ebook img

Constraints on Primordial non-Gaussianity from Future HI Intensity Mapping Experiments PDF

0.58 MB·
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Constraints on Primordial non-Gaussianity from Future HI Intensity Mapping Experiments

Constraints onPrimordialnon-Gaussianity from Future HI Intensity Mapping Experiments YI-CHAO LI∗ and YIN-ZHE MA† School of Chemistry and Physics, University of KwaZulu-Natal, Westville Campus, Private Bag X54001, Durban, 4000, South Africa and NAOC-UKZN Computational Astrophysics Centre (NUCAC), University of KwaZulu-Natal, Durban, 4000, South Africa The primordial non-Gaussianity induces scale-dependent bias of the HI with respect to the underlying dark matter, which exhibits features on the very large scale of 21-cm power spectrum potentially observ- able with HI intensity mapping observations. We forecast the prospective constraints on the four funda- mental shapes of primordial non-Gaussianity (local, equilateral, orthogonal and enfolded), with the cur- 7 rent and future HI intensity mapping experiments, BINGO, FAST and SKA-I. With the current config- 1 uration of the experiments and assumed one-year observation time, we find that the SKA-I will pro- 0 vide tighter constraints on local shape of primoridal non-Gaussianity than Planck. The results are 2 n (σfNloLcal,σfNeqLuil,σfNorLth,σfNenLfold)SKA−I = (0.54,86,25,43), (σfNloLcal,σfNeqLuil,σfNorLth,σfNenLfold)BINGO = a (17,100,128,164), (σfNloLcal,σfNeqLuil,σfNorLth,σfNenLfold)FAST = (9.5,44,75,94). If thelower frequency band J ofFASTcanbeused, theconstraintonlocal-typeprimordialnon-GaussianitywillbeσfNL ∼ 1.62whichis betterthanPlanck. Inaddition,iftheobservationtimeforFASTcouldbeextendedtotwoyears,theconstraint 1 on equilateral shape of primordial non-Gaussianity will be improved to be σ ∼ 32. Similarly, if obser- fNL vationaltimeofSKA-Icouldbeextendedtotwoyears, theconstraintonlocalandorthogonal shapescanbe ] O improvedto0.43and20respectively,achievingbetterconstraintsthanPlanck. C . h I. INTRODUCTION from SDSS survey and obtain the limit for local type PNG p as 31( 96) < flocal < +70(+96) at 95% (99.7%) CL, - wh−ichwa−scomparaNblLetothethenmeasurementsfromWilkin- o Thestatisticalpropertiesoftheprimordialfluctuationoffer son Microwave Anisotropy Probe (WMAP) five-year results. r richinsightsintothephysicsofinflationandtheearlyuniverse t Ref. [19] used radio sourcesfromthe NRAO VLA Sky Sur- s [1]. One of the wildly discussed questionsis whetheror not a vey(NVSS),thequasarandMegaZ-LRG(DR7)cataloguesof theprimordialfluctuationsaredeviatedfromGaussiandistri- [ theSDSS,andthefinalSDSSIILuminousRedGalaxy(LRG) bution. The simple single-field slow-roll inflationary model photometricredshiftsurveyandfoundflocal = 48 20(1σ v1 tpiroendic[t2s]p.riHmoowrdeivaelrfl,umcatunaytiaolntewrnitahtivaelmmosotdGelasuossfiasinndgilset-rfiibeuld- CL). Ref. [20] found fNloLcal = 90 ± 30NaLt 1σ CL±by using 1 photometricSDSS data, but due to unaccountedsystematics slow-roll inflation can produce different types of primordial 2 thisresultmaybebetterinterpretedasflocal < 120at84per 2 non-Gaussianity [3–10] (hereafter PNG), which leaves dis- centCL.Ref.[21]usedSDSS-IIIBaryonNOLscillationSpectro- 0 tinctivefeaturesinstatisticalpropertiesofcosmicmicrowave scopicSurvey(BOSS)datatoconstraintheflocal,andfound 0 background(CMB)andthelarge-scalestructure(LSS)ofthe NL 45 < flocal < 195 at 2σ CL. In addition, Ref. [22] used 1. Universe. −thecorrelaNtLionoftheresidualpeculiarvelocitiesondifferent 0 IftheprimordialfluctuationisGaussian,thetwo-pointcor- directionstoconstrainPNGandfound flocal <25.7at68% 7 relation function (i.e. power spectrum in Fourier space) can CL.Theselimitsarecurrentlyconsisten|twNLith|butweakerthan 1 describe all of the statistical properties of PNG. Therefore, the measurementsfromthe PlanckCMB observation. How- : v themoststraightforwardwaytomeasurethePNGisthrough ever, forecasts indicate that the constraint errors could de- Xi the higher-order correlation of CMB or LSS. Current mea- creaseoneortwoordersofthemagnitudewiththefutureLSS surementsof the temperatureand polarizationof CMB from survey,especiallyforthefutureradiosurvey.(see[23]andits r a Plancksatelliteprovidesthestate-of-the-artconstraintsonlo- referencesforreview). cal,equilateralandorthogonaltypesofPNG[11]asfNloLcal = The scale-dependent bias not only affects the large-scale 0.8 5.0,fequil = 4 43andfortho = 26 21at68% galaxy bias, but also affects the HI distribution. A more ± NL − ± NL − ± confidencelevel(CL). efficient method of the radio survey is to map out a large volume of the Universe through the intensity mapping tech- Besides the constraints from CMB, there have been a lot ofeffortsofmeasuringf throughlarge-scalestructuresur- nique,whichmeasuresthecombinedHIemissionoftheunre- NL solvedgalaxies.Therefore,inprincipleonecanobtainathree- veys.ThisisbecausethePNGinducesascale-dependentbias ofthegalaxywithrespecttotheunderlyingdarkmatterdistri- dimensional HI distribution that can provide more modes of fluctuations than the CMB two-dimensional sphere. There butiontracer [12–18]. Ref.