v5.6.0
What's Changed
Major Changes 🛠
- Rewrite logp graph before taking the gradient by @dehorsley in #6736
- Support automatic imputation for multivariate and symbolic distributions by @ricardoV94 in #6797
New Features 🎉
- Probabilty inference for arc transformations by @LukeLB in #6775
- Add icdf function for Cauchy and Logistic distributions by @amyoshino in #6747
- Add icdf functions for Lognormal, Half Cauchy and Half Normal distributions by @amyoshino in #6766
- Add icdf functions for Moyal, Gumbel, Triangular and Weibull distributions by @amyoshino in #6802
- Allow non-scalar measurable switch mixtures by @ricardoV94 in #6796
Bugfixes 🪲
- squeeze dim_0 dimensions for scalars away by @TimOliverMaier in #6764
- NotImplementedError for icdf of non-injective MeasurableTransforms by @ricardoV94 in #6793
Documentation 📖
- Logprob docs by @ricardoV94 in #6762
- Simplify Potential docstrings and examples by @ricardoV94 in #6772
- Fix broken link by @eeriksp in #6767
- speed-up doc building and fix several issues by @OriolAbril in #6781
Maintenance 🔧
- Removed **kwargs from sample_numpyro_nuts and sample_blackjax_nuts by @fonnesbeck in #6768
- Sponsor: add ODSC logo to README.rst by @reshamas in #6770
- Fix typos in PolyaGamma's docstring by @aleicazatti in #6672
- fixing links to images in the pymc_pytensor.ipynb notebook by @jaharvey8 in #6739
- Better coverage for float32 tests by @ferrine in #6780
- Allow creating
SymbolicRandomVariable
s insideCustomDist
by @ricardoV94 in #6805
New Contributors
- @jaharvey8 made their first contribution in #6739
- @eeriksp made their first contribution in #6767
Full Changelog: v5.5.0...v5.6.0