Extracting clusters and determining the selection function for SZ surveys Jean-Baptiste Melin U.C. Davis J. Bartlett J. Delabrouille APC – College de France
Contents I. Theoretical selection function of SZ cluster surveys II. Fast SZ extraction algorithm III. Selection function results
Contents I. Theoretical selection function of SZ cluster surveys II. Fast SZ extraction algorithm III. Selection function results
The selection function ? Contamination Completeness (c,Y)= Number of false detections Total number of detections (recovered clusters + false detections) (c,Y)= Number of recovered clusters True number of clusters If you don’t know and y, don’t expect to do science !
Matched filters (1/2) AMPLITUDE ? c NORMALIZED TEMPLATE NOISE Haehnelt & Tegmark 96 Herranz et al. 2002a, 2002b AMPLITUDE ? c c=2 arcmin CMB CMB+beam Instrumental noise NORMALIZED TEMPLATE NOISE
Matched filters (2/2) in Fourier space in real space Aest linear estimator unbiased <Aest-A>=0 minimize the variance =<(Aest-A)2> in Fourier space in real space [arbitrary unit] [arbitrary unit] Aest (S/N)est= Aest/ single-frequency & multi-frequency
Y = Aest Tc > 5 . . Tc AMI-like : =15GHz, beam=2arcmin inst. noise=5µK/beam, pt. sources : S<100Jy Aest/>5 Y = Aest Tc > 5 . . Tc
Contents I. Theoretical selection function of SZ cluster surveys II. Fast SZ extraction algorithm III. Selection function results
Cluster extraction in 3 steps Simulations 3º 3º 15 GHz 30 GHz 90 GHz pix=30’’ Primary CMB anisotropies Instrumental beam (fwhm=2 arcmin) Insrumental white noise (DT=20 K/pix) Radio sources (S<0.1mJy at 15 GHZ) Multifrequency (n=15, 30, 90 GHz) Cosmology : LCDM
Cluster extraction Filtered map sample Step 1 … … c(filter)=3.0 arcmin c(filter)=0.1 arcmin c(filter)=1.6 arcmin
Cluster extraction Cluster candidates Step 2 S/Nthreshold = 3, 5, … S/Ncarte> S/Nthreshold … … c(filter)=3.0 arcmin c(filter)=0.1 arcmin c(filter)=1.6 arcmin
Cluster extraction c and Y recovery Step 3 c given by the node having the highest S/N in a given branch Y derived from the filtered map at scale c c=3.0 arcmin . c=0.3 arcmin c=0.2 arcmin c=0.1 arcmin
Contents I. Theoretical selection function of SZ cluster surveys II. Fast SZ extraction algorithm III. Selection function results
Single frequency – 15 GHz 50 simulations (3 deg × 3 deg each) Cl perfectly known simulations detection theoretical selection fit CBI excess simulations detection theoretical selection fit
Single frequency – 15 GHz Cluster counts Clusters with Y>5.10-5arcmin2
Single frequency – 15 GHz Cosmological parameters BIAS !
Conclusions Multi-frequency/Single-frequency Selection function non-trivial depends on instrument, observation strategy, confusion, cluster physics & data pipeline Bias Additional source of error ‘Survey calibration’ &
The method A Monte Carlo triangle Fast SZ simulation tool Fast SZ detection tool Input catalog Output catalog Comparison Selection function
Single frequency – 15 GHz A non-trivial selection function Clusters with Y>5.10-5arcmin2 Y>10-4arcmin2 Y>3.10-4arcmin2
Single frequency – 15 GHz Theoretical selection curves détecté detected non détecté not detected