[15]usedspectroscopicandpho- havebeenseveralworkstoforecastthedetectabilityofPNG tometric luminous red galaxy samples, and quasars samples through HI intensity mapping technique [24–26], but those forecastsareexclusivelyonlyforlocalandequilateraltypeof PNG and limited experimental cases (SKA and Tianlai). In ∗Electronicaddress:[email protected] this work, we will calculate the scale-dependent bias of all †Electronicaddress:[email protected] fourtypicaltypesof PNG by using the halo model, and cal- 2 culatetheirimprintsonthepowerspectrumof HI . Thenwe A. Localshape forecastthedetectabilityofallofthethreeon-goingHIimag- ing surveys, i.e. BAO as Integratd Neutral Gas Observation The standard single-field inflation predicts the local-type (BINGO)[27],Five-Hundred-MetreApericalSphericalTele- PNG [37]. But the parameter flocal is expected to be very NL scope (FAST) [28, 29] and Square Kilometre Array Phase-I closetozero[2]. However,largeamountofnon-Gaussianity (SKA-I)[30]. can be produced with different inflationary model, such as Thispaperisorganizedasfollows. InSect.IIwesumma- multifieldsmodel[3]orcurvatonmodels[4]. Inthesecases, rize the primordialbispectrumand discuss differenttypesof flocalcanbesubstantiallydifferentfromzero. NL PNG to be forecasted in this work. In Sect. III we calculate Thepotentialbispectrumofthelocal-typePNGhasthesim- thescale-dependentbiasoftheLSSinducedbythePNG,and pleform, thenthepowerspectrumofHI. InSect.IVweintroducethe Fisher matrix forecast method that used in our analysis. In B (k ,k ,k )=2flocal[P (k )P (k )+(cyc.)], (3) φ 1 2 3 NL φ 1 φ 2 Sect.V,the detailedexperimentparametersarediscussed. In Sect.VIwepresentourresultsandsomediscussion. Conclu- inwhich,Pφ(k)=2π2As(k0)(k/k0)ns−4isthepowerspec- sionwillbeinthelastsection. trumoftheGaussianBardeenpotential. BesidesthePNGparameters,wewilladoptaspatiallyflat UniversewithcosmologicalparametersfixedasPlanck2015 B. Equilateralshape best-fittingvalues[31],i.e. Ω = 0.309;Ω = 0.691;σ = m Λ 8 0.809; and h = 0.68, where the Hubble constant is H = 0 100hkms−1Mpc−1. Theamplitudeandtiltofscalarpower Theequilateral-typeofPNGcanbeproducedbytheinfla- spectrumareA (k )=2.141 10−9andn =0.961,where tionary models with higher-derivative interactions. Usually s 0 s pivotscaleisk =0.002Mpc×−1. therearetwodominantinteractiontermsoftheinflationfield 0 givingrisetothePNG.Oneofthedominanttermsleadstothe equilateral-typeofPNG,andanotherleadstotheorthogonal- type. II. PRIMORDIALBISPECTRUM Theprimordialbispectrumoftheequilateraltypetakesthe form[6], Mostofthesingle-fieldinflationarymodelspredictthepri- B (k ,k ,k )=6fequilγ(k ,k ,k ) mordialcurvaturefluctuationswiththedeviationfromGaus- φ 1 2 3 NL 1 2 3 sian distribution [32–34]. The deviation is particularly de- P (k )P (k )+(cyc.) φ 1 φ 2 scribed by writing the gauge-invariantBardeen’s potential φ ×h−(cid:16) (cid:17) 2/3 asthesumofaGaussianrandomfieldandaquadraticcorre- 2 P (k )P (k )P (k ) φ 1 φ 2 φ 3 lation[33,35], − (cid:16) (cid:17) + P1/3(k )P2/3(k )P (k )+(cyc.) , φ 1 φ 2 φ 3 φ=φG+fNL(φ2G−hφ2Gi), (1) (cid:16) (cid:17)i(4) inwhich,fNL isadimensionless,phenomenologicalparame- inwhich,functionγ(k1,k2,k3)takesintoaccountoftherun- terdescribingthemagnitudeofthePNG. ningoffequilandreads [38], NL To extract more information of the non-Gaussian primor- dialfluctuations,weneedtogobeyondthestatisticsofpower γ(k ,k ,k )= k1+k2+k3 −2κ, (5) 1 2 3 spectrum. The lowest-order statistics sensitive to the PNG (cid:20) kCMB (cid:21) is the three-point function or bispectrum B (k ,k ,k ), in which,φistheprimordialBardeenpotentialwφhic1hi2sdi3rectly wherekCMB =0.086hMpc−1andκ= 0.2. − relatedtothecurvatureperturbation[36].Thepotentialofpri- mordialcurvatureperturbationisrelatedtotheNewtonianpo- C. Orthogonalshape tentialduringthematterdominationviathetransferfunction T(k)which satisfies T(k 0) = 1. By applyingthe Pois- sonequation,φisrelatedto→thematterdensityfieldδ (k)by TheorthogonalshapeofPNGbispectrumisnearlyorthog- m δ (k)= (k)φ(k),where, onaltoboththelocalandequailateralformsofPNG,whichis m M wellapproximatedbythefollowingtemplate[39] 2k2T(k) (k)= . (2) B (k ,k ,k )=6forth 3 P (k )P (k )+(cyc.) M 3 ΩmH02 φ 1 2 3 NL h− (cid:16) φ 1 φ 2 (cid:17) 2/3 8 P (k )P (k )P (k ) The configuration shape of Bφ(k1,k2,k3) is related to the − (cid:16) φ 1 φ 2 φ 3 (cid:17) physicalmechanismsduringtheinflation. Inouranalysis,we +3 P1/3(k )P2/3(k )P (k )+(cyc.) , consider four classes of bispectrum shape characterizing the (cid:16) φ 1 φ 2 φ 3 (cid:17)i (6) local,equilateral,enfoldedandorthogonaltypesofPNG. 