On the mass mismatch between simulations and weak-lensing measurements. (arXiv:1906.00975v1 [astro-ph.CO])
<a href="http://arxiv.org/find/astro-ph/1/au:+Svensmark_J/0/1/0/all/0/1">Jacob Svensmark</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Martizzi_D/0/1/0/all/0/1">Davide Martizzi</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Agnello_A/0/1/0/all/0/1">Adriano Agnello</a>
The recently discovered discrepancy between galaxy mass measurements from
weak lensing and predictions from abundance matching questions our
understanding of cosmology, or of the galaxy-halo connection, or of both. We
re-examined this tension by considering, as models, different cosmological
simulations in the Illustris suite. We produced excess profiles $RDeltaSigma$
from subhalo snapshots at different redshifts in Illustris-1 and IllustrisTNG
(TNG100 and TNG300) simulations, enabling a direct comparison with weak-lensing
measurements. We separate the individual contributions of stars, dark matter
and gas within $approx1$ Mpc (comoving length), beyond which correlated
two-halo terms dominate. The mismatch between measurements and predictions is
more severe than in previous studies: $RDeltaSigma$ profiles from
IllustrisTNG are $approx2$ times higher than the measured ones. Contrary to
abundance matching results, the mismatch is mostly unchanged with increasing
redshifts. The contribution of gas to the $RDeltaSigma$ profiles is $5-10%$
over the scales dominated by one-halo terms. Different procedures to link
stellar and halo masses (abundance matching, cosmological simulations) are
still significantly discrepant with weak lensing measurements, but their trends
are different. Therefore, the change in cosmological parameters advocated
through abundance-matching arguments may not resolve this tension. Also,
current criteria to select isolated massive galaxies in simulations are
susceptible to resolution issues and may not correspond to observational
criteria. The (currently subdominant) contribution of gas is non-negligible,
and even if the major discrepancy within stellar and halo masses is resolved,
it will be an appreciable source of systematics in the LSST era, when
uncertainties on the $RDeltaSigma$ profiles are expected to be $approx10$
times smaller.
The recently discovered discrepancy between galaxy mass measurements from
weak lensing and predictions from abundance matching questions our
understanding of cosmology, or of the galaxy-halo connection, or of both. We
re-examined this tension by considering, as models, different cosmological
simulations in the Illustris suite. We produced excess profiles $RDeltaSigma$
from subhalo snapshots at different redshifts in Illustris-1 and IllustrisTNG
(TNG100 and TNG300) simulations, enabling a direct comparison with weak-lensing
measurements. We separate the individual contributions of stars, dark matter
and gas within $approx1$ Mpc (comoving length), beyond which correlated
two-halo terms dominate. The mismatch between measurements and predictions is
more severe than in previous studies: $RDeltaSigma$ profiles from
IllustrisTNG are $approx2$ times higher than the measured ones. Contrary to
abundance matching results, the mismatch is mostly unchanged with increasing
redshifts. The contribution of gas to the $RDeltaSigma$ profiles is $5-10%$
over the scales dominated by one-halo terms. Different procedures to link
stellar and halo masses (abundance matching, cosmological simulations) are
still significantly discrepant with weak lensing measurements, but their trends
are different. Therefore, the change in cosmological parameters advocated
through abundance-matching arguments may not resolve this tension. Also,
current criteria to select isolated massive galaxies in simulations are
susceptible to resolution issues and may not correspond to observational
criteria. The (currently subdominant) contribution of gas is non-negligible,
and even if the major discrepancy within stellar and halo masses is resolved,
it will be an appreciable source of systematics in the LSST era, when
uncertainties on the $RDeltaSigma$ profiles are expected to be $approx10$
times smaller.
http://arxiv.org/icons/sfx.gif
