A direct probe ofcosmologicalpower spectra ofthe peculiar velocityfield and thegravitationallensing magnificationfrom photometric redshift surveys Adi Nusser,1,2,∗ Enzo Branchini,3,4,5,† and Martin Feix1,‡ 1Department of Physics, Israel Institute of Technology - Technion, Haifa 32000, Israel 2Asher Space Science Institute, Israel InstituteofTechnology - Technion, Haifa32000, Israel 3Department of Physics, Universita` RomaTre, ViadellaVascaNavale84, 00146, Rome, Italy 4INFN Sezione di Roma 3, Via della Vasca Navale 84, 00146, Rome, Italy 5INAF, Osservatorio Astronomico di Brera, Via Brera 28, 20121 Milano, Italy (Dated:July26,2012) Thecosmologicalpeculiarvelocityfield(deviationsfromthepureHubbleflow)ofmattercarriessignificant 2 informationondarkenergy,darkmatterandtheunderlyingtheoryofgravityonlargescales. Peculiarmotions 1 ofgalaxiesintroducesystematicdeviationsbetweentheobservedgalaxyredshiftszandthecorrespondingcos- 0 mologicalredshiftsz .Anovelmethodforestimatingtheangularpowerspectrumofthepeculiarvelocityfield 2 cos basedonobservationsofgalaxyredshiftsandapparentmagnitudesm(orequivalentlyfluxes)ispresented.This l methodexploitsthefactthatameanrelationbetweenz andmofgalaxiescanbederivedfromallgalaxies u cos inaredshift-magnitude survey. Givenagalaxymagnitude, itisshownthatthez (m)relationyieldsitscos- J cos 4 mologicalredshiftwitha1σerrorofσz ∼ 0.3forasurveylikeEuclid(∼ 109 galaxiesatz . 2),andcanbe usedtoconstraintheangularpowerspectrumofz−z (m)withahighsignal-to-noiseratio. Atlargeangular 2 cos separationscorresponding tol . 15, weobtain significant constraintsonthepower spectrumof thepeculiar velocityfield.At15.l.60,magnitudeshiftsinthez (m)relationcausedbygravitationallensingmagnifica- ] cos O tiondominate,allowingustoprobetheline-of-sightintegralofthegravitationalpotential.Effectsrelatedtothe environmentaldependenceintheluminosityfunctioncaneasilybecomputedandtheircontaminationremoved C fromtheestimatedpowerspectra. Theamplitudeofthecombinedvelocityandlensingpowerspectraatz∼1 h. canbemeasuredwith.5%accuracy. p - PACSnumbers:98.80.-k,98.80.Es o r t s I. INTRODUCTION galaxydistanceswhichareavailableforonlyasmallfraction a ofgalaxies. Presently,thenumberofgalaxieswithmeasured [ Darkenergyandthetheoryofgravitationdictatetheevolu- distancesisseveralordersofmagnitudebelowthatofgalaxies 1 tionoflarge-scalestructureintheUniverse.Thephysicalcon- in redshift surveysused for clustering studies. Furthermore, v ditionsallowingfortheformationofgalaxies,ultimatelylead although the underlyingpeculiar velocity is an honest tracer 0 to a bias between the distribution of galaxies and the under- ofthegeneralmatterflow,theinferenceofpeculiarvelocities 0 8 lyingmassdensity. However,large-scalemotionsofgalaxies fromobservationaldata is plaguedwith observationalbiases 5 are most certainly locked to the peculiar velocity field asso- [1]. Traditionalpeculiarvelocitycatalogsareexpectedtoim- . ciatedwith the gravitationaltugof the totalunderlyingmass prove within the next few years, but it remains questionable 7 fluctuations. This assumes that gravity is the only relevant how well observational biases will be controlled, especially 0 2 large-scale force and neglects the contribution from decay- atlargedistances. Analternativeprobeofthepeculiarveloc- 1 inglinearmodes.Despitethebiasbetweenthedistributionof ityfieldmaybeastrometricmeasurementsofgalaxiesbythe : galaxiesandtheunderlyingmatterfield,theclusteringproper- Gaia space mission [2, 3]. This probe is essentially free of v tiesofgalaxieshavebeenthemaintoolfortestingcosmologi- the classic biases contaminating traditional peculiar velocity i X calmodels.Currentlyplannedgalaxysurveyswillallowusto measurements,butitisalsolimitedtonearbygalaxieswithin r quantifythe clustering of galaxieson hundredsof comoving ∼100h−1Mpc. a Mpcsandtoevenmeasurecoherentdistortionsofgalaxyim- Herewe describeamethodforderivingstrongconstraints ageswhicharisefromgravitationallensingbytheforeground onthepowerspectrumofthegalaxies’peculiarvelocityfield matter. independent of direct distance measurements and any bias- On the other hand, the peculiar motions of galaxies have ing relation between galaxies and mass. The method is an traditionallybeenlesssuccessfulasacosmologicaltool,and extension of the approaches we have recently proposed [4– thereareseveralreasonsforthat. Neglectingotherpotentially 6], and it relies on using the observed fluxes of galaxies as importanteffects,peculiarvelocitiesareapproximatelyequal a proxy for their cosmological distance [7]. Although this to the redshifts less the correspondingHubble expansionre- mostbasic“distanceindicator”isverynoisy,thelargenumber cession velocities. The latter requiredirectmeasurementsof of galaxiesavailable in futuresurveyswill allow oneto beat down this noise to a sufficiently low level. Planned galaxy redshiftsurveyssuchasEuclid[8,9]willprobethestructure ∗[email protected] of the Universe over thousands of (comoving) Mpcs, com- †branchin@fis.uniroma3.it prising . 