3 D. Enfoldedshape It is well studied that if the inital vacuum state for the in- flationdeviatesfromthestandardBunch-Daviesvacuum,the resultingbispectrumtakestheenfolded-shape [7–10],which canbeapproximateby B (k ,k ,k )=6fenfold P (k )P (k )+(cyc.) φ 1 2 3 NL φ 1 φ 2 h(cid:16) (cid:17) 2/3 +3 P (k )P (k )P (k ) φ 1 φ 2 φ 3 (cid:16) (cid:17) P1/3(k )P2/3(k )P (k )+(cyc.) . −(cid:16) φ 1 φ 2 φ 3 (cid:17)i (7) FIG.1: ThreemodelsofLagrangianbiasbL(z),i.e.Matarreseand Verde[13],MoandWhite[42],andMoandWhite[43]. III. HIBIASANDPOWERSPECTRAOF21-CM TheHIbiasisthebiasofHIdistributionwithrespecttothe otherdarkmattertracers,itcanbenegative. Thetracersanti- underlyingdarkmatterdistributionandthe HI biasfunction, correlated with the initial dark matter fields will lead to the bHI,canbeobtainedbyassumingamodelfortheamountof less clustered distribution than the dark matter field at later HImassinadarkmatterhaloofmassM,MHI(M),andinte- time. gratingoverthehalomassfunctiondn/dM.Hereweusethe Itthepastthirtyyears,peoplehavebeendevelopingdiffer- Sheth-Tormenhalomassfunction[40]withmassrange[108, ent analytical, semi-analytical and parametric models of the 1016]M⊙ bias function. In below, we list the three most typical and commonly-usedones. 1 Mmax dn Basingon thePress& Schechter(hereafterPS) halomass b (z)= dM (M,z)M (M)b(M,z), HI ρHI(z)ZMmin dM HI function [44] and its extensions, MoandWhite (1996) [42] (8) givesthebiasfactorforhalosofmassM, inwhich,b(M,z)isthereal-spacehalobiasandρ (z)is, HI 1 Mmax dn bL(M,z)= δ ν2(M,z)−1 , (11) ρHI(z)=Z dMdM(M,z)MHI(M). (9) c (cid:2) (cid:3) Mmin whereν(M,z) = δ (z)/σ . δ (z) = δ /D(z),whereD(z) c R c c Forthe HI intensitymappingexperiments,we followthe as- is the linear growth function which we use eq. (10) in [43] sumptiondiscussed in [41] andconsidera simple powerlaw to computeit. δ 1.686is the critical density contrastfor c ≃ modelfortheamountofHImass, sphericalcollapse. Withtheapproximationofhigh-peak,the abovebiasfactorcanbeexpressedas b (M,z) = δ (z)/σ2 MHI(M)=AMα, α≃0.6, (10) (MatarreseandVerde 2008 [13]). WithL the ellipsoicdal coRl- lapsemodel[45],MoandWhite(2002)[43]givesanotherex- which is a redshift independent function. The pre-factor A pression, willbecanceledwiththenormalizationofρ . HI 1 b (M,z) = ν′2+bν′2(1−c) L δ (z) A. TheLagrangianbias c h ν′2c/√a , (12) TheLagrangianbiasdescribesthestatisticalbiasofthehalo − ν′2c+b(1 c)(1 c/2)(cid:21) − − distributiontotheprimordialdarkmatterfields. ThePNGaf- inwhich,ν′ =√aν anda=0.707, b=0.5, c=0.6. fectstheinitialconditionsoftheprimordialdensityfields,soit Figure1showsthethreemodelsofLagrangianbiaswedis- ismoreconvenienttostudysucheffectsinLagrangianspace. cussedabove. Ontheotherhand,itisalsonecessarilytostudythestatistics of the evolved halo field at low redshifts in Eulerian space, whichisconvenientlyrelatedtotheobservation. Thebiasin Lagrangianspace, b , relates to the Eulerian space bias, b , B. Thescale-dependentbias L E via b = b +1 [42]. The extra unity factor of b reflects E L E themotionsofprimordialpeaksatlatertimes[22]. Theuni- Asweanalyzedbefore,PNGaffectsthedistributionofthe formly distributed halos in the initial epoch, which have the peaksattheinitialstageofmatterfluctuations,thereforeitis b = 0, will lead to unbiased distributionto the darkmatter correlatedwiththeLagrangianbias. InthepresenceofPNG, L fieldatlatertime.Theb forhalosispositivedefined.Butfor the halo bias can be written as the combination of a usual L 4 FIG.2: Theabsolutevalue of scale-dependent bias|∆b(z,k)|(Eq.(15))for differentPNGshapes atz = 0(leftpanel) andz = 2(right panel)withassumedfNL =1.ThefourshapesofPNGareshownindifferentcolouranddashedlineslistedinthelegend. Thereasontoplot absolutevalueisbecauseorthogonalshapeof∆bisnegative(seealsofig.1in[46]).TheapproximationoflocalshapeofPNGbyDalaletal. [12](Eq.(16))isshowninthebrowndashedline,whichisconsistentandalmostcompletelyoverlappedwiththecomputationfromthehalo model(Eqs.(3)and(17))inredsolidline. scale-invariant bias, b(M,z), and a scale-dependent modifi- in which, δ (z) = δ /D(z) 1 and (k) is the Eq. (2) c c R M cation,∆b(M,z,k), smoothedwithwindowfunctionW (k), R bNG(M,z,k)=b(M,z)+∆b(M,z,k). (13) 2T(k)k2 (k)= W (k), (18) MR 3 H2Ω R BysubstitutingEq.(13)intoEq.(8),wecanobtainthescale- 0 m dependentHIbias,whichcanbeexpressedas, where R denotes a smoothing radius which defines the halo bNG(z,k)=b (z)+∆b (z,k), (14) massM by, HI HI HI in which bNHIG(z,k) is the total bias, bHI(z) is the scale- M = 3H02Ωm4πR3. (19) independent term, and ∆b (z,k) is the scale-dependent 8πG 3 HI term, which is obtained by integrating ∆b(M,z,k) over the So∆bMV isalsoafunctionofhalomass,M. (k)isrelated halomassfunctionandHImassmodel, F to thebispectrumof primordialpotentialfield B (k ,k ,k), φ 1 2 1 Mmax andthepowerspectrumPφ(k), ∆b (z,k)= dM HI ρHI(z)ZMmin (15) 1 dn (M,z)M (M)∆b(M,z,k), F(k)= 16π2σ2 Z dk1k12MR(k1) × dM HI R (20) 1 B (k ,k ,k) φ 1 2 whereρHI(z)iscalculatedinEq.(9). ×Z dµMR(k2) P (k) , Dalal et al. [12] firstly derived the expression of scale- −1 φ dependent correction to the bias of galaxies and halos for wherek2 =k2+k2+2kk µandσ isthermsoftheunder- 2 1 1 R local-shapebispectrum, lyingdarkmatterfluctuationfieldssmoothedonscaleRgiven inEq.(19). 