DRAFTVERSIONJANUARY23,2017 PreprinttypesetusingLATEXstyleemulateapjv.04/17/13 THEVULTURESURVEYI:ANALYZINGTHEEVOLUTIONOFMgIIABSORBERS NIGELL.MATHES1,CHRISTOPHERW.CHURCHILL1,ANDMICHAELT.MURPHY2 DraftversionJanuary23,2017 ABSTRACT We present detailed measurements of the redshift path density, equivalent width distribution, column den- sity distribution, and redshift evolution of MgII absorbers as measured in archival spectra from the UVES spectrograph at the Very Large Telescope (VLT/UVES) and the HIRES spectrograph at the Keck Telescope 7 (Keck/HIRES)toequivalentwidthdetectionlimitsbelow0.01Å.Thissurveyexamines432VLT/UVESspec- 1 trafromtheUVESSQUADcollaborationand170Keck/HIRESspectrafromtheKODIAQgroup,allowingfor 0 detectionsofinterveningMgIIabsorbersspanningredshifts0.1<z<2.6. Weemployanaccurate,automated 2 approach to line detection which consistently detects redshifted absorption lines. We measure the equivalent n widths, apparent optical depth column densities, and velocity widths for each absorbing system. Using our a complete sample of all detectable MgII absorbers, we can accurately determine the redshift path density of J absorbersacrosscosmictime. WemeasureevolutioninthecomovingMgIIlinedensity,dN/dX,findingmore 9 high equivalent width absorbers at z=2 than at present. We also measure evolution in the equivalent width 1 distribution, parameterized by a Schechter function fit, finding a shallower weak-end slope (α) for absorbers atredshiftsbetween1.53<z<2.64,withα=−0.81 0.12,comparedtoabsorbersbetween0.14<z<0.78, ] whereα=−1.09 0.09.Finally,wecalculatethecosm±icmassfractionofMgIIusingthecolumndensitydistri- A bution,findingth±atΩ increasesfromΩ =(0.9 0.2) 10−8at<z>=0.49toΩ =(1.4 0.2) 10−8 G MgII MgII ± × MgII ± × at<z>=2.1. WefindthatweakMgIIabsorbers,thosewithequivalentwidthslessthan0.3Å,arephysically . distinct and evolve separately from very strong MgII absorbers, which have equivalent widths greater than h p 1.0Å.WecompareourobservedevolutionarytrendsinthedistributionsofMgIIabsorberstopreviouslystud- - ied cosmic trends in metallicty, the ionizing ultraviolet background, and star formation in order to conclude o thatgalaxiesejectmoremetalenrichedgasintotheirhalosaroundz=2thanatlowerredshiftsintheformof tr highequivalentwidthMgII-absorbingoutflows. Overtimefromz=2tothepresent,thesefeedbackprocesses s declineandtheevolvingconditionsinthecircumgalacticmediumgiverisetoapopulationoflowequivalent a width,passiveMgIIabsorbers. [ Keywords:galaxies: halos—quasars: absorptionlines 1 v 4 1. INTRODUCTION automatedsearches. 2 One of the most important questions in modern studies of TheoriginofMgIIabsorbinggasisstilldebated. Assum- 6 marizedinKacprzaketal.(2011)andMatejeketal.(2013), galactic evolution asks, how do baryons cycle into and out 5 two separate interpretations exist to explain the origin of of galaxies, and how does this cycle determine the growth .0 andevolutionofgalaxiesthemselves? Morespecifically,how strong,highequivalentwidth(Wr)MgIIabsorbers(Wrλ2796> 1 does the process of gas accretion, star formation, and sub- 0.3 Å) and weak, low equivalent width (Wλ2796 < 0.3 Å) r 0 sequent supernovae-driven feedback shape both the galaxies MgII absorbers. Forthestrong,higherequivalentwidthsys- 7 themselvesandtheircircumgalacticmedium(CGM)?Byus- tems,multiplecorrelationsexistbetweentherestframeMgII 1 ing spectroscopic observations of quasars, we can identify equivalent width around galaxies and the host galaxy’s star v: and analyze metal line absorbers in and around the halos of formation properties. Zibetti et al. (2007), Lundgren et al. i foregroundgalaxies.Thoughabsorptionlinestudiesbythem- (2009),Noterdaemeetal.(2010),Bordoloietal.(2011),and X selves cannot directly answer these questions, the statistical Nestor et al. (2011) all found a correlation betwen Wλ2796 r r results from such studies can provide vital information from and blue host galaxy color, showing that galaxies with more a whichfurtherprogresscanbemade. active star formation have more metal enriched gas in their One of the most prolific absorption features, the halos. Bordoloi et al. (2014) also found that MgII equiva- MgII λλ2796,2803 doublet, traces cool (T 104 K; lent width increases with increasing star formation rate den- Churchilletal.(2003))metalenrichedgasinthedis(cid:39)ksandha- sity. In addition, spectroscopic observations of star form- losofgalaxies. Itisoneofthebesttracersofthisgasbecause inggalaxieshaverevealedstrongMgIIabsorptionblueshifted it can exist in a wide range of ionizing conditions, ranging 300−1000kms−1 relativetothehostgalaxy(Tremontietal. inionizationparameterfrom−5<logU <1(Churchilletal. 2007;Weineretal.2009;Martin&Bouché2009;Rubinetal. 1999),itisobservableinopticalwavelengthsforredshiftsbe- 2010). Bouché et al. (2006) found an anti-correlation be- tween0.1<z<2.6,andithaspredictablelinecharacteristics tween galaxy halo mass, derived from the cross-correlation definedbyitsresonantdoubletnaturewhichmakeitidealfor betweenMgIIabsorptionsystemsandluminousredgalaxies, and MgII equivalent width, showing that individual clouds of a MgII system are not virialized in the halos of galax- 1NewMexicoStateUniversity,LasCruces,NM88003,UnitedStates ies. They interpreted their results as a strong indication that 2CentreforAstrophysicsandSupercomputing,SwinburneUniversity high equivalent width absorbers withWλ2796 (cid:38)2 Å arise in ofTechnology,Victoria3122,Australia r 2 MATHESETAL. galactic outflows. Marginal anti-correlations between MgII anexponentialfittotheoveralldistribution. equivalentwidthandgalaxyhalomassusingthesamecross- The most recent studies have employed new multi-object correlation method were also reported by Gauthier et al. spectrographs such as the Sloan Digital Sky Survey (SDSS) (2009)andLundgrenetal.