Astronomy&Astrophysicsmanuscriptno.szlens_conta_cib (cid:13)c ESO2017 February1,2017 LettertotheEditor Predicting the CIB-φ contamination in the cross-correlation of the tSZ effect and φ G.Hurier1 Institutd’AstrophysiqueSpatiale,CNRS(UMR8617)andUniversitéParis-Sud11,Bâtiment121,91405Orsay,France e-mail:[email protected] Preprintonlineversion:February1,2017 Abstract 7 1 TherecentreleaseofPlanckdatagivesaccesstoafullskycoverageofthethermalSunyaev-Zel’dovich(tSZ)effectandofthecosmic 0 microwave background (CMB) lensing potential (φ). The cross-correlation of these two probes of the large-scale structures in the 2 Universeisapowerfultoolfortestingcosmologicalmodels,especiallyinthecontextofthedifferencebetweengalaxyclustersand n CMBforthebest-fittingcosmologicalparameters. a However,thetSZeffectmapsarehighlycontaminatedbycosmicinfra-redbackground(CIB)fluctuations.Unlikeotherastrophysical J components, the spatial distribution of CIB varies with frequency. Thus it cannot be completely removed from a tSZ Compton 1 parametermap,whichisconstructedfromalinearcombinationofmultiplefrequencymaps.Wehaveestimatedthecontaminationof 3 theCIB-φcorrelationinthetSZ-φpower-spectrum.WeconsideredlinearcombinationsthatreconstructthetSZComptonparameter fromPlanckfrequencymaps.Weconcludethateveninanoptimisticcase,theCIB-φcontaminationissignificantwithrespecttothe ] tSZ-φsignalitself. O Consequently,WestressthattSZ-φanalysesthatarebasedonComptonparametermapsarehighlylimitedbythebiasproducedby CIB-φcontamination. C . Keywords.galaxyclusters,CMB,cosmology,power-spectrum,modelling h p - o 1. Introduction estimate the CIB-φ contamination in tSZ-φ cross-correlation. r In Sect. 4, we discuss the cleaning process proposed by Hill & t s Modern cosmology extensively used cosmic microwave back- Spergel (2014) and estimate the residual contamination level. a ground (CMB) data. During their propagation along the line FinallyinSect.5,weconclude. [ of sight, photons are affected by several physical processes 1 such as the thermal Sunyaev-Zel’dovich effect (tSZ, Sunyaev Throughoutthepaper,weconsiderH =67.1,σ =0.80and 0 8 v & Zeldovich 1969, 1972) and CMB gravitational lensing Ω =0.32asourfiducialcosmologicalmodel,unlessotherwise m 7 (Blanchard & Schneider 1987), which trace the gravitational specified. 4 potentialintegratedalongthelineofsight,φ. 0 At microwave frequencies, foreground emissions contribute to 9 2. ModellingthetSZ,φ,andCIBcross-correlations the total signal of the sky, for example the cosmic infra-red 0 background (CIB, Puget et al. 1996; Fixsen et al. 1998). The . Thetotalpower-spectrumbetweentheAandBquantities(tSZ, 1 tSZ effect, gravitational lensing, and the CIB are tracers of φorCIB)canbeexpressedas 0 the large-scale structures in the matter distribution of the 7 Universe. They have been powerful sources of cosmological C(cid:96)A×B =C(cid:96)A×B−P+C(cid:96)A×B−C, (1) 1 and astrophysical constraints (see e.g., Planck Collaboration withCA×B−PthePoissoniancontributionandCA×B−Cthecontri- v: results. XX 2014; Planck Collaboration results. XXX 2014; bution(cid:96)fromlargeangularscalecorrelations. (cid:96) i PlanckCollaborationresults.XVII2014). X r 2.1. Poissonianterm a Alltheseprobespresentcorrelationsthroughtheirrelationto thelarge-scaledistributionofmatterintheUniverse.TheCIB-φ ThePoissonianorone-halotermofthepower-spectrumcanbe correlation has already been detected with high significance writtenusingahalomodelasthesumofthecontributionsfrom (Planck