Efficient and Unbiased Sampling from Boltzmann Distributions via Variance-Tuned Diffusion Models
Fengzhe Zhang, Laurence I. Midgley, José Miguel Hernández-Lobato · 2025
Read Efficient and Unbiased Sampling from Boltzmann Distributions via Variance-Tuned Diffusion Models by Fengzhe Zhang, Laurence I. Midgley, José Miguel Hernández-Lobato (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-21005.txt.
Description
Fengzhe Zhang, Laurence Midgley, and José Miguel Hernández-Lobato improve a way to draw fair samples from physics distributions, in a paper published in 2025. Diffusion models can sample such distributions, but small errors in them bias later averages. The standard fix weighs each sample, yet weighing needs a costly equation solve that grows worse in high dimensions.
Their method tunes the noise at each step of a trained model, before use, to shrink the gap between its forward and reverse paths. Each full path then gets one weight, giving unbiased averages with little extra cost. On three test problems, from toy particles to a small molecule, useful sample shares beat plain fixes at a fraction of the cost. The gain was largest on the simplest problem and smallest on the molecule.
Catalog record
| CREATOR | Fengzhe Zhang, Laurence I. Midgley, José Miguel Hernández-Lobato |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 59 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-21005