ebook img

Capturing Halos at High Redshifts PDF

0.16 MB·English
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 Capturing Halos at High Redshifts

LA-UR-05-9198 DRAFTVERSIONFEBRUARY4,2008 PreprinttypesetusingLATEXstyleemulateapjv.26/01/00 CAPTURINGHALOSATHIGHREDSHIFTS KATRINHEITMANN1,ZARIJALUKIC´2,SALMANHABIB3,ANDPAULM.RICKER2,4 1ISR-1,ISRDivision,TheUniversityofCalifornia,LosAlamosNationalLaboratory,LosAlamos,NM87545 2Dept.ofAstronomy,UniversityofIllinois,Urbana,IL61801 3T-8,TheoreticalDivision,TheUniversityofCalifornia,LosAlamosNationalLaboratory,LosAlamos,NM87545 4NationalCenterforSupercomputingApplications,Urbana,IL61801 DraftversionFebruary4,2008 ABSTRACT 6 WestudytheevolutionofthemassfunctionofdarkmatterhalosintheconcordanceΛCDMmodelathighred- 0 shift. Weemployoverlapping(multiple-realization)numericalsimulationstocoverawiderangeofhalomasses, 0 107- 1015h- 1M⊙,withredshiftcoveragebeginningatz=20. ThePress-Schechtermassfunctionissignificantly 2 discrepantfromthesimulationresultsathighredshifts.Ofthemorerecentlyproposedmassfunctions,ourresults areinbestagreementwithWarrenetal. (2005). Thestatisticsofthesimulations–alongwithgoodcontrolover n a systematics–allowforfitsaccuratetothelevelof20%atallredshifts.Weprovideaconcisediscussionofvarious J issuesindefiningandcomputingthehalomassfunction,andhowtheseareaddressedinoursimulations. 1 Subjectheadings:methods:N-bodysimulations—cosmology:halomassfunction 1 1 1. INTRODUCTION carried out over differentmass and redshiftranges. The clos- v est to the presentwork are Reed et al. (2003)and Springelet Dark matter halos occupy a central place in the paradigm 3 al. (2005);incomparisontotheirresults, ourhalomassrange of gravitationally-driven structure formation arising from the 3 goesdeeperbythreeordersofmagnitude,withgoodstatistics nonlinearevolutionofprimordialGaussiandensityfluctuations. 2 and control of systematics out to z=20, substantially higher Gascondensation,resultantstarformation,andeventualgalaxy 1 than in these papers. (We review results from other work be- 0 formation occurswithin halos. Consequently, the halo profile low.)Essentially,theearlierresultsareinverygoodagreement 6 andmassfunctionarecentralingredientsinphenomenological withtheSheth-Tormenmassfunction(Sheth&Tormen1999), 0 modelsofnonlinearclusteringofgalaxies. Thedistributionof atredshiftsz≤10.Asweshowbelow,variousfittingformulae / halomasses–thehalomassfunction–anditstimeevolution, h givenintheliterature–mosttunedtosimulationresultsatz=0 arealsosensitiveprobesofcosmology. p – can differsubstantially in their predictionsathigh redshifts, The halo mass function at the high-mass end (cluster mass - byasmuchasafactoroftwo.Therefore,itisimportanttocarry o scales)isexponentiallysensitivetotheamplitudeoftheinitial r density perturbations,the meanmatter density parameter,Ω , outsimulationsofsufficientdynamicrangeandaccuracytotest t m thesepredictions. s andtothedarkenergycontrolledlate-timeevolutionoftheden- a In order to extractthe mass function from simulations, dif- sity field. Thelastfeature,particularlyatlowredshifts,z<2, : ferentquestionshavetobeaddressed,suchas:Howisthemass v allowsclusterobservationstoconstrainthedarkenergycontent, Xi ΩΛ,andtheequationofstateparameter,w(Holderetal. 2001). fuulantciotino?ntWohbeendmefiunsetdt?heWsihmeunldatoiotnhebfierssttahrtaeldosinfoorrmdeirntoasciamp-- Thehalomassfunctionisalsoofconsiderableinterestathigh r ture these halos? What force and mass resolution is required a redshift, relating to questions such as predictions for quasar to capture halos of a certain mass at a specific redshift? We abundanceandformationsites(Haiman&Loeb2001),thefor- havederivedandtestedcertaincriteriatoensurethatoursimu- mation history of collapsed baryonic halos, and the reion- lationscapturethehalosofinterest; detailswillbegivenelse- ization history of the Universe (Furlanettoetal. 