Collaboration results. XVIII 2014). More recently, the each halo. The number of halos per unit of mass and redshift, tSZ-φ correlation estimated from a Compton parameter map d2N ,isgivenbythemassfunction(seee.g.,Tinkeretal.2008). (Hill&Spergel2014)hasbeenusedtoconstraincosmological dMdV Usingthelimberapproximation,wecanwritetheone-haloterm parameters. as (cid:90) (cid:90) dV d2N We discuss the contamination from the CIB-φ correlation C(cid:96)A×B−P =4π dzdzdΩ dMdMdVWAPWBP. (2) intothetSZ-φcorrelationestimatedfromaComptonparameter ThetSZcontributioncanbewrittenas map.InSect.2,wepresentacoherentmodellingofthetSZ,φ, and the CIB spectra and cross-correlation. Then in Sect. 3, we WP =Y y , (3) tSZ 500 (cid:96) 1 G.Hurieretal.:CIB-φcontaminationinthecross-correlationofthetSZeffectandφ withY thetSZfluxoftheclusters,relatedtothemass, M Table1.Cosmologicalandscaling-lawparametersforourfidu- 500 500 viathescalinglawpresentedinEq.4,andy theFouriertrans- cial model for both Y − M (Planck Collaboration results. (cid:96) 500 500 form on the sphere of the cluster pressure profile from Planck XX2014)andtheL −M relationfittedonCIBspectra. 500 500 Collaborationresults.XX(2014)perunitoftSZflux. We used the M − Y scaling law presented in Planck Fiducialcosmo M −Y M −L 500 500 500 500 500 500 Collaborationresults.XX(2014), Ωm 0.32 logY(cid:63) -0.19±0.02 T0 24.42 E−βsz(z)D102an−g4(Mz)Yp5c020=Y(cid:63)(cid:34)0h.7(cid:35)−2+αsz(cid:34)6(1×−1b0)1M4M50(cid:12)0(cid:35)αsz, (4) Hσ80 06.870 αβsszz 10..7696±±00..058 αγβccciiibbb 011..37.765222 δ 3.22 with E(z) = Ωm(1+z)3 +ΩΛ.ThecoefficientsY(cid:63),αsz andβsz, (cid:15)cciibb 1.1±0.2 taken from Planck Collaboration results. XX (2014), are given in Table 1. The mean bias, (1 − b), between X-ray mass and the true mass was estimated from numerical simulations, b (cid:39) 2.2. Large-scalecorrelationterms 0.2 (see Planck Collaboration results. XX 2014, for a detailed We now focuss on the contribution from large angular scale discussionofthisbias). correlations in the matter distribution to the tSZ-φ-CIB cross- correlationpower-spectramatrix.Weexpresthiscontributionas Thelensingcontributioncanbewrittenas (cid:90) (χ(cid:48)−χ)χ dV WP = −2ψ , (5) CA×B−C =4π dz WCWCP , (10) φ (cid:96) χ(cid:48) (cid:96) dzdΩ A B k with χ the comoving distance, χ(cid:48) the comoving distance of the withP ,thematterpower-spectrum. k CMB,andψ(cid:96) the3DlensingpotentialFouriertransformonthe In the following subsections we present the window functions, sky. W,relatedtoeachcontribution. Wecanexpressthepotentialψasafunctionofthedensitycon- trast, For the tSZ effect, we considered the two-halo large-scale ∆ψ= 3Ω H2δ, (6) correlation. We neglected the contribution produced by diffuse 2 m 0a tSZemissionfromthewarmhotinterstellarmedium.Thetwo- withatheuniversescalefactorandδthedensitycontrast.Then, halocontributionreads thelensingcontributionreads (cid:90) d2N WP = 3ΩmH02(1+z)(χ(cid:48)−χ)χδ , (7) WtCSZ = dMdMdVY500y(cid:96)blin, (11) φ (cid:96)((cid:96)+1) χ(cid:48) (cid:96) where b is the bias relating the halo distribution to the over- whereδ istheFouriertransformofthedensitycontrastprofile. lin (cid:96) density distribution. We considered the bias from Mo & White (1996) and Komatsu & Kitayama (1999), which is realistic at TheCIBcontributioncanbewrittenas galaxyclusterscale. WP (ν)=S (ν)c , (8) CIB 500 (cid:96) with S (ν) = aL500(ν), the infra-red flux at frequency ν of the For the lensing potential, we accounted for the large-scale 500 