Large-scale Structure Simulations A.E. Evrard, R Stanek, B Nord (Michigan) E. Gaztanaga, P Fosalba, M. Manera (Barcelona) A. Kravtsov (Chicago) P.M Ricker (UIUC/NCSA) R. Wechsler (Stanford) D. Weinberg (OSU)
core science areas non-linear evolution of the matter density P(k) for weak lensing, BAO halo characterization for clusters, BAO, weak lensing gas dynamic simulations of clusters g(y SZ, N gal, … | M halo,z) : form of observable-mass relation sensitivity to galaxy/AGN physics mock sky surveys of galaxies and clusters SZ + optical cluster finding : test self-calibration multiple techniques to model galaxy formation and evolution empirical: halo occupation, ADDGALS first principle: SAM’s, direct gas dynamic 100 sq deg now, several x 1000 sq deg by mid-2007
methods and resources mpi-based large-scale structure codes GADGET: tree-PM N-body + Lagrangian hydro (SPH) ART: tree N-body + Eulerian, adaptive-grid hydro FLASH: PM N-body + Eulerian, adaptive-grid hydro compute resources BCN (10 4 cpus,10 6 hours Tb) NCSA allocations of cycles and storage local compute clusters (~100 cpu’s) and storage (~10 Tb) each billion particle run generates ~10Tb of output NASA AISR proposal to grid-enable this work (follow DM lead)
Millennium Simulation (MS) L=500 Mpc/h Ω m =0.25, Ω =0.75, h=0.73, 8 = particles m p =8.7e8 Msun/h halo/sub-halo catalogs semi-analytic galaxies Springel et al 2005 test red-sequence cluster finding
workflow view of galaxy formation star / SMBH formation
galaxy samples redshiftz-magNumber Croton et al galaxy types in a halo: central - accrete gas + form stars satellite - no gas accretion or star formation red sequence in halos w/ N gal ≥ 4: width of r–z color grows with redshift factor ~ 2 wider than observed
halo occupation of red-sequence galaxies z = 0.41 regular behavior slope slightly steeper than 1 no funny `dark’ clusters
simple cluster finder based on mean sky density (parallels 3D algorithm used to define halos) for brightest galaxy – re-center volume on galaxy – apply line-of-sight color gradient for z-evolution – grow disc until mean RS number density threshold is reached – assign group members if N gal ≥N min (=4) repeat for next available (non-assigned) galaxy apply simple cluster finder to volume projections Aim: lower-bound on blending due to supercluster projections –use periodic BC’s to re-center volume around each galaxy -- apply linear color gradient to fore/background r–z color redshift
cluster classification based on halo matching f best = N gal (halo) / N gal (cluster) for the halo contributing the largest number of galaxies 2 classes: clean : f best ≥ 0.5 (plurality is majority) blended : f best < 0.5 (plurality is minority)
cluster richness-mass relation red sequence cluster finding recovers well the intrinsic halo occupation clean : f best ≥ 0.5 blended : f best < 0.5 halo cluster
conditional likelihood of halo mass at fixed richness clusters halos
conditional likelihood of halo mass at fixed richness clean clusters halos blended clusters Next step: test whether SZ signatures will remove blends consider bi-modal likelihood p(M|N gal ) ?
MS w/ gas: halo space density 5x10 8 particles m dm =1.4x10 10 Msun/h m gas =2.9x10 9 Msun/h 3 simulations: 0. gravity only 1. cooling + heating I 2. cooling + heating II F. Pearce, L. Gazzola (Nottingham) + Virgo Consortium collaborators R. Stanek, B. Nord (Umich) M 200 mass function : run 0 open: DM only filled: DM + gas Evrard et al (2002) `prediction’
gravity only cool+heat 1 thermal SZ gas mass fraction DM velocity dispersion gas temperature MS w/ gas: scaling relations
MS w/ gas: covariance of observables high Lx systems are likely to be gas rich correl. coeff. r = 0.5 deviation in X-ray luminosity deviation in gas mass fraction
MS galaxies match b+K band LF’s