Marginal Bayesian Statistics Using Masked Autoregressive Flows and Kernel Density Estimators with Examples in Cosmology

TL;DR

This paper introduces a computationally efficient method for combining cosmological data constraints by generating marginal probability density estimators, enabling faster and nuisance-free Bayesian analysis across multiple experiments.

Contribution

It presents a novel approach combining nested sampling and the MARGARINE code to produce lossless marginal likelihoods, simplifying complex high-dimensional Bayesian analyses in cosmology.

Findings

Method produces identical posteriors to full nested sampling

Approach is faster than evaluating full likelihoods

Successfully applied to combine DES and Planck data

Abstract

Cosmological experiments often employ Bayesian workflows to derive constraints on cosmological and astrophysical parameters from their data. It has been shown that these constraints can be combined across different probes such as Planck and the Dark Energy Survey and that this can be a valuable exercise to improve our understanding of the universe and quantify tension between multiple experiments. However, these experiments are typically plagued by differing systematics, instrumental effects and contaminating signals, which we collectively refer to as `nuisance' components, that have to be modelled alongside target signals of interest. This leads to high dimensional parameter spaces, especially when combining data sets, with > 20 dimensions of which only around 5 correspond to key physical quantities. We present a means by which to combine constraints from different data sets in a computationally efficient manner by generating rapid, reusable and reliable marginal probability density estimators, giving us access to nuisance-free likelihoods. This is possible through the unique combination of nested sampling, which gives us access to Bayesian evidences, and the marginal Bayesian statistics code MARGARINE. Our method is lossless in the signal parameters, resulting in the same posterior distributions as would be found from a full nested sampling run over all nuisance parameters, and typically quicker than evaluating full likelihoods. We demonstrate our approach by applying it to the combination of posteriors from the Dark Energy Survey and Planck.