2005). Re- where(Lukic´etal. 2006). cent results from the Wilkinson Microwave Anisotropy Probe The paper is organizedas follows. In Section 2 we discuss (WMAP)(Kogutetal. 2003;Spergeletal. 2003)indicatethat popularmassfunctionformulae,previouswork,andstrategies reionization could have begun at redshifts as high as z∼20. for determiningthe halo mass function at high redshifts. The Much of the work on possible reionization scenarios is based simulations and mass function results are discussed in Sec- on the simple Press-Schechter (PS) mass function (Press & tion3.Criteriaformassandforceresolutionandinitialredshift Schechter 1974, Bond et al. 1991) the use of which can lead needed to span the desired mass and redshift range are given toimprecisepredictionsforthereionizationhistory. here.WepresentourconclusionsinSection4. Simulations play a dual role in characterization of the halo mass function. If only a few fixed sets of cosmological pa- 2. THEMASSFUNCTION rameters and a finite dynamic range are required, simulations can producevaluable results. In order to investigatea variety Overthelastthreedecadesdifferentfittingfunctionsforthe of cosmologiesanddifferentscenariosfor physicalprocesses, mass function have been suggested. The first analytic model e.g.,reionization,itisneverthelessveryconvenient,ifnotnec- for the mass function was developed by Press & Schechter essary, to have accurate analytic fitting relations. Simulations (1974). Their theoryconsidersa sphericallyoverdenseregion canbeusedtovalidatethesefitsoverawide(albeit,discretely in an otherwise smooth background density field. The over- sampled)rangeofparameters. density evolves as a Friedmann universe with positive curva- Various numerical studies of the mass function have been ture. Initially, the overdensity expands, but at a slower rate thanthebackgrounduniverse(thusenhancingthedensitycon- 1 2 CapturingHalosatHighRedshifts trast), until it reaches the ‘turnaround’ density, after which it The determination of mass functions at high redshifts is a collapses. Althoughformally this collapse ends with a singu- nontrivialtask. Highredshifthaloshaveverylowmasses,plac- larity,itisassumedthatinrealitytheoverdenseregionwillviri- ingheavydemandsonthemassandforceresolutionneededto alize. ForanEinstein-deSitteruniverse,thedensityofsuchan resolvethem.Theserequirementscanbeachievedintwoways. overdenseregionatthevirializationredshiftisρ≈180ρ (z).At First, a simulation with a very large number of particles and c thispoint,thedensitycontrastfromthelineartheoryofpertur- highforceresolutioncanbeperformed. Thisisexpensive,and bationgrowth[δ(~x,z)=D+(z)δ(~x,0)]isδ (z)≈1.686(1+z).For onlyaverylimitednumberofsuchsimulationscanbecarried c Ω <1,δ (z)evolvesdifferently(seeLacey&Cole1993),but out. Second, since determining the mass function is simply a m c the dependenceoncosmologyis weak(see e.g., Jenkinsetal. questionofstatistics, manyrelativelymodestsimulationswith 2001).Thus,weadoptδ =1.686=δ (0). moderateparticleloadingcanbeperformed:thisisthestrategy c c Followingtheabovereasoningandwiththeassumptionthat weadopthere. Assimulationscanonlybetrusteduntila red- theinitialdensityperturbationsaregivenbyaGaussianrandom shiftatwhichthelargestmodeisclosetobecomingnonlinear, field,thePSmassfunctionisgivenby: multipleoverlappingboxsizesmustbeused. Springel et al. (2005)have recently followed the evolution f (σ)= 2δcexp - δc2 , (1) of 21603 particles in a 500h- 1Mpc box. The high mass and PS rπ σ (cid:20) 2σ2(cid:21) forceresolutionallowthemtostudythemassfunctionreliably where σ is the variance of the linear density field, f(σ,z) ≡ out to a redshift of z=10, covering a mass range of roughly (MU/ρsibn)g(denm/pdilrnicσa-l1a)r,gaunmdeρnbtsisSthheethba&ckTgorromunend(d1e9n9s9it,yh.ereafter l1a0ti1o0hn-s1iMnc⊙lutdoe1J0a1n6gh--C1Mon⊙d.elElx&amHperlensqoufissti(n2g0l0e1s)m(a1lhl-- b1Moxpscimboux- ST)proposedanimprovedfitofthefollowingform: with 1283 particles evolved to z=10) and Cen et al. (2004) (4h- 1Mpc box, 5123 particles, evolved to z = 6). Results in f (σ)=0.3222 2aδcexp - aδc2 1+ σ2 p , (2) both papers are claimed to be consistent with PS but without ST rπ σ (cid:20) 2σ2(cid:21)(cid:20) (cid:18)aδ2(cid:19) (cid:21) detailed quantification. The simulation of Reed et al. (2003) c is a compromise between the two extremes: a box size of witha=0.707,and p=0.3. Shethetal. (2001)interpretedthis 50h- 1Mpc with 4323 particles and a concomitant halo mass fittheoreticallybyextendingthePSapproachtoanellipsoidal rangeof roughly1010h- 1M to 1014.5h- 1M . Reed etal. find collapsemodel.Inthismodel,thecollapseofaregiondepends ⊙ ⊙ goodagreement(betterthan20%)withtheSTfituptoz≃10. notonlyon its initialoverdensity,butalso on the surrounding For higher redshifts they find that the ST fit overpredicts the shearfield.ThedependenceischosentorecovertheZel’dovich number of halos, at z=15 up to 50%. At this high redshift, approximation(Zel’dovich1970) in the linear regime. A halo however,theirresultsbecomestatistics-limited,themassreso- isconsideredvirializedwhenthethirdaxiscollapses(seealso lutionbeinginsufficienttoresolvetheverysmallhalos. Lee&Shandarin1997). In this paper we analyze a suite of 50 N-body simulations Jenkins et al. (2001, hereafter Jenkins) combine high res- withvaryingboxsizesbetween4h- 1Mpcand126h- 1Mpcwith olutionsimulationsfordifferentcosmologiesspanninga mass rangeofoverthreeordersofmagnitude[∼(1012- 1015)h- 1M⊙], multiplerealizationsofallboxestostudythemassfunctionat and includingseveralredshiftsbetween z=5 and z=0. They findthatthefollowingfittingformulaworksexceptionallywell 0.2 Warren et al. (within20%),independentoftheunderlyingcosmology: Jenkins et al. 0 fJenkins(σ)=0.315exp - |lnσ- 1+0.61|3.8 . (3) -0.2 PSrheestsh--STcohremchenter z=0 Theaboveformulaisveryclose(cid:2)tothenominalST(cid:3)fit. 9 10 11 12 13 14 15 By performing 16 nested-volume simulations Warren et al. 0.4 0.2 (2005,hereafterWarren)obtainsignificanthalostatisticsspan- 0 ning a mass range of five orders of magnitude [∼ (1010- -0.2 z=5 1015)h- 1M⊙]. Theirbestfitemploysafunctionalformsimilar uals-0.49 10 11 12 13 toanimprovedversionofST(Sheth&Tormen2002): d si 0 e fWarren(σ)=0.7234 σ- 1.625+0.2538 exp(cid:20)- 1.1σ9282(cid:21). (4) ve R--00..48 z=10 (cid:0) (cid:1) ati 7 8 9 10 11 12 ThediscrepancybetweenPSandthemoreaccuratefitsisev- el R identinFigure1wheretheredshiftevolutionofthemassfunc- 0 tion is shown. The redshift dependence in the analytic mass -0.4 z=15 functionsentersonlythroughσ(z)=σ(0)d(z),whered(z)isthe -0.8 growthfactornormalizedsuchthatd(0)=1. Asthefunctional 7 8 9 10 11 dependenceon σ is differentin the differentfits, this leads to 0 substantialvariationacrossthefitsasafunctionofredshift.For -0.4 z=0 the Warren fit agrees – especially in the low mass range z=20 -0.8 below1013M –tobetterthan5%withtheSTfit. Atthehigh ⊙ 7 8 9 10 mass end the difference increases up to 20%. The Jenkins fit log(M [M./h]) O leads to similar results over the considered mass range. Note FIG. 1.