108 − 109 galaxies at z ∼ 1 and beyond. These ‡[email protected] observations will provide redshifts and fluxes, z and f, re- 2 spectively. The observed redshifts deviate from the cosmo- as[10] logicalredshiftswhichwouldbeobservedinapurelyhomo- geneousuniverse. Thelargenumberofgalaxiesandthelarge Θ= V(t,r) − Φ(t,r) − 2 t0dt∂Φ[rrrˆr(t),t], (1) sky coverage of these surveys can be used to derive a mean c c2 c2 ∂t Zt(r) globalrelationbetweenthemeanredshiftofagalaxyandits whererrrˆ is a unit vectoralong the line-of-sightto the object. apparent magnitude m = −2.5logf + const. We interpret Here the radial peculiar velocity V and the usual metric po- this relation as yielding the cosmologicalredshift z (m) for cos tential Φ are assumed as relative to their present-day values agivenapparentmagnitudem. Angularpowerspectraofthe at r = 0 (t = t ). The first and second terms on the right- difference z − z (m) between the observed redshift z of a 0 i cos i i handsideofEq. (1)aretheDopplerandgravitationalshifts, galaxyanditsexpectedcosmologicalredshiftz shouldcon- cos respectively,whilethethirdtermdescribestheenergychange tainvaluableinformation,mainlyonthepeculiarvelocityfield of light as it passes througha time-varyinggravitationalpo- whichisthemaincosmologicalsourceforz −z (m).Inthis i cos i tential. Notethatthisthirdtermisequivalenttothelate-time work,wewillshowthatthevelocitypowerspectrumonlarge integrated Sachs-Wolfe effect experiencedby photonsof the scales of a few 100 Mpcs could be constrained with signifi- cosmic microwave background(CMB). In what follows, we cant signal-to-noise ratio (S/N) at the effective depth of the survey. Another contributionto z −z (m) results from the will denote the three terms as ΘV, ΘΦ and ΘΦ˙, respectively. timeevolutionofthegravitationalipoteconstiali alongthephoton Also, we will consider angular power spectra (equivalent to path, but is significantly smaller than that induced by pecu- angular correlations) of Θ on large scales where the corre- liar velocities as we will show below. There are two addi- spondingsignalissignificantonlyrelativetotheexpecteder- tional,indirecteffectswhichmodifytherelationz (m)along ror. Throughoutthispaper,wethereforerelyonlineartheory a given line of sight. The first effect is related tcoos the envi- ina ΛCDM modelwhereperturbationsonallscales growat ronmental dependence between galaxy luminosities and the thesame rate. IntroducingD(t) asthegrowthrateoftheun- large-scale structure in which they reside. Since this depen- derlyingmassdensitycontrastδ=ρ/ρ¯−1,lineartheoryyields dence is closely connectedto the underlyingdensity field, it thewell-knownrelations canbeself-consistentlyremovedinouranalysis. Thesecond δ(t,rrr)= D(t)δ (rrr), (2) effectiscaused bygravitationallensing magnificationwhich 0 changestheapparentmagnitudesofgalaxiesinagivendirec- D(t) Φ(rrr,t)= Φ (rrr), (3) 0 tion.Thislattercontributioncouldactuallybeveryrewarding a since gravitationallensing providesa directprobeof the un- 2aD˙(t) ∂Φ V(t,rrr)=− 0, (4) derlyingmassdistribution.Consideringtheanalysispresented 3Ω H2 ∂r below,wewillthereforetreatitaspartofthesoughtsignal. 0 0 The paper is structured as follows: We begin with a de- where D(t0) = 1. The second relation is obtained from the taileddescriptionofthemethodanditsapplicationtogalaxy first using Poisson’s equation, i.e. ∇∇∇2rΦ = 3H02Ω0δ/2a. For redshiftsurveysinSec. II.InSec. III,weconsiderpredictions thespecialcaseΩ0 =1,wehaveD=aandfluctuationsinthe forthestandardΛCDMmodelanddiscussthemethod’svia- gravitationalpotentialremainconstantwithtime. bilityaswellasitsexpectedperformance.Finally,wepresent ourconclusionsinSec. IV.Forclarity,someofthetechnical A. Cosmologicalredshiftversusapparentmagnitude material is given separately in an appendix. In the follow- ing, we adoptthe standard notation. The matter density and the cosmologicalconstant in units of the critical density are WeaimtoderivethefieldΘsampledatthepositionsofall denotedbyΩ andΛ, respectively. The scale factora is nor- galaxiesinafluxlimitedredshiftsurveycoveringasignificant malizedto unity atthe presenttime (t = t ), andthe Hubble regionoftheskytogetherwithalargenumberofgalaxies.As 0 functionisdefinedasH =a˙/a.Further,r =c t0dt′/a(t′)will an example, we consider the planned Euclid redshift survey t [8,9].InordertoobtainanestimateofΘ foreachgalaxy,we bethecomovingdistancetoanobjectandzitscorresponding i R shalluseapparentgalaxymagnitudes(orequivalentlyfluxes) redshift, assuming a homogeneousand isotropic cosmologi- asaproxytothecosmologicalredshiftordistance. Although calbackground. Throughoutthepaper,thesubscript“0”will refer to quantities given at t = t , and a dot symbol denotes this most trivial distance indicator is very noisy, we will see 0 partialderivativeswithrespecttotimet,i.e. A˙ ≡∂A/∂t. thatthehighnumberofavailablegalaxiesallowsonetobeat downitsscatter. Themeancosmologicalredshiftz (m)correspondingtoa cos givenapparentmagnitudemis II. METHODOLOGY z2 z (m)=z¯ = z P(z |m)dz , (5) cos cos cos cos cos Zz1 Inaninhomogeneousuniverse,theobservedredshiftzofa andtherootmeansquare(rms)scatteraroundthisrelationis givenobjectdiffersfromitscosmologicalredshiftz (defined forthe unperturbedbackground). Therelativediffceosrencebe- z2 σ (m)= (z −z¯ )2P(z |m)dz , (6) tweentheseredshifts,Θ≡(z−z )/(1+z )canbeexpressed z cos cos cos cos cos cos Zz1 3 where z and z are the limiting redshifts in the survey, and forallgalaxiesinthesurvey. Notethatinthedenominatorof 1 2 P(z |m)denotestheprobabilitythatagalaxywithmeasured theabove,wehaveusedz insteadofz (m).Thissubstitution cos i cos apparent magnitude m has a cosmological redshift z . The is consistent at linear order and