zalandoresearch / JaxRK

RKHS feature vectors, operators, and statistical models using JAX for automatic differentiation

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

JaxRK

JaxRK is a library for working with (vectors of) RKHS elements and RKHS operators using JAX for automatic differentiation. This library includes implementations of kernel transfer operators and conditional density operators.

Installation

First you have to make sure to have jax and jaxlib installed. Please follow the JAX installation instructions depending on whether you want a CPU or GPU/TPU installation. After that you only need

$ pip install jaxrk

Quick start examples

For some examples of what you can do with JaxRK, see examples/Quick_start.ipynb.

Development

To help in developing JaxRK, clone the github repo and change to the cloned directory on the command line. Then

$ pip install -e .
$ pytest tests/

will install the package into your python path. Changes to files in the directory are reflected in the python package when loaded.

About

RKHS feature vectors, operators, and statistical models using JAX for automatic differentiation

License:MIT License


Languages

Language:Jupyter Notebook 50.7%Language:Python 49.3%