Slicing cluster mass functions with a Bayesian razor

TL;DR

This paper uses a Bayesian razor approach to compare different galaxy cluster mass function models, determining the simulation size needed for strong evidence favoring one model over another based on cluster mass.

Contribution

It introduces a Bayesian razor method to forecast model comparison strength and applies it to galaxy cluster mass functions using N-body simulations.

Findings

Minimum simulation size for strong evidence varies with cluster mass threshold.

Two-parameter models can be strongly favored over simpler models with sufficient data.

The approach guides future simulation design for model discrimination.

Abstract

We apply a Bayesian "razor" to forecast Bayes factors between different parameterizations of the galaxy cluster mass function. To demonstrate this approach, we calculate the minimum size N-body simulation needed for strong evidence favoring a two-parameter mass function over one-parameter mass functions and visa versa, as a function of the minimum cluster mass.