3Ω ∆bD(z,k)=2(b 1)f δ m , (16) If we substitute the local-shape bispectrum into Eq. (20), E− NL c2a(z)g(z)r2 k2 H and take the limit of k 0, then2 the dependence of in which, δ is the critical density, a(z)g(z) = D(z) is the ∆bMV(M,z,k)onhaloma→ssautomaticallydropsoff c lineargrowthfactorandr =1/H . Equation(16)isderived H 0 (k 0) 1 by only considering the high peaks of the density contrast, F → → whichmeansthattheexpressiononlyworksatthelargescales T(k 0) 1 → → withk 0. (k 0) (2/3)k2/(H2Ω ), More→accurate analytical expressions for the scale- MR → → 0 m dependentbiashavebeenstudied[13–18]. Awildlyusedex- pressionisderivedbyMatarreseandVerde[13], δ2(z) (k) 1Thisisconsistentwitheq.13in[13].The“∆c”definedin[13]isequalto ∆bMV(M,z,k)=2f c F , (17) δcinthispaper. NL(cid:18) σR2 (cid:19)MR(k) 2InRef.[13]bE−1=bL=δc/σR2 5 and, δ 3H2Ω Redshift Window(i) = 3.06 ∆bMV(z,k→0)→2(bE −1)fNLa(z)gc(z)2 0k2m 2K] 102 3.0 , µ =∆bD(z,k) )[ 101 π 2.7 2 k−2, /( ∼ ijCℓ 100 2.4 (21) ) i.e. thegeneralexpressionofscale-dependentbiasinEq.(17) ℓ(ℓ+1 10-1 2.1ow(j) roefcuosvienrgsEthqe.b(1ia7s)pisrotphoasteidtcinanDbaelaulseetdatlo.[c1a2l]c.uTlahteeaadnvyasnhtaapgee 10-2 1.8Wind ofPNG,providedthatthebispectrumBφ functionisgiven. 1.0 1.5hift The scale-dependent bias for equilateral, orthogonal and s d enfolded shapes of PNG can be obtained by substituting 0.5 1.2Re Eqs.(4),(6)and(7)intoEq.(20).InFig.2,weshowtheabso- iiCℓ lutevalueofthescale-dependentpartofthebias,i.e. Eq.(15) ij/Cℓ 0.0 0.9 for the four shapes of PNG at z = 0 (left panel) and z = 2 −0.5 0.6 (rightpanel).Onecanseethatthelocalshapehasmostpromi- nent feasture of scale-dependent bias at large scales, which −1.0 canbeconstrainedwith21-cmintensitymappingobservation 101 102 onlargeangularscales. Theorthogonalandenfoldedshapes ℓ haveless prominentfeaturesbutpossiblydetectableatsmall k. The scale-dependentbias induced by equilateral shape is FIG.3: Upperpanel: Cross-correlatedangularpowerspectrumbe- toosmallonlargescalessoitwillbehardtobedetected.The tween redshift zi = 3.06 and zj, which ranges from0.37 to 3.06 resultsshowninFig.2areconsistentwiththeanalysisin[18] shownwithdifferentcolors.Lowerpanel:Theradiooftomographic andfig.1in[46]. angularcross-powerspectrumbetweenzi andzj totheauto-power Wecanseetheasymptoticbehaviorofscale-dependentbias spectrumofzi. (Eq.(15)) onlargescalesbytakingthelimitofk 0, then → ∆b ( / ). Therefore, R → F M ∆b(Local)∼k−2 1w0h−e3re[4Ω8H].I TishtehweifnradcotwionfuanlcHtiIodnensℓi(tyk)asiss,umedtobe0.62× W ∆b(Equilateral) const ∼ (22) ∆b(Enfolded) k−1 (k)= dχdNg(χ)j (kχ)bNG(χ(z),k)T (χ,k), (25) ∼ Wℓ Z dχ ℓ HI δ ∆b(Orthogonal) k−1. ∼ wherej isasphericalBesselfunction,dN (χ)/dχisthered- These asymptotic behaviors of ∆b is consistent with the ℓ g shift distribution of galaxy number, T (χ,k) is the transfer computationofhalomodelsinFig.2. δ function for the galaxy number over-density, and bNG is the HI total bias of HI (Eq. (14)). To calculate the angular power C. Powerspectrum spectrum,weusetheCAMB SOURCESpackage[49]. Figure 3 shows the tomographic angular power spectrum. The Upper panel shows the cross-power spectrum between WeemploytheHItomographicangularpowerspectrumas redshift z = 3.06 and z , which ranges from 0.37 to 3.06 theobservableinouranalysis, Theexpressionoftheangular i j shownwithdifferentcolors. TheLowerpanelshowstheratio powerspectrumofthei-thandthej-thredshiftbinsis, of the cross-power spectrum of different redshift bins to the auto-powerspectrumofthesameredshiftbin.Wecanseethat Cij =4πTij d lnk i(k) j(k)∆2(k), (23) ℓ b Z Wℓ Wℓ ζ the cross-power spectrum decreases as the redshift deviates from z = 3.06. This is what we expected, since the cross- inwhich,∆2(k)isthedimensionlesspowerspectrumofpri- i ζ correlatedsignalshoulddropifthefrequencywindowsmove mordial curvature perturbation and Tij = T (z )T (z ) is awayfromeachother. b b i b j the multiplication of HI mean brightness temperature of the i-thandj-thredshiftbins. WeusetheexpressionofT (z)in b Changetal.(2008)[47], IV. FISHERMATRIXFORECAST Ω 1+z 0.5 Tb(z) = 0.39(cid:18)10H−I3(cid:19)(cid:18) 2.5 (cid:19) ToforecastthepotentialforconstrainingfNL,weperform theFishermatrixanalysis. Ifweassumethatthemodellike- Ω +(1+z)−3Ω −0.5 lihood surface in parameter space can be well approximated m Λ mK, (24) × (cid:18) 0.29 (cid:19) by a multivariant Gaussian, the Fisher matrix F is then a 6 good approximationfor the inverse of the parameter covari- a. BINGO The BINGO experiment is a single-dish ance. In the 21-cm tomography, each frequency band will HI intensity mapping experiment, which aims at map- providea map of 21-cm intensities, so we need to sum over ping the HI emission at frequencies between 960MHz and theFishermatrixinbothℓ-spaceandfrequencyspace. Since 1260MHz [27, 53]. The telescope of BINGO experiment ν = 1420MHz/(1 + z), each frequency corresponds to a hasnomovingpartsanditconductsadrift-scanstrategy. To uniqueredshiftslice. TheFishermatrixis, achieveenoughsurveyarea,awideinstantaneousfieldofview (FOV)withmultiplefeedsisrequired.Atotalof60feedslaid ℓmax 2ℓ+1 outinarectangleof16m 15matthefocalplane. Thiswill Fαβ =fsky (cid:18) 2 (cid:19)tr[Cℓ,αΣℓCℓ,βΣℓ], (26) form a FOV of about 10×◦(in Declination direction) 9◦(in ℓXmin Right Ascension direction). With the 10◦ wide stri×p cen- inwhich,C isann n matrix,inwhicheachelementisthe tering at Declination of 45◦, the total survey area is about ℓ z× z 2500deg2. − HI crossangularpowerspectrumbetweenthetwofrequency bins. Σ = (C +N )−1 isthetotalnoiseinversematrix,in b. FAST FAST is the largest single-dish telescope, ℓ ℓ ℓ whichN isthen n experimentalnoisepowerspectrum. which also has the multi-beam system of 19 feed-horns ar- ℓ z z × ray [28, 29]. The multi-beamsystem is proposedto work at Herewemakeasimpleassumptionthatthenoisesindifferent frequency(redshift)binsareuncorrelated,thereforetheN is frequenciesfrom1.05GHzto1.45GHzwithsystemtempera- ℓ adiagonalmatrix. Inreality,21-cmintensitymapsarehighly tureof25K. Inouranalysis,weonlyincludethefrequencies contaminatedbytheforeground,suchasGalacticsynchrotron up to 1.35GHz. With the 300m illuminated aperture, each ofthefeed-hornhasthebeamsize(FullWidthatHalfMaxi- emission, extragalactic point sources, and atmospheric sig- nal. One needs to apply foreground removale technique to mum)of2.9′,andforma26′FOVwith19beams. Duetothe longslewing time, FAST can only workon drift-scanobser- reducetheforegroundcontamination[50–52].However,there vationmode. SimilartotheBINGOexperiment,FASTscans always be some level of residual Galactic foreground after applying such techniques to the maps. Therefore the cross- a26′ widestripalongtheRightAscensiondirectionforeach sidereal day. But the zenith angle of FAST can be adjusted correlationof noisesbetweendifferentfrequencybandsmay notcompletelybezero. fromDec: 14◦12′ to Dec:65◦48′. Withoutoverlappingbe- Underoursimplifiedassumption,theelementofN matrix tweenscan−ningstrips,ittakesabouthalfyeartocoverall80◦ ℓ is Declination range . With one-year observation (3.15 107 second),themaximumsurveyareaisabout24000deg2×. Nij = δijNHI c. SKA-I The SKA Phase I (SKA-I) plans to construct ℓ ℓ = δijT2 S /(N N t ∆ν). (27) 190movable15mdishes[26]. Themaximumsurveyareais sys survey ant feed TOT about 25000deg2. A efficient survey area is need to be ex- T = T + T is the system temperature, which plored to minimal the constraint errors. In our analysis, we sys rec sky is contributed from the sky temperature, T = 60 onlyconsidertheautocorrelationofeachdishes,whichmeans sky (300MHz/ν)2.55, andreceivertemperatureT foreachex×- that the SKA-I works as 190 single dishes. Without the in- rec periment. N andN arethenumberofantennaandthe terferometry, the SKA-I has very low resolution and is only ant feed number of feed horn in each antenna respectively. The de- sensitivetothelow-ℓmodes. tailedexperimentalparametersforFAST,SLA-IandBINGO Figure 4 shows the noise power spectra of different ex- arelistinTableI. periments in at redshift bin z = 0.37 (left upper panel) and z = 3.06 (right upper panel). The black solid line in the upperpanelofeachfigureshowsthestandardangularpower V. EXPERIMENTPARAMETERS spectrum of 21-cm (f = 0); The black dash-dotted, dot- NL tedanddashedlinesshowthenoisepowerspectraofSKA-I, FAST and BINGO experiments. One-year observation time and 2500deg2 survey area are assumed for all the experi- TABLEI: TheexperimentparametersforFAST,SKA-IandBINGO. ments.ThepartialderivativesofCiiwithrespecttoparameter D istheilluminatedaperture. ℓ dish f areshowninthelowerpanel. Thedifferentcolorscorre- NL FAST SKA-I BINGO spondtodifferenttypesofPNG. νmin[MHz] 1050 350 960 Comparing to the BINBO experiment, FAST and SKA-I νmax[MHz] 1350 1050 1260 canhaveverylargesurveyarea. However,withthelimitinte- grationtime,largesurveyareamaynotbeabletobeatdown ∆ν[MHz] 10 10 10 theconstrainterror.WewilldiscussthedetailsinSect.VI. nν(nz) 30 70 30 D [m] 300 15 25 dish Nant×Nfeed 1×19 190×1 1×60 VI. RESULTSANDDISCUSSION tTOT[yr] 1 1 1 Trec[K] 25 28 50 Figure 5 shows the σ contoursfor local-shape PNG in Ssurvey[deg2] <24000 <25000 2500 the plane of the surveyfaNreLa and total observationtime. The leftandmiddlepanelsofFig.5showthecontoursforSKA-I 7 104 104 Redshift Window(i) = 0.37 Redshift Window(i) = 3.06 103 103 2] 2] K K )[µ 102 )[µ 102 π π 2 2 /( 101 /( 101 iiCℓ iiCℓ 1) 100 1) 100 + fNL=0 + fNL=0 ℓ(ℓ 10-1 Noise Level SKAI ℓ(ℓ 10-1 Noise Level SKAI Noise Level FAST Noise Level FAST Noise Level BINGO Noise Level BINGO 10-2 10-2 2] 2] µK 10-1 µK 10-1 [L [L fN 10-2 fN 10-2 ∂ ∂ ii/Cℓ10-3 iiC/ℓ10-3 π)∂10-4 π)∂10-4 )/(2 10-5 LEoqcuaillateral )/(2 10-5 LEoqcuaillateral 1 1 + 10-6 Orthogonal + 10-6 Orthogonal ℓ Enfolded ℓ Enfolded ( ( ℓ ℓ 101 102 101 102 ℓ ℓ FIG. 4: Upper panels–Comparison between the noise power spectra of different experiments and the 21-cm power spectrum in standard model (fNL = 0) for thetwo representative redshift bins (left and right panels). Inboth panels, one-year observation time(equivalent to 3.15×107sec)and2500deg2surveyareaareassumedforalltheexperiments. Lowerpanels–ThepartialderivativesofCiiwithrespectto ℓ parameterfNLforfourshapesofPNG. 