(2009).Itisimporanttonote,how- and the FIRE spectrograph on the Magellan Baade Tele- ever,thatChurchilletal.(2013a)andChurchilletal.(2013b) scope(Nestoretal.2005;Matejek&Simcoe2012;Chenetal. findnocorrelationbetweenWλ2796 andhalomasswhenhalo 2016). Nestor et al. (2005), who examined over 1300 inter- r massisderivedfromabundancematching. Theyinsteadfind veningMgIIabsorbersinSDSSquasarspectrawithWλ2796> r thatgalaxiesinhabitingmoremassivedarkmatterhaloshave 0.3Å,foundthattheequivalentwidthdistributionfunctionis strongerabsorptionatagivendistance. well fit by an exponential. They did not find evidence for For the weak, lower equivalent width systems, it seems redshift evolution in systems with 0.4<Wλ2796 <2 Å, but none of the above correlations hold. Chen et al. (2010), r observedanenhancementinthenumberofWλ2796>2Åab- Kacprzak et al. (2011), and Lovegrove & Simcoe (2011) r sorberspercomovingredshiftpathlengthasafunctionofin- found little evidence for a correlation between galaxy color creasingredshiftfromz 0uptoz 2. Matejek&Simcoe and MgII equivalent width when restricting their samples to ∼ ∼ (2012) and Chen et al. (2016), analyzing 279 MgII absorb- weakabsorbers. Kacprzaketal.(2011)measuredtheorienta- ing systems from 2 < z < 7 in infrared FIRE spectra, also tionofgalaxiesrelativetoMgIIdetectionsinthesightlinesof found that the equivalent width distribution function is well backgroundquasarsandidentifiedlowmetallicity,lowequiv- fit by an exponential. They also observed that systems with alentwidthMgIIabsorbersco-planarwithsomesomegalaxy Wλ2796 <1.0 Å show no evolution with redshift, but higher disks,implyingstructuresassociatedwithaccretingfilaments r equivalent width systems grow in number density from low as opposed to outflows, which are more often observed per- redshift to z 3, after which the number density declines. pendiculartothegalaxydisk(Bordoloietal.2011;Kacprzak ∼ Collectively, thesesurveysimplyphysicalchangesintheas- etal.2012). Finally, thesimulationsofStewartetal.(2011) trophysical processes or in the state of the gas structures in andFordetal.(2013)revealedareservoiroflow-ionization, metalenriched,co-rotatinggasaroundmassivegalaxies. To- the environments giving rise to MgII absorption as the uni- verseages. gether, these studies imply that weak MgII absorption sys- tems may preferentially trace low metallicity infall and co- The properties of a given MgII absorbing cloud are gov- ernedbythetotalamountofgaspresent,thegasphasemetal- rotatinggasinthecircumgalacticmedium. licity, and the nature of the background radiation incident Nielsen et al. (2013b) constructed a sample of MgII ab- on the cloud. As shown by Quiret et al. (2016), studying sorbers and their associated galaxies and examined both a large sample of damped Lyα absorbers (DLAs; neutral strong and weak MgII absorbers from 0.07 z 1.1. In ≤ ≤ hydrogen absorbers with log(N(HI))>20.3) and sub-DLAs thesubsequentanalysisoftheirsample,Nielsenetal.(2013a) found a more extended MgII absorbing CGM around higher (19.0 cm−2 < log(N(HI)) < 20.3 cm−2), the average metal- luminosity, bluer, higher redshift galaxies. In addition, licity of the circumgalactic medium decreases from z=0 to inNielsenetal.(2016),theyfoundthatbluergalaxiesreplen- z=5. In addition, Ménard & Chelouche (2009) show that a ishtheirMgIIabsorbingCGMthroughoutflows,whereasred correlationexistsbetweentheneutralhydrogencolumnden- galaxies do not. Finally, in Nielsen et al. (2015), it is made sityofanabsorberandtherestframeMgIIequivalentwidth, clear that the largest velocity dispersions in MgII absorbing with stronger MgII absorbing systems having large HI col- systems are measured around blue, face-on galaxies probed umn densities. As MgII absorbers are often found associ- along their minor axis, strongly suggesting that these MgII atedwithnearbygalaxies,onecanexpectthatthemetallicity absorbersoriginateinbi-conicaloutflows. evolution of DLAs and sub-DLAs should be reflected in the Many surveys have been undertaken to inventory MgII evolutionofMgII absorbingsystems(Kulkarni&Fall2002; absorbers and examine their evolution. The earliest stud- Prochaskaetal.2003;Kulkarnietal.2005,2007). ies(Lanzettaetal.1987;Tytleretal.1987;Sargentetal.1988; Thecosmicionizingbackgroundalsochangesdramatically Steidel & Sargent 1992) found that MgII systems with rest from z = 0 to z = 2.5. Haardt & Madau (2012) showed that the slope and intensity of the diffuse UV/X-ray ioniz- equivalentwidthsabove0.3ÅshownoevolutionindN/dzbe- ing background increase as redshift increases, with a harder, tweenredshifts0.2<z<2.15. Thesestudiesalsofoundthat the equivalent width distribution function, f(Wλ2796), could more intense ionizing background present at higher redshift. r Specifically, the comoving 1 Rydberg emissivity increases befitequallywellwitheitheranexponentialorapower-law from 2 1023 ergs−1Mpc−3Hz−1 at z = 0 to 70 function. It remains uncertain whether the cosmic distribu- tion of MgII in galactic halos exhibits a fractal, self-similar 1023er∼gs−1×Mpc−3Hz−1atz=2.5.MgIIabsorbersare∼subje×ct nature, or if f(Wλ2796) flattens at equivalent widths below primarilytothisUVbackground,withverylittlecontribution r from stellar radiation from a nearby galaxy (Churchill et al. Wλ2796<0.3Å. r 1999;Charltonetal.2000;Rigbyetal.2002). MgII absorption surveys have taken one of two different Behroozi et al. (2013) showed that the cosmic star forma- approaches to try to analyze the global distribution of MgII tion rate peaks around z 2. At this point in time, galaxies absorbinggasacrosscosmictime. Churchilletal.