Reproduce “AutoStep: Locally adaptive involutive MCMC” 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
Tiange Liu, Nikola Surjanovic, Miguel Biron-Lattes, Alexandre Bouchard-Côté, Trevor Campbell
AutoStep selects a step size each iteration from the current state. It includes its tuning parameter in an augmented state and retains a Metropolis correction, preserving the target distribution conditionally. In an RWMH experiment, its symmetric rule kept acceptance above 10% across norms, unlike the earlier asymmetric rule; this result applies only to that experiment.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.