Constructing high-fidelity halo merger trees in <scp>abacussummit</scp>

Tracking the formation and evolution of dark matter haloes is a critical aspect any analysis cosmological $N$-body simulations. In particular, mass assembly halo its progenitors, encapsulated in form merger tree, serves as fundamental input for constructing semi-analytic models galaxy and, more generally, building mock catalogues that emulate surveys. We present an algorithm trees from AbacusSummit, largest suite simulations performed to date consisting nearly 60 trillion particles, which has been designed meet Cosmological Simulation Requirements Dark Energy Spectroscopic Instrument (DESI) survey. Our method tracks cores determine associations between objects across multiple timeslices, yielding lists progenitors descendants several tens billions identified entire suite. application these means enhance fidelity AbacusSummit by flagging "merging" deemed exhibit non-monotonic past histories. show this cleaning technique identifies portions population have deblended due choices made finder, but could feasibly part larger aggregate systems. demonstrate post-processing step, we remove potentially unphysical features default catalogues, leaving behind robust can be used create highly-accurate realisations AbacusSummit.