(1999)and ∼ are on average forming stars at a rate ten times greater than Narayanan et al. (2007) aimed to determine more precisely at z=0. In conjunction with the fact that galactic-scale out- howdN/dzand f(Wλ2796)evolvewithredshiftbysurveying r flows can be driven by star formation (Zhu et al. 2015), and weakMgIIabsorbers. Theyfoundthat,fortheselowequiva- these outflows can eject MgII absorbing gas to large galac- lentwidthabsorbers,dN/dzincreasesasafunctionofincreas- tocentric radii (Sharma & Nath 2013; Kacprzak et al. 2012; ingredshiftupuntilz=1.4. Athigherredshifts, dN/dzfalls Nestor et al. 2011), it follows that more metal enriched gas, to lower values, though the uncertainties are large. In addi- tracedbyMgIIabsorbers,shouldbedrivenoutofgalaxiesat tion,theyfoundtheequivalentwidthdistributionfunctionfor z 2. OnegoalofTheVultureSurveyistofindandanalyze weakabsorbersisbestfitbyapower-law,stronglydisfavoring ∼ definitiveobservationalsignaturesofthisenergeticepoch. THEVULTURESURVEYI 3 Wenowaimtobetterunderstandthecomplexrelationship whichdiscoveredMgIIabsorbersdidnothavethesensitivity betweenabsorbinggasintheCGM/IGMandthephysicalpro- todetectweak,lowequivalentwidthsystems,quasarspectra cessesshapinggalaxyformationastheuniverseages. Forour wereneverselectedbaseduponthepresenceofweakabsorp- survey,wewillanalyzethelargest,mostcomprehensivesam- tion. However,somespectrawereselectedbasedonthepres- pleofhighresolution,highS/N quasarspectratouniformly enceofstrong,Wλ2796>0.3Åsystems. Inordertoproperly r observe both strong and weak MgII absorbers. We hope to quantifyanybias,wecompareourworktothemassive,unbi- finally bridge the equivalent width dichotomy in prior MgII asedsampleoftheSloanDigitalSkySurvey(SDSS).Specifi- absorption line surveys by analyzing large numbers of both cally,weturntotheworkofNestoretal.(2005)whoinvento- strongandweakabsorbers. Todoso,wewillexaminequasar riedstrongMgIIabsorbersinSDSSquasarspectra. InNestor spectra observed with either the VLT/UVES (Dekker et al. etal.(2005),theauthorsconstructedtheequivalentwidthfre- 2000) or Keck/HIRES (Vogt et al. 1994) spectrographs. We quency distribution for absorbers with Wλ2796 >0.3 Å and aimtocharacterizetheevolutioninthenumberdensityofall r foundthebestfittothisdistributionwasanexponentialfunc- MgIIabsorbersfrompresenttobeyondthepeakofthecosmic tionoftheform, star formation rate. We interpret these results in the context ofglobalevolutioninmetallicityaroundgalaxies,theionizing N∗ background,andcosmicstarformation. f(Wrλ2796)=W∗e−(Wrλ2796/W∗) (1) We begin by explaining the methods of acquiring and an- alyzing the quasar spectra in Section 2. In Section 3, we where N∗ andW∗ are constants. In order to compare to the present the results showing the evolution of the MgII equiv- SDSSdata,welimitoursampletoabsorberswithequivalent alent width distribution, dN/dX, and the MgII column den- widthsWλ2796>0.3Åandcalculatetheequivalentwidthfre- r sity distribution across redshift. We also analyze the func- quency distribution. In Figure 1(a), we show f(Wλ2796) for r tional fit to both the equivalent width and column density The Vulture Survey and an exponential fit to our data, along distributions. In Section 4 we discuss the redshift evolution with the exponential fit of Nestor et al. (2005). Nestor et al. of all types of MgII absorbers and derive the relative mat- (2005)foundthebest-fitparametersandcorresponding1σun- ter density contributed to the universe by MgII, ΩMgII. In certaintiestobeN∗=1.187 0.052andW∗=0.702 0.017. Section 5 we summarize our results and look to future stud- When we perform the sam±e analysis, fitting Equat±ion 1 to ies using this rich data set, including a companion analysis The Vulture Survey data, we derive N∗ = 1.39 0.10 and of intervening CIV absorbers and detailed kinematic analy- W∗ =0.74 0.05. By eye, there appears a sligh±t excess in sis of intervening absorbing systems. For all calculations, thenumber±ofMgIIabsorbersaboveWλ2796>3Å. weadoptthemostrecentlypublishedPlanckcosmology,with r In order to statistically determine the bias in our sample, H0=67.81kms−1Mpc,ΩM=0.308,andΩΛ=0.692(Planck we perform a Kolmogorov–Smirnov (KS) test to quantita- Collaborationetal.2016). tivelymeasurethesimilaritybetweenoursampleofabsorbers 2. DATAANDANALYSIS withWrλ2796 >0.3 Å and the SDSS sample of Nestor et al. (2005). We first sample a population of 1331 measured ab- 2.1. QuasarSpectraSample sorber equivalent widths, matching the SDSS sample size, We have assembled a sample of 602 archival quasar spec- from the exponential fit to the equivalent width ditribution tra observed with the VLT/UVES and Keck/HIRES spectro- of Nestor et al. (2005), incorporating the reported 1σ scatter graphs. The data originate from two archival data mining inthefitparameters.Thisallowsustodirectlycomparetothe efforts - the UVES SQUAD collaboration (432 spectra) led sampleofmeasuredMgIIequivalentwidthsfromTheVulture by Michael Murphy, and the KODIAQ Survey (170 spec- Survey. In Figure 1(b), we show the cumulative distribution tra) led by John O’Meara (O’Meara et al. 2015). The spec- ofthestrongabsorbersinTheVultureSurveyinblack,along tra range in signal-to-noise ratio (S/N) from 4 to 288 per withtheexponentialfittoourdataandthefitofNestoretal. 1.3−2.5kms−1 pixel,withthepixelsizedependentuponthe (2005). We then perform a two-sample KS test, calculating resolution of the spectrum. The mean S/N for the sample theP-value,whichistheprobabilitythatthetwosamplesare is 38 per pixel. Quasar emission redshifts span 0.014<z< drawnfromdifferentunderlyingdistributions,definedasP(K- 5.292. Wavelength coverage for each spectrum varies based S).ToavoidissuesduetotherandomresamplingoftheNestor upon the settings used for each spectrograph. VLT/UVES etal.