Allow accessing wrapped function attributes in PointFunc by @ricardoV94 in #7823
• implement display method for pm.Model for builtin marimo support by @williambdean in #7830
• Allow var_names to be propagated to nutpie sampler by @PabloRoque in #7850
• Better guesses for why logp has RVs by @ricardoV94 in #7663
• Allow debug evaling IR logp graphs by @ricardoV94 in #7666

New Contributors

• @PabloRoque made their first contribution in #7850

Full Changelog: v5.23.0...v5.24.0

v5.23.0

What's Changed

Major Changes 🛠

• Bump PyTensor dependency by @ricardoV94 in #7792
  • This release raises error for some forms of runtime broadcasting in advanced set_subtensor operations that were previously allowed.

New Features 🎉

• Transform to remove Minibatch from model by @zaxtax in #7746
• Implement reversed and strictly positive ordered transform by @velochy in #7759
• Add public testing function to mock sample by @williambdean in #7761

Bugfixes 🪲

• Default to SVD for MvNormal in Latent GP conditionals by @michaelosthege in #7755

Documentation 📖

• Add gp util module to docs by @williambdean in #7763
• Update docs theme by @OriolAbril in #7781
• Tweak Model.profile docstring and type hint by @williambdean in #7795

Maintenance 🔧

• Do not include rvs in symbolic normalizing constant by @ricardoV94 in #7787

New Contributors

• @star1327p made their first contribution in #7650

Full Changelog: v5.21.2...v5.23.0

v5.22.0

What's Changed

Major Changes 🛠

• Bump PyTensor dependency by @ricardoV94 in #7740
  • This release changes how JAX dispatch of random variables are implemented. Custom dispatches will have to be updated

Documentation 📖

• Add return type overload for sample_posterior_predictive by @nataziel in #7710
• Replaced "log-likelihood" with "distribution" in for all the classes where distributions is more appropriate rather than log-likelihood. by @tanishy7777 in #7573

Maintenance 🔧

• Allow passing graph level attributes to graphviz by @zaxtax in #7741
• Add warning if a Minibatched variable is used without total_size by @zaxtax in #7742

New Contributors

• @tanishy7777 made their first contribution in #7573

Full Changelog: v5.21.1...v5.22.0

v5.21.2

Important

This was an erroneous duplicate release of v5.22.0.

v5.21.1

What's Changed

New Features 🎉

• Support PyTensor deterministic operations as observations by @wd60622 in #7656
• Allow draws from Weibull, MvStudentT, LKJCorr and LKJCholeskyCovRV in alternative backends by @ricardoV94 in #7685

Maintenance 🔧

• Make do interventions shared variables by default by @ricardoV94 in #7596
• Remove setupegg.py by @Armavica in #7697
• Expand logging test cases for sample_prior_predictive and add return type overloads by @nataziel in #7707
• Allow compile_kwargs in sample_smc by @jessegrabowski in #7702

Full Changelog: v5.21.0...v5.21.1