Gas around galaxy haloes - III: hydrogen absorption signatures around galaxies and QSOs in the Sherwood simulation suite

0.38 MB·

by  Avery Meiksin

Mon.Not.R.Astron.Soc.000,1–10(2016) Printed25January2017 (MNLATEXstylefilev2.2)
Gas around galaxy haloes - III: hydrogen absorption
signatures around galaxies and QSOs in the Sherwood
simulation suite
7
1
0 Avery Meiksin1⋆, James S. Bolton2, Ewald Puchwein3
2
1SUPA†, Institute forAstronomy, Universityof Edinburgh, Blackford Hill, Edinburgh EH9 3HJ, UK

n 2School of Physics and Astronomy, Universityof Nottingham, UniversityPark, Nottingham NG7 2RD, UK
a 3Kavli Institute for Cosmology and Institute of Astronomy, Madingley Road, Cambridge, CB3 0HA, UK
J

4
2 Accepted .Received;inoriginalform

]
A
ABSTRACT
G
. Moderntheoriesofgalaxyformationpredictthatgalaxiesimpactontheirgaseous
h
surroundings,playingthe fundamental roleof regulatingthe amountofgas converted
p
intostars.Whilestar-forminggalaxiesarebelievedtoprovidefeedbackthroughgalac-
-
o tic winds, Quasi-Stellar Objects (QSOs) are believed instead to provide feedback
r through the heat generated by accretion onto a central supermassive black hole. A
t
s quantitativedifference in the impactoffeedback onthe gaseousenvironmentsof star-
a
forminggalaxiesandQSOshasnotbeenestablishedthroughdirectobservations.Using
[
the Sherwood cosmological simulations, we demonstrate that measurements of neu-

