Reproduce “Stochastic Gradient MCMC for Nonlinear State Space Models” 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
Christopher Aicher, Srshti Putcha, Christopher Nemeth, Paul Fearnhead, Emily B. Fox
The authors introduce buffered stochastic-gradient Langevin dynamics for parameter inference in nonlinear state-space models. The method filters a wider window around a short subsequence to provide hidden-state context, but accumulates gradient terms only within the subsequence. On a synthetic Gaussian model with 10⁶ observations, buffering lowers log₁₀ KSD from 4.68 to 3.43 under fixed compute; exchange-rate GARCH does not improve.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.