Scalable inference with Autoregressive Neural Ratio Estimation

Abstract

In recent years, there has been a remarkable development of simulation-based inference (SBI) algorithms, and they have now been applied across a wide range of astrophysical and cosmological analyses. There are a number of key advantages to these methods, centred around the ability to perform scalable statistical inference without an explicit likelihood. In this work, we propose two technical building blocks to a specific sequential SBI algorithm, truncated marginal neural ratio estimation (TMNRE). In particular, first we develop autoregressive ratio estimation with the aim to robustly estimate correlated high-dimensional posteriors. Secondly, we propose a slice-based nested sampling algorithm to efficiently draw both posterior samples and constrained prior samples from ratio estimators, the latter being instrumental for sequential inference. To validate our implementation, we carry out inference tasks on three concrete examples: a toy model of a multi-dimensional Gaussian, the analysis of a stellar stream mock observation, and finally, a proof-of-concept application to substructure searches in strong gravitational lensing. In addition, we publicly release the code for both the autoregressive ratio estimator and the slice sampler.

12 pages. 6 figures. Codes: swyft is available at https://github.com/undark-lab/swyft , torchns is available at https://github.com/undark-lab/torchns - v2: version accepted by MNRAS

References in corpus (19)

• Array Programming with NumPy
• The James Webb Space Telescope
• PolyChord: nested sampling for cosmology
• MADE: Masked Autoencoder for Distribution Estimation
• The shape of the inner Milky Way halo from observations of the Pal 5 and GD-1 stellar streams
• Analytic models of plausible gravitational lens potentials
• Nested sampling for physical scientists
• Nuisance hardened data compression for fast likelihood-free inference
• Gaps in globular cluster streams: giant molecular clouds can cause them too
• Milky Way Mass and Potential Recovery Using Tidal Streams in a Realistic Halo
• Large Synoptic Survey Telescope: Dark Energy Science Collaboration
• Truncated proposals for scalable and hassle-free simulation-based inference
• Debiasing Standard Siren Inference of the Hubble Constant with Marginal Neural Ratio Estimation
• Solving high-dimensional parameter inference: marginal posterior densities & Moment Networks
• Targeted Likelihood-Free Inference of Dark Matter Substructure in Strongly-Lensed Galaxies
• Overview of the European Extremely Large Telescope and its instrument suite
• What to do when things get crowded? Scalable joint analysis of overlapping gravitational wave signals
• Detection is truncation: studying source populations with truncated marginal neural ratio estimation
• The Cherenkov Telescope Array

Cited by in corpus (7)

• The Science of the Einstein Telescope
• Cosmology with Persistent Homology: a Fisher Forecast
• Mapping reionization bubbles in the JWST era II: inferring the position and characteristic size of individual bubbles
• Near-instantaneous Atmospheric Retrievals and Model Comparison with FASTER
• STAR NRE: Solving supernova selection effects with set-based truncated auto-regressive neural ratio estimation
• Observable Optimization for Precision Theory: Machine Learning Energy Correlators
• On the Energy Distribution of the Galactic Center Excess' Sources