Implicit Priors in Galaxy Cluster Mass and Scaling Relation Determinations

TL;DR

This paper examines how fully parametric models of the intracluster medium in galaxy clusters inherently impose priors that bias mass scaling relation slopes towards the self-similar model, affecting the interpretation of observational data.

Contribution

It reveals the implicit priors in parametric ICM modeling and discusses their impact on scaling relation analyses, proposing alternative approaches to avoid bias.

Findings

Parametric models favor self-similar slope values.

Non-parametric methods do not impose such priors.

Literature shows clustering of results based on modeling approach.

Abstract

Deriving the total masses of galaxy clusters from observations of the intracluster medium (ICM) generally requires some prior information, in addition to the assumptions of hydrostatic equilibrium and spherical symmetry. Often, this information takes the form of particular parametrized functions used to describe the cluster gas density and temperature profiles. In this paper, we investigate the implicit priors on hydrostatic masses that result from this fully parametric approach, and the implications of such priors for scaling relations formed from those masses. We show that the application of such fully parametric models of the ICM naturally imposes a prior on the slopes of the derived scaling relations, favoring the self-similar model, and argue that this prior may be influential in practice. In contrast, this bias does not exist for techniques which adopt an explicit prior on the form of the mass profile but describe the ICM non-parametrically. Constraints on the slope of the cluster mass--temperature relation in the literature show a separation based the approach employed, with the results from fully parametric ICM modeling clustering nearer the self-similar value. Given that a primary goal of scaling relation analyses is to test the self-similar model, the application of methods subject to strong, implicit priors should be avoided. Alternative methods and best practices are discussed.