Reproduce “Jarzynski Reweighting and Sampling Dynamics for Training Energy-Based Models: Theoretical Analysis of Different Transition Kernels” 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
Davide Carbone
The authors develop a path-reweighting method for estimating expectations in energy-based models when short-run Markov-chain samples may not match the current model distribution. The method tracks transition-path weights across time-dependent kernels. Under stated assumptions, weighted expectations recover the model expectation and mean weights give the partition-function ratio. This is a population identity, not a finite-sample accuracy or training-speed guarantee.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.