Brancher

A user-centered Python package for differentiable probabilistic inference

Here is an example of the Brancher syntax:

a = Normal(0, 1, name="a")
b = Normal(0, 1, name="b")
c = Normal(a**2 + sin(b), exp(b), name="c")

Brancher is:

Easy to extend modeling framework with
GPU-accelerated PyTorch backend

Easy to use modern toolbox with Pandas and Seaborn support

Easy to learn symbolic interface with math-like syntax

Developed at Donders Institute, Radboud University, Nijmegen, Netherlands.

Learn Brancher in Google Colab with the following tutorials (more to come soon!):

After installing PyTorch, Brancher can be installed from either PyPI:

pip install brancher

Explore Brancher in Google Colab with the following examples (more to come soon!):

Meet the Brancher developers and maintainers:

Assistant Professor at
Donders Inst, Radboud Uni.

Ph.D. Candidate at
Utrecht Uni and Radboud Uni.

ASSISTANT PROFESSOR AT
DONDERS INST, RADBOUD UNI.

ASSISTANT PROFESSOR AT
DONDERS INST, RADBOUD UNI.

ASSISTANT PROFESSOR AT
DONDERS INST, RADBOUD UNI.

PROFESSOR AT
DONDERS INST, RADBOUD UNI.