v5.19.1

What's Changed

Bugfixes 🪲

• Bump numpy version due to use of Generator.spawn only available in >=1.25 by @ricardoV94 in #7607

Maintenance 🔧

• Exponential scale default to 1.0 by @ricardoV94 in #7604

Full Changelog: v5.19.0...v5.19.1

v5.19.0

What's Changed

Major Changes 🛠

• Speedup sample and allow specifying compile_kwargs by @ricardoV94 in #7578. This introduces several major changes related to step samplers:
  • internal uses of logp_dlogp_function now work with raveled inputs. External use will issue a warning unless ravel_inputs is specified explicitly. Eventually it will only be possible to use ravel_inputs=True.
  • Step samplers arguments besides vars must be passed by keyword
  • RaveledVars.point_map_info is now a 4-n tuple, with size introduced.
  • assign_step_method does not call instantiate_steppers, but returns arguments needed for the latter.
  • Allow passing compile_kwargs to sample which is then forwarded to the step samplers functions

Bugfixes 🪲

• Fix error in find_measurable_bitwise by @ricardoV94 in #7585

Documentation 📖

• Explain difference between BinaryMetropolis and BinaryGibbsMetropolis by @ricardoV94 in #7586
• Add example on freeze_data_and_dims by @ricardoV94 in #7594

Maintenance 🔧

• Register the overloads added by CustomDist so it works with multiprocess with SMC by @EliasRas in #7241

New Contributors

• @EliasRas made their first contribution in #7241

Full Changelog: v5.18.2...v5.19.0

v5.18.2

What's Changed

Maintenance 🔧

• Allow interdependent initial points from same OpFromGraph node by @ricardoV94 in #7569

Full Changelog: v5.18.1...v5.18.2

v5.18.1

What's Changed

Bugfixes 🪲

• Harmonize HSGP.prior dimension names and order by @juanitorduz in #7562
• Fix bug in implicit_size_from_params by @ricardoV94 in #7564
• Do not mutate Scan inner graph when deriving logprob by @ricardoV94 in #7575

Maintenance 🔧

• Make pytensorf.constant_fold unconditional by @ricardoV94 in #7568

Full Changelog: v5.18.0...v5.18.1

v5.18.0

What's Changed

Major Changes 🛠

• Add step method state and make step results deterministic with respect to it by @lucianopaz in #7508
• Remove deprecated features by @Armavica in #7533

New Features 🎉

• Allow copy and deepcopy of PYMC models by @Dekermanjian in #7492
• Add mean dispatch for pymc distributions by @lucianopaz in #7530
• Allow for passing of backend and gradient_backend to nutpie by @fonnesbeck in #7535
• Derive logprob of matmul by @ricardoV94 in #7542

Maintenance 🔧

• Simplify Model new and metaclass by @thomasaarholt in #7473

Full Changelog: v5.17.0...v5.18.0

v5.17.0

What's Changed

Major Changes 🛠

• Cleanup logprob module by @ricardoV94 in #7443
• Deprecation warning for find_constrained_prior by @aloctavodia in #7458
• Remove deprecated Distribution kwargs by @thomasaarholt in #7488
• Allow Minibatch of derived RVs and deprecate generators as data by @ricardoV94 in #7480

New Features 🎉

• Implement specialized MvNormal density based on precision matrix by @ricardoV94 in #7345
• Infer logcdf of discrete transformations by @ricardoV94 in #7444
• Allow more distributions to be truncated by @ricardoV94 in #7476
• Add multi-output support to GP Latent by @AlexAndorra in #7471

Bugfixes 🪲

• Workaround PyTensor bug in vectorize of logp graph by @ferrine in #7415
• Print OP name for unnamed RVs instead of raising AssertionErrors by @lucianopaz in #7428
• Introduce value variables in logprob IR by @ricardoV94 in #7491
• Fix latex rendering of variables with underscore in name by @Dekermanjian in #7501

Documentation 📖

• Update Twitter Link to X by @Demon-Sheriff in #7451
• Changed PytTensor import alias to pt in PyMC overview by @Krupakar-Reddy-S in #7452
• Clarified 0-based indexing requirement for OrderedLogistic and OrderedProbit by @kdotmanoj in #7457
• Fix typo in the docstring of the Beta distribution by @erik-werner in #7469
• Update GLM_linear.ipynb to correct the URL for statsmodels by @seyedrezamirkhani in #7490

Maintenance 🔧

• Reduce blackjax sampling memory usage by @junpenglao in #7407
• Enforce custom initval in SMC by @tvwenger in #7439
• Blackjax sampler fix for breaking change / enable progress bar under parallel chain_method by @andrewdipper in #7453
• Add ability to set mode in check_start_vals by @lucianopaz in #7482
• Add compile_kwargs to compute_log_density functions by @lucianopaz in #7483
• Make zip strict in apply_function_over_dataset by @ricardoV94 in #7485
• Return InferenceData when there are no variables sampled and extend=True by @OriolAbril in #7509
• Build the package with python -m build by @Armavica in #7522

New Contributors

• @Demon-Sheriff made their first contribution in #7451
• @Krupakar-Reddy-S made their first contribution in #7452
• @kdotmanoj made their first contribution in #7457
• @seyedrezamirkhani made their first contribution in #7490
• @Dekermanjian made their first contribution in #7501

