what can the cmb sll tell us?• praccally cv limited for determining the integrated reionizaon...
TRANSCRIPT
WhatcantheCMBs-lltellus?
ElenaPierpaoliUniversityofSouthernCalifornia
Summary
– AbroadlookatPlanckparameteres-ma-on
– Reioniza-on– DiffuseForegroundscharacteriza-on– Tensor‐to‐scalar(i.e.res-ma-on)forPlanckandfutureexperiments
From:Colombo&Pierpaoli(arXiv:0804.0278) Colombo,PierpaoliPritchard(arXiv:0811.2622) Betouleetal(arXiv:0901.1056)Macellarietal(tobesubmiZed)
+WhitePapersforCMBPol
I:NearPerspec-ves:thePlancksatellite
• Smallerbeam• Lowernoise• Polariza-on• BeZerfrequencycoverage
COBE WMAP PlanckYear data received 1992 2003 2009Spatial resolution (deg) 7 0.23 0.08Frequencies (GHz) 30-90 22-94 30–857Polarization no yes yesSensitivity (muK/30' pix) 10.5 (8yrs) 1.4
WherePlanckwilldobeZer
• New information coming from: – second and third peak in TT spectrum – greater l range – reionization “bump” in EE and second peak
(L.ColomboE.PierpaoliJ.Pritchard2008)
Comparisonofparameteres-ma-onperformances(WMAPorPlanckonly)
(ColomboPierpaoliPritchard08)
Wheredoestheinforma-oncomesfrom?
• Temperature: 217 GHz and 143GHz playing similar roles at high resolution
• Polarization: CV limited to l= 8-12, possibly sensitive to 3rd peak
Impactonparameters
• Major information coming from 70-143 GHz. Adding 217GHz iproves parameter determination by ~10%.
• Giving up 143Ghz,pol results in a 73% (or factor 2) degradation for r estimate. • LFI useful for improving r estimates.
Frequencycoverageanddegeneracies
• Frequencycoveragehelpssomedegeneracies,notall….
Impactofbeammodeling
• Assumingawrongbeambytheamountconsistentwithcurrentuncertain-esmaybiasimportantparametersby~sigma
Under‐es-ma-ngorover‐es-ma-ngthebeamby0.05%( )or0.2%()
Tensortoscalarra-orfromPlanck
Forr>0.05,halfoftheinforma-oncomesfromtheBBspectrum
Issueswiththetensorspectralindex
• Forr~0.1,Planckwillnotbeabletoconstrainrandn_tsimultaneosly
• EvenforanEPICtypeofexperiment,marginalizingonn_tincreasestheerrorsonrbyafactor3.
Synergieswithothersurveys
Planckonly
BAO~JDEM(z=0.5‐2,10000sqdeg)
(CIP:z=2–6.5,1000sqdeg,2Mgal)
Thelow‐lparameters:Thetensor‐to‐scalarra-orandreioniza-onτ
• Reioniza-onimpactstheCMBtempertatureandpolariza-onspectra.• TheOp-caldepthistheonlyparametercurrentlyconstrainedanditis
determinedbyEEspectra.
Constraints(WMAP3):Alldata:τ = 0.09 +/‐0.03EEonly:τ = 0.1 +/‐0.03(Pageetal2006)(WMAP5:0.087+/‐0.017)
II.REIONIZATION
Lookingmorecarefully….
• Differencesatl<30onEEandBBspectraare~10%‐20%
t=0.0845
(seee.g.HolderHaiman,Kaplinghat,Knox03,Hu&Holder03Colomboetal05)
Ques-ons
• Currentanalysismodelsassume“sharp”reioniza-on(possiblyincorrect).Howmuchthisassump-onaffectsPlanckparameteres-ma-on?
• Canyouavoidmakingthisassump-on?Atwhichcost?Areotherparametersaffectedbythechoice?
