CombineHarvesterFlow: Joint Probe Analysis Made Easy with Normalizing Flows

TL;DR

CombineHarvesterFlow introduces a method using normalizing flows to efficiently sample and reweight joint posteriors of multiple experiments, enabling rapid combined analysis with significant computational savings.

Contribution

The paper presents a novel approach employing normalizing flows to jointly analyze multiple experiments, reducing computational cost and enabling new combined constraints.

Findings

Achieved joint constraints on cosmological parameters from multiple datasets.

Reduced computational time and carbon footprint for joint posterior sampling.

Released an open-source package for the community to perform these analyses.

Abstract

We show how to efficiently sample the joint posterior of two non-covariant experiments with a large set of nuisance parameters. Specifically, we train an ensemble of normalizing flows to learn the posterior distribution of both experiments. Once trained, we can use the flows to reweight $\mathcal{O} (10^9)$ samples from both measurements to compute the joint posterior in seconds -- saving up to $\mathcal{O}(1)$ ton of $\text{CO}_2$ per Monte Carlo run. Using this new technique we find joint constraints between the Dark Energy Survey $3 \times 2$ point measurement, South Pole Telescope and Planck CMB lensing and a BOSS direct fit full shape analyses, for the first time. We find $\Omega_{\rm m} = 0.32^{+0.01}_{-0.01}$ and $S_8 = 0.79 ^ {+0.01}_ {-0.01}$. We release a public package called {\tt CombineHarvesterFlow} (https://github.com/pltaylor16/CombineHarvesterFlow) which performs these calculations.