Reproduce “Diffusion-based Annealed Boltzmann Generators : benefits, pitfalls and hopes” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Louis Grenioux, Maxence Noble
The authors test deterministic transitions along diffusion paths for Monte Carlo sampling. On controlled multimodal Gaussian mixtures with an exact diffusion model, a probability-flow ODE map with Hutchinson-estimated density correction outperforms first-order stochastic transitions and is competitive with second-order methods. With learned models, these methods generally do not beat tempering and often underperform direct reverse diffusion sampling, implicating density errors.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.