Three-point intrinsic alignments of dark matter halos in the IllustrisTNG simulation

Abstract

We use the IllustrisTNG suite of cosmological simulations to measure intrinsic alignment (IA) bispectra of dark matter subhalos between redshifts 0 and 1. We decompose the intrinsic shear field into E- and B-modes and find that the bispectra $B_{\delta\delta\mathrm{E}}$ and $B_{\delta\mathrm{EE}}$, between the matter overdensity field, $\delta$, and the E-mode field, are detected with high significance. We also model the IA bispectra analytically using a method consistent with the two-point non-linear alignment model. We use this model and the simulation measurements to infer the intrinsic alignment amplitude $A_\mathrm{IA}$ and find that values of $A_\mathrm{IA}$ obtained from IA power spectra and bispectra agree well at scales up to $k_\mathrm{max}= 2 \, h \mathrm{Mpc}^{-1}$, for example at $z=1$ $A_\mathrm{IA} = 2.13 \pm$ 0.02 from the cross power spectrum between the matter overdensity and E-mode fields and $A_\mathrm{IA} =2.11 \pm$ 0.03 from $B_{\delta\delta \mathrm{E}}$. This demonstrates that a single physically motivated model can jointly model two-point and three-point statistics of intrinsic alignments, thus enabling a cleaner separation between intrinsic alignments and cosmological weak lensing signals.