Offsets between member galaxies and dark matter in clusters: a test with the Illustris simulation

Abstract

Dark matter with a non-zero self-interacting cross section ($\sigma_{\rm SIDM}$) has been posited as a solution to a number of outstanding astrophysical mysteries. Many studies of merging galaxy clusters have given constraints on $\sigma_{\rm SIDM}$ based on the spatial offset between the member galaxy population and the dark matter distribution. Assuming $\sigma_{\rm SIDM} = 0$, how likely is it for us to see the galaxy-DM offset values observed in merging clusters of galaxies? To answer this question, we formulate a hypothesis test using data from Illustris, a $\Lambda$CDM cosmological simulation. We select 43 Illustris clusters and their galaxy members at z~0 and examine the accuracy of commonly used galaxy summary statistics, including kernel-density-estimation (KDE) luminosity peak, KDE number density peak, shrinking aperture, centroid and the location of the brightest cluster galaxy (BCG). We use the dark-matter particles to reproduce commonly adopted methods to identify dark-matter peaks based on gravitational lensing cluster maps. By analysing each cluster in 768 projections, we determine the optimistic noise floor in the measurements of the galaxy-DM offsets. We find that the choice of the galaxy summary statistics affects the inferred offset values substantially, with the BCG and the luminosity peak giving the tightest 68-th percentile offset levels, $\lesssim$ 4 kpc and $\lesssim$ 32 kpc, respectively. Shrinking aperture, number density and centroid give a large offset scatter of about 50-100 kpc at the 68-th percentile level, even for clusters with only one dominant mass component. Out of the 15 reported offsets from observed merging clusters that we examined, 13 of them are consistent with Illustris unrelaxed cluster offsets at the 2-sigma (95-th percentile) level, i.e. consistent with the hypothesis that $\Lambda$CDM is the true underlying physical model.