motivated by the fact that z cos i z (m) relation can be linked to certain characteristics of the isactuallyabetterestimateofthetruez thanz (m)which cos cos cos galaxysurvey.Theluminosityofagalaxyatz isL=4πfd2 additionally includes deviations from the actual value of z cos L cos whered (z )istheluminositydistancetothegalaxy,andits duetolargerandomerrorswithanrmsofσ . L cos z absolute magnitude is defined as M = −2.5logL+const = Environmentaldependencesof the luminosity distribution m − 5logd . Let Φ(M) be the underlying luminosity func- onthelarge-scaledensityfieldmaysystematicallyshiftallΘ L i tion suchthatΦdM is the numberdensityof galaxieswithin for galaxies lying in the direction of a certain line of sight. themagnitudeinterval[M,M+dM]. Generally,thefunction However, we will show below that the resulting signal con- Φ(M)dependsoncosmictime,butinfavorofasimplifiedde- tamination is small and in any case, it can be removedfrom scription, we will assume for the momentthat it varies little the correlations since information on the underlying density throughoutthedepthofthesurveyconsidered. Notethatitis field attherelevantscaleswill bedirectlyavailablefromthe trivialtoincludeatime-dependentevolutionaryterminΦ(M). observations(seeSec. IIC). TheprobabilityP(z |m)satisfies cos P(z |m)∝ P(m|z )n(z ), (7) 1. ExpectationsforEuclid cos cos cos wheren(z )istheunderlyingmeannumberdensityofgalax- cos Asatestcaseforafuturesurvey,weconsiderEuclidwhich iesat z . In the absenceofgalaxypopulationevolution,we cos aimsatmeasuringphotometricandspectroscopicredshiftsof have galaxieswithz ∼ 1over15,000deg2. Moredetailsonthe n(z )∝d (z )2ddc(zcos), (8) Euclid missioncocsan be found in [9]. Here we focus on the cos c cos dz photometric redshift survey since we aim at good statistics cos rather than precise redshift measurements. Photometric red- where d is the comoving distance from the observer to z . The relaction in Eq. (7) can be easily derived using Bayecoss’ shiftswillbemeasuredwithanerrorofσphot ≤0.05(1+zphot) for ∼ 30 galaxies per arcmin2, with z ≥ 0.7 and magni- theoremwhichyields cos tudesinthebroadR+I+Zband(550-920nm)RIZ ≤ 24.5. AB P(z |m)P(m)= P(m|z )P(z ), (9) Estimatesofthephotometricredshiftswillrelyonthreenear- cos cos cos infrared(NIR) bands(Y, J, and H in the range 0.92-2.0µm) where P(m)isanobservedquantity, P(z )isproportionalto forobjectswithY ≤ 24, J ≤ 24,andH ≤ 24. These cos AB AB AB n(z ),andP(m|z )isrelatedtoΦ(M)through willfurtherbecomplementedbyground-basedphotometryin cos cos thevisiblebandsderivedfrompublicdataorthroughengaged Φ(M) P(M|z )= . (10) collaborations. cos Ml(zcos)Φ(M)dM TheexpectednumberdensityofEuclidgalaxieswithmea- −∞ sured photometric redshift can be parametrized as follows HereMl =ml−5logdL(zcosR)istheabsolutemagnitudewhich [11]: corresponds to the limiting apparent magnitude m of the l galaxysurvey. n(zphot)∝z2phote−(zphot/z0)3/2, (12) Note that the scatter of z about the mean relation is not cos wherez =z /1.412isthepeakoftheredshiftdistribution Gaussian. However, the associated 1σ error in the derived 0 mean andz themedian. Herewe assume thatz = 0.9. We angular power spectra depends only on the rms quantity σ . mean mean z have compared the expected distribution n(z) of Eq. (12) to Moreover,thecentrallimittheoremimpliesthattheerrorsin that measured from galaxies with H ≤ 24 and RIZ ≤ thepowerspectratendtobeGaussian.Consideringtheactual AB AB 24.5 in the zCOSMOS catalog [12–14]. Indeed, we have observations,thetwoquantitiesz (m)andσ (m)canbeesti- cos z foundthatEq. (12)doesprovideagoodfittothedatainthe matedfromthefullsurveybydividingthedatainmagnitude range0.7<z <2.0whichwewillconsiderinouranalysis. binswithoutactuallycomputingP(z |m). Inthiscase,σ (m) phot cos z The observed z − H relation of zCOSMOS galaxies willincludeapositivecontributionfromthecosmologicalde- phot hasbeen used to derivez¯ (H) and σ (H) in differentH- viations,butthisisoverwhelmedbyboththeintrinsicscatter phot zphot magnitude bins. To match Euclid constraints, we have only in the z (m) relation and the uncertainties in the photomet- cos considered galaxies with RIZ ≤ 24.5, H ≤ 24, and ric redshifts. The underlying assumption is that the sought AB AB 0.7 < z < 2.0. The additionalconstraintsY ≤ 24 and cosmologicaldeviationsbetweenobservedandcosmological phot AB J ≤ 24 do not significantly modify our results and have redshiftscanceloutwhenallgalaxiesoftheentiresurveyare AB therefore not been enforced. Our results for the zCOSMOS used. Clearly, a global constant mode should persist in this dataareshowninFig. 