Skip to content

Latest commit

 

History

History
20 lines (13 loc) · 811 Bytes

README.md

File metadata and controls

20 lines (13 loc) · 811 Bytes

Easy Rejection Sampling

This repository hosts our research code for the Easy Rejection Sampling (ERS) algorithm. The goal is that ERS be easy from a user's perspective - in that they only need to pass a function that computes the (unnormalized) density they wish to draw samples from, and optionally the domain of that function if not infinite. In this code any function defined in a JAX compatible way should work.

This is research quality code with no warranty or support.

Citations

If you use the ERS algorithm or code, please cite our work!

@inproceedings{ERS,
author = {Raff, Edward and McLean, Mark and Holt, James},
booktitle = {European Conference on Artificial Intelligence (ECAI)},
title = {{An Easy Rejection Sampling Baseline via Gradient Refined Proposals}},
year = {2023}
}