(2005)distribution, werepeatthisexerciseonemillion has 3 CCD chips available, offering large wavelength cover- times as a Monte-Carlo method, generating an ensemble of agefrom 3000to 10000Å.However,wavelengthcover- P(K-S) values. We take the mean P(K-S) value as the true age availa∼ble for eac∼h quasar spectum varies based upon the probabilitythatthesamplesareinconsistent,andthestandard selected cross-disperser settings. The exposures used from deviationaboutthismeanasthe1σuncertainty. Ourcriterion Keck/HIRESweretakenfrom2004topresent,whena3chip toassertthatoursampleisnotinconsistentwithanunbiased CCD mosiac was installed, also allowing wavelength cov- sample requires P(K-S)>0.0027, which means that it could not be ruled out at the 3σ level that the two populations are erage from 3000 to 10000 Å. Again, though, individ- ∼ ∼ consistent with one another. With P(K-S)= 0.112 0.049, ual quasar observations vary in wavelength coverage based ± and only one instance out of the million Monte-Carlo runs upon cross-disperser angle. We detect 1180 MgII absorb- exhibitingaP(K-S)valuebelow0.0027,weconcludethatour ing systems from 0.14 < z < 2.64 to a detection limit of sample is not inconsistent with an unbiased sample, even at Wλ2796 0.01ÅforregionswithS/N>40perpixel. r (cid:39) the2σlevel,andthatthestrongabsorbersinTheVultureSur- The archival quasar spectra used to construct The Vulture veycouldverywellhaveoriginatedfromthesameunderlying Survey were observed for a multitude of reasons. In cases populationastheunbiasedSDSSquasarspectrasample. where observations were taken to target a previously known absorptionlinesystem,thesedetectionscouldbiasoursample 2.2. DataReductionandLineDetection when calculating dN/dz and dN/dX. Because early studies 4 MATHESETAL. 101 1.2 (a) Nestoretal.2005:W∗=0.702±0.017,N∗=1.187±0.052 Vulturesurveyfit (b) FittoData:W∗=0.74±0.05,N∗=1.39±0.10 Nestoretal.2005fit 100 VultureSurvey,Wrλ2796≥0.3A˚ 1.0 Vulturesurvey,Wrλ2796≥0.3A˚ n o W uti0.8 b /∆ 10−1 stri dNdz Di0.6 P(K-S)=0.112 0.049 = e ± v W) 10−2 ati f( mul0.4 u C 10−3 0.2 10−40 1 2 3 4 5 6 0.00 1 2 3 4 5 6 Wλ2796[A˚] Wλ2796[A˚] r r Figure1. (a)TheWλ2796 0.3ÅequivalentwidthfrequencydistributionforTheVultureSurveyinblackcomparedtotheexponentialfitofNestoretal. (2005),shownastherorange≥dashedline,andanexponentialfittoTheVultureSurveydatainpurple.TheexponentialfitsareoftheformshowninEquation1. (b)ThecumulativedistributionofTheVultureSurveydataandtwocomparativeexponentialfits.TheP(K-S)valueshowncomparesoursurvey’sdatatothefit fromNestoretal.(2005). TheKODIAQdatasampleisreducedandfullycontinuum didates detected above a certain S/N threshold. The filter fit,deliveredasnormalizedspectraaccordingtotheprescrip- is a top hat function centered at the wavelength of the de- tionsofO’Mearaetal.(2015). Tosummarize,observingruns siredredshiftedabsorptionline. Itswidthisselectedtomatch aregroupedtogetheranduniformlyreducedusingHIRedux3 theresolutionofthespectrum,whichisafunctionoftheslit as part of the XIDL4 suite of astronomical routines in IDL. widthusedduringtheexposure. Alargevarietyofslit-widths ContinuumfitsareappliedoneorderatatimeusingLegendre wereusedtoachievedifferentresolutionsforvaryingscience polynomialsbyasinglememberoftheKODIAQteam,John drivers,but,characteristicallyfora1.0arcsecslit,R 40,000 O’Meara,tominimizebiasandvariation. for VLT/UVES and R 45,000 for Keck/HIRES.∼We con- ∼ The UVES SQUAD sample also comes reduced and con- volve the filter with the normalized spectrum to generate a tinuumfitaccordingtotheprescriptionsofKingetal.(2012); normalizedpowerspectruminredshiftspace,withabsorption Bagdonaite et al. (2014); Murphy et al. (2016); Murphy (in featureshavingpositivepower. prep). Reduction was carried out using the ESO Com- The error spectrum in both instruments is complex, irreg- mon Pipeline Language data-reduction software.5 The con- ular, and has frequent single-pixel spikes which makes uni- tinuum is fit automatically with a low order polynomial in form normalization impossible. Therefore, we cannot con- smallsectionsusingUVES_popler,anESO/VLTUVESpost- volve the filter with the error spectrum to derive normalized pipeline echelle reduction program written by Michael T. noise estimates, as is often done in matched filter analysis. Murphy(Murphy2016). Thisfitcanincorrectlyestimatethe Instead,weexaminethenoiseinthederivedpowerspectrum. continuumaroundnarrowemissionregionsandbroadabsorp- To derive the noise, we first sigma-clip chunks of the power tion features. Using UVES_popler, we add a higher order spectrumtoremoveabsorptionfeatures,leavingonlythecon- continuum fit to such regions of each spectrum, always pre- tinuumpowerspectrum. Next, wecalculatethestandardde- serving the continuity of the continuum with non-absorbing viation of this continuum. Finally, we use the standard de- regions. viation as the noise to calculate the S/N of the absorption The next step involves detecting all MgII absorption fea- featuresinthepowerspectrumastheratioofthenormalized tures. We first limit the search range to regions of the spec- power(S)tothenormalizednoise(N). Aflaggedabsorption trum redward of the Lyα emission, as Lyα forest contam- feature has S/N > 5. A confirmed doublet detetection for ination would render automatic detection of weaker metal MgII λλ2796,2803 requires detection of S/Nλ2796 >5 and lines nearly impossible. We also do not search 5000 kms−1 S/Nλ2803 >3. In addition, our automated routines remove blueward of the quasar emission redshift in order to avoid detectionswithnon-physicaldoubletratiosinunsaturatedre- absorbers associated with the quasar