Reproduce “ENFR: Evidence Networks with Flow-Matched References for Amortized Simulation-Based Model Inference” 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

Kai Lehman, Jameson Dong, Carolina Cuesta-Lazaro, Niall Jeffrey, Benjamin D. Wandelt, Francisco Villaescusa-Navarro, Sven Krippendorf, Jochen Weller

ENFR estimates absolute Bayesian evidence for simulator-based models whose likelihoods cannot be evaluated. It fits a flow to simulator samples, then learns a density-ratio correction by classifying simulated and flow-generated data. At inference, it combines the correction with the flow’s density. On 100 held-out BAO ΛCDM simulations, MultiENFR’s log-evidence RMSE is 0.52, versus 1.34 for flow matching; this is benchmark-specific.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.