—RelativeresidualsofthePS,Jenkins,andWarrenmassfunction that at higher redshifts and intermediate mass ranges around fitswithrespecttoSTforfivedifferentredshifts. Notethattherangesofthe 109M⊙ the disagreement between the Warren and ST fits in- axesaredifferentinthedifferentpanels.WedonotshowtheJenkinsfitbelow creasesupto40%. massesof1011h- 1M⊙sinceitisnotvalidinthismassrange. Heitmann,Lukic´,Habib,Ricker 3 redshiftsuptoz=20andtocoveralargemassrangebetween 16h- 1Mpcboxes.Thehalogrowthfunctionisparticularlyvalu- 107h- 1M and1015h- 1M evenathighredshifts.Significantly, able for determining when the halos at a certain mass should ⊙ ⊙ atz=20,gasinhaloswithamassscaleabove∼107h- 1M⊙can firstform. Thisisagoodtestforproblemsinsimulationsaim- coolviaatomiclinecooling(Tegmarketal. 1997). ingto capturehaloswitha givenminimummassatsome red- shift. Anexampleofthisisinsufficientforceresolutioninthe 3. SIMULATIONSANDMASSFUNCTIONRESULTS base-gridsofadaptive-mesh-refinement(AMR)codes. Allsimulationsinthispaperarecarriedoutwiththeparticle- Oncethenumberofparticlesforasimulationanda desired mesh code MC2 (Mesh-based Cosmology Code). MC2 has massforthesmallesthaloaredecided,therequiredboxsizeis been extensively tested against other cosmologicalsimulation fixed.Theforceresolutionneededtoresolvethesmallesthalos codes(Heitmannetal. 2005).Thechosenvaluesofcosmolog- has then to be determined. Our aim here is not to precisely icalparametersare: measurethe haloprofilebutsimply to becertain thatthe total Ω =1.0, Ω =0.253, Ω =0.048, halomass is correct. As shownin Heitmannet al. (2005)the tot m baryon halomassisarelativelyrobustquantityanda simpleestimate σ =0.9, H =70km/s/Mpc, (5) 8 0 oftheforceresolutionisallthatisneeded.Theforceresolution as set by the latest cosmic microwave background and large mustbesmallcomparedtothecomovinghalovirialradiusr∆ scalestructureobservations(MacTavishetal. 2005). Themass (withtheoverdensityrelativetothecriticaldensity,∆∼200)at transferfunctionsaregeneratedwithCMBFAST(Seljak&Zal- allredshifts.Theresultinginequalitycanbestatedintheform darriaga 1996). We summarize the different runs, including δ n Ω(z) 1/3 theirforceandmassresolutioninTable1. f <0.62 h , (6) Weidentifyhaloswiththefriends-of-friendsalgorithm(FOF), ∆p (cid:18) ∆ (cid:19) basedonfindingneighborsofparticlesatacertaindistance(see whereδ istheforceresolutionandn isthenumberofparticles f h e.g., Einasto et al. 1984; Davis et al. 1985). The halo mass per halo. In the simulations performedhere we use a ratio of is defined simply by the sum of particles which are members oneparticleper64gridcells,whichallowshaloswithroughly of the halo. (For connections between differentdefinitions of 50particlestobe captured. Ithasbeenshownin Heitmannet halo masses, see White 2001.) Despite several shortcomings al. (2005)thatthisratiodoesnotcausecollisionaleffectsand of the FOF halo finder, e.g., halo-bridging (see, e.g., Gelb & leadstoconsistentresultsincomparisonwithothercodes.Mass Bertschinger 1994, Summers et al. 1995) or statistical biases functionconvergencetests with differentforce resolutionsare foundbyWarrenetal.(2005),theFOFalgorithmitselfiswell- nicely consistent with the above estimate as shown in Lukic´ definedandveryfast. etal. (2006); time-stepcriteria andconvergencetests are also There are two sourcesof possible biases in determiningin- describedthere. dividual halo masses using FOF. First, the halo may be sam- Thelargesetofsimulationswehavecarriedoutallowsusto pledwithaninsufficientnumberofparticles(seeWarrenetal. study the mass functionat redshiftsbetween z=20 and z=0. 