June 5 2003 K. Romer XCS XMM Cluster Survey Xray,SZE yield cleaner mass fn, sel. fn. than optical SZ will be cheaper by x10 Use XMM for now, but area too small for LSS want survey with large z,mass range + T_x Romer et al 01: Flux limit of 10^-14 gets you "all" clusters; higher z sensitive Lambda where "all" is >4keV Data reduction: use archive (>1000 pointings) custom pipeline 60 sq. deg. of unique area at high |b|, similar to RASS aarchival surveys Source detection: Wavelets (MR1) use extent classification to select cands RCS method on optical imaging (to be obtained) XMM-SDSS matching applied to 30 XMM pointings: 27 known clusters, 28 good cands, 27 trash photo-z vs. spectro-z look ok Sel. Fn. Assume 800 sq. deg. to RDCS limit->cosmo params to few % assumes no evol. in L-T, no scatter in M-T, and know z,T perfectly updated to include better knowledge of sensitivity & observing effects little difference; if area down to 100 sq. deg. gets much worse with high s/n can see iron line to get z, and can measure T, but not realistic Scharf: many XMM fields probably already observed by previous missions, so not much new area are you using hardness ration? Not yet... ========================================================================================= ========================================================================================= C. Miller Radial Profiles of Galaxies in Clusters How are galaxies distributed in clusters vs. galaxy props, cluster props 250 clusters with >3*10^14 Msun 140k galaxies with photo,spec deprojection: invert 2d->3d or project 3d to 2d radial profile of composite clusters compares well to CNOC; truncated NFW & Hernquist models fit well substructure : veldisp vs. radius in each cluster shows L_x vs. veldisp -those clusters with good veldisp correlate much better -compare radial profiles of clusters with good vs. bad veldisps -those with bad veldisps appear flattened. -could be misestimate of r_200 (which comes from veldisp) or actual difference Using ONLY "clean" (no substructure) clusters: -radial profiles of bright vs. dim galaxies: nothing different (top 1/3 vs bottom 1/3, only down to M*+0.5) -concentration of clusters : same for high & low veldisp systems. May not have enough mass range -blue vs. red galaxies: fit E/S0 ridgeline, take spectroscopic galaxies only by 2 R_vir, there are many fewer blue galaxies than red; blue pop flattens in core -star forming galaxies: similar to blue pop; even more extreme flattening. Core contains little or no star forming gals. ========================================================================================= ========================================================================================= Z. Haiman see astro/ph 0306053 Using power spectrum to constrain cosmology P(k)=Ak^n T^2(k) where T(k) is transfer function; unaffected by w but there are baryon wiggles - small features in P(k) which are w-dependent How does P(k) evolve with redshift? P(k) depends on perp & transverse modes -> P(k_p,k_t) power spec of sources biased relative to matter P(k), plus has redshift-space distortion, and growth function bias matter z-space dist growth P_s(k_p,k_t) = b^2 P_m(k_p,t_t)*[1+B(z)(k_t/k)^2]*D(z) k_t ~ H(z) k_p ~ 1/D_A(z) where D_A is angular diameter distance cosmology enters in many terms of observed power spectrum in k_t vs. k_p space can define ellipsoidal shapes which deviate from circles -size has h dependence, but shape contains ~50% of cosmological info -very weak dependence on redshift-space distortion observable depends on volume ~ P_s/volume ~ P_s/(H^-1 D_A^2) For galaxies (SDSS spectro) for 10,000 square degrees main survey: M(halo)>10^12.1 Msun z<0.1 bias~1 LRG: M>10^13.5 0.110^14.2 z<1.2 bias~4 Assume bias known only to 50%; B (z-space dist) only to 50% add priors from CMB based on Planck estimates k 0.0005 - 0.15 Mpc^-1 assume know all redshifts constraints: galaxies: sig(w) = 0.025 clusters: sig(w) = 0.04 Planck alone will give sig(w)~0.1 clusters similar to galaxies due to larger bias and higher z range Jain: what about w(z)? A: not tested issues that could change errors on w: how much information from wiggles vs. just turnover in P(k): factor of 2. So want contiguous large survey, not separate fields how much do CMB priors add? degrade from Planck to WMAP - adds only 10% to errors. means that low-z measurement of H^-1 , D_A(z) pins down these relations well (if you have wiggles) how much would redshift-dependent or no knowledge of bias add? marginalize over b in redshift bins. Degrade errors by factor of ~2 ==>half of info from wiggles, half from bias (growth fn.) how much z precision needed in cluster survey to get constraints? -depends on k modes containing info -2% photo-zs would be good enough -if only have clusters to z<0.7, error up by factor of 1.7 Similar to Szalay & Matsubara, but addition of wiggles is important Jain: No scale-dependent bias; could be impt Berlind: does galaxy power come from main or LRG sample? A: probably main, but unknown Nichol: LRG program w/ 2dF to go to 0.7, does that help? A: probably not much if use CMB, without CMB will help a lo