Reproduce “Provable Convergence and Limitations of Geometric Tempering for Langevin Dynamics” 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
Omar Chehab, Anna Korba, Austin Stromme, Adrien Vacher
The authors analyze Langevin sampling along geometric paths from an easy proposal to a target whose normalizing constant may be unknown. The path’s score combines endpoint scores, making it tractable. Yet intermediate distributions can develop sampling obstacles: for a specified unimodal target, a lower bound requires exponentially large accumulated Langevin time while temperature remains short of its endpoint.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.