Non-universality of halo profiles and implications for dark matter experiments

TL;DR

This study analyzes the variability in dark matter halo profiles from simulations, revealing significant scatter that impacts dark matter detection efforts and suggesting that stacking halos for signals is unbiased.

Contribution

It demonstrates that halo-to-halo density profile scatter follows a Gaussian distribution, affecting dark matter detection interpretations and providing insights into profile variability.

Findings

25% uncertainty in local dark matter density

Factor of 3 uncertainty in Galactic annihilation flux

Gaussian distribution of halo profile scatter

Abstract

We explore the cosmological halo-to-halo scatter of the distribution of mass within dark matter halos utilizing a well-resolved statistical sample of clusters from the cosmological Millennium simulation. We find that at any radius, the spherically-averaged dark matter density of a halo (corresponding to the "smooth-component") and its logarithmic slope are well-described by a Gaussian probability distribution. At small radii (within the scale radius), the density distribution is fully determined by the measured Gaussian distribution in halo concentrations. The variance in the radial distribution of mass in dark matter halos is important for the interpretation of direct and indirect dark matter detection efforts. The scatter in mass profiles imparts approximately a 25 percent cosmological uncertainty in the dark matter density at the Solar neighborhood and a factor of ~3 uncertainty in the expected Galactic dark matter annihilation flux. The aggregate effect of halo-to-halo profile scatter leads to a small (few percent) enhancement in dark matter annihilation background if the Gaussian concentration distribution holds for all halo masses versus a 10 percent enhancement under the assumption of a log-normal concentration distribution. The Gaussian nature of the cluster profile scatter implies that the technique of "stacking" halos to improve signal to noise should not suffer from bias.