An improved probabilistic approach for linking progenitor and descendant galaxy populations using comoving number density

TL;DR

This paper introduces a probabilistic method for linking galaxy progenitors and descendants across cosmic time, accounting for their evolving and dispersing number densities, leading to more accurate predictions of galaxy properties.

Contribution

The paper presents a novel probabilistic approach that models galaxy number density evolution and dispersion, improving upon previous constant or evolving number density methods.

Findings

Probabilistic method outperforms constant and evolving number density methods.

Improved accuracy in predicting galaxy properties like stellar mass and star formation rate.

Method is applicable to observational data with an available code package.

Abstract

Galaxy populations at different cosmic epochs are often linked together by comoving cumulative number density in observational studies. Many theoretical works, however, have shown that the number densities of tracked galaxy populations evolve in bulk and spread out over time. We present a number density method for linking progenitor and descendant galaxy populations which takes both of these effects into account. We define probability distribution functions that capture the evolution and dispersion of galaxy populations in comoving number density space, and use these functions to assign galaxies at one redshift $z_f$ probabilities of being progenitors or descendants of a galaxy population at another redshift $z_0$. These probabilities are then used as weights for calculating distributions of physical properties such as stellar mass, star formation rate, or velocity dispersion within the progenitor/descendant population. We demonstrate that this probabilistic method provides more accurate predictions for the evolution of physical properties then either the assumption of a constant number density or the assumption of an evolving number density in a bin of fixed width by comparing the predictions against galaxy populations directly tracked through a cosmological simulation. We find that the constant number density method performs most poorly at recovering galaxy properties, the evolving number method density slightly better, and the probabilistic number density method best of all. The improvement is present for predictions of both stellar mass as well as inferred quantities such as star formation rate and velocity dispersion which were not included in the number density fits. We demonstrate that this method can also be applied robustly and easily to observational data, and provide a code package for doing so.