4πχ2(z) correlationbyconsideringstructuresinthelineargrowthregime hosthaloandc theFouriertransformoftheinfra-redprofile. (cid:96) as TtioonmdoedrievletdhefroLm500th−eMre5l0a0tiorenlaptrioonpowseeduisnedShaanpgareatmaelt.ri(c20r1e2la)-. WC = 3ΩmH02 (1+z)(χ(cid:48)−χ). (12) φ c2(cid:96)((cid:96)+1) χ(cid:48)χ TocomputethetSZ-CIBcorrelation,weneedtorelatetSZhalos withCIBemissions.Consequently,wecannotuseaparametriza- tionthatrelatesgalaxyhaloswiththeCIBflux.Weneedarela- For the CIB at frequency ν, we considered the two-halo tion that relates the galaxy cluster halo to the associated total large-scalecorrelationas CIBflux. We chose to parametrize the CIB flux, L500, at galaxy cluster (cid:90) d2N scalewithapowerlawofthemass1,M , WC (ν)= dM S (ν)c b . (13) 500 CIB dMdV 500 (cid:96) lin (cid:34) M (cid:35)(cid:15)cib L500(ν)= L0 1×1051040M (1+z)δcibΘ[(1+z)ν,Td(z)], (9) It is worth noting that the CIB-φ cross-correlation is domi- (cid:12) natedbythetwo-haloterm3,incontrasttothetSZ-CIBandtSZ- where L0 is a normalization parameter, Td(z) = T0(1+z)αcib is φcorrelationswhichpresentasignificantcontributionfromboth thethermaldusttemperature,andΘ[ν,Td]isthetypicalspectral oneandtwo-haloterms. energydistribution(SED)ofagalaxythatcontributestothetotal CIBemission, (cid:40) 2.3. Redshiftdistribution Θ[ν,T ]= νβcibBν(Td) ifν<ν0 d ν−γcib ifν≥ν0, In Fig. 1, we present the redshift distribution of power for the tSZ,CIB,andlensingpotentialauto-correlationspectra.ThetSZ with ν0 the solution of dlog[dνlβocigb(Bνν)(Td)] = −γcib. The coefficients effectisproducedbystructuresatlowredshift(z<2).TheCIB T ,α ,β ,γ ,δ ,and(cid:15) aregiveninTable1. andφspectraaredominatedbyobjectsathigherredshift. 0 cib cib cib cib cib 1 We consider the CIB-φ and tSZ-CIB cross-correlations, conse- 3 Thiscanbeexplainedbythehigh-zandlow-masshalosensitivity quentlythereisnoneedtomodeloftheCIBatgalaxyscale. of this cross-correlation, which favours the two-halo, whereas low-z, 2 FixedvaluefromPlanckCollaborationresults.XXX(2014) high-massobjectsfavourtheone-haloterm. 2 G.Hurieretal.:CIB-φcontaminationinthecross-correlationofthetSZeffectandφ reads (cid:88) Cconta = w Cφ×CIB(ν). (15) (cid:96) ν (cid:96) ν Figure1.DistributionsofthetSZ,φ,andCIBpowerasafunc- tionoftheredshiftat(cid:96) = 1000.Theblacksolidlinedepictsthe lensingpotential,φ,thegreydashedlinethetSZeffect,thedark blue,lightblue,green,yellow,orange,andreddashedlinesplot theCIBat100,143,217,353,545,and857GHz,respectively. Figure2. Residual contamination before cleaning in light blue and after cleaning in dark blue. The red solid line presents the The mean redshift of structures that dominate in the CIB is a tSZ-φcorrelationforourfiducialmodel.Thesolidbluelinesare function of the considered frequency. These redshift windows the contamination considering the set of weights from Hill & highlightthehighdegreeofcorrelationbetweentheCIBandφ Spergel(2014)for fsky =0.3.Thedashedbluelinesarethecon- andthelowdegreeofcorrelationbetweentSZandtheothertwo tamination considering the set of weights from Hill & Spergel probesoflarge-scalestructures. (2014)for fsky =0.2. Forexample,thehighdegreeofcorrelationbetweentheCIBand φhaverecentlybeenusedtodetectthelensingBmodesofpo- larizationbytheSPTcollaboration(Hansonetal.2013). InFig.2,wepresentourestimationoftheCIB-φcontamina- Consequently,smallCIBresidualsinatSZComptonparameter tion considering the set of weights from Hill & Spergel (2014) mapcanleadtoasignificantbiasinthetSZ-φcorrelationanaly- andcompareitwiththeexpectedtSZ-φcross-correlationpower- sis. spectrumforourfiducialmodel. Thecontaminationishigh,withasimilaramplitudeasthetSZ-φ 3. CIB-φcontaminationintSZ-φcross-correlation correlation. Furthermore, the contamination level is highly de- pendentonthesetofweights.Fordifferentsetsofweights,the 3.1. tSZComptonparametermap contaminationvariesanthesameorderofmagnitudeasthetSZ- φsignal. The recently released Planck data allow constructing full-sky ThisvariabilityhighlightsthedifficultyofcontrollingtheCIB-φ tSZComptonparametermap(PlanckCollaborationresults.XXI contaminationthatcanleadtoahighbiasintSZ-φanalysis. 2014)usinginternallinearcombination(ILC)componentsepa- However,thecontaminationlevelishighlydependentontheex- ration methods (see e.g., Remazeilles et al. 2011; Hurier et al. actCIBfrequencycorrelationmatrix.Consequently,thederived 2013). These methods reconstruct the tSZ signals using linear valuehastobeconsideredasanorder-of-magnitudeestimateof combinations y=(cid:88)w T , (14) theeffectivecontamination. ν ν ν where Tν is the intensity at the frequency ν and wν are the 4. CleaningtheCIB-φcontamination weightsofthelinearcombination.ForILC-basedmethods,these weightsarecomputedbyminimizingthevarianceoftherecon- In their analysis, Hill & Spergel (2014) proposed a method structedy-mapunderconstraints. of cleaning for CIB-φ contamination in tSZ-φ spectrum. Their ThelevelofCIB-φcontaminationinatSZ-φanalysisbasedona methodconsistsofacorrectionoftheform tSZComptonparametermapdependsontheweights,w ,used forthelinearcombination. ν C(cid:96)φ×y,corr =C(cid:96)φ×y−αcorrC(cid:96)φ×CIB(857), (16) We here considered the weights provided in Table 2 of Hill & Spergel (2014), which have been used to constrain the cosmol- whereCφ×yisthemeasuredcorrelationbetweenthetSZandpo- (cid:96) ogyfromthetSZ-φcross-correlationpower-spectrum. tential maps. The factor α is computed from an adjustment corr ofCCIB(857)×CIB(857) onthecross-correlationofthetSZCompton (cid:96) parameterandthe857GHzmaps. 3.2. PropagationofCIB-φspectra Weproposetotesttheaccuracyofthisprocedureusingourmod- We stress that the CIB is composed of different populations of ellingofthetSZ,φ,andCIBcross-correlations.Intheiranalysis objects at different frequencies. The associated power-spectra Hill&Spergel(2014)neglectedthetSZ-CIBcorrelationcontri- present variations of shape with respect to the frequency. The butionforcomputingα .However,attheconcernedfrequen- corr CIB-φ spectra present the same variations of shape with fre- cies, the tSZ-CIB correlation factor can reach about 20% and quency.TheCIB-φcontaminationinthetSZ-φcross-correlation therefore cannot be neglected. Neglecting tSZ-CIB correlation 3 G.Hurieretal.:CIB-φcontaminationinthecross-correlationofthetSZeffectandφ forthecleaningprocessleadstoanunderestimationoftheCIB- Acknowledgment φcontamination.Inanoptimisticcase4,α reads corr We are grateful to N.Aghanim and M.Douspis for usefull (cid:42)(cid:16)(cid:80)w CCIB(ν)×CIB(857)(cid:17)+CtSZ×CIB(857)(cid:43) discutions. α = ν (cid:96) (cid:96) . (17) WeacknowledgethesupportoftheFrenchAgenceNationalede corr CCIB(857)×CIB(857) laRechercheundergrantANR-11-BD56-015. (cid:96) Wecomputedthisvalueforeachsetofweightsat(cid:96) = 600.The valueofα variesby3%dependingonthevalueof(cid:96)usedfor corr the computation (300 < (cid:96)<1500), with a higher level of con- References tamination using higher (cid:96) values. Hill & Spergel (2014) found Blanchard,A.