1. The(blue)solidcurveandthe(red) procedure,butweshallneglectitinthispaper,assumingthat dashedcurverepresenttheexpectedz (H)andσ (H)forEu- itdoesnotaffectmodesinthepowerspectraonsmallerscales. cos z clidgalaxies,respectively. Sinceerrorsonthemeasuredred- Onceweobtainz (m),wecompute cos shift, i.e. σ , dominatethe scatterforH< 20, theywillbe phot z −z (m) added in quadrature to determine the effective scatter in the Θ = i cos (11) i 1+z z −H relation. i cos 4 1.2 1 0.8 z cos z 0.6 σ /(1+z) z 0.4 0.2 0 18 19 20 21 22 23 24 H FIG.1. Themean relationz (m) andthecorresponding rmsscatter σ(m)(1+z) for Euclidgalaxies withRIZ ≤ 24.5, H ≤ 24, and cos z AB AB photometricredshiftsintherange0.7<z <2.0:TheshownresultsarebasedonzCOSMOSdata. phot B. Sphericalharmonicsdecompositionfordiscretenoisydata observedpartofthesky,andthenumberoftermsinthesumof thesecondrelationis(2l+1)f . Asisobvious,theserelations sky Asusual,angularpowerspectraaredefinedintermsofthe shouldbeunderstoodtoholdintheapproximatesense[17]. spherical harmonics Y (rrrˆ). For an all-sky continuous field Considering a function f(rrrˆ) with limited sky coverage in lm f(rrrˆ),thedecompositionis thecontinuouslimit,wedefine 1 flm = dΩf(rrrˆ)Ylm(rrrˆ), flm = f1/2 dΩf(rrrˆ)Ylm(rrrˆ), (15) Z sky Z ∞ +l (13) wheretheintegrationisagaintakenovertheobservedregion f(rrrˆ)= f Y∗ (rrrˆ), lm lm only.TheangularpowerspectrumC isdefinedasthevariance l Xl=0 mX=−l ofthe flm’sandgivenby Inourcase,however,wehavetodealwithpartialskycoverage C =<|f |2 > , (16) (around30%oftheskyforEuclid)andconsiderthatthefield l lm ens is sampled at discrete points given by the galaxy positions. wheretheaverageistakenovermanydifferentrealizationsof The limited sky coveragecould formallybe describedby an a field with the same power spectrum as f. For brevity, we appropriatemaskingofthesphere[15,16]. However,thede- willusethesymbol<·>withoutanysubscripttorefertothis scriptionintermsofmaskswillunnecessarilycomplicatethe kind of averaging. Up to cosmic variance, the values of C l notationandsomewhatobscurethephysicalinterpretationof obtainedfrom the ensemble average < |f |2 > should equal lm the results. We shall thereforeresortto a simplified descrip- those computedby averagingoverthe (2l+1)f modesfor tionandassumethatweareprovidedwithasurveycovering sky eachl. Thus,theangularpowerspectrummaybeestimatedas 4πf steradiansofthesky.Foreachdegreel,weassumethat tfhuellrseskykayreco(2vle+rag1e).fsCkyoinnsdiedpeernindgenmtomdoedsewsi,tihnsatnegaudlaorfr2elso+lu1tifoonr Cl =<|flm|2 >m≡ (2l+11)f |flm|2. (17) sky m muchsmallerthantheextentofthesurvey,wethenhave X Inthefollowing,weshallinterchangebetweenthesedifferent Y (rrrˆ)Y∗ (rrrˆ)dΩ= f δKδK , kindsofaveragingwheneverappropriate.Thefactor1/f1/2in lm l′m′ sky ll′ mm′ Eq. (15)andtherulesinEq. (14)guaranteethat sky Z (14) (2l+1)f Y (rrrˆ)Y∗ (rrrˆ′)= skyP(rrrˆ·rrrˆ′), 2l+1 m lm lm 4π l C(cosθ)= Cl 4π Pl(cosθ), X Xl (18) whereP istheLegendrepolynomialofdegreel. Throughout l C = dΩC(cosθ)P(cosθ), thepaper,theangularintegrationiscarriedoutonlyoverthe l l Z 5 which can be shown by decomposing P into Y according galaxy according to the σ in the sum of Eq. (20). To min- l lm i to the second relation in Eq. (14) and assuming thatC(θ) is imize the effects of shot noise, we weight each galaxy by a negligibleforangularseparationslargerthantheextentofthe factorofw whichisgivenby i survey(seeAppendixA). Foradiscretesamplingof f atthepositionsof N galaxies Nσ−2 w2 = i . (26) distributedovertheobservedpartofthesky,wewrite[18] i σ−2 j j N ThisparticularweightingschePmeyields f(rrrˆ)≈ n(rrrˆ)f(rrrˆ)dΩ i Xi Z (19) N σ2 = , (27) ≈n¯ f(rrrˆ)dΩ, f σ−2 j j Z wheren(rrrˆ)istheprojectednumberdensityofobjectsandn¯ = wherewehaveassumedthatthPattheunderlyingsignalmakes a negligible contribution to σ2. The weighting does not af- N/(f 4π)isthecorrespondingmeannumberdensityoverthe f sky observedpartofthesky. Herewehaveassumedthat f itself fect the ensemble average of |f |2, and its net effect is that lm dependsonthedensitycontrastδ=n/n¯−1,sothatthelaststep σ should be computed as given by Eq. (27), i.e. it merely f appliestolinearorderinthefluctuations. Again,theangular reducesshotnoiseerrors(foradditionaldetails,seeAppendix integration is carried out only over the observed part of the B).Inprinciple,onemayuseanyweightingscheme. sky.InanalogytoEq. (15),wefurtherdefine 1 N C. Environmentaldependencesintheluminosityfunctions f = f(rrrˆ)Y (rrrˆ). (20) lm n¯f1/2 i lm i andmagnificationbygravitationallensing sky Xi=1 In this case, one can show that the angular power spectrum So far, we have assumed that the systematic shifts in the takestheform(seeAppendixB) z (m)relationforgalaxiesinagivendirectionaresolelydue cos tothetermsappearingontheright-handsideofEq.(1).How- σ2 ever, additional shifts may arise from changes in the mean C =<|f |2 >− f, (21) l lm magnitudesdue to large-scaledensityfluctuationsin a given n¯ direction,i.e.environmentaldependencesandevolutioninthe where the second term represents the contribution of shot luminosity function, and the magnification effect caused by noiseduetothediscretesamplingof f andσ2f = i f(rrrˆi)2/N. gravitationallensing. In what follows, we shall denote their ToestimatethevaluesofCl,wethereforeuse contribution as Θenv and Θlens, respectively. Both effects re- P sultinamagnitudeshiftwhichtranslatesintoadditionalcor- σ2 Cl =<|flm|2 >m − n¯f. (22) rTeolamteodddeelvtihaetiocnosntorfibtuhteioenstoimfathteedsezceosfffercotms iintstmheeacnorrreellaattiioonn. functionofΘ,weneedtotranslateamagnitudeshift∆mintoa If f =S(rrrˆ)+ǫ whereS isanunderlyingcosmologicalsignal i i i correspondingshift∆z. Thiscandirectlybereadoffthe(blue) andǫ isanuncorrelatedrandomerror,thissimplyreflectsthe i solidcurvez (m)showninFig.1inwhichashift∆m = 0.2 factthatC =< |S |2 >. The expectederrorin this estimate cos l lm leadstomeanshift∆z ≈0.017[19]. ofC isgivenby(again,seeAppendixB) cos l To quantifythe impactof environmentaldependences,we 2 σ2 2 assumethatthesystematicshiftingalaxymagnitudedepends Σ2 = f +C (23) on the density contrast δ in regions where they reside, irre- (2l+1)f n¯ l spective of the