Reproduce “Neural Markov chain Monte Carlo: Bayesian inversion via normalizing flows and variational autoencoders” 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
Giacomo Bottacini, Matteo Torzoni, Andrea Manzoni
The authors use a learned-likelihood framework for two simulated inverse problems. A variational autoencoder compresses readings into probabilistic features, and a conditional normalizing flow estimates their likelihood for candidate parameters, guiding Bayesian sampling without repeated simulator evaluations. In synthetic bridge tests, damage-class accuracy reaches 93% with posterior means and 97% with MAP; this is not real-world validation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.