Sunyaev-Zel'dovich contribution in CMB analysis

TL;DR

This paper analyzes how different models of the Sunyaev-Zel'dovich effect impact cosmological parameter estimation from CMB data, highlighting biases and improvements in confidence intervals.

Contribution

It compares three SZ modeling approaches and demonstrates the benefits and sensitivities of a full cosmological dependency model in joint CMB-SZ analysis.

Findings

SZ template methods bias parameters if the assumed cosmology differs.

Full cosmological dependency model reduces biases and improves confidence intervals.

Sensitivity to intra-cluster gas parameters necessitates additional cluster data.

Abstract

The Sunyaev-Zel'dovich (SZ) effect has long been identified as one of the most important secondary effects of the Cosmic Microwave Background (CMB). On the one hand, it is a potentially very powerful cosmological probe providing us with additional constraints and on the other hand it represents the major source of secondary fluctuations at small angular scales (l > 1000). We investigate the effects of the SZ modelling in the determination of the cosmological parameters. We explore the consequences of the SZ power spectrum computation by comparing three increasingly complex modelling, from a fixed template with an amplitude factor to a calculation including the full cosmological parameter dependency. We also examine the dependency of the cosmological parameter estimation on the intra-cluster gas description used to calculate the SZ spectrum. We show that methods assuming an SZ template bias the cosmological parameters (by up to 2 sigmas on sigma_8) when the cosmology used in the template deviates from the reference one. A joint CMB-SZ analysis with a full cosmological dependency of the SZ spectrum does not suffer from such biases and moreover improves the confidence intervals of sigma_8 and Omega_{dm}h^2 (2.5 and 2 times respectively) with respect to a pure CMB analysis. However, the latter method is quite sensitive to the intra-cluster gas parameters and hence requires extra information on the clusters to alleviate the induced biases.