• Canweconstrainsomethingmorethanjustτ?• Ifyoudogiveupthatassump-on,whichisthebeststrategytoincorporateamoregeneralreioniza-onhistory?
Colombo&Pierpaoli08
• GetCl'sformodelswithdifferentxe(z)butsame t andothercosmologicalparameters• BuildsimplePlancksimulateddatafor143GHzwithfsky=80%skycoverage
• MCMCanalysisofsimulateddataassumingasharpreioniza-on
Howbigistheproblem?
fiducialmodels:sharp,extended,two‐step,doublewitht=.0845
t=.0845modelmean sbiasSR .0844.0045 ‐.02ER .0850.0044 .112R .0895.0046 1.09DR .0861.0046 .35log[1010As]=3.135modelmeansbiasSR 3.1349.0092 ‐.01ER 3.1356.0092 .072R 3.1452.0094 1.09DR 3.1382.0093 .34r=0.1modelmeansbiasSR .104.029 .14LR .096.029 ‐.142R .088.028 ‐.43DR .092 .030 ‐.26
“Agnos-c”approachestoreioniza-on• Binnedreioniza-on(Lewis,Weller,BaZye06)
• PrincipalComponentofreioniza-onhistory(Mortonson&Hu07)
– Assumeafiducialbinnedreioniza-onhistoryxe
fid(zi),i=1,2,...,Nzzmin≤zi≤zmax– EvaluatetheFisherMatrixfortheEEmodes
– TheprincipalcomponentsSn(zi)aretheeigenvectorsoftheFishermatrix
– Agenericreioniza-onhistoriycanbespecifiedonthisbasisbyasetofcoefficientsµnas:
– Evaluateasubsetofthesefromthedata
xe(z) = xfid + ∑ µnSn(z)
Xefid=1,z<60.15,6<z<30
Reconstruc-ngreioniza-on• Underlyingmodel:sharpreioniza-on(i.e.:wewanttobenasty)• Analyzedwith6binsbetweenz=6andz=30(Δz=4)
Black:binnedreioniza-onRed:5Eigenmodes
• Samebiasinrobservedinthetwomethods(otherparametersarerecoveredequallywell)
• Binnedspectralesswellreconstructed(alotofcorrela-onbetweenbins)
• Computa-onally:4-mesmoreexpensive
• BinnedmodelingmaybebeZersuitedtointroduceexternalconstraintsonreioniza-on
Colombo&Pierpaoli08
EPIC,PlanckandWMAPreioniza-on
WMAP5PlanckEPIC
Zaldarriagaetal2008
PlanckwillgreatlyimproveonWMAPonhighredshi{reioniza-on.
“Experimental”Heliumcontribu-ontothemeasurementofreioniza-onhistory
BeingabletousemorecoolingheliumwouldresultinbeZerconstraintsonthereioniza-onhistoryathighredshi{s.
Howwellwillafutureexperimentperform?
• Prac-callyCVlimitedfordeterminingtheintegratedreioniza-onsignal• Notverygoodindeterminingtheextentofreioniza-on• WillhelpinrelievingsomemilddegeneraciesPlanckmayhave.
Zaldarriagaetal08
III:Whatwewantandwhatweget:foregroundsissues
=+
CMB
Foregrounds
ObservaPons(realWMAPmaps!)
Galaxyclusters
GalacPcdust
Synchrotronemission
Thesearchforr
• B‐modepolariza-onwouldbeasmokinggunforinfla-onbutpolarizedforegroundsmaybealimi-ngfactor.
Pageetal07• Howmuchdoweknowaboutpolarizedforegrounds
• Howwellcanwehandlethem
• Ul-mately,howlowofanrcanwemeasure,givenourcurrentknowledgeofforegrounds
Characterizingdiffuseforegrounds:Cross‐correla-onmethod
• Idea:assessforegroundcontribu-ontoWMAP5mapandtheirproper-esviacross‐correla-onswithdifferenttemplates(seeDaviesetal06)
• Objec-ves:performananalysisatdifferentla-tudes,alsoincludingpolariza-on
• Advantages:non‐parametric• Disadvantages:notlocalized,dependentontherepresenta-onoftheforegroundbythetemplate.
