The XMM-LSS cluster sample and its cosmological applications. Prospects for the XMM next decade
aa r X i v : . [ a s t r o - ph ] D ec Astron. Nachr. / Solicited talk given at
XMM-Newton: The Next Decade , ESAC, Madrid 4-6 June 2007 , No. 00, 1 – 5 (0000) /
DOI please set DOI!
The XMM-LSS cluster sample and its cosmological applications.Prospects for the XMM next decade
M. Pierre ,⋆ , F. Pacaud , , J.B. Melin , and the XMM-LSS consortium DAPNIA/Service d’Astrophysique, Laboratoire AIM CNRS, CEA-Saclay, F-91191 Gif-sur- Yvette, France Argelander-Institut f¨ur Astronomie, University of Bonn, Auf dem H¨ugel 71, 53121 Bonn, Germany DAPNIA/Service de Physique des Particules, CEA Saclay, F-91191 Gif-sur-Yvette, France.Received : September 2007, accepted : November 2007
Key words
X-rays: galaxies: clusters – cosmological parametersThe well defined selection function of the XMM-LSS survey enables a simultaneous modelling of the observed clusternumber counts and of the evolution of the L-T relation. We present results pertaining to the first 5 deg for a well con-trolled sample comprising 30 objects : they are compatible with the WMAP3 parameter set along with cluster self-similarevolution. Extending such a survey to 200 deg would (1) allow discriminating between the major scenarios of the clusterL-T evolution and (2) provide a unique self-sufficient determination of σ and Γ with an accuracy of ∼
5% and 10%respectively, when adding mass information from weak lensing and S-Z observations. c (cid:13) It has been recognised for a long time that clusters of galax-ies, as the most massive entities of the universe, can be usedas cosmological probes. They provide key information onthe normalisation of the power spectrum and are potentiallysuitable for studying the properties of dark energy. Theyrepresent important and independent constraints in additionto those from the CMB and supernovae because they in-volve very different physics. It is also critical to ensure con-sistency between the cosmological constrains from the earlyand local universe.Main statistical tools for cluster studies are the clusternumber counts ( dn/dz ) and the cluster two-point correla-tion function ( ξ ). This requires that, whatever the detectionwavelength, the samples must be in some sense, completeand uncontaminated and thus, requires well understood de-tection and selection procedures. Cluster physics evolutionis a key ingredient in interpreting the observed cluster den-sity as function of redshift. It is usually modelled in the formof scaling laws relating observable quantities such as flux,richness, luminosity or temperature to cluster masses. Thecluster scaling laws are, however, still poorly known beyondthe local universe.With its mosaic of overlapping XMM pointings ( s),the XMM Large-Scale Structure survey (XMM-LSS, Pierreet al 2004) has been designed to detect a significant frac-tion of the cluster population out to z = 1 , over an area ofseveral tens of deg , so as to constitute a sample suitablefor cosmological studies. We present below the proceduresdeveloped to detect the clusters and to further analyse their ⋆ Corresponding author: e-mail: [email protected] number counts along with their temperature and luminositydistribution in a self-consistent approach. In light of the re-sults obtained so far, we discuss the cosmological impact ofa future 200 deg XMM wide survey.