3 SKA-I Local FAST Local 104 .00 104 101 SKA-I Local FAST Local 0 SKA-I Equilateral FAST Equilateral SKA-I Orthogonal FAST Orthogonal ××Nt[year]feedTOT110023 1.500.5000 1100NL01fσ ××Nt[year]feedTOT110023 16..500000 3.000 10NL1fσ /σσ)(minffNLNL SKA-I Enfolded FAST Enfolded Nant101 Nant101 15.000 100 6.000 15.00030.000 30.000 100 102 103 104 105 102 103 104 105 102 103 104 105 Ssurvey[deg2] Ssurvey[deg2] Ssurvey[deg2] FIG.5: Theleft(forSKA-I)andmiddle(forFAST)panelsshowtheσ contoursforlocal-shapePNGintheparameterspaceofthesurvey fNL areaandtotalobservationtime. ThedashedcontouristheerrorofconstraintwithPlancktemperatureandpolarizationdata[11]. Theright panelshowstheσfNL/(σfNL)minasafunctionofsurveyareaforvariousPNGtypes.ThesolidlinesshowtheresultsforSKA-Iwithone-year observationaltimeand190dishes;thedashedlinesshowtheresultsforFASTwithone-yearobservationaltimeand19beams. and FAST experiments respectively. The colour going from optimization is about 6000deg2 for FAST experiment. For red to blue meansthat the constraintsbecomestronger. Dif- other shapes, the optimized survey areas are approaching to ferent black solid lines are the contours of the same error themaximumskycoverageofSKA-IorFAST.Thelargesur- of flocal. Therefore, the error tends to become smaller if veyareacanhelptobeatthecosmicvarianceonlargescales, NL N N t becomesbigger.Thusthemostefficient buttheintegrationtimeperpixelbecomessmaller,leadingto ant feed TOT × × way toreducetheconstrainterroristo increasethe observa- largerpixelnoise. tiontimeorthenumberofdishes(feeds). Assumingone-year One can see from the right panel of Fig. 5 that generally observation time and the maximum dish(feeds) number for speaking the larger survey area is, the smaller the error of SKA-I and FAST experiments, the constraint errors of vari- f is, except for measuring equilateral shape of PNG us- NL ous PNG types as function of survey area are shown in the ing FAST survey. This is different from the situation of us- rightpanelof Fig. 5. In orderto have a clear view, the con- ing21-cmintensitymappingtomeasuretheangularscale of straint errors, σ , are divided by the their minimal values. BAOacousticoscillation,whichhavetheoptimalsurveyarea fNL Itistruethattheoptimalsurveyareamaynotbethemaximal around6000deg2(ForBINGOseefig.7in[27],andforFAST surveyarea.Forexample,inthecaseofEquilateralshape,the seefig.1in[54]).Thereasonisbecausescale-dependentbias 8 fromPNGisalwaysprominentonverylargescales,sobeat- tween350MHzand1050MHz, whichisthesameasthefre- ing down cosmic variance is more important than lowering quency range of SKA-I experiment, the constraint for local downthepixelnoise. However,BAOscaleissub-horizonfor shapePNG willbeσ 1.62withtheoptimizedsurvey which one needs to find a balance between lowering down areaof6000deg2.ThfeNlocLcoalns∼trainterrors(σ )fororthogonal pixel noise and beating down cosmic variance. We use dif- fNL andenfoldedshapesbecome39and64respectively,whichare ferent optimized survey areas for different cases in the later allhighlyreduced. analysis. Figure 6 shows the σ as a function of ℓ if we fix fNL min ℓ = 600. Different PNG shapes are shown in different max panels.Ineachpanel,differentcolorsindicatedifferentexper- imentsas shown in the legends. The optimizedsurveyareas areappliedto theanalysis. Theconstrainterrorsof different VII. CONCLUSION PNG shapes from Planck satellite are shown with the black dashed lines [11]. The σ of different PNG shapes fore- fNL In this work, we explored the constraining power on the castedwithdifferentexperimentsarelistinTableII. primordial non-Gaussianity (PNG), with the future single- Wecanseethat,forthelocalshapePNG,theSKA-Iexper- dish HI intensity mappingobservationswith BINGO, FAST iment is potentially able to constrain f better than Planck NL and SKA-I. Four fundamentalshapes of PNG are studied in experiment. But we should realize that it is only the most our analysis, including local, equilateral, orthogonaland en- idealcase. Itiswellknownthat,oneofthebigchallengesfor folded. Wefocusontheeffectofscale-dependentbiastothe observationsof HI intensity mapping is the foregroundsub- underlyingdarkmattertracer,inducedbytheprimordialnon- traction, and the low-ℓ modes may not be detectable due to Gaussinaity. The properties of such scale-dependentbias at the foregroundcontamination. Our results show that, to ob- large-scale limit are discussed in our analysis. The forecast tainaremarkableconstraintonf withtheSKA-Iintensity NL resultsarelistinTableII. mappinginthefuture,weneedtorecovertheangularpower spectrumof HIwiththeminimalℓmax 50. Thisistheaim Our forecasts show that with the current configuration of ofseveralrecenteffortsofrestoringlarg≃eangularpowerwith the experimentsone-yearobservationtime, the constrainton cross-correlationwithweakgravitationallensing[55,56].We local shape of PNG from SKA-I intensity mapping experi- also find that the constraint error for orthogonalshape PNG ment can be better than the current Planck experiment. The with SKA-I is 25, which is at the same level of current optimized survey area of 25000deg2 is applied in the anal- Planck limit. If∼the observation can be extended to 2 years, ysis of SKA-I, but the results are more sensitive to the total theerrorwillbereducedto 20. observationtimethanthesurveyarea.However,theHIinten- Theconstrainterrorfore∼quilateralshapePNG withFAST sity mapping experimentsmay be contaminatedby the fore- is 44,whichisbetterthantheresultsofSKA-IandBINGO groundandthelow-ℓmodesmaybebedetectable.Ouranaly- ex∼periments. The