pathint is a jax implementation of the path integral sampler,
a method based on the Schrödinger bridge problem for sampling from (unnormalized)
probability densities. Behind the scenes it relies on diffrax
to handle the stochastic differential equations, equinox
for the neural networks and optax for training.
Check out the docs badge above for more details, or try out the scripts applying the method to some low-dimensional problems (runnable on a laptop).
Running
git clone git@github.com:adam-coogan/path-integral-sampler.git
cd path-integral-sampler
pip install .
will install the pathint package.