smoothingscale [20, 21]. Here we adoptthe sky which includes contributions from both shot noise Σsn and flirnoemarthreelavtiisoibnle∆mban=d[02.22]δ.wThhiechacitsuaolbdseeprveantdioennaclelyisinaffeurrnecd- cosmicvarianceΣcv, tion of the photometric band, and it is expected to be much weakerinEuclid’sH-bandasindicatedbytheweakenviron- Σ = 2 σ2f, Σ = 2 C (24) mental dependence of the Schechter parameter M∗ fitted to sn (2l+1)f n¯ cv (2l+1)f l NIR luminosities [23]. The remaining Schechter parameter, s sky s sky usually dubbed α, exhibits a stronger dependence, but since Thevarianceofthescatterσ2, this parameter fixes the shape at the faint end, it will have i onlyasmallimpactindeepsurveyslikeEuclid. Nonetheless, σ2 (z)+σ2(m) itwillbepossibletoremovemostofthecontaminationcaused σ2i =<ǫi2 >= phot(1i+z)2z i , (25) bythiseffectusingtheobserveddistributionofgalaxiesinthe i Euclidsurvey. dependsonbothmagnitudeandredshift. Notethatthefactor Intheweak-fieldlimit,themagnificationinducedbygrav- of(1+z)arisesfromthedefinitionofΘinEq. (1). Consid- itational lensing is proportionalto 1+2κ, where κ is the ef- i eringtheapplicationtorealdata,itisprudenttoweighteach fectiveconvergencefield,i.e. anintegraloverthe(weighted) 6 densitycontrastalongthelineofsight[24]. ForaflatΛCDM where j is the usual first-kind spherical Bessel function of l modelandafixedsourceredshiftcorrespondingtoacomov- degreel,weget ingsourcedistancer,onefinds il r2 Θ˜Φ = drW d3kΦ j(kr)Y (kkkˆ). (33) κ(θ,r)= 3H02Ω0 rdr′r′(r−r′)δ(r′θ,r′), (28) lm 2π2c2 Zr1 φZ kkk l lm 2c2 a(r′)r Z0 Therefore,wefinallyarriveat wherethetwo-dimensionalangularvectorθ isperpendicular CΦ =<|Θ˜Φ |2 > to the lineof sight. Therefore,itis straightforwardto model l lm themagnificationanditseffectinouranalysis. Themagnifi- 2 r2 2 (34) cationfield containsvaluableinformationsince it probesthe = πc4 dkk2PΦ(k) drWφjl(kr) , growth of the angular derivatives of the gravitational poten- Z (cid:12)(cid:12)Zr1 (cid:12)(cid:12) (cid:12) (cid:12) tial[e.g.24,25]. Itcanalsobeusedtoconstrainthegravita- wherewehaveused<ΦkkkΦkkk′ >=(cid:12)(cid:12)(cid:12)(2π)3δD(kkk−kkk′)P(cid:12)(cid:12)(cid:12)Φ(k). Sim- tionalslipwhicharisesincertainmodificationsofthegeneral ilarly,usingthelinearrelationinEq.(4),weobtain theory of relativity [e.g. 26, 27]. Given a theory of gravity, much of the contribution to the power spectra from gravita- CV = 2 dkk2P (k) r2drW ljl −kj 2, (35) tionallensingcan, in principle, be removedusing the under- l πc2 Φ V r l+1 lyinglarge-scaledensityfieldwhichisinferredfromthefore- Z (cid:12)(cid:12)Zr1 !(cid:12)(cid:12) (cid:12) (cid:12) groundgalaxydistribution. Consideringthefollowinganaly- where W = 2aD˙p(r)/3Ω (cid:12)(cid:12)H2. As for ΘΦ˙, the last(cid:12)(cid:12) term ap- sis,however,wewilltreattheeffectsoflensingmagnification V 0(cid:12) 0 (cid:12) pearing in Eq. (1), we will assume that the signal is mostly aspartofthesignal. Aswewillseebelow,itscontributionis causedbythelarge-scalestructurebetweentheobservedhigh notnegligible,andcanbeconstrainedtogetherwiththepower redshiftgalaxysampleandtheobserver.Therefore,ΘΦ˙ isap- spectrumofthevelocityfield. proximatelythesameforallgalaxiesalongacommonline-of- sightsuch thatΘ˜Φ˙ ≈ ΘΦ˙. The angularpowerspectrumthen reads III. THEORETICALANGULARPOWERSPECTRA CΦ˙ =<|ΘΦ˙ |2 > l lm Inthefollowing,wewillpresentpredictionsfortheangular 8 r 2 (36) powerspectra correspondingtothe varioustermsinEq. (1). = dkk2P (k) dr′W j(kr′) , Sincetheobservedgalaxiescoveraredshiftrangez1 <z<z2, πc6 Z Φ (cid:12)Z0 φ˙ l (cid:12) wdeefidnoednaottacospnesicdifierctrheedsahnigfut,labrutpionwsteerasdpwecetruaseofaquantity f whereWφ˙(t)/a=(d/dt)[D/a].Th(cid:12)(cid:12)(cid:12)(cid:12)eintegrationove(cid:12)(cid:12)(cid:12)(cid:12)rr′istaken fromr = 0outtoadistancebeyondwhichΦbecomesnearly r2 constantwithtime. Forsimplicity,wewillassumethatmost f˜(rrrˆ)= f [rrrˆr,t(r)]p(r)dr, (29) ofthecontributiontotheintegralcomesfromtheinneredge Zr1 oftheconsideredgalaxysample,andadoptz = 1inEq. (36) forallgalaxiesinthesurvey. ForEuclid,thisseemstobethe where r and r are the comoving distances at the survey’s 1 2 casesincethemeanredshiftisexpectedtobez∼1.Notethat limiting redshifts, z and z , respectively, and p(r)dr is the 1 2 the same assumptionsare used when calculating the angular probabilityofobservingagalaxywithintheinterval[r,r+dr]. powerspectrumClens. Note that the function f represents any of the terms on the right-handside of Eq. (1), i.e. ΘV, ΘΦ, and ΘΦ˙. Beginning withΘΦ,wewrite A. PredictionsoftheΛCDMscenario Θ˜Φ = dΩΘ˜ΦY (rrrˆ) lm lm Havingderivedtherelevantexpressionsabove,wearenow Z1 r2 (30) readytomakepredictionsfortheframeworkofΛCDM.Here = dΩY (rrrˆ) W (r)Φ (rrrˆr)dr, weadoptaspatiallyflatmodelwithbest-fitparametersbased c2 lm φ 0 Z Zr1 on the CMB anisotropies measured by the Wilkinson Mi- crowave Anisotropy Probe (WMAP) [28]. In this case, the wherewehaveusedthelinearrelationΦ(rrr,t)=(D/a)Φ (rrr,t ) 0 0 totalmassdensityparameterisΩ =0.266,thebaryonicden- and defined W = Dp(r)/a, with D(t) and a(t) evaluated at m φ sity parameter Ω = 0.0449, the Hubble constant h = 0.71 t=t(r). ExpandingΦ (rrr)inFourierspace, b 0 in units of 100 kms−1Mpc−1, the scalar spectral index n = s Φ0(rrr)= (21π)3 d3kΦkkkeikkk·rrr, (31) 0w.i9t6h3in,asnpdheσre8s=of08.8h0−1foMrpthce.rWmeswofolriknewairthdeanpsaitryamfluectrtiucaftioornms Z ofthepowerspectrumtakenfromRef. [29](seeEqs. 