Learning Rate Free Sampling in Constrained Domains

NeurIPS 2023

Louis Sharrock   ·   Lester Mackey   ·   Christopher Nemeth  

Description

This repository contains code to reproduce the results of the numerical experiments contained in Sharrock et al. (2023). The code for each of two algorithms can be found in separate directories:

• The code for Coin MSVGD is contained in coin-msvgd.
• The code for Coin MIED is contained in coin-mied.

Citation

If you find the code in this repository useful for your own research, please consider citing our paper:

@InProceedings{Sharrock2023,
  title = 	 {Learning Rate Free Sampling in Constrained Domains},
  author =       {Sharrock, Louis and Mackey, Lester and Nemeth, Christopher},
  booktitle = 	 {Proceedings of The 37th Conference on Neural Information Processing Systems},
  year =         {2023},
  city =         {New Orleans, LA},
}

Acknowledgements

Our implementations of Coin MSVGD and Coin MIED are based on existing implementations of MSVGD and MIED. We gratefully acknowledge the authors of the following papers for their open source code:

• J. Shi, C. Liu and L. Mackey. Sampling with Mirrored Stein Operators. ICLR, 2022. [Paper] | [Code].
• L. Li, Q. Liu, A. Korba, M. Yurochkin and J. Solomon. Sampling with Mollified Interaction Energy Descent. ICLR, 2023. [Paper] | [Code].