Title: The FLAMINGO project: galaxy clusters in comparison to X-ray observations
Authors: Joey Braspenning, Joop Schaye, Matthieu Schaller, Ian G. McCarthy, Scott T. Kay, John C. Helly, Roi Kugel, Willem Elbers, Carlos S. Frenk, Juliana Kwan, Jaime Salcido, Marcel P. van Daalen, Bert Vandenbroucke
First Author’s Institution: Leiden Observatory, Leiden University, PO Box 9513, 2300 RA Leiden, the Netherlands
Status: Published in MNRAS [open access]
*A group of flamingos is called a flamboyance
Galaxy clusters– collections of hundreds or thousands of galaxies– are the most massive objects in the universe and their abundance and distribution throughout the universe are important for understanding cosmology. Clusters are usually discovered in the optical or X-ray; optical observations locate the galaxies in the cluster, while X-rays trace the hot gas of the intracluster medium (ICM). However, their large size makes them rare, so the sample of clusters is small. Even in simulations the number of galaxy clusters is typically low because of the enormous volume required to produce them.
From observations, astronomers calculate two major metrics for clusters: cluster scaling relations and thermodynamic radial profiles. Cluster-scale metrics quantify general information about the cluster (like mass, X-ray luminosity, and temperature) as they compare to other clusters in what’s called a scaling relation. Radial profiles of individual thermodynamic quantities, like temperature, metallicity, density, entropy, and pressure, provide insight to the physical processes going on at different distances from the cluster center. Radial profiles are especially important to validate the physics being used in simulations; a simulated cluster might fall along scaling relations but have an unrealistic radial profile.
In today’s paper, the authors compare the scaling relations and radial profiles of galaxy clusters of the FLAMINGO simulations to an observational sample of clusters.
The FLAMINGO simulations are a set of cosmological hydrodynamic simulations that contain several million clusters and groups of galaxies. They cover a huge volume – (2.8 Gpc)^3 – and use a Smoothed Particle Hydrodynamics (SPH) method that breaks the total gas volume into interacting “particles”. The simulations include radiative cooling, star formation, stellar mass loss, and supernova, black hole, and AGN feedback. Machine learning is used to calibrate the feedback to reproduce the observed gas fractions and stellar mass functions in low redshift (or relatively nearby) clusters. They produce several sets of simulations with slightly varied physics to compare with the observations.
The authors select all clusters with M_500c > 1e13 solar masses (M_sun). M_500c is the mass enclosed within the sphere defined by R_500c, which is the radius of a halo inside which the density is 500 times the critical density of the universe. “Virial quantities” referred to in this paper are defined by this radius and the virial theorem. Physical quantities were computed using a FLAMINGO tool and an example cluster’s quantities is shown in Figure 1.
First, the authors test the FLAMINGO clusters’ scaling relations. The simulations are only calibrated to match the observed gas fractions at low redshifts, which doesn’t guarantee the reproduction of scaling relations across redshifts. The simulated and observed scaling relations for the mass and temperature are shown in Figure 2. The simulated relation, and others involving the X-ray luminosity, generally agree well with the observational data.
The authors then test the radial profiles. In each profile, the weighted average of the thermodynamic quantity is calculated; the quantity chosen to act as the weight could be the mass, volume, or X-ray luminosity of the individual SPH particles. The differences between the profiles produced by the different weights are small except at small radii from the cluster center, with X-ray weighting tending to be the best match to the observational data.
They normalize each profile to the virial quantity and take median profiles for each cluster mass bin. If all clusters were the simplest model – a size-invariant constant-temperature sphere in hydrostatic equilibrium – then the profiles would be independent of the cluster mass. This is not true in reality, and indeed FLAMINGO sees a strong mass dependence that agrees with other simulations like MilleniumTNG and TNG-Clusters. The profiles from FLAMINGO agree with observations down to all but the smallest radii except for the metallicity, which is ~0.3 dex higher than observed. This is likely due to an overestimation of the material released by supernovae or with the fraction of matter estimated to be in the form of stars in the simulated clusters.
To study the evolution of clusters over redshifts, the authors chose halos with the same mass at z = 0, 0.5, 1, and 1.5 to compare their properties. When normalized over virial quantities, the authors found little or no evolution of the density, pressure, or entropy over redshift, as seen in Figure 3. This is expected as the expansion of the universe should be accounted for by the normalization. The metallicity sees a clear evolution with redshift, with lower metallicity in high redshift clusters as fewer supernovae have had time to go off and deposit their metals into the environment. The metallicity curves are generally consistent with other simulations, but FLAMINGO finds that the core (r < 0.15 R_500c) metallicity decreases with redshift, which is not seen in other simulations.
Overall, the FLAMINGO-simulated clusters are a good match to real clusters, with comparable radial profiles and scaling relations except in the case of metallicity. The FLAMINGO clusters evolve across redshift in line with the observational expectations, and can therefore be an effective tool for studying cluster evolution.
Astrobite edited by Ivey Davis
Featured image credit: Braspenning+24 Figure 1, WikiMedia Commons
Discover more from astrobites
Subscribe to get the latest posts to your email.
.jpg){kind=link}