29-31 andusing intheirpaper). Forthecalculationofangularpowerspectra, we use the expressionsfromSec. III togetherwith p(r) cor- eikkk·rrr =4π ilj(kr)Y∗ (nnnˆ)Y (kkkˆ), (32) respondingtotheredshiftdistributionofgalaxiesappropriate l lm lm forEuclidwhichisgivenbyEq. (12). l,m X 7 −9 10 Ctot Clens −10 10 Cenv Cl Σ CV sn CΦ −11 10 CΦ˙ 10Σ env −12 10 5 10 15 20 25 30 l FIG.2. AngularpowerspectraintheΛCDMmodelasexplainedinthetext: TheshadedarearepresentsthecosmicvarianceuncertaintyΣ cv onCtot,andΣ isthe1σshot-noiseerrorfortheEuclidsurvey. ThedesiredsignaliscontaminatedbyCenvwhichresultsfromenvironmental sn dependences in the luminosity function and can be removed from the actual Euclid data with high precision represented by Σ . Only env contributions from ΘV (peculiar velocity), ΘΦ (gravitational shift), and Θlens (lensing magnification) are included inCtot. At l ∼ 60 (not shown),Σ andCtotbecomecomparable. sn Theangularpowerspectraandassociatederrorsareplotted B. Signal-to-noiseandexpectederroronmodelnormalization inFig. 2. Theshot-noiseerrorΣ iscomputedusingEq.(27) sn and assuming σ = 0.17 as inferred from the (red) dashed f SupposethatobservationsyieldanestimateCobsfortheto- curve in Fig. 1. For completeness, we also show the power l spectrumCenvwhichresultsfromenvironmentaldependences talpowerspectrumCltot. LetClH0 =0andClH1 betheexpected totalpowerspectraforthenullhypothesiswithnocorrelations in the luminosity function. As explained in Sec. IIC, the (H )andoftheΛCDMmodel(H ),respectively.Theratioof contaminationarising from this effect can be removedgiven 0 1 probabilitiesforH andH isgivenby theobserveddistributionofgalaxiesinthesurvey. Theaccu- 1 0 racy to which this can be achieved for Euclid is represented byΣ , the1σerrorwithinwhichCenv canbeestimatedex- −2lnP(H )/P(H )= env 0 1 pclleicairtlfyrofmromCletnhseindathtae(fisgeuereA,pmpaegnndiifixcBatifoonrcdaeutsaeilds)b.yAgrsavis- Clobs−ClH0 2 − Clobs−ClH1 2 +2ln Σsn . (37) i(tzait−iozncaosl)l/e(n1s+inzgi)i.ntroducessignificantangularcorrelationsin Xl (cid:16) Σ2sn (cid:17) (cid:16) Σ2 (cid:17) Σ ! The quantityCtot is the power spectrum of the sum of the If H holds, it followsthatCH1 is equalto theCtot shown in threesignalsΘlens,ΘV,andΘΦ,correspondingtothelensing Fig.12,andCobs−CH1 isaralndomvariablewithlvarianceΣ2 magnification, the Doppler shift, and the gravitational shift, l l givenbyEq. (23). Therefore,thesignal-to-noiseratio(S/N) respectively. Note that the calculation of Ctot does include forrejectingH is 0 covariancebetween the three individualsignals. The contri- butionfromΘΦ˙ isnegligibleasisindicatedbyCΦ˙,andwedo S 2 not include it in Ctot. At l∼< 10, CV is the dominant contri- =−2lnP(H )/P(H )= 0 1 bution, but the lensing term Clens takes over at l & 10−15, N 0 rTohueghshlyaduendtilalre∼a6r0epwrehseernetisttdhreop1sσbeelrorworthoef sChtoott-dnuoeisetolecvoesl-. (cid:18) (cid:12)(cid:12)(cid:12)(cid:12) ((cid:19)2l+1)fskyCl2 + 1+ n¯Cl 2−2ln 1+ n¯Cl , (38) mtoicprvoavriidaencme.eaFsourrepmraecnttiscaolfpCurVpoasneds,Citleniss tbhyeraenfoarpepproospsriibaltee Xl 2σ4f/n¯2 σ2f σ2f fittingprocedureofthetwocorrespondingcurvestothemea- whereC = CH1 = Ctot. Substituting the relevantquantities l l l suredCtot. Unfortunately,atlowl,cosmicvarianceissolarge into the above, we obtain (S/N)| = 101, where the above 0 thataccurateconstraintsonCΦdonotseempossible. sumrapidlyconvergesbyl = 30. Ontheotherhand,ifH is 0 8 1.4 9.2 6.17 1.3 2.3 1.2 1.1 v a 1 0.9 0.8 0.7 0.8 0.85 0.9 0.95 1 1.05 1.1 1.15 1.2 a lens FIG. 3. Confidence levels for measuring the amplitudes of the velocity and lensing power spectra by fitting a2CV +a2 Clens to random v lens realizationsofCtotwhichincludebothshotnoiseandcosmicvariance: Thefittingprocedureignoresphysicalcross-correlationsbetweenthe usedquantities.Contoursareshownfor∆χ2 =2.3,6.17,and9.2,correspondingtoconfidencelevelsof68%,95.4%and99%,respectively. true,thentheS/N forrejectingH isgivenby two-parametermodel 1 S 2 =−2lnP(H1)/P(H0)= Ctot,m =a2VCV+a2lensClens. (41) N 1 (cid:18) (cid:12)(cid:12)(cid:12)(cid:12) (cid:19)(2l+1)fskyCl2 + 1+ n¯Cl −2+2ln 1+ n¯Cl , (39) aTshwisemlloadsealnfyorcCovtoat,rmiannecgelbecettswteheenctohnetrreibmuatiionninogfsΘigΦnatols,CΘobVs l 2(σ2f/n¯ +Cl)2 σ2f σ2f and lensing magnification Θlens. Using the above expres- whXichquicklyconvergesatl∼ 50andyields(S/N)|1= 14.7. sion, we have repeated the procedure described above for The fact that (S/N)|1 ≪ (S/N)|0 is a result of cosmic vari- constraining σ8 with the 1000 random realizations of Cobs. ancewhichiszerofortheH hypothesis,butsignificantinthe The result is presented in Fig. 3 which shows contours of 0 ΛCDMscenario,i.e. Σ2 =2C2/(2l+1)f . ∆χ2 = −2lnP + 2lnmax(P) as a function of aV and alens. InadditiontoS/Ncocnvsideraltions,wecaskny assesshowwella The contourshave beencomputedforone of the randomre- measurementofCtotwouldconstrainthenormalizationofthe alizations,butthevaluesofthebest-fitparameters(givingthe ΛCDMpowerspectrumintermsofσ . Tothisend,wewrite lowest∆χ2)obtainedforthisparticularrealizationhavebeen 8 the model’s total power spectrum as Ctot,m = (σ /0.8)2Ctot, shifted to their underlyingvalue, i.e. unity for both. This is 8 whereCtotasillustratedinFig.2isobtainedforσ =0.8.The reasonablesincetheaveragebest-fitparametersof1000real- 8 expected1σerroron σ8 isthen (−∂2lnP(H1)/∂σ28)−1, where izationsareessentiallyunbiased.Asseenfromthefigure,alens P(H )isnowexpressedas is constrained with better accuracy than the velocity ampli- 1 tudea . ThisisconsistentwithFig. 