Theano-PyMC is a Python library that allows you to define, optimize, and efficiently evaluate mathematical expressions involving multi-dimensional arrays. It can use GPUs and perform efficient symbolic differentiation.

This is a fork of the original Theano library that is being maintained by the PyMC team.

Warning

The name of this repository/project may change in the near future.

Features

A hackable, pure-Python codebase
Extensible graph framework suitable for rapid development of custom symbolic optimizations
Implements an extensible graph transpilation framework that currently provides compilation to C and JAX JITed Python functions
Built on top of one of the most widely-used Python tensor libraries: Theano

Getting started

The documentation is located here. If the theano-doc package is installed, it is also available locally.

Warning

As development progresses, the legacy documentation may become less applicable.

Installation

The latest release of Theano-PyMC can be installed from PyPI using pip:

pip install Theano-PyMC

Or via conda-forge:

conda install -c conda-forge theano-pymc

The current development branch of Theano-PyMC can be installed from GitHub, also using pip:

pip install git+https://github.com/pymc-devs/Theano-PyMC

For platform-specific installation information see the legacy documentation here.

Support

The PyMC group operates under the NumFOCUS umbrella. If you want to support us financially, you can donate here.