itself. Finally, we ex- gions;specifically,weexcludecaseswhereWλ2803>Wλ2796, cluderegionsofstrongtelluricabsorptionbands,specifically orWλ2803 <(cid:0)0.3 Wλ2796(cid:1). The latter consrtraint is cornser- from 6277−6318 Å, 6868−6932 Å, 7594−7700 Å, and r × r 9300−9630Å,becausewefoundthatthemolecularlinesep- vativeforunsaturatedMgIIabsorbers,asWrλ2803israrelyob- arationsandratioscanleadtonumerousfalsepositiveswhen servedlessthan0.5×Wrλ2796. Werelaxthisconstraintinsat- searchingforMgIIdoublets. uratedfeatures. Thissystemcouldpotentiallyexcludedetec- To find all intervening MgII λλ2796,2803 absorbers, we tionswhereeithertheMgII2796orMgII2803lineisblended employ a techinque outlined in Zhu & Ménard (2013), in withanothertransitionbutdoesnotsaturate;however,confir- whichweperformamatchedfiltersearchforabsorptioncan- mationofthesecasesrequiresextraverificationfromseparate absorbing features, such as FeII, which are weaker and not alwayscoveredinthespectrum. 3http://www.ucolick.org/xavier/HIRedux/ 4http://www.ucolick.org/xavier/IDL/index.html All absorption features are visually verified upon comple- 5http://www.eso.org/observing/dfo/quality/UVES/pipeline/pipe_reduc.html tion of the detection algorithm. Multiple feature detections THEVULTURESURVEYI 5 within 500 kms−1 of each other are grouped together to ± generateabsorptionsystems,designatedasasingleabsorber, (cid:88) to be analyzed. Once absorption systems are identified, we g(Wj,zk)= H(zk−zmnin)H(zmnax−zk)H[Wj−5σk/(1+zk)], calculate the optical depth-weighted median absorption red- n (2) shifttodefinethecenteroftheentireabsorptionsystem. The whereH istheHeavisidestepfunction,zmin andzmax arethe formalderivationofthisredshiftisdescribedintheappendix n n minimumandmaximumredshiftsobservedforthenthquasar ofChurchill&Vogt(2001). spectrum,andwherethesumextendsoverallquasarspectra We also derive an equivalent width detection limit across in the sample. This heat map details the number of spectra the spectrum. To do so, we insert modelled Gaussian ab- sorptionfeaturesacrossthespectrumandassumeafull-width inwhichaMgIIλλ2796,2803doubletcouldbedetectedasa function of the equivalent width detection limit and redshift. at half maximum (FWHM) defined by the resolution of the The vertical stripes with no redshift path coverage represent instrument to represent unresolved lines. We then solve for the omitted telluric absorption regions for our survey. The theheightoftheGaussian,definedasthevalueatthecurve’s integralalongagivenW2796 slicegivesthetotalredshiftpath peak,requiredtodetecttheunresolvedlinewithourmatched r filteringtechniqueataS/N=5. Finally,weintegratetofind lengthavailableforthesample(∆Z). theequivalentwidth,andtakethatvalueastheminimumde- tectableequivalentwidthatagivenwavelength.Thedetection 3.3. dN/dzanddN/dX algorithm is therefore self-monitoring. This full equivalent The largest sample of quasar spectra originates from the width detection limit spectrum also allows us to accurately Sloan Digital Sky Survey (SDSS), with more than 105 spec- characterize the completeness of our sample, along with the tra at present, which employs a spectrograph with an instru- fullredshiftpathlengthsearched. mental resolution around 69 kms−1, limiting SDSS absorp- tionsurveystostrongabsorbers,withWλ2796 0.3Å(Nestor 2.3. MeasuringAbsorptionProperties r ≥ etal.2005;Zhu&Ménard2013). Conversely,previousstud- For each absorption system, we automatically define the ies of weak absorbers used small samples of quasar spectra, wavelength bounds of an absorbing region by finding where neverexceeding100quasarspectra(Steidel&Sargent1992; the flux recovers to within 1σ of the continuum, with σ de- Narayananetal.2007;Kacprzaketal.2011).Inthispaper,we fined by the error spectrum, for three pixels on either side aimtocharacterizetheevolutionoftheincidencerate,number of the absorption trough. Within these regions we calcu- of absorbers per redshift path length, comoving line density, late rest-frame equivalent widths (Wr), velocity widths (∆v), andcosmicmassdensityofallMgIIabsorbersfromredshifts optical depth-weighted kinematic spreads (ωv), apparent op- 0.18<z<2.57. tical depth (AOD) column densities (log(N)), and absorp- ThenumberofMgIIabsorbersperredshiftpathlengthand tion asymmetries. The functional forms of these parame- itsassociatedvariancearedefinedas tersaredetailedintheappendixofChurchill&Vogt(2001), equationsA3−A7. dN (cid:88) 1 (cid:88)(cid:104) 1 (cid:105)2 = , σ2 = , (3) dz ∆Zi(Wr) ddNz ∆Zi(Wr) 3. RESULTS i i 3.1. SampleCharacterization where we count the number of MgII absorbers, dividing by thetotalsearchedredshiftpathlength(∆Z),definedas Figure2showstherelationshipsbetweenthemeasuredab- sorptionparameters,characterizingthedistributionofabsorp- (cid:90) z2 tion properties for our survey. With redshift, there are no ∆Z (W)= g (W,z)dz, (4) i r i r obvious trends other than the highest equivalent width ab- z1 sorbers, withW2796 >4 Å, existing mainly at z>1.5. The r where gi(Wr,z) is the equivalent width sensitivity function data gaps at z = 1.7 and z = 2.4 represent the larger omit- at a given equivalent width detection limit shown in Equa- ted search regions which overlap with the stronger telluric tion2.Thefunctiong(W,z),firstformulatedinLanzettaetal. r absorption bands. With column density, we see the normal (1987), detailsthenumberofspectrainwhichanabsorption trends of higher column density systems exhibiting higher featurewithagivenequivalentwidthmay bedetectedatthe equivalentwidthsandvelocityspreads,withthedistributions 5σlevelinagivenredshiftinterval. asymptoting near logN 15 cm−2 due to saturation effects ThecomovingMgIIlinedensityanditsassociatedvariance (cid:39) and the nature