1whichshowsthatClens V Cobs−Ctot,m 2 dominatesthetotalsignaloveralargerangeoflwhileCV is lnP(H1)=− (cid:16) l 2Σ2l (cid:17) − lnΣl, (40) significantonlyforlowlwherecosmicvariancebecomesin- l l l creasingly important. Still, both parametersare constrainted X X andΣ isgivenbyEq. (23)withC = Ctot,m. Usinganormal withgoodaccuracy. l l l probability distribution of the form given in Eq. (40) with fixedσ = 0.8,wehavegenerated1000randomrealizations 8 of Cobs. For each of these realizations, we have maximized IV. DISCUSSION P(H ) givenby Eq. (40) with respectto σ . Asa result, we 1 8 findthatthetruevalueσ =0.8isrecoveredwithinarelative In this paper, we have presented a novel method for de- 8 1σ error of less that4%, withoutany statistically significant rivingdirectconstraintsonthe peculiarvelocityand gravita- bias. tionalpotentialpowerspectrafromcurrentlyplannedgalaxy We have also inspected the possibility of constraining the redshift surveys. The large number of galaxies with photo- velocity and lensing signal amplitudes with the help of the metricredshiftsinthesesurveysallowsonetoexploitappar- 9 entgalaxymagnitudesasaproxyfortheircosmologicalred- wide-fastsurveyisexpectedtoobserve∼ 20,000deg2inthe shiftssinceitbeatsdownthelargescatterinthez (m)relation ugrizybands. Afterabout10yearsofoperation,itwillreach cos andtheuncertaintyinthephotometricredshifts. Themethod muchdeeperdepths(downtoaco-addedmagnituder = 27), aimsatdirectlyconstrainingpowerspectraoftheunderlying detectingabout3×109galaxies[40]. fluctuationfieldsindependentofthewaygalaxiestracemass. Othermethodsforextractingcosmologicalinformationfrom redshiftsurveysrelyonaccuratemeasurementsofthegalaxy V. ACKNOWLEDGMENTS power spectrum in redshiftspace (since galaxy distances re- main unknown). The power spectrum and other statistical We are grateful to Micol Bolzonella for computing all measuresbasedonthedistributionofgalaxieshavebeensuc- redshift-magnitude relations used in this papers from the cessful at probing the nature of dark matter and placing im- zCOSMOS data. We also thank Henry McCracken for pro- portantconstraintsonneutrinomasses [30–32]. Having said viding us with the H-band magnitudesof zCOSMOS galax- that, however, they depend on a very accurate knowledgeof ies. E.B. thanks Gianni Zamorani and Massimo Meneghetti the relation between galaxies and the full underlying matter for useful discussions and suggestions. This work was sup- distribution. Themethodwehaveconsideredhereislesspre- ported by THE ISRAEL SCIENCE FOUNDATION (grant cise,butitiscompletelyindependentofthegalaxyformation No.203/09),theGerman-IsraeliFoundationforResearchand process and offers a much more sensitive assessment of the Development, the Asher Space Research Institute and the underlying physical mechanism driving cosmic acceleration WINNIPEG RESEARCH FUND. E.B. acknowledges the andstructureformation. Thisapproachisparticularlyworth- supportprovidedbyMIURPRIN2008“Darkenergyandcos- whileifsuchconstraintsonthevelocityfieldandthegravita- mologywith largegalaxysurveys”andby Agenzia Spaziale tionalpotentialarecontrastedwithlocalconstraintsobtained Italiana (ASI-Uni Bologna-Astronomy Dept. ’Euclid-NIS’ fromdataatlowredshifts. Forexample,peculiarmotionsof I/039/10/0). M.F.is supportedin partatthe Technionbythe galaxieswithinadistanceof∼ 100h−1Mpccanbemeasured LadyDavisFoundation. usingtightrelationsbetweenintrinsicobservablesofgalaxies [e.g. 33, 34], and also using astrometric observations of the Gaia space mission which is currently scheduled for launch AppendixA:ProofoftherelationsinEq.(18) in2013. Thesepeculiarmotionsofgalaxieshavebeenuseful for constrainingcosmologicalparameters[35] as well as the Foranyfunction f(rrrˆ),wehavedefined amplitudeofthevelocityfieldinthenearbyUniverse[5,36]. 1 AlthoughwehavepresentedpredictionsfortheEuclidsur- flm = f1/2 dΩf(rrrˆ)Ylm(rrrˆ). (A1) vey, the science proposedin this paper will nothave to wait sky Z for this space mission. In fact, several ground-based photo- UsingtherelationsgiveninEq. (14),weget metric surveysin the optical and near-infraredbands, which willconstitutethebackboneofEuclid’sphotometry,willpro- 1 1 C = <|f |2 >= vide photometric redshift catalogs that can be used for our l (2l+1)f lm (2l+1)f2 purposeswellbeforethelaunchofthesatellite. Onashorter sky Xm sky timescale, the VLT Survey Telescope (VST) will be used to × dΩdΩ′ < f(rrrˆ)f(rrrˆ′)> Y (rrrˆ)Y∗ (rrrˆ′) lm lm carry out the Kilo Degree Survey (KiDS), one of the ESO Z m publicsurveys. Itwillcover1,500deg2 tou = 24,g = 24.6, 1 X r=24.4,andi=23.1,andwillprobablycontain∼108galax- = 4πf dΩdΩ′C(rrrˆ·rrrˆ′)Pl(rrrˆ·rrrˆ′) (A2) ieswithmeasuredphotometricredshifts[37]. Also,theDark sky Z 1 EnergySurvey(DES)willstartitsoperationssoon,anditwill = dΩ′ dΩC(rrrˆ·rrrˆ′)P(rrrˆ·rrrˆ′) cover5000deg2oftheSouthernskywithin5years,reaching 4πf l sky Z Z magnitudesupto∼24inSDSSgrizfilters,comparabletothe = dΩC(cosθ)P(cosθ), limitingmagnitudesofEuclidandwitharedshiftdistribution l dN/dz similar to that of Euclid galaxies. DES will measure Z photometricredshiftsof∼3×108galaxieswithσ ∼0.12 where for the last step, we have used that the integral over photo