Statistical Identification of Mergers in Galaxy Surveys

Mergers play a central role in the evolution of galaxies and are key drivers of star formation and AGN. Since 2018, merger classification has been dominated by machine learning (ML) techniques, while these techniques do achieve high performance, they are resource intensive and do not transfer well to previously unseen datasets. Classical techniques for merger classification use morphological statistics. In this talk I compared the performance of morphological classifiers to ML visual classifiers. For this comparison the statistical classifiers are applied to the merger challenge by Margalef-Bentabol et al. 2024. I will also present a potential link between the merger time and Gini-m20 statistics; predicting merger times is currently only attempted with ML techniques. For modern surveys such as LSST, fast and reliable methods will be necessary to produce the largest and purest merger catalogues possible in order to study the true impact of mergers on a large scale.