FAST error on fequil is close to the cur- sisshowsthattheSKA-Iexperimentcanstillhavetheremark- rent limit of Planck experiment. ThNiLs is because the scale- ableconstraintwithoutthemodesof ℓ < 50 With two-years dependent bias induced by the equilateral shape PNG has observation,theconstraintonorthogona∼lshapePNGis 20, ∼ higher signal-to-noise ratio at small scales and the FAST whichisalsobetterthantheconstraintfromPlanckmeasure- experiment is more sensitive to the small-scale modes than ment. SKA-IsingledishmodeandBINGO. The FAST experimenthasits advantageof higherangular We also test the possible extensionsof the currentconfig- resolution,andmoresensitivetothesmall-scalemodes,which uration by adding more integration time. If the observation is good for constraining the equilateral shape of PNG. With time for SKA-I and FAST could be extendedto 2 years, the thecurrentconfigurationandtwo-yearsobservation,thecon- constraintsonfNL canbeimprovedquantitatively. Thefore- strainterrorforEquilateralshapeofPNGwillbeσfNL = 32, castedconstraintondifferentshapesofPNGarelistedinTa- whichisbetterthanthecurrentlimitofPlanckobservation. bleII.Itisworthnoticingthattheconstrainterroronorthog- Similar constraint on the local shape of PNG can be onalshapePNGwithSKA-IandequilateralshapePNGwith FASTbecomesmallerthanlimitsofPlanckwithextendedob- achievedbytheFAST HI intensitymapping,ifitsfrequency bandwidthcouldbeextendedtothelowerfrequencies(ultra- servationaltime. wideband).Assumingthesameworkingfrequencyrange,the A good extension for FAST experiment is to extend its bestconstraintfromFASTonlocalshapeofprimordialnon- bandwidth to the lower frequencies, which are correspond- Gaussianiyisσ 1.62. ingtothehigherredshifts.SofartheFASTtelescopehasone fNL ∼ ultra-wide band receiver working on 270MHz 1.62GHz. The studieswe conducthere are the standardpowerspec- ∼ Unfortunately, the ultra-wide band receiver has only one tra analysis of 21-cm. There have been efforts on using the beam. Itwilltakequitelongtimetoachievethesameobser- multi-tracertechniquetobeatthecosmicvarianceandobtain vationtime asthemulti-beamreceiver. Nowthe multi-beam tighter constraintson f [57–59]. In addition, using three- NL system of the FAST telescope is designed to work on fre- point correlation function is another way to measure PNG. quencies between 1050MHz and 1350MHz. Assuming that Thesemethodswillbeexploredtomeasureallshapesoff NL the FAST multi-beam system works on the frequencies be- inthefuturework. 9 Local Shape Equilateral Shape 104 Sum over ℓ to ℓmax=600 104 Sum over ℓ to ℓmax=600 SKA-I 25000deg2 SKA-I 25000deg2 FAST 24000deg2 FAST 6000deg2 103 BINGO 2500deg2 103 BINGO 2500deg2 Planck Planck NL NL σf 102 σf 102 101 101 100 100 101 102 101 102 ℓ ℓ Enfolded Shape Orthogonal Shape 104 Sum over ℓ to ℓmax=600 104 Sum over ℓ to ℓmax=600 SKA-I 25000deg2 SKA-I 25000deg2 FAST 24000deg2 FAST 24000deg2 103 BINGO 2500deg2 103 BINGO 2500deg2 Planck NL NL σf 102 σf 102 101 101 100 100 101 102 101 102 ℓ ℓ FIG.6: TheσfNL asafunctionofℓminforvariousexperimentsandPNGshapes. TheblackdashedlineisthecurrentconstraintwithPlanck temperatureandpolarizationdata[11]. TABLEII: σ ofdifferentPNGshapesforecastedwithdifferentexperiments. Theoptimizedsurveyareasareappliedintheanalysis. The fNL “Planck2015”columnshowstheconstrainterrorwithPlanktemperatureandpolarizationdata[11]. Thenumbersinboldcharacterarethe constraintsbetterthanPlanck. CurrentConfiguration Extentions Planck2015 FAST SKA-I BINGO SKA-I2yr† FAST2yr†† FASTlow‡ Local 5 9.5 0.54 17 0.43 7.4 1.6 Equilateral 43 44 86 100 66 32 53 Orthogonal 21 75 25 128 20 59 39 Enfolded – 94 43 164 36 70 64 †SKA-Iwithtwo-yearobservation;††FASTwithtwo-yearobservation;‡FASTwithlowfrequenciesrange from350MHzto1050MHz Acknowledgments on CAMB SOURCES . This work is supported by National Research Foundation of South Africa with grant no.105925 WethankNealDalal,DiLi,RoyMaartensandXiao-Dong andUniversityofKwaZulu-Natalstaffstart-upgrant. Xu for helpful discussions and Stefano Camera for his help [1] N.Bartolo, E.Komatsu, S.Matarrese, and A.Riotto, Physics [2] T.Falk, R.Rangarajan, andM.Srednicki, PhysicalReviewD Reports402,103(2004),astro-ph/0406398. 10 46,4232(1992),astro-ph/9208002. [28] R. Nan, D. Li, C. Jin, Q. Wang, L. Zhu, W. Zhu, H. Zhang, [3] A. Linde and V. Mukhanov, Physical Review D 56, R535 Y.Yue,andL.Qian,InternationalJournalofModernPhysicsD (1997),astro-ph/9610219. 20,989(2011),1105.3794. [4] D.H.Lyth,C.Ungarelli,andD.Wands,PhysicalReviewD67, [29] D.LiandZ.Pan,RadioScience51,1060(2016). 023503(2003),astro-ph/0208055. [30] P.E.Dewdney,P.J.Hall,R.T.Schilizzi,andT.J.L.W.Lazio, [5] M.Alishahiha,E.Silverstein,andD.Tong,PhysicalReviewD IEEEProceedings97,1482(2009). 70,123505(2004),hep-th/0404084. [31] PlanckCollaboration, P.A.R.Ade, N.Aghanim, M.Arnaud, [6] P.Creminelli,A.Nicolis,L.Senatore,M.Tegmark,andM.Zal- M.Ashdown, J.Aumont, C.Baccigalupi, A.J.Banday, R.B. darriaga, Journal of Cosmology and Astroparticle Physics 5, Barreiro,J.G.Bartlett,etal.,AstronomyandAstrophysics594, 004(2006),astro-ph/0509029. A13(2016),1502.01589. [7] X. Chen, M.