In the redshift range of interest, although the cluster ap-parent sizes ( ′′ < R c < ′′ ) are significantly largerthan the XMM PSF and source confusion can be consid-ered as negligible, cluster detection is a very specific tasksince our objects are weak sources (count-rate from 0.3 to 3counts/min). We developed a two-step procedure combiningwavelet multi-resolution analysis and maximum likelihoodfits both using Poisson statistics. The pipeline was exten-sively tested using simulations which allowed us to definea sub-region in the extent vs extent likelihood parameter space, where the contamination level by point-sources is lower than 1%. This constitutes the class one (C1)cluster sample. Strictly speaking, this selection is not fluxlimited, but allows the construction of well controlled anduncontaminated cluster samples significantly larger than asimple flux limit would allow (Pacaud et 2006). Our C1sample shows a density of ∼ clusters per deg . The XMM-LSS currently covers 10 deg . It is located in theW1 area of the Canada-France-Hawaii Telescope LegacySurvey (CFHTLS) and associated with a number of surveysin the radio, infrared and UV domains (Fig. 1). In these pro- c (cid:13) M. Pierre: The XMM-LSS survey
Fig. 1 Layout of the 98 XMM-LSS pointings includingthe Subaru Deep Survey. The colour scale indicates the ef-fective exposures, from 80 ks (SDS01) to ∼
0. The greenrectangle delineates data obtained prior to the AO5. In ad-dition to the coverage by the SWIRE and CFHT LegacySurveys, observations from the VLA, Integral, UKIDSS andGalex are available in the field. Full coverage by SCUBA2and Herschel is planned.ceedings, we summarise the results from the first 5 deg which are presented in detail by Pacaud et al 2007 (P07). Some 30 C1 clusters are found in the first 5 deg . They allhave been spectroscopically confirmed. Their redshift dis-tribution is displayed on Fig. 2. P07 performed an ab initiomodelling of the observed number counts as follows: as-sume a (1) cosmological framework ( Λ CDM) and a powerspectrum along with a transfer function; (2) a mass func-tion, (3) a halo model, (4) various scaling evolutionary re-lations for cluster physics. Then, for each redshift and massrange, the predicted luminosities are transformed into XMMcount-rates using a dedicated plasma code. The C1 selec-tion criteria are finally applied, resulting in a simulated ob-served redshift distribution (Fig. 2). The effect of the degen-eracy between cluster scaling law evolution and cosmologyis clearly illustrated on the figure; however, our results, de-spite the still small size of the sample, favour the WMAP3parameter sets along with cluster self-similar evolution.
Each C1 cluster undergoes dedicated spatial and spectral fitsin order to derive reliable luminosity and temperature es-timates. Willis et al (2005) have shown that using a welladapted binning procedure, it is statistically possible to ob-tained a 20% temperature accuracy with only 200 photons
Fig. 2 The current C1 cluster redshift distributionover the first 5 deg of the XMM-LSS . The colour scaleindicates the cluster mean temperature for each bin (un-weighted mean of the individual cluster temperatures inkeV.) The solid green histogram shows the expectations ofour cosmological model (WMAP3: σ = 0.74 and self- sim-ilar evolution for the Lx-T relation) along with the Poissonerror bars. The dash-line histogram shows the expectationsfor a model with WMAP 1st year cosmological parameters( σ = 0.85) and a non-evolving Lx-T relation. Fluctuationsaround the mean expectation are represented by the solidand dotted error bars for the shot noise and sample vari-ance (estimated from Hu & Kravtsov 2003) respectively.The grey error bars are for 5 deg , while the black onesfor 200 deg .for groups up to 2 keV. It turns out that for all C1 clus-ters, we obtain temperature measurements with satisfactoryaccuracy. The average temperature of the C1 clusters as afunction of redshift is displayed on Fig. 2. Because of thevery tight relation between X-ray temperature and luminos-ity, the mean temperature of the detected clusters appearsto increases with redshift (Malmquist bias). This redshift -temperature distribution shows that the XMM-LSS surveyunveils for the first time the population of low-mass groups(T= 2 keV) around z = 0 . , which constitute the buildingblocks of the present day clusters.Assuming that all clusters, whatever their mass, followthe same evolutionary scaling laws, we have used our datato constrain the evolution of the L-T relation. The observedluminosity enhancement, with respect to the local expec-tation, is computed for the 30 clusters distributed in fourredshift bins (Fig 3). The raw data suggest a rather strongevolution, best fitted with a two-parametre model. However,the inclusion of the survey selection function in the fit sug-gests a much more mild evolution, fully compatible with c (cid:13) stron. Nachr. / Solicited talk given at XMM-Newton: The Next Decade , ESAC, Madrid 4-6 June 2007(0000) 3
