An Empirical Mass Function Distribution

TL;DR

This paper introduces an empirical mass function model for dark matter halos that achieves high accuracy and can incorporate measurement errors, aiding survey design and galaxy formation studies.

Contribution

It presents a new four-parameter empirical mass function form that is accurate, interpretable, and adaptable within Bayesian analyses, with applications to survey optimization.

Findings

Achieves better than 4% accuracy in the medium-mass regime.

Effectively incorporates measurement errors and biases in hierarchical Bayesian models.

Provides open-source Python and R tools for implementation.

Abstract

The halo mass function, encoding the comoving number density of dark matter halos of a given mass, plays a key role in understanding the formation and evolution of galaxies. As such, it is a key goal of current and future deep optical surveys to constrain the mass function down to mass scales which typically host $L_\star$ galaxies. Motivated by the proven accuracy of Press-Schechter-type mass functions, we introduce a related but purely empirical form consistent with standard formulae to better than 4\% in the medium-mass regime, $10^{10}-10^{13} h^{-1}M_\odot$. In particular, our form consists of 4 parameters, each of which has a simple interpretation, and can be directly related to parameters of the galaxy distribution, such as $L_\star$. Using this form within a hierarchical Bayesian likelihood model, we show how individual mass-measurement errors can be successfully included in a typical analysis whilst accounting for Eddington bias. We apply our form to a question of survey design in the context of a semi-realistic data model, illustrating how it can be used to obtain optimal balance between survey depth and angular coverage for constraints on mass function parameters. Open-source {\tt Python} and {\tt R} codes to apply our new form are provided at \url{http://mrpy.readthedocs.org} and \url{https://cran.r-project.org/web/packages/tggd/index.html} respectively.