5–7 Mar 2025 Conference
Nagoya University
Asia/Tokyo timezone

A Neural Network Approach to Consider Secondary Dependence of Halo Bias

5 Mar 2025, 18:15
1h 45m
Sakata and Hirata Hall (Nagoya University)

Sakata and Hirata Hall

Nagoya University

Science South bulding, Furo-cho, Chikusa, Nagoya, Aichi, 464-8602, Japan

Speaker

Keitaro Ishikawa (Nagoya University)

Description

Galaxies form in dark matter haloes. The spatial distribution of dark matter haloes, and the distribution and the number of galaxies within a dark matter halo, depend primarily on the halo mass. However, they are also known to depend on halo properties other than mass, such as halo formation history (Wechsler et al. 2006). This secondary dependence is called assembly bias. Due to dramatic improvements in the statistical accuracy of upcoming galaxy surveys, ignoring the assembly bias would bias cosmological parameter constraints (Miyatake et al. 2022). In this work, to construct a halo statistics emulator that also predicts the assembly bias, we focus on the concentration of haloes as a representative secondary parameter, and measure the cross-correlation function of various halo samples selected according to the mass and concentration using Dark Quest II simulation data. We construct an emulator for these statistics based on a feed-forward neural network. As a result, we find that the halo correlation function can be emulated with a typical accuracy of 1%. In this poster, we will mention that the result emulated in redshift space not only real space, and the outlook relative to the Stage IV survey; LSST, Roman projects.

Presentation materials

There are no materials yet.