2005). Second, the effective slope of the halo density profile The main results are shown in Figure 3, where the simulation close to the virialradiusr , atfixed particlenumber,also in- data for the mass function are shown along with the Warren, vir fluences the FOF mass. If the force resolution of the N-body PS,andSTfitsatdifferentredshifts.AtallredshiftstheWarren code affectsthe profile, this too, addsa systematic bias. Here fithasthebestagreementwiththesimulationswithascatterof werecordthemassfunctionforthelinkinglengthb=0.2FOF approximately 20% and is a numerically significant improve- massincludingonlythecorrectionofWarrenetal. (2005).Ina mentoverST.Suchaclosematchisquitegratifyinggiventhe follow-uppaper(Lukic´etal.2006)wewilladdresssystematics overalldynamicrangeoftheinvestigation.ThePSfit,overthe issuesindetermininghalomassesindetail. massrangeconsidered,isapoorfitatz≥10,deviatingbymore Wenowdiscusscriteriafoundtobeveryimportantfordemon- thanafactoroftwofromthenumericalresults. stratingtheconvergenceandrobustnessofourresults. Details willbepresentedinLukic´ etal. (2006). Thefirstissuerelates totheinitialredshiftofthesimulation. Twoconditionsareim- 10000 portant:(i)thesimulationmustbeginsufficientlyearlythatthe 109 - 1010MO./h initialZel’dovichdisplacementisasmallenoughfractionofthe 1010 - 1011MO./h meaninterparticleseparation∆p; onaverageaparticleshould 1000 1011 - 1012MO./h notmovemorethan∼∆ /3;(ii)thehighestredshiftwherethe p mass functionis to be evaluatedmust be sufficiently removed halos 100 fifrrosmt cthhaenrceedstohifftoormffihrsatl-ocsr.osTsihnegsztcrrionssgewnhceyreopfatrhteicsleescrhiatevreiathies ber of m such that the small boxes require very high starting redshifts, Nu 10 e.g.,the4h- 1Mpcboxhadaninitialredshiftz =500.Thisisa in muchearlierstartingredshiftthanthoseusedinprevioussimu- lations;theconventionalrequirementthatallmodesinthebox 1 be linear at the initial redshiftprovesto be much weaker, and thereforeinadequate,asaconvergencecriterion. 0.1 A simple test of how well the simulations track the mass 5 10 15 20 Redshift function formulae is to follow the number of halos in a spec- ified mass bin at a given redshift. For this purpose we con- FIG. 2.—Halogrowthfunctionforthreemassbinsforthe16h- 1Mpcbox. TheWarren(solid),ST(long-dashed),andPS(short-dashed)fitsarecompared vertthemassfunctionfitintoafunctionofz,definingthehalo tosimulationdatawithPoissonerrorbars. Notethequalityoftheagreement growth functionas shown in Figure 2. The evolution of three withtheWarrenfit. massbinsisshownasafunctionofzalongwithresultsfromthe 4 CapturingHalosatHighRedshifts TABLE1 SUMMARYOFTHEPERFORMEDRUNS Mesh BoxSize Resolution z z ParticleMass SmallestHalo #ofRealizations in final 10243 126h- 1Mpc 120h- 1kpc 50 0 9.94×109h- 1M⊙ 3.98×1011h- 1M⊙ 10 10243 64h- 1Mpc 62.5h- 1kpc 80 0 1.30×109h- 1M⊙ 5.2×1010h- 1M⊙ 5 10243 32h- 1Mpc 31.25h- 1kpc 150 5 1.63×108h- 1M⊙ 6.52×109h- 1M⊙ 5 10243 16h- 1Mpc 15.63h- 1kpc 200 5 2.04×107h- 1M⊙ 8.16×108h- 1M⊙ 5 10243 8h- 1Mpc 7.81h- 1kpc 250 10 2.55×106h- 1M⊙ 1.02×108h- 1M⊙ 20 10243 4h- 1Mpc 3.91h- 1kpc 500 10 3.19×105h- 1M⊙ 1.27×107h- 1M⊙ 5 Note.—Massandforceresolutionsofthedifferentruns.Thesmallesthalosweconsidercontain40particles.Allsimulationsarerunwith2563particles. 3 quired,asWhite(2002)pointsout,“itmaynotbesufficientto useasimpleparametrizedform”inconstrainingcosmological Warren fit 2 4 Mpc/h parameterswiththemassfunction. 