&Schneider,J.1987,A&A,184,1 αcorr =(6.7±1.1)×10−6KC−1MBfor fsky =0.3.Consideringtheas- Fixsen,D.J.,Dwek,E.,Mather,J.C.,Bennett,C.L.,&Shafer,R.A.1998,ApJ, sociatedsetofweightwepredictα =(7.1±0.2)×10−6K−1 508,123 corr CMB Hanson,D.,Hoover,S.,Crites,A.,etal.2013,PhysicalReviewLetters,111, for300<(cid:96) <1500.Withthis,weestimatetheassociatedresid- 141301 ualcontaminationinthetSZ-φspectrumas Hill,J.C.&Spergel,D.N.2014,J.CosmologyAstropart.Phys.,2,30 Hurier,G.,Macías-Pérez,J.F.,&Hildebrandt,S.2013,A&A,558,A118 C(cid:96)res =(cid:88)wνC(cid:96)φ×CIB(ν)−αcorrC(cid:96)φ×CIB(857). (18) KMoom,Hat.suJ.,&E.W&hKitiet,aySa.mDa.,MT..11999969,,MApNJR,5A2S6,,2L812,347 PlanckCollaborationresults.XVII.2014,A&A,571,A17 ν PlanckCollaborationresults.XVIII.2014,A&A,571,A18 In Fig. 2, we present the obtained residuals for the set of PlanckCollaborationresults.XX.2014,A&A,571,A20 PlanckCollaborationresults.XXI.2014,A&A,571,A21 weights from Hill & Spergel (2014). The CIB-φ residual con- PlanckCollaborationresults.XXX.2014,A&A,571,A30 tributeswith(20±10)%ofthetSZ-φsignalfor(cid:96)rangingfrom Puget,J.-L.,Abergel,A.,Bernard,J.-P.,etal.1996,A&A,308,L5 500to2000. Remazeilles,M.,Delabrouille,J.,&Cardoso,J.-F.2011,MNRAS,410,2481 This level of contamination is dependent on the CIB luminos- Shang,C.,Haiman,Z.,Knox,L.,&Oh,S.P.2012,MNRAS,421,2832 ityfunction(seeSect.2.2).Theuncertaintylevelwasestimated Sunyaev,R.A.&Zeldovich,Y.B.1969,Nature,223,721 Sunyaev,R.A.&Zeldovich,Y.B.1972,CommentsonAstrophysicsandSpace by propagating the CIB luminosity function uncertainties (see Physics,4,173 PlanckCollaborationresults.XXX2014,fordetailsonestimat- Tinker,J.,Kravtsov,A.V.,Klypin,A.,etal.2008,ApJ,688,709 ing the CIB luminosity function parameters). This uncertainty levelisdominatedbyδ and(cid:15) ,whichsettheamountofcor- cib cib relationbetweentheCIBmapsobservedatdifferentfrequencies. Other parameters, such as b or CIB SED parameters, have a lin weakereffectontheCIB-φresidualestimation. 5. Conclusionanddiscussion We have presented a modelling of the tSZ-φ-CIB cross- correlations.Basedonthismodelling,wepredictedtheexpected CIB-φcontaminationinthetSZ-φpower-spectradeducedfrom a Compton parameter map built by linearly combining Planck channelsfrom100to857GHz. Wedemonstratedthattheexpectedlevelofcontaminationfrom theCIB-φisatthesamelevelasthetSZ-φsignalitself. We also demonstrated that the CIB-φ contamination level is highlydependentonthesetofweightsusedforconstructingof theComptonparametermap.Todothis,weusedrealisticvalues deducedfromthesky,fortheweights. We tested a cleaning method for this bias and showed that the level of residual bias can reach about 20% of the tSZ-φ spec- trumfortheusedsetofweights.Consequently,westressthatan tSZ-φanalysisthatis basedonamulti-frequencyComptonpa- rameter map may present a high level of bias. A careful CIB-φ analysishastobeperformedsimultaneouslytoatSZ-φanalysis toavoidhighbiasinthefinalresults. Hill&Spergel(2014)constrainedσ (Ω /0.282)0.26 = 0.824± 8 m 0.029 and found that the tSZ-φ power-spectrum scales as σ6.1. 8 Thus,abiasof(20±10)%onthethetSZ-φamplitudeproduces a bias of (3.3±1.7)% on the σ amplitude. As a consequence, 8 theCIB-φinducedresidualbiasinthismeasurementcontributes about1σoftheuncertainty.Suchabiasmayexplainthediffer- encebetweenthetSZ-φcross-correlationandcosmologicalpa- rameterestimationmadewithatSZspectrum.Thelatterfavours lowervaluesforσ ,withσ (Ω /0.28)0.395 =0.784±0.016. 8 8 m 4 Withoutuncertaintiesorsystematicsforthecomputationofα corr 4