Great Lakes Cosmology Workshop 8, Columbus OH June 2, 2007 Sources of scatter in cluster mass-observable relations Paul Ricker University of Illinois National Center for Supercomputing Applications With Karen Yang (UIUC) Zarija Lukić (UIUC) Ramesh Balakrishnan (NCSA)
Counting galaxy clusters for cosmology Cluster abundance as a function of mass and redshift d2n dv = n M,z dm dz dz 2 b n M, z d exp 2 M 2 c Depends on: Volume-redshift relation dv/dz Linear growth factor ( (z)) Power spectrum ( (M,z)) Mohr (2005) 2 2
Cluster surveys Dark Energy Survey (DES) Optical redshift catalog of ~ 107 galaxies to z ~ 1.4 5000 deg2 survey of southern sky 500 Mpixel 4-color camera South Pole Telescope (SPT) Microwave catalog of ~ 104 clusters 4000 deg2 survey at < 1' resolution 3 5 frequencies (95 350 GHz) Survey volumes ~ 1 Gpc3 3 3
Cluster mass-observable relations Observable X = LX, TX, LIR, y500, Ngal, etc. Controlling parameters M, z1, z2, etc. P X cosmology P X M, z 1, z 2,... P M, z 1, z 2,... cosmology Mass-observable relation X-ray temperature Popesso et al. (2005) Mass function X-ray luminosity Stanek et al. (2006) 4 4
Options for constraining cosmology Directly form stars, AGN, etc. (perhaps with subgrid models) Simulate observations, including light travel time and response Compare with observations in data space N-body + mass-observable relation* Mass function from simulations (e.g., Lukić et al. 2007) Assign observables based on observed scalings Self-calibration* (Levine et al.; Lima & Hu; Majumdar & Mohr) Parametrize mass-observable relation Fit parameters along with cosmology * Need to know the form and evolution of the intrinsic scatter! Ease Physics content Direct N-body/gasdynamics + mock skies 5 5
Role of merging vs. nongravitational physics What can create intrinsic scatter? Anisotropy, formation time variation from initial perturbations Merger effects shocks, turbulence, displacement from equilibrium Rapid cooling Ricker & Sarazin (2001), Randall et al. (2002) Episodic heating (AGN)...? How important is each, and what form does its contribution take? O'Hara et al. (2006) 6 6
Distinguishing contributions to scatter Large-volume N-body simulation Controlling parameters Yang & Ricker (2007) M V tcool PAGN Correlations in scatter High-resolution resimulations with gas and extra physics Simulated observations 7 7
Physics in our simulations FLASH (Fryxell et al. 2000) Oct-tree adaptive mesh refinement Gasdynamics (PPM) Metal-dependent radiative cooling Star formation w/upper SFR cutoff Type Ia and II supernovae Dark matter (particle-mesh) Adaptive smoothing Initial conditions externally generated by CMBFAST/GRAFIC 8 8
Making simulated Chandra X-ray observations Dark matter and gas halo finding via parallel FOF MARX with user source to get ACIS-I photon event file Compute MEKAL emissivity in each cell Project to make energy-dependent sky intensity map 0.7 2.0 kev XSPEC fit to determine spectral parameters 9 9
Simulation details CDM WMAP3 parameters h = 0.73, m = 0.238, b = 0.044, 8 = 0.74 Ti = 6.1 103 K (zi = 81) S(z=3) = 220 kev cm2 (cf. Run S5 of Bialek et al. 2001) 256h-1 Mpc, xmin = 250h-1 kpc, 5123 particles (mp = 6.8 109 h-1 M ) log gas slice (z=0) log Tgas slice (z=0) 1010
X-ray mass-temperature scaling TX M500c (z = 0) log (TX/keV) 853 clusters with M500c 7 1012 h-1 M (out of 2059) at z = 0 log TX = -8.58 + 0.624log M500c log (M500c/M ) 1111
Dynamical state evolution in idealized mergers Virialization parameter (e.g., Ricker 1998) 2 T U V 1 W Ricker & Sarazin (2001) R500 R200 2 T U V 1 W S Poole et al. (2006) 1212
X-ray scatter vs. dynamical state: V Virialization parameter (e.g., Ricker 1998) log(tx/kev) 2 T U V 1 W V200 1313
X-ray scatter vs. dynamical state: centroid offset log(tx/kev) Offset between projected luminosity peak and friends-offriends halo density peak (Mohr et al. 1995) Result: no significant variation with w on resolvable scales w/r500c 1414
X-ray scatter vs. dynamical state: power log(tx/kev) Power in multipole moments of the projected gas density distribution (Buote & Tsai 1995, 1996) Lowest-order moments: P(1) P2 P3 log(tx/kev) log(tx/kev) P(1) ellipticity bimodality triplets P2/P0 Result: no apparent trends P3/P0 1515
Simulation details Small box CDM problem from Heitmann et al. (2005) h = 0.71, m = 0.314, b = 0.052, 8 = 0.8 Ti = 1.6 104 K (zi = 50) S(z=3) = 220 kev cm2 (cf. Run S5 of Bialek et al. 2001) 64h-1 Mpc, xmin = 125h-1 kpc, 2563 particles (mp = 1.1 109 h-1 M ) 1616
P(1)/P0 P2/P0 64h-1 Mpc box, 2563 particles 63 clusters at z=0 50 clusters at z=1 log(tew/kev) log(tew/kev) log(tew/kev) X-ray scatter vs. dynamical state: evolution P3/P0 1717
Conclusions To use cluster surveys for cosmology, need to know Form and evolution of mass-observable scatter Dominant physical parameters driving scatter Dynamical state influences on M-T scatter: very weak Virialization parameter, centroid offset, power ratios show little evidence of influence on scatter in this quantity Supports conclusions of O'Hara et al. (Preliminary) no evolution in scatter seen Next steps Additional measures of dynamics: merger tree extraction Radiative cooling and feedback 1818