of measuring column densities with the AOD aredefinedas method. Measured column densitites of saturated lines are lower limits. With respect to kinematic spread, we observe dN (cid:88) 1 (cid:88)(cid:104) 1 (cid:105)2 the saturation line in the ωv vs. Wr2796 relationship, showing dX = ∆Xi(Wr), σ2ddNX = ∆Xi(Wr) , (5) the maximum ω for a flat-bottomed absorption profile of a i i v givenequivalentwidth. where we count the number of MgII absorbers, dividing by thetotalsearchedabsorptionpath(∆X),definedas 3.2. SampleCompletenessandSurveyPathCoverage Toevaluatethecompletenessandcalculatetheredshiftpath (cid:90) z2 (1+z)2 ∆X (W)= g (W,z) dz, (6) coverage of our survey, we use the derived 5σ equivalent i r i r (cid:112) width detection limit described at the end of Section 2.2 to z1 ΩM(1+z)3+ΩΛ determinethenumberofspectrainwhichwecoulddetectan where Ω is the cosmic matter density, and Ω is the cos- M Λ absorber of a given equivalent width and redshift. Figure 3 micdensityattributedtodarkenergy. Countingwithrespect showsthefunctiong(Wλ2796,z),definedas to ∆X accounts for both cosmological expansion along the r 6 MATHESETAL. 15 14 N 13 g o l 12 11 400 300 v ω 200 100 0 8 6 9 7 6 2 λr 4 W 2 0 0.5 1.0 1.5 2.0 2.5 11 12 13 14 15 0 100200300400500 z log N ω v Figure2. CorrelationsbetweenmeasuredabsorptionpropertiesforTheVultureSurvey. logN istheMgIIAODcolumndensity,ωv isthekinematicspread, Wr2796istherestframeMgII2796equivalentwidth,andzistheabsorptionredshift. lineofsightandthetransverseseparationofobjectswithun- mospheres.Thefullproductof c n σ thenrepresentstheco- changingnumberdensityandcrosssection,allowingformore movingHubbleopticaldepthforHoMg0II0absorbers.Wefindthat consistentcomparisonsacrossredshift. thebest-fitvalueof(cid:15)isnegativewhenanalyzingthefullsam- In Figure 4, we plot dN/dz and dN/dX, respectively, as a pleofMgIIabsorbers,includingalldetectionswithmeasured function of redshift for different minimum equivalent width equivalentwidthsaboveWλ2796>0.01Å.Theevolutionpa- thresholds,suchthatdetectedMgIIabsorbershaveequivalent rameter,(cid:15),thenincreaseswrithsubsequentlylargerminimum widths greater thanWrλ,m2i7n96. Error bars in each bin represent equivalentwidththresholds,becomingpositiveforabsorbers 1σ uncertainties calculated according to Equations 3 and 5. withWλ2796>1.0Å.Thistrendisdrivenprimarilybyanen- Dottedlinesarefitaccordingtotheanalyticalformwhichal- r,min hancementindN/dX forthestrongestMgIIabsorbersaround lowsforredshiftevolutionindN/dX,definedas, z 2, relativetolowerredshifts. Conversely, atlowredshift ∼ c c we observe more weak MgII absorbers per absorption path f(z)≡ H n(z)σ(z)= H n0σ0(1+z)(cid:15), (7) length. We show in Table 1 the values of the fit parameters o o forvaryingWλ2796,alongwiththeir1σuncertainties. where c is the speed of light, Ho is the Hubble Constant, n0 InFigure5,r,mwineshowthevaluesof c n σ and(cid:15)asafunc- is the comoving number density of MgII absorbers at z=0, Ho 0 0 σ0 istheabsorbing cross-sectionatz=0, and(cid:15) istheevolu- tionofWrλ,m2i7n96.Theshadedredareasrepresentthe1σstandard tionparameter,definedasthepowerdependenceofdN/dXon deviationsderivedfromthefitstothedN/dX distribution.We redshift. Theproductn0σ0 representsacomovingopacityof showfirstthatthecomovingHubbleopticaldepthofMgIIab- MgII-selectedabsorptionlinesystemsbyvirtueoftheunits, sorbersdecreasesasafunctionofWλ2796. Thisimpliesthat, r which are an inverse path length, and the analogous absorp- per unit absorption path length, there are fewer high equiva- tioncoefficienttodescribetheopacityofmaterialinstellarat- THEVULTURESURVEYI 7 g(Wλ2796, z) r 0.10 ) A ˚ ( t 0.08 i m i L n 0.06 o i t c e t 0.04 e D 6 9 7 0.02 2 λ r W 0.5 1.0 1.5 2.0 2.5 z 0 50 100 150 200 250 300 350 400 Number of Spectra Figure3. Thefunctiong(Wλ2796,z)shownasaheatmapwiththecolorsrepresentingthevalueofg(W2796,z). Thisisthenumberofspectrainwhichan r r absorptionlineofagivenequivalentwidthandagivenredshiftmaybedetectedaccordingtothedetectionlimitofthespectrum. Theverticalblackbars representingnoredshiftpathlengthcoverageshowtheomittedwavelengthregionsofthesurveybaseduponcontaminatingtelluricabsorptionfeatures. sorberswithequivalentwidthsbetween0.3<Wλ2796<1Å. r Table1 Weprovideaparameterizedfitto c n σ and(cid:15)asafunc- ParameterizationofdN/dX Ho 0 0 tionofWλ2796. InFigure5(a), weadoptapower-lawwitha r,min Wr2,m79in6 Hcon0σ0 (cid:15) generalizedexponentialdecaytomodelthe Hcon0σ0 distribu- [Å] tion,definedas, 0.01 2.583 0.827 -1.04 0.38 c ± ± n σ (ψ)=Ψ∗(ψ)αe−ψβ, (8) 0.30 0.446 0.076 -0.14 0.21 H 0 0 ± ± o 1.00 0.116 0.043 0.31 0.44 0.01 0.019±±0.014 0.94±±0.85 fiwthepraeraψm=etWerrsλ,m2ai7nr9e6/WΨr∗∗,m=in0to.2s4imp0li.f0y1,thWe e∗qua=tio1n..19The0b.0e2st, ± r,min ± α=−0.49 0.01,andβ=1.50 0.05. Thisparameterization ± ± lent width MgII absorbers, and/or that they exist in smaller resemblesaSchechterfunction,butwerequiredanexponen- absorbing structures. We also show that the slope of the tial drop-off at the high end faster than e−x, which manifests redshift dependence, (cid:15), increases as a function of increasing itself in the form of β. Next, in Figure 5(b), we fit a broken Wλ2796. This evolution parameter, (cid:15), changes from negative power-lawtothe(cid:15)distribution,definedas, r,min to positive toward higher equivalent width MgII absorbers, implyingthatstrongMgIIabsorbersevolveaway,decreasing a1(Wrλ,m2i7n96)γ1+b1 if Wrλ,m2i7n96<1.1Å in relative number per absorption path length, from z=2 to (cid:15)(Wλ2796)= (9) r,min present.Conversely,weakMgIIabsorbersbuildupovertime, a (Wλ2796)γ2+b if Wλ2796 1.1Å, increasinginrelativenumberperabsorptionpathlengthfrom 2 r,min 2 r,min ≥ z=2topresent. Weobservenoevolutionwithredshiftinab- where