at z ∼ 1 [38]. Furthermore, the first of the four planned dΩisindependentofrrrˆ′,assumingthatthecoherenceangular Pan-STARRStelescopes[39]hasbeenoperationalsinceMay length is smaller than the spatial extent of the survey. Sim- 2010. The planned 3π area of the sky will be considerably ilarly, the first relation in Eq. (18) follows from substituting shallower, detecting galaxies below a limiting magnitude of it into the second and exploiting the (near) orthogonality of ∼24inthegrizbands.AdeepersurveyinvolvingthePS1and Pl(cosθ). PS2telescopesiscurrentlybeingplanned. Thesurveyshould cover ∼ 7,500deg2 with limiting fluxes g = 24.7, r = 24.3, i = 24.1, and z = 23.6. Photometric redshifts will then be AppendixB:Estimatingangularpowerspectrafordiscretedata measured for ∼ 4.5 × 108 galaxies with similar errors. Fi- nally, on the long run, the Large Synoptic Survey Telescope Justasinthemainbodyofthepaper,wewilltakethesym- (LSST)isexpectedtostartoperationsin2020.Itsmaindeep- bol< (·) > ≡ (1/(2l+1)f ) (·)todenoteaveragingover m sky m P 10 all indices m corresponding to the degree l of the spherical Notetheremarkablefactthatduetotheuncorrelatednatureof harmonic decomposition. Again, the symbol < · > without theerrorsandbecauseα, βaswellasα′ , β′,onlysecond- anysubscriptwillrefertothepreviouslyintroducedensemble ordermomentsofnand f contributetoΣ2. Thisisimportant average,andweshallinterchangebetween< · > and< · > sinceitimpliesthatthenon-Gaussiannatureofthescatterin m whenever appropriate. For brevity of notation, we also use thez (m)relationdoesnotaffectthevarianceoferrorsinthe cos thedefinitionYi ≡ Y (rrrˆ),thusomittingthesubscriptlsince estimatedC. Assuming that fn is a Gaussian random field, m lm i l all calculationswillrefer to a givendegreel of the spherical wefurtherhave harmonics. Startingfromthedefinitionof flm inEq. (20),we < frnr frnrfrnr frnr > obtain α α β β α′ α′ β′ β′ = < frfrnrnr >< fr frnr nr > 1 α β α β α′ β′ α′ β′ <|f |2 >= < f f >YiY∗j (B6) lm n¯2f i j m m + < fαrfαr′nrαnrα′ >< fβrfβr′nrβnrβ′ > sky i,j X + < frfrnrnr >< fr frnr nr >. = n¯21f < fi2 > Ymi 2+ n¯21f < fifj >YmiYm∗j (B1) Considering the aboαveβ,′ aαveβr′ages wα′ithβ nαo′ eβqual indices will sky i sky i,j σ2 X (cid:12)(cid:12)(cid:12) (cid:12)(cid:12)(cid:12) X contribute terms proportionalto Cl2 in the sum of Eq. (B5). f Neglectingthecontributionofshotnoise,thiswouldyieldthe = +C, n¯ l usual cosmic variance expression 2C2/(2l + 1)f . For the l sky wherethevalueofσ isinferredfromthez (m)relation,i.e. case α = α′ and β = β′, the second term on the right-hand σ2 = σ2/N. AsoufrestimateforC,wethcoesreforetake side of Eq. (B6) turns into σ4(n¯δΩ)2, where we have again f i i l f used that < n2 >=< n >= n¯δΩ. The other combination P C =<|f |2 > −σ2f. (B2) which makes tαhe same αcontribution is α = β′ and α′ = β. l lm m n¯ A bit of algebra shows that the contribution to the variance Another way of arriving at this result is to partition the ob- in Eq. (B5) of these two combinationsis 2σ4/(2l+1)n¯2f . f sky servedskyintoinfinitesimallysmallcellsofangularsizeδΩ, For α = α′ and β , β′, the second term on the right-hand witheachcellcontainingatmostonegalaxy[16].Inthiscase, side of Eq. (B6) simplifies to σ2 < frfr > (n¯δΩ)3, and the f β β′ wewrite totalresultofsimilarpermutationswithonlytwoequalindices f = 1 f n Yα, (B3) reads 4σ2fCl/(2l+ 1)n¯fsky. Finally, summing up all relevant lm n¯f1/2 α α m termsleadsto where nα, the number of gsakylaxXiαes in cell α, is either 0 or 1. Σ2 = 2 σ2f +C 2. (B7) Usingthat< nα >= n¯δΩinthisrepresentation,weobtainthe (2l+1)fsky n¯ l expression 1 UsingtheweightsdefinedinEq.(26),itiseasytosaythatthe <|flm|2 >= n¯2f < fαfβ ><nαnβ >YmαYm∗β samInetrheesumltsaianrebovdalyidowf itthheσpfapgeivr,enwbeyhEavqe. (a2s7su).med that en- sky α,β X vironmental dependences of the galaxy luminosity function 1 = n¯2f < fα2 ><n2α > Ymα2+Cl ariseduetovariationsinthelarge-scaledensitycontrast. The sky Xα (cid:12) (cid:12) (B4) signalcontaminationassociatedwiththiseffecthasanangu- = 1 < f2 >n¯δΩ Yα(cid:12)(cid:12) 2+(cid:12)(cid:12) C larpowerspectrumwhichisproportionaltothatofthedensity n¯2f α m l contrast. Inthefollowing,wegiveanestimateforthispower sky α X (cid:12) (cid:12) spectrumandthe1σerrorwithinwhichitcanbeconstrained = σ2f +C, (cid:12)(cid:12) (cid:12)(cid:12) fromtheobservedgalaxydistribution.Tobeginwith,wecon- n¯ l sider[16] wherewehaveusedtherelationn2α =nα. Wefindthispartic- δ = 1 n Yα. (B8) ular approacha little moreintuitiveto computethe expected lm n¯f1/2 α m error of the estimate in Eq. (B2). Let fr and nr be possible sky Xα randomrealizationsofthedatainanotheruniverse. Thevari- Forthe f ’s,wehaveassumedthatn isadiscretesampling lm α ance of the error in C may then be written as an ensemble ofauniformdistribution.Thatwasconsistentwithlinearthe- l averageovertheserealizations,i.e. orysincewehaveassumedthat f isafunctionofthedensity contrast. However, now we are interested in the autocorre- Σ2 = <(Cr−C)2 > l l lation functionof n , and thus the treatmentof δ is a little α lm 2 different. To obtain an estimate for the angular power spec- 1 = n¯2f fαrfβrnrαnrβ <YmαYm∗β >m −Cl trumofthedensitycontrast,wewrite = *1 sky Xα,β < frnr frnrfrnr frnr > + (B5) <|δlm|2 >= n¯21f <nαnβ >YmαYmβ n¯4fs2ky α,Xβ,α′,β′ α α β β α′ α′ β′ β′ 1 sky Xα,β (B9) × <YmαYm∗β >m<Ymα′′Ym∗β′′ >m′ −Cl2. = n¯ +Clδδ.