8 Mpc/h Theerrorcontrolcriteriadevelopedinthisworkhaveanat- 1 16 Mpc/h uralapplicationinhigh-resolutionAMRsimulationsintheset- -3h)]) 0 3624 MMppcc//hh ttiinognoisfirnefipnroegmreesnst.anderrorcontrolcriteria.Workinthisdirec- c/ 126 Mpc/h p M PS fit M[( -1 ST fit WethankKevAbazajian,DanHolz,LamHui,GerardJung- man,SavvasKoushiappas,AdamLidz,SergeiShandarin,Ravi g o dl-2 Sheth, and Mike Warren for useful discussions. The authors dn/ acknowledge support from IGPP, LANL. S.H. and K.H. ac- log( -3 knowledge support from the DOE via the LDRD program at LANL. P.M.R. and Z.L. acknowledge UIUC, NCSA, and a DOE/NNSA PECASE award (LLNL B532720). S.H., K.H., -4 and P.M.R. acknowledge the hospitality of the Aspen Center forPhysicswherepartofthisworkwascarriedout. We espe- -5 z=20 z=15 z=10 z=5 z=0 cially acknowledge supercomputing support under the LANL InstitutionalComputingInitiative. 7 8 9 10 11 12 13 14 15 log(M./h])) O FIG.3.—Themassfunctionat5differentredshiftswithPoissonerrorbars. REFERENCES The red line is the Warren fit, blue is Press-Schechter, and black is Sheth- Bond,J.R,Cole,S.,Efstathiou,G,Kaiser,N. 1991,ApJ,379,440 Tormen. Cen, R., Dong, F., Bode, P., & Ostriker, J.P. 2004, astro-ph/0403352, ApJ, submitted Davis,M.,Efstathiou,G.,Frenk,C.S.1985,ApJ,292,371 4. CONCLUSIONSANDDISCUSSION Einasto,J.,Klypin,A.A.,Saar,E.,Shandarin,S.F. 1984,MNRAS,206,529 Furlanetto,S.R.,McQuinn,M.,Hernquist,L., 2005astro-ph/0507524 Inthispaperwehavestudiedtheevolutionofthemassfunc- Gelb,J.M.&Bertschinger,E. 1994,ApJ,436,467 Haiman,Z.,Loeb,A. 2001,ApJ,552,459 tion starting at a redshift of z=20 and covering a halo mass Heitmann,K.,Ricker,P.M.,Warren,M.S.,&Habib,S.2005,ApJS,160,28 rangeof107 to1015h- 1M⊙. Ourresultsincorporatenewhalo- Holder,G,Haiman,Z.,&Mohr,J. ApJ,560,L111 based N-bodyerrorcontrolcriteria that are describedin more Jang-Condell,H.&Hernquist,L. 2001,ApJ,548,68 Jenkins,A.etal. 2001,MNRAS,321,372 detailin Lukic´ etal. (2006). We findthatthePress-Schechter Kogut,A.etal.2003,ApJS,148,161 massfunctiondeviatessignificantlyfromourresults. Morere- Lacey,C.G.&Cole,S. 1993,MNRAS,262,627 Lukic´,Z.,Heitmann,K.,Habib,S.,&Ricker,P.M.2006,inpreparation cent mass function fits are in better agreement; in particular, MacTavish,C.J.etal.2005,astro-ph/0507503,ApJ,submitted therecentlyintroducedfittingfunctionofWarrenetal. (2005) O’Shea, B.W., Nagamine, K., Springel, V., Hernquist L., & Norman, M.L. agreesatthe20%levelovertheentireredshiftrange. 2005,ApJS,160,1 Press,W.H.&Schechter,P. 1974,ApJ,187,425 Thepreciseagreementofthenumericallyobtainedhalogrowth Reed,D.etal. 2003,MNRAS,346,565 functionaswellastheevolutionofthemassfunctionwiththe Seljak,U.&Zaldarriga,M. 1996,ApJ,469,437 Sheth,R.K.&Tormen,G. 1999,MNRAS,308,119 (evolved z=0) Warren fit demonstrates the remarkable result Sheth,R.K.&Tormen,G. 2002,MNRAS,329,61 thattheevolutionofthemassfunctioniscompletelycontrolled Sheth,R.K.,Mo,H.J.,&Tormen,G. 2001,MNRAS,323,1 bythelineargrowthofthevarianceofthelineardensityfield. Spergel,D.N.etal.2003,ApJS,148,175 Springel,V.etal. 2005,Nature,435,629 In order to find a mass function fit relevant to observa- Summers,F.J.,Davis,M.&Evrard,A.E.1995,ApJ,454,1 tions,severalhurdlesremaintobeovercome,includingreach- Tegmark,M.etal. 1997,ApJ,474,1 ingagreementonanappropriatedefinitionofhalomass(White Warren, M.S., Abazajian, K., Holz, D.E., and Teodoro, L. 2005, ApJL, submitted 2001) and improving the precision and accuracy of N-body White,M. 2001,A&A,367,27 codes beyondthe currentstate of the art (O’Shea et al. 2005; White,M. 2002,ApJS,143,241 Zel’dovich,Y.B.1970,A&A,5,84 Heitmannetal. 2005). Dependingonthelevelofprecisionre-

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.