Syhchrotron:HaslamDust:FDS8Free‐free:Hα(Dickinsonetal03)
Macellarietal,inprep.
Templates:
Intensityresults
• Evidenceforanomalousdustemissionplusthermaldust,Free‐freeandsynchrotronspectralindexwithnosurprises.
CfrDaviesetal06
Intensityresultsbyla-tude
• Dustcorrelatedcomponentdominantonsynchrotronatlowandintermediatela-tudes
Polariza-onemissivity
• Synchrotrondominantatlowfrequencies,dustat94GHz.
• Totaldustfrac-onalpolariza-onbelow5%.• Trendinla-tudeandfrequency.
Synchrotronspectralindex
• Consistentresultsintemperatureandpolariza-on,flaZeningathigherla-tudes
• Synchrotronfrac-onalpolariza-onsmallonthegalac-cplane,reaching40‐50%atintermediate/highgalac-cla-tudes
IV:Impactofforegroundsonrdetermina-on
• Methodforcomponentsepara-on:SMICA(Cardosoetal2008)
• Maps:implemen-ngBforthethePlanckSkyModel• Advantages:“bestpowerspectrumreconstruc-on”• Disadvantages:noforegroundreconstruc-on.• Foregroundconsidered:
– Synchrotron(withrunning)– Thermaldust(differentamplitedes/powerspectra)– (extra‐galac-csources)– (Lensing)
X=vectorinfrequencyofobserva-ons:
Thelikelihood
• Insteadoftheusual:Thelikelihoodisgivenby:
WhereKisthesprectralmismatchbetweenthematrices:
Wemaximizethelikelihoodwithrespectto
PreviousaZempts:
• Verdeetal06:Fishermatrixapproachwithinstrumentalnoiseand~10%foregroundresiduals
• Amblardetal07:Wienerfilteredmapsandes-mateoferrorsincludingforegroundresiduals(needtoknowcovariancesofyourforegrounds!)
Foregroundlevelsandmodeling
ForegroundmodeledoutofthePSM
Forecastsforfutureexperiments
VariantstothemainmodelPointsources(EPIC‐CS)
Lensing(r=0.001)
Varia-onsinthesynchrotronspectralindex
Varia-onsinthedust
Conclusions• WMAPnaïveparameteres-matesabout30%lowerthanrealones.
Planckmayimprovebyseveralfactors,especiallyonr(with1/2oftheinforma-oncomingfromtheBBpowerspectrum).Es-matesconsideringforegroundsuggestthatatr~0.1foregroundsshouldnotdegradethesignificanceofthedetec-on.Planckwillcomplementfuturesurveysin…….
• Allowingforageneralreioniza-onmodelwithinPlanckparameteres-ma-onisnecessarynottobiasrandpowerspectrumamplitude.Theimplementa-onisprac-callydoable.Bothprincipalcomponentorredshi{binsmaybeused.FutureCMBpolariza-onmissionsallperformwellNogoodreconstruc-onofreioniza-onhistoryistobeexpected,apartfromdiscrimina-ngbroadlybetweenearlyandlatecontribu-ontotheop-caldepth.
• WMAP5Cross‐correla-onanalysis:dustpolariza-onbelow6%,dependentonfrequencyandla-tude.Synchrotronpolariza-onincreasingwithla-tudeupto40‐50%.Synchrotronspectralindexconsistentintemperatureandpolariza-on,flaZeningathigherla-tudes.
• AsforfutureCMBpolariza-onmissions,foregroundshouldnotlimitdetec-onifr~0.001,withactualperformancesdependingonconfigura-on.Varia-onsinthediffusecomponentsisirrelevantfortheresults.Pointsources,ifneglected,biastheresults.