-x. Huang, S. Kachru, and G. Shiu, Journal of [32] T. J. Allen, B. Grinstein, and M. B. Wise, Physics Letters B Cosmology and Astroparticle Physics 1, 002 (2007), hep- 197,66(1987). th/0605045. [33] A. Gangui, F. Lucchin, S. Matarrese, and S. Mollerach, The [8] R. Holman and A. J. Tolley, Journal of Cosmology and As- AstrophysicalJournal430,447(1994),astro-ph/9312033. troparticlePhysics5,001(2008),0710.1302. [34] S. Renaux-Petel, Comptes Rendus Physique 16, 969 (2015), [9] P. D. Meerburg, J. P. van der Schaar, and P. S. Corasaniti, 1508.06740. JournalofCosmologyandAstroparticlePhysics5,018(2009), [35] E.KomatsuandD.N.Spergel,PhysicalReviewD63,063002 0901.4044. (2001),astro-ph/0005036. [10] A.Achu´carro,J.-O.Gong,S.Hardeman,G.A.Palma,andS.P. [36] J.M.Bardeen,PhysicalReviewD22,1882(1980). Patil, Journal of Cosmology and Astroparticle Physics 1, 030 [37] J. Maldacena, Journal of High Energy Physics5, 013 (2003), (2011),1010.3693. astro-ph/0210603. [11] PlanckCollaboration, P.A.R.Ade, N.Aghanim, M.Arnaud, [38] X. Chen, Physical Review D 72, 123518 (2005), astro- F. Arroja, M. Ashdown, J. Aumont, C. Baccigalupi, M. Bal- ph/0507053. lardini,A.J.Banday,etal.,AstronomyandAstrophysics594, [39] L.Senatore,K.M.Smith,andM.Zaldarriaga,JournalofCos- A17(2016),1502.01592. mologyandAstroparticlePhysics1,028(2010),0905.3746. [12] N.Dalal, O.Dore´, D.Huterer, andA.Shirokov, PhysicalRe- [40] R.K.ShethandG.Tormen,MonthlyNoticesoftheRoyalAs- viewD77,123514(2008),0710.4560. tronomicalSociety329,61(2002),astro-ph/0105113. [13] S.MatarreseandL.Verde,AstrophysicalJournal Letters677, [41] M. Santos, P. Bull, D. Alonso, S. Camera, P. Ferreira, L77(2008),0801.4826. G.Bernardi,R.Maartens,M.Viel,F.Villaescusa-Navarro,F.B. [14] C.Carbone,L.Verde,andS.Matarrese,AstrophysicalJournal Abdalla,etal.,AdvancingAstrophysicswiththeSquareKilo- Letters684,L1(2008),0806.1950. metreArray(AASKA14)19(2015),1501.03989. [15] A. Slosar, C. Hirata, U. Seljak, S. Ho, and N. Padmanabhan, [42] H. J. Mo and S. D. M. White, Monthly Notices of the Royal JournalofCosmologyandAstroparticlePhysics8,031(2008), AstronomicalSociety282,347(1996),astro-ph/9512127. 0805.3580. [43] H. J. Mo and S. D. M. White, Monthly Notices of the Royal [16] V. Desjacques, D. Jeong, and F. Schmidt, Physical Review D AstronomicalSociety336,112(2002),astro-ph/0202393. 84,063512(2011),1105.3628. [44] W.H.PressandP.Schechter, TheAstrophysicalJournal 187, [17] T. Matsubara, Physical Review D 86, 063518 (2012), 425(1974). 1206.0562. [45] R.K.Sheth,H.J.Mo,andG.Tormen,MonthlyNoticesofthe [18] A. Raccanelli, M. Shiraishi, N. Bartolo, D. Bertacca, RoyalAstronomicalSociety323,1(2001),astro-ph/9907024. M.Liguori,S.Matarrese,R.P.Norris,andD.Parkinson,ArXiv [46] C.Fedeli,C.Carbone,L.Moscardini,andA.Cimatti,Monthly e-prints(2015),1507.05903. Noticesof theRoyal Astronomical Society414, 1545 (2011), [19] J.-Q.Xia,C.Baccigalupi,S.Matarrese,L.Verde,andM.Viel, 1012.2305. JournalofCosmologyandAstroparticlePhysics8,033(2011), [47] T.-C.Chang,U.-L.Pen,J.B.Peterson,andP.McDonald,Phys- 1104.5015. icalReviewLetters100,091303(2008),0709.3672. [20] N.Nikoloudakis, T.Shanks, andU.Sawangwit, MonthlyNo- [48] E. R. Switzer, K. W. Masui, K. Bandura, L.-M. Calin, T.-C. tices of the Royal Astronomical Society 429, 2032 (2013), Chang, X.-L.Chen, Y.-C.Li,Y.-W.Liao,A.Natarajan, U.-L. 1204.3609. Pen,etal.,MonthlyNoticesoftheRoyalAstronomicalSociety [21] A. J. Ross, W. J. Percival, A. Carnero, G.-b. Zhao, M. Man- 434,L46(2013),1304.3712. era, A. Raccanelli, E. Aubourg, D. Bizyaev, H. Brewington, [49] A. Challinor and A. Lewis, Physical Review D 84, 043516 J.Brinkmann,etal.,MonthlyNoticesoftheRoyalAstronomi- (2011),1105.5292. calSociety428,1116(2013),1208.1491. [50] L. Zhang, E.F. Bunn, A. Karakci, A. Korotkov, P. M. Sutter, [22] Y.-Z. Ma, J. E. Taylor, and D. Scott, Monthly Notices of the P.T.Timbie,G.S.Tucker,andB.D.Wandelt,TheAstrophysi- RoyalAstronomicalSociety436,2029(2013),1308.2673. calJournalSupplementSeries222,3(2016),1505.04146. [23] V.Desjacques,D.Jeong,andF.Schmidt,ArXive-prints(2016), [51] M.-A.Bigot-Sazy,C.Dickinson,R.A.Battye,I.W.A.Browne, 1611.09787. Y.-Z. Ma, B. Maffei, F. Noviello, M. Remazeilles, and P. N. [24] S. Camera, M. G. Santos, P. G. Ferreira, and L. Ferramacho, Wilkinson,MonthlyNoticesoftheRoyalAstronomicalSociety PhysicalReviewLetters111,171302(2013),1305.6928. 454,3240(2015),1507.04561. [25] Y.Xu,X.Wang,andX.Chen,TheAstrophysicalJournal798, [52] L. C. Olivari, M. Remazeilles, and C. Dickinson, Monthly 40(2015),1410.7794. Noticesof theRoyal Astronomical Society456, 2749 (2016), [26] P.Bull, P.G.Ferreira, P.Patel,andM. G.Santos, TheAstro- 1509.00742. physicalJournal803,21(2015),1405.1452. [53] R.Battye,I.Browne,T.Chen,C.Dickinson,S.Harper,L.Oli- [27] R.A.Battye,I.W.A.Browne,C.Dickinson,G.Heron,B.Maf- vari,M.Peel,M.Remazeilles,S.Roychowdhury,P.Wilkinson, fei, and A. Pourtsidou, Monthly Notices of the Royal Astro- etal.,ArXive-prints(2016),1610.06826. nomicalSociety434,1239(2013),1209.0343. [54] M.-A. Bigot-Sazy, Y.-Z. Ma, R. A. Battye, I. W. A. Browne,

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.