Cosmoglobe: Towards end-to-end CMB cosmological parameter estimation without likelihood approximations

TL;DR

This paper demonstrates the feasibility of performing full Bayesian cosmological parameter estimation for CMB data without likelihood approximations, highlighting current computational costs and the need for further optimization.

Contribution

It implements a likelihood-free Bayesian estimation method in Commander and assesses its computational efficiency for realistic CMB simulations.

Findings

Single sample costs about 20 CPU-hours.

Effective cost per independent sample is around 2,000 CPU-hours.

Current methods are feasible but require optimization for practical use.

Abstract

We implement support for a cosmological parameter estimation algorithm as proposed by Racine et al. (2016) in Commander, and quantify its computational efficiency and cost. For a semi-realistic simulation similar to Planck LFI 70 GHz, we find that the computational cost of producing one single sample is about 20 CPU-hours and that the typical Markov chain correlation length is $\sim$100 samples. The net effective cost per independent sample is $\sim$2 000 CPU-hours, in comparison with all low-level processing costs of 812 CPU-hours for Planck LFI and WMAP in Cosmoglobe Data Release 1. Thus, although technically possible to run already in its current state, future work should aim to reduce the effective cost per independent sample by one order of magnitude to avoid excessive runtimes, for instance through multi-grid preconditioners and/or derivative-based Markov chain sampling schemes. This work demonstrates the computational feasibility of true Bayesian cosmological parameter estimation with end-to-end error propagation for high-precision CMB experiments without likelihood approximations, but it also highlights the need for additional optimizations before it is ready for full production-level analysis.