HIKER: a halo-finding method based on kernel-shift algorithm

TL;DR

HIKER is a new halo/subhalo finder that uses a kernel-shift algorithm with machine learning techniques to accurately locate density peaks and identify structures in cosmological simulations.

Contribution

The paper introduces HIKER, a novel halo finder based on the mean-shift algorithm with Plummer kernel, improving accuracy in locating halo centers and reproducing halo properties.

Findings

HIKER accurately recovers input halo properties in mock tests.

HIKER outperforms most halo finders in locating halo/subhalo centers.

HIKER's halo/subhalo mass functions agree well with established finders.

Abstract

We introduce a new halo/subhalo finder, HIKER (a Halo fInder based on KERnel-shift algorithm), which takes advantage of a machine learning method -- the mean-shift algorithm combined with the Plummer kernel function, to effectively locate density peaks corresponding to halos/subhalos in density field. Based on these density peaks, dark matter halos are identified as spherical overdensity structures, and subhalos are bound substructures with boundaries at their tidal radius. By testing HIKER code with mock halos, we show that HIKER performs excellently in recovering input halo properties. Especially, HIKER has higher accuracy in locating halo/subhalo centres than most halo finders. With cosmological simulations, we further show that HIKER reproduces the abundance of dark matter halos and subhalos quite accurately, and the HIKER halo/subhalo mass functions and $V_{max}$ functions are in good agreement with two widely used halo finders, SUBFIND and AHF.