Full Changelog: v5.16.2...v5.17.0

v5.16.2

What's Changed

Bugfixes 🪲

• Do not consider dims without coords volatile if length has not changed by @JasonTam in #7381
• Fix bug with multiple minibatch variables by @ricardoV94 in #7408

Documentation 📖

• Add Myst cross reference link example to Jupyter style guide by @AlexAndorra in #7235

Maintenance 🔧

• Add Conda Downloads Badge by @PatriceJada in #7378
• Refactor model graph and allow suppressing dim lengths by @wd60622 in #7392
• Reduce JAX post-processing memory usage by @andrewdipper in #7311

New Contributors

• @PatriceJada made their first contribution in #7378

Full Changelog: v5.16.1...v5.16.2

v5.16.1

What's Changed

Bugfixes 🪲

• Avoid spurious deprecation warning in CustomDist by @ricardoV94 in #7391
• Assert ndim and number of dims match by @ricardoV94 in #7390

Full Changelog: v5.16.0...v5.16.1

v5.16.0

What's Changed

Major Changes 🛠

• Standardize draws as parameter in sample_prior_predictive by @wd60622 in #7366
• Allow opting out of model nesting by setting model=None by @ricardoV94 in #7352
• Bump PyTensor dependency (which changes signature of RandomVariables) by @ricardoV94 in #7370
• Move CustomDist logic to dedicated module and docs section by @markgreene74 in #7363

New Features 🎉

• HSGP improvements by @AlexAndorra in #7335
• Add more idata attributes for JAX samplers by @osyuksel in #7360

Bugfixes 🪲

• Fix bug with compute_p for partially observed OrderedLogistic and OrderedProbit variables by @ricardoV94 in #7349
• Avoid repeated status polling in smc by @aseyboldt in #7351
• Fix bug when freezing model with partially observed RVs by @ricardoV94 in #7388

Documentation 📖

• Fix broken links by @Armavica in #7347

Maintenance 🔧

• Allow customizing truncation max_n_steps in Hurdle Mixtures by @tjburch in #7339
• Add refresh call to progress bar in apply_function_over_dataset by @fonnesbeck in #7343
• HSGP misc fixes by @bwengals in #7342
• Fixed myst version to less than 1.0.0 (fixes graph squashing) by @hevansDev in #7356
• Update readme example by @kiramclean in #7358
• Fix import pymc.testing by @bomtall in #7357
• Minor text improvements to the Introductory Overview of PyMC notebook (pymc_overview.ipynb) by @alimanfoo in #7361
• Remove default initval for Flat variables by @ricardoV94 in #7379
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in #7371

New Contributors

• @hevansDev made their first contribution in #7356
• @kiramclean made their first contribution in #7358
• @bomtall made their first contribution in #7357
• @alimanfoo made their first contribution in #7361
• @osyuksel made their first contribution in #7360
• @markgreene74 made their first contribution in #7363

Full Changelog: v5.15.1...v5.16.0

What's Changed

Major Changes 🛠

• Rename samples argument to draws in sample_prior_predictive by @wd60622 in #7366
• Allow opting out of model nesting by setting model=None by @ricardoV94 in #7352
• Bump PyTensor dependency (which changes signature of RandomVariables) by @ricardoV94 in #7370
• Move CustomDist logic to dedicated module and docs section by @markgreene74 in #7363

New Features 🎉

• HSGP improvements by @AlexAndorra in #7335
• Add more idata attributes for JAX samplers by @osyuksel in #7360

Bugfixes 🪲

• Fix bug with compute_p for partially observed OrderedLogistic and OrderedProbit variables by @ricardoV94 in #7349
• Avoid repeated status polling in smc by @aseyboldt in #7351
• Fix bug when freezing model with partially observed RVs by @ricardoV94 in #7388

Documentation 📖

• Fix broken links by @Armavica in #7347
• Fixed myst version to less than 1.0.0 (fixes graph squashing) by @hevansDev in #7356
• Minor text improvements to the Introductory Overview of PyMC notebook (pymc_overview.ipynb) by @alimanfoo in #7361

Maintenance 🔧

• Allow customizing truncation max_n_steps in Hurdle Mixtures by @tjburch in #7339
• Add refresh call to progress bar in apply_function_over_dataset by @fonnesbeck in #7343
• HSGP misc fixes by @bwengals in #7342
• Update readme example by @kiramclean in #7358
• Fix import pymc.testing by @bomtall in #7357
• Remove default initval for Flat variables by @ricardoV94 in #7379

New Contributors

• @hevansDev made their first contribution in #7356
• @kiramclean made their first contribution in #7358
• @bomtall made their first contribution in #7357
• @alimanfoo made their first contribution in #7361
• @osyuksel made their first contribution in #7360
• @markgreene74 made their first contribution in #7363

Full Changelog: v5.15.1...v5.16.0

v3.11.6

What's Changed

• Add legacy messages to PyMC3 by @aloctavodia in #7338
• Bump release pymc3 by @aloctavodia in #7340

Full Changelog: v3.11.5...v3.11.6