the fit parameters for minimum equivalent width 8 MATHESETAL. (a) (b) (cid:15)=-0.18 100 (cid:15)=0.74 (cid:15)=-1.04 100 (cid:15)=1.16 (cid:15)=-0.14 z X (cid:15)=0.31 dN/d (cid:15)=1.67 dN/d10−1 (cid:15)=0.94 10−1 Wrλ2796≥0.01A˚ Wrλ2796≥0.01A˚ WWrrλλ22779966≥≥01..30AA˚˚ 10−2 WWrrλλ22779966≥≥01..30AA˚˚ 10−02.0 0.5 1.0 1.5 2.0 2.5 Wrλ2739.60≥2.0A˚ 3.5 0.0 0.5 1.0 1.5 2.0 2.5 Wrλ2739.60≥2.0A˚ 3.5 z z Figure4. (a)dN/dzand(b)dN/dXasafunctionofredshiftfordifferentminimumequivalentwidththresholds,Wλ2796.ColorsrepresentdifferentWλ2796.The r,min r,min blackdottedlinesarefitstothedistributionofthefunctionalform f(z)= Hcon0σ0(1+z)(cid:15),withthebestfit(cid:15)valuelabelled. Weseeincreasingvaluesof(cid:15)with increasingequivalentwidth,drivenbyanenhancementofstrongerMgIIabsorbersaroundredshift2comparedtolowerredshifts. Verticalerrorbarsrepresent 1σuncertaintiesineachbin. 4 (a) (b) 3 100 2 nσ00 10−1 (cid:15) 1 cHo 0 10−2 1 − 2 10−2 10−1 100 − 10−2 10−1 100 Wλ2796 xA˚ Wλ2796 xA˚ r ≥ r ≥ Figure5. (a)Thecomovingnumberdensityofabsorbersmultipliedbytheabsorbingcross-section,derivedbyfittingEquation7todN/dX,asafunctionof Wrλ,m2i7n96withshaded1σuncertainties.AsweexaminesampleswithincreasingminimumMgIIequivalentwidththresholds,eitherthespacedensityofabsorbing cloudstructuresdecreases,theabsorbingcross-sectiondecreases,orbothparametersdecrease.(b)Theredshiftevolutionparameter,(cid:15),asafunctionofWλ2796. r,min WeakMgIIabsorbersaremoreabundantatlowredshift,leadingtoanegativecoefficient(cid:15). Absorberswithequivalentwidthsnear0.3Ådonotevolve,with (cid:15) 0.StrongMgIIabsorbersevolveawayatlowredshift,showingalargepositive(cid:15)increasingtowardsz 2. (cid:39) ∼ thresholdsbelow1.1Åarea =3.13 1.49,γ =0.09 0.05, To calculate the equivalent width frequency distribution 1 1 and b =2.87 1.49. The fit param±eters for the pow±er-law f(W), the number of absorbers of a given equivalent width 1 withWλ2796 ±1.1 Å are a =0.07 0.01, γ =3.73 0.25, perunitpathdensity,wecalculatedN/dzordN/dX foreach andb =r,m−in0.2≥1 0.03.Com2biningth±efitsto c2n σ an±d(cid:15),we equivalent width bin and divide by the bin width. We split 2 ± Ho 0 0 thesampleintofourredshiftregimes,ensuringthatthenum- nowhaveananalyticparameterizationofdN/dXasafunction berofabsorbersineachredshiftsubsampleremainsconstant. ofWλ2796,andz,oftheform, r,min The result is a characteristic number of MgII absorbers per redshiftorabsorptionpathlengthperequivalentwidth. dN dX(Wrλ,m2i7n96,z)=Ψ∗ψαe−ψβ(1+z)(cid:15)(Wrλ,m2i7n96). (10) buItnionFigwuirthe6re,swpeecptltootethiteheerqudiNva/ldezntowriddNth/dfrXeq.uWenecyfitdiesatrcih- distributionwithaSchechterfunctionoftheform, This function can be used in future semi-analytic models to parameterize the physical properties of MgII absorbers in (cid:18)W (cid:19)α galaxyhalos(e.g.Shattowetal.(2015)). Φ(Wr)=Φ∗ Wr∗ e−Wr/Wr∗, (11) r 3.4. EquivalentWidthFrequencyDistribution whereΦ∗ isthenormalization,αisthelowequivalentwidth THEVULTURESURVEYI 9 103 (a) (b) 102 102 101 101 W W ∆ ∆ 100 / / dNdZ 100 dNdX )= )= 10−1 f(W 10−1 WWW∗∗∗===122...730695,,,ααα===---101...090993 f(W 10−2 WWW∗∗∗===122...730260,,,ααα===---101...090992 10−2 Wz=∗[=0.124.2-2,0α.78=]-0.80 Wz=∗[=0.124.2-5,0α.78=]-0.81 z=[0.78-1.09] 10−3 z=[0.78-1.09] z=[1.09-1.53] z=[1.09-1.53] 10−3 z=[1.53-2.64] z=[1.53-2.64] 10−2 10−1 100 101 10−4 10−2 10−1 100 101 Wλ2796[A˚] Wλ2796[A˚] r r Figure6. (a)TheequivalentwidthdistributionofMgIIabsorbers,definedastheredshiftpathdensity(dN/dz)ineachequivalentwidthbindividedbythebin width. (b)Theequivalentwidthdistribution,definedasthecomovinglinedensity(dN/dX)ineachequivalentwidthbindividedbythebinwidth. Errorbars represent1σuncertaintiesineachbin.WefiteachdistributionwithaSchechterfunction,capturingtheself-similarpower-lawbehaviorofweakMgIIabsorbers andtheexponentialpower-lawcutoffwhenobservingthestrongestMgIIsystems. Table2 Table3 SchechterFitto f(W)= dN/∆W SchechterFitto f(N)= dN/∆N dX dX RedshiftRange Φ∗ W∗ α RedshiftRange Φ∗ N∗ α [Å] [ 10−16] [ 1014cm−2] × × 0.14−0.78 0.15 0.10 1.72 0.68 −1.09 0.09 0.14−0.78 8.79 8.19 1.71 0.97 −1.14 0.08 ± ± ± ± ± ± 0.78−1.09 0.12 0.06 2.36 0.81 −0.99 0.06 0.78−1.09 6.41 2.43 2.29 0.56 −1.10 0.03 ± ± ± ± ± ± 1.09−1.53 0.11 0.03 2.00 0.35 −1.02 0.04 1.09−1.53 4.15 2.40 2.65 1.00 −1.12 0.04 ± ± ± ± ± ± 1.53−2.64 0.12 0.08 2.25 0.87 −0.81 0.12 1.53−2.64 7.98 6.51 2.35 1.32 −0.91 0.08 ± ± ± ± ± ± 15 cm−2, the measured column densities are lower limits as power-lawslope,andW∗istheturnoverpointinthedistribu- r theAODmethodcannotconstrainthetruecolumnwhenthe tion where the low equivalent width power-law slope transi- absorptionlinebecomessaturated. tionsintoanexponentialcutoff. Table2showsthevaluesof InFigure7,weplotthecolumndensityfrequencydistribu- Φ∗,W∗, and α, along with their associated 1σ uncertainties r tion using either dN/dz or dN/dX. Again, we fit this distri- derived from the fitting routine. This functional fit is moti- bution with a Schechter function of the same form as Equa- vatedbypaperssuchasKacprzak&Churchill(2011),where tion 11, except with equivalent width replaced with column theauthorsseektocombineprevioussurveysofstrongMgII density.Table3showsthevaluesofΦ∗,N∗,andα,alongwith absorbers, in which exponential fits were preferred, and sur- their associated 1σ uncertainties. We find again that the low veys of weak MgII absorbers, where power-laws best fit the columndensityslopeisshallowernearz 2thanatz 0.5. equivalent width distribution. The power-law nature of the ∼ ∼ Duetosaturationeffects,thehighestcolumndensitymeasure- distributionofweakabsorbersinoursurveyisapparent,and mentsarelowerlimits; therefore, thefinalhighcolumnden- the exponential cutoff is motivated by physical limits to the sitybinin f(N)shouldberegardedasanupperlimit. These size, density, and velocity widths of MgII absorbing clouds. limits are taken into account in the functional fitting proce- Examining the distribution as a function of redshift, we find dures. thelowequivalentwidthslopebecomesmoreshallowatz 2 ∼ comparedwiththepresentepoch,withα=−1.09inoursub- 3.6. Ω samplewith0.14 z<0.78andα=−0.81inoursubsample MgII with1.53<z 2.