Fig. 3 The L-T relation . The graph shows the cluster X-ray luminosity enhancement with respect to the expectationat z = 0 . Our clusters are sorted in 4 redshift intervals. Thedash line is the result of an ad hoc two-parameter fit to theraw data points: (1+ z ) α × E ( z ) β with α = 4 . , β = − . .When the survey selection function is taken into account,the best fit is the solid line, (1 + z ) α E ( z ) with α = − . ,which is very close to the evolution predicted by the self-similar model (dotted line, α = 0 ). The grey region de-lineates the 1 σ confidence interval for the surveyed 5 deg .The dot-dash and triple-dot-dash lines are the evolutionarymodels by Voit (2005) including non gravitational physics.The thick error bars at the end of the three models indicatethe expectations for a 200 deg survey.the predictions of the self-similar model. The fact that selec-tion effects were not systematically considered in the formerL-T(z) studies may explain the discordant results obtainedso far (see a compilation in P07): because the cluster massfunction is so steep, most clusters are detected around thesurvey limiting sensitivity, hence favouring the compilationof over-luminous objects at any redshift. This is certainlynot a feature unique to the XMM-LSS survey, but rather ofany X-ray cluster survey (provided that the detection is per-formed down to the capabilities of the survey). This biasis to affect any sub-sample “randomly” selected from all-sky or serendipitous surveys for subsequent deep XMM orChandra temperature observations. We emphasise that theXMM-LSS survey represents the first attempt, not only todetermine the L-T relation with survey data alone but also toexplicitly include the selection effects in the determinationof the evolution of the scaling laws. The latter was possiblethanks to the very well modelled survey selection function. The XMM-LSS provides a source density of ∼ / deg down to a flux limit of 4 − erg s − cm in the [0.5-2]keV band (95% level completeness limit). This constitutes the largest deep AGN X-ray sample over a single field. Thespecial relevance of the sample also comes from its uniquemulti- λ coverage. The XMM-LSS data set allowed, for the first time, the studyof the angular distribution of faint AGNs over an area of5 deg . We found a significant clustering in the soft bandand none in the hard band. A sub-sample of ∼ sourceswith hard X-ray count ratios, likely dominated by obscuredAGNs, does show a positive signal allowing for a large an-gular correlation length at the σ level (Gandhi et al 2006). A dedicated study of some 100 AGN selected in the hardband reveals a mismatch between the classification basedon the characteristics of the optical emission lines and theclassification given by the X-ray spectroscopy. This led toquestion some aspects of the AGN unified scheme (Garcetet al 2007). The many CFHTLS and SIWRE flux data pointsallowed us to perform SED fits on the XMM-LSS pointsource population. From this, it was possible to classify theAGN (star formation, type 1 or 2 AGN, Seyfert) and to ob-tain photometric redshifts. Combining with the X-ray spec-tral data points, we demonstrate that the SED properties arecontinuous through the various classes (Tajer et al 2007,Polletta et al 2007).
Cluster data (positions, redshifts, L X , T X , mass estimates)are published for the first 5 deg along with the scientificanalysis (Valtchanov et al 2004 , Willis et al 2005, P07).The complete source catalogue with optical data and thumbnail images is also public (Pierre et al 2007). Data can beretrieved from the CDS or, in a more extensive form, viathe consortium data bases for the cluster and source cata-logues. deg widesurvey During the past years, we explored a number of issues re-garding cluster detection and science with the XMM-LSSsurvey. In addition to the C1 clusters, a complementary clus-ter sample of about the same size has been identified inthe XMM-LSS, but for which the selection criteria are lesswell defined (Adami et al in prep). We have also detecteda number of z > clusters (Andreon et al 2005, Bremeret al 2006). The determination of the luminosity, temper-ature and mass (hydrostatic hypothesis) of the C1clusters http://l3sdb.in2p3.fr:8080/l3sdb/ http://cosmos.iasf-milano.inaf.it/ lssadmin/Website/LSS/Query/ c (cid:13) M. Pierre: The XMM-LSS survey
