Effective Window Function for Lagrangian Halos

TL;DR

This paper introduces an effective window function for protohalos in Lagrangian space, improving the modeling of halo-matter correlations and mass functions by better matching the measured protohalo profiles.

Contribution

It proposes a new effective window function that accurately describes protohalo profiles and their scale dependence, enabling improved predictions of halo abundances.

Findings

The effective window is a rounded tophat convolved with a Gaussian.

It accurately models the scale dependence of halo-matter correlations up to high wavenumbers.

The mass function estimate matches simulations within 20% accuracy.

Abstract

The window function for protohalos in Lagrangian space is often assumed to be a tophat in real space. We measure this profile directly and find that it is more extended than a tophat but less extended than a Gaussian; its shape is well-described by rounding the edges of the tophat by convolution with a Gaussian that has a scale length about 5 times smaller. This effective window $W_{\rm eff}$ is particularly simple in Fourier space, and has an analytic form in real space. Together with the excursion set bias parameters, $W_{\rm eff}$ describes the scale-dependence of the Lagrangian halo-matter cross correlation up to $kR_{\rm Lag} \sim 10 $, where $R_{\rm Lag}$ is the Lagrangian size of the protohalo. Moreover, with this $W_{\rm eff}$, all the spectral moments of the power spectrum are finite, allowing a straightforward estimate of the excursion set peak mass function. This estimate requires a prescription of the critical overdensity enclosed within a protohalo if it is to collapse, which we calibrate from simulations. We find that the resulting estimate of halo abundances is only accurate to about 20%, and we discuss why: A tophat in `infall time' towards the protohalo center need not correspond to a tophat in the initial spatial distribution, so models in which infall rather than smoothed overdensity is the relevant variable may be more accurate.