≤64.WeobservefewerweakMgIIabsorbers We now aim to calculate the matter density of MgII ab- and more stro≤ng MgII absorbers per redshift/comoving ab- sorbersacrosscosmictime. Todoso,weemploythefollow- sorptionpathlengthatz 2thanwedoatz 0.5. ingcustomaryequationrelatingthemassdensityofanionas ∼ ∼ a fraction of the critical density today to the first moment of 3.5. ColumnDensityDistribution thecolumndensitydistribution, To calculate the column density distribution, the number ofabsorbersofagivencolumndensityperunitpathdensity, Ω = H0mMgII(cid:90) Nmax f(N )N dN , (12) we calculate dN/dz or dN/dX for each column density bin MgII cρ MgII MgII MgII and divide by the bin width. The result is a characteristic c,0 Nmin numberdensityof MgII absorbersperredshift orabsorption whereH0istheHubbleconstanttoday,mMg=4.035 10−23g, pathlengthasafunctionoftheircolumndensities. Itshould c is the speed of light, ρ is the critical density a×t present, c,0 be noted that at high column densities near log(N(MgII))= f(NMgII)isthecolumndensitydistributionofMgIIabsorbers, 10 MATHESETAL. 10−11 (a) 10−11 (b) 10−12 10−12 N 10−13 N 10−13 ∆ ∆ dN/dz10−14 dN/dX10−14 = = N) 10−15 N∗=1.84e+14,α=-1.14 N) 10−15 N∗=1.71e+14,α=-1.14 f( 10−16 NNN∗∗∗===222...763782eee+++111444,,,ααα===---110...119020 f( 10−16 NNN∗∗∗===222...263955eee+++111444,,,ααα===---110...119021 z=[0.14-0.78] 10−17 z=[0.14-0.78] 10−17 z=[0.78-1.09] z=[0.78-1.09] zz==[[11..0593--12..5634]] 10−18 zz==[[11..0593--12..5634]] 10−18 1011 1012 1013 1014 1015 1011 1012 1013 1014 1015 N(MgII) N(MgII) Figure7. (a)ThecolumndensitydistributionofMgIIabsorbers,definedastheredshiftpathdensity(dN/dz)ineachcolumndensitybindivididedbythe binwidth. (b)Thecomovinglinedensity(dN/dX)ineachcolumndensitybindividedbythebinwidth. WefitthisdistributionwithaSchechterfunctionto accuratelyparameterizethelowcolumndensitypower-lawslopeandtheexponentialcutoffandhighcolumndensities. andN isthecolumndensity. Usingourderivedfittothe 4. DISCUSSION MgII column density distribution, we are able to numerically in- WehaveshownacosmicinventoryofMgII absorbinggas tegrate the first moment from 0 < logN(MgII) < 20 cm−2. from 0.1<z<2.6, measuring dN/dz, dN/dX, the equiva- The results are shown below in Figure 8. 1σ uncertainties lent width distribution, and the column density distribution arederivedwithabootstrapMonte-Carlomethod. Weselect downto detectionlimitsas lowasWλ2796=0.01Å. Weaim random column densities, with replacement, from the sam- r nowtorelatethepropertiesofMgII absorbersandtheirevo- ple of measured column densities for all of our MgII ab- lution across cosmic time to other known evolutionary pro- sorbers until we reach the sample size. We then recalculate cesses with the hope of gaining insight into the mechanisms the column density distribution, find the best parameterized whichgiverisetoMgIIabsorbinggas. Schechter fit, and then integrate and compute Equation 12. We perfom this task 1499 times to develop a statistical en- 4.1. EvolutionofMgIIDistributions semble of values for Ω, with this number of samples repre- sentingtheunderlyingscatterintheΩ distributionatthe Narayanan et al. (2007) measured the evolution of weak MgII 99% confidence level according to Davidson & MacKinnon MgII absorbers from 0.4<z<2.4 in VLT/UVES spectra. (2000). Wetakethestandarddeviationaboutthemeanofthis They compared to Churchill et al. (1999), who fitted the ensembleofsimulatedmeasurementsasthe1σuncertaintyin equivalentwidthfrequencydistributionwithapower-law,and ΩMgII.WefindthatthecosmicmassdensityofMgIIincreases toNestoretal.(2005),whofittedanexponentialto f(Wr). In from Ω 0.8 10−8 at z 0.5 to Ω 1.3 10−8 at thecaseofweakabsorbersatz<1.4,Narayananetal.(2007) z 2. MgII (cid:39) × ∼ MgII (cid:39) × found that a power-law with a slope of α=−1.04 is a sat- ∼ isfactoryfit,confirmingtheresultsofChurchilletal.(1999). Whentheysplittheirsampleintolowredshift,withdetections 2.0 ΩMgII between0.4<z<1.4,andhighredshift,withdetectionsbe- tween1.4<z<2.4, theyfoundthatthelowredshfitsample remained consistent with a power-law but the high redshift samplewasbestfitbyanexponentialfunction.Ourdatashow that a faint end power-law slope of α=−0.81 is appropriate 1.5 forthehigherredshiftsubsample,andwenotethatthisisalso 810)× coNnsairsateynatnwanithetthael.da(t2a0o0f7N) aarlasyoanaannaleytzaeld.(t2h0e07e)v.olution of ( 1.0 dN/dz with redshift for weak MgII absorbers. They found MgII that the distribution follows the “no evolution” expectation; Ω that is, the expected number density for a nonevolving pop- ulation of absorbers in a ΛCDM universe, at redshifts less 0.5 than z=1.5. At higher redshift, they found that dN/dz for weakabsorbersdecreasesbelowthenoevolutionexpectation. InFigure9(a)wemakeadirectcomparisonwithNarayanan 0.0 0.5 1.0 1.5 2.0 2.5 3.0 et al. (2007), showing dN/dz for 0.02 Wλ2796 < 0.3 Å z binned in the same manner as their Figu≤re 4.r Here, we ob- Figure8. ΩMgIIasafunctionofredshift.ThecosmicmassdensityofMgII servethattheapparentpeakindN/dzforweakabsorbersoc- stays roughly flat near a value of 1 10−8, with a 0.5 dex increase from curs near z=0.75, as opposed to z=1.2 in Narayanan et al. z 0.5toz 2. × (2007). However, the overall shape of dN/dz for this low ∼ ∼ equivalentwidthpopulationremainsingoodagreement,with