Fig. 4 Comparison between the XMM-LSS andSunyaev-Zel’dovich sensitivities in terms of limiting mass.The red lines show various measured detection probabilitythresholds for the C1 clusters. The blue lines are the predic-tions for the 10 µ K APEX survey, currently observing theXMM-LSS field. For the regime of interest ( z < ), the X-ray observations are at least as efficient as the S-Z ones interms of cluster detection.from survey data appears to provide quantities reliable forstatistical cosmological studies . A weak lensing analysisover part of the XMM-LSS area shows promising prospectsfor constraining independently the slope and the normalisa-tion of the M-T relation (Berge et al 2007). A further stepin constraining cluster masses is to be reached by the up-coming generation of Sunyaev- Zel’dovich surveys which,in principle, with a sensitivity of µK , are well matchedto mass range of the XMM-LSS survey (Fig. 4).Further, we have shown that, in order to increase theprecision of the L-T relation evolution, it is more efficientto increase the cluster sample than the accuracy on the tem-perature measurements - a useful tip for a proper use of fu-ture XMM observing time. This is due to the large intrinsicdispersion of the L-T relation itself (P07).In this way, a 200 deg survey with 10 ks XMM point-ings , would allow definitively discriminating not only be-tween cluster self-similar evolution and no evolution, butalso between other theoretically justified models based onnon-gravitational physics (Fig. 3). Such a 200 deg surveywould moreover overcome sample variance problems andprovide more than a thousand C1 clusters as well as some100 clusters at z > . Having determined the cluster evo-lution rate, an important degeneracy would be removed inthe cosmological interpretation of the cluster number counts(Fig. 2) leaving a handle on the equation of state of the DarkEnergy (P07). The cluster sky distribution provides addi- The mass of XLSS 29 at z = 1 . was measured to be . M ⊙ with survey data (P07) and . ± . M ⊙ from a subsequent 80 ksXMM pointing (Maughan et al 2007) such a survey would require some 24 Ms with the current observingsettings, which could be decreased by about 1/3 when the foreseen mosaic-ing mode with a reduced pn overhead is implemented Fig. 5 Top: Constraints on the σ − Ω m planefrom dn/dz + ξ ( r ) . All contours are 1 σ for C1 clusters,marginalised over Ω Λ . (dots) The C1 X-ray cluster popu-lation alone (6/ deg ) over 200 deg ; adding informationfrom S-Z (dash-dot) and weak lensing (solid) mass mea-surements; (dash) same as solid but for 10 deg coverage.X-ray masses are taken to be accurate to 50%, adding S-Zthen weak lensing data reduces this to 20% then to 10%;the latter giving an accuracy on σ of 6 %. Bottom: Con-straints on the σ − Γ plane from ξ ( r ) + dn/dz . Con-tours are 1 σ for the C1 population, 200 deg , mass accuracyof 10% and marginalisation over Ω m and Ω Λ . The expectedaccuracy on Γ is 10%.tional constraints on the cosmological parameters and a 200 deg survey will not only allow a self-sufficient and accu-rate determination of σ , i.e. 5 %, but also of the slope ofthe matter power spectrum, Γ , i.e. 10% (Fig. 5). References
Andreon, S., et al: 2005, MNRAS 359, 1250Berge J. et al: 2007, MNRAS submittedBremer, M., et al: 2006, MNRAS 371, 1427 c (cid:13) stron. Nachr. / Solicited talk given at XMM-Newton: The Next Decade , ESAC, Madrid 4-6 June 2007(0000) 5Gandhi, P., et al : 2006, A&A 457, 393Garcet, O., et al : 2007, A&A in press (astroph/0709.2027)Hu & Kravtsov: 2003, ApJ 584, 702Maughan, B., et al: 2007, MNRAS submitted (astroph/07092300)Pacaud, F. et al : 2006, MNRAS 372, 578Pacaud, F et al (P07): 2007, MNRAS in press (astroph/0709.1950)Pierre, M., et al : 2004, JCAP 9, 11Pierre, M., et al: 2007, MNRAS in press (astroph/0708.3299)Polletta, M., et al: 2007, ApJ 663, 81Tajer, M., et al 2007: A&A 467, 73Valtchanov I., et al : 2004 A&A 423, 75Voit, M.: 2005, Rev. Mod. Phys 77, 207Willis, J., et al : 2005, MNRAS 363, 675 c (cid:13)(cid:13)