Reproduce “VESPER: Online Learning of Free Energies with a Multi-Tempered Variational Objective” 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
Jay L. Kaplan, Gustavo R. Pérez-Lemus, Mengjian Hua, Juan J. de Pablo, Eric Vanden-Eijnden
VESPER couples molecular dynamics sampling and free-energy learning in a neural model that biases molecular walkers and guides auxiliary walkers in collective-variable space. Its tempering levels share the physical free-energy minimizer, allowing training without inter-level reweighting. At 10 ns per replica, alanine dipeptide RMSE was 2.02 ± 0.24 kJ/mol for the four-level ladder, close to S-CFF’s 2.04 ± 0.51; this does not establish faster or universal performance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.