1 tralhydrogeninthe vicinityofstar-forminggalaxiesandQSOsduringtheeraofpeak
v galaxy formationshow excess Lyman-α absorptionextending up to comoving radiiof
8 ∼150kpcforstar-forminggalaxiesand300−700kpcforQSOs.Simulationsincluding
4 supernovae-drivenwinds account for the absorptionaround star-forming galaxies but
9
not QSOs.
6
0 Key words: cosmology:large-scalestructure of Universe – quasars:absorptionlines
1. – galaxies: formation – intergalactic medium
0
7
1
:
v 1 INTRODUCTION (QSOs), corresponds to a dark matter halo mass of about
Xi The paradigm of cosmological structure formation through 1012M⊙ (Kauffmann et al. 2003; Trainor & Steidel 2012).
Thisis thehalo massfor which theefficiencyof gas conver-
r thegrowthofprimordialdensityfluctuationspredictsgalax- sion into stars peaks (Behroozi et al. 2013).
a ies reside within accreting haloes at the interstices of a
cosmic web of dark matter and gas. The most straight- The opportunity to detect direct evidence for any dif-
forward estimates of the expected stellar content of galax- ference in the environmental impact between star-forming
ies, however, suggest a conversion rate of infalling gas into galaxies and QSOs, however, has been wanting. This has
stars that is too high. Galaxies should be much brighter now changed with the advent of observational programmes
than measured. It is widely believed some form of feed- that probe thegaseous surroundings of galaxies and QSOs.
back must regulate the efficiency of star formation. Both Utilizing the large-scale QSO surveys from the Sloan Digi-
semi-analytic models and cosmological hydrodynamicssim- talSkySurvey(Yorket al.2000;Schneideret al.2005)and
ulations suggest two forms of feedback are in play: stellar its extension to the Baryon Oscillation Spectroscopic Sur-
andsupernovae-drivengalacticwindsinstar-forminggalax- vey (Croom et al. 2004), the Quasar-Probing-Quasar sur-
ies and heating generated through accretion onto a super- vey (QPQ) (Hennawi et al. 2006) has used lines of sight to
massive black hole in galaxies with an Active Galactic Nu- 149 QSOs to probe the gaseous surroundings of both fore-
cleus (AGN) (see the review by Somerville & Dav´e 2015). groundQSOs(Prochaska & Hennawi2009;Prochaska et al.
The estimated galactic stellar mass dividing star-forming 2013) and galaxies (Prochaska et al. 2013). Similarly, the
galaxiesfromthemostmassiveAGN,Quasi-StellarObjects KeckBaryonicStructureSurvey(KBSS)(Rakic et al.2012;
Rudieet al. 2012; Trainor & Steidel 2012) has used lines of
sightto15hyperluminousQSOstomeasuregaseousabsorp-
⋆ E-mail:[email protected](AM) tion through the extended haloes of more than a thousand
† ScottishUniversitiesPhysicsAlliance galaxies.Interesthasparticularlyfocussedonthecentralfew
(cid:13)c 2016RAS
2 Avery Meiksin, James S. Bolton, Ewald Puchwein
hundred kiloparsecs, the Circumgalactic Medium (CGM), columndensitiesexceeding1015.5cm−2 aroundstar-forming
since this is the arena in which supernovae-driven winds in galaxies for lines of sight with a range of impact param-
star-forming galaxies and heating driven by black hole ac- eters both inside and outside the virial radii. They were
cretion in QSOsare expected to have their greatest impact similarlyabletoreproducetheprofilesforthecoveringfrac-
(e.g.Kay et al.2002;Schayeet al.2010;van deVoort et al. tions of optically thick systems around QSOs. As noted by
2011;Haas et al. 2013). Faucher-Gigu`ere et al. (2016), however, the covering frac-
The QPQ and KBSS results demonstrate galaxies and tion of optically thickabsorbers increases rapidly with halo
QSOs are surrounded by excess amounts of Lyman-α ab- mass and Rahmatiet al. (2015) confined their comparison
sorptionoverthatarisingfromthelarge-scaledistributionof to QSO halo virial masses Mh > 1012.5 M⊙. Observations
gas, the Intergalactic Medium (IGM) (Hennawi et al. 2006; favour a somewhat lower mass range (Trainor & Steidel
Steidel et al.2010;Rakicet al.2012;Prochaska et al.2013). 2012; Whiteet al. 2012; Eftekharzadeh et al. 2015). Since
Measuring the departures from the absorption predictions the simulation used included AGN feedback, it is also not
of models that exclude feedback provides a novel means of known whether supernovae-driven winds would have been
quantifying the impact of the winds and black hole accre- sufficient. The mass resolution of the simulation was more-
tion.Matchingtheexcessabsorptionprovidesatestoffeed- over intermediate between the FIRE simulations that suc-
back models. Accounting for the covering fractions of neu- ceeded in reproducing the covering fraction around QSOs
tralhydrogenabsorption systems,especially thoseoptically without AGN feedback and those that failed, so it is not
thick to photoionizing radiation, has proven particularly knownif thisis acritical factor aswell. Ontheother hand,
challenging. Motivated by the search for evidence of cold anadvantageofcosmological simulationsoverzoom-in sim-
streams feeding the growth of galaxies (Birnboim & Dekel ulations is the large sample size, with over a hundred QSO
2003; Kereˇs et al. 2005), early simulations produced sub- halomasssystemsatz=2intheEAGLEsimulation,which
stantially smaller covering fractions in massive haloes than maybemorerepresentativeofthepopulationthanthe∼15
measured. More recent simulations have had greater suc- systems available in the FIRE simulation, and fewer in the
cess, but still have not fully accounted for the observa- Fumagalli et al. (2014) analysis.
tions.UsingtheFIRE“zoom-in”simulations(Hopkinset al. The covering fractions do not provide a full descrip-
2014), which include feedback from stellar winds and su- tion of hydrogen absorption around galaxies and QSOs.
pernovae but not from AGN, Faucher-Gigu`ere et al. (2015) Thecollectiveabsorptionmaybequantifiedadditionallyby
reproduced the average covering fraction of Lyman Limit theintegratedequivalentwidthofabsorption alongtheline
Systems, absorbers optically thick at the Lyman edge of sight, or, analogously, the excess in the mean absorbed
(N > 1017.2cm−2), for lines of sight passing within flux over the contribution from the diffuse IGM. Using a
HI
the virial radius of star-forming galaxies, as measured by zoom-in simulation ofasingle galaxy includingsupernovae-
Rudieet al. (2012), but not for the QSO measurements drivenwindfeedback,Shenet al.(2013)matchedtheequiv-
of Prochaska et al. (2013). The model prediction for star- alentwidthsmeasuredoutsidethevirialradiiofstar-forming
forminggalaxiesfortheaveragecoveringfractionwithinthe galaxiesintheKBSSsurvey(Steidel et al.2010;Rakic et al.
virial radius of absorption systems with N >1015.5cm−2 2012), but the predictions fell nearly a factor of two short
HI
was a factor ∼2 short of measurements. On increasing the for lines of sight with impact parameter at about half the
mass resolution, Faucher-Gigu`ereet al. (2016) were able to virial radii, not much different from the case with no feed-
reproduce the covering fraction of optically thick absorbers back (Meiksin et al. 2015). The simulations successfully re-
withinthevirialradiusofQSOs.NotablyAGNfeedbackwas produced the mean covering fraction of optically thick ab-
stillnotincluded.Usingadifferentzoom-insimulationwith sorbers within the virial radii of star-forming galaxies, but
feedback from stellar winds and supernovae, but not AGN wereafactoroftwolowattwicethevirialradius.Thecover-
(Dekelet al. 2013, and references therein), Fumagalli et al. ingfractionforsystemswithNHI>1015.5cm−2wasalsolow
(2014)foundacoveringfractionforopticallythickabsorbers compared with themeasured values. The virial mass of the
within the virial radius of star-forming galaxies about half galaxyhalowas2.6×1011M⊙,somewhat lowforthepopu-
that measured, but consistent within the broad observa- lation of star-forming galaxies in the KBSS survey, but the
tionalerror.ThepredictionforQSOsfellwellshortofmea- resultsillustratethatavarietyofabsorptionmeasurements,
surements. The dark matter particle mass was only twice including profile information, must be used to demonstrate
thatofthehigherresolutionFIREsimulationsused,butthe thata simulation modelgivesan accurate representation of
baryonmassresolutionachievedmayhavebeenconsiderably thedistribution of neutral hydrogen around galaxies.
poorer.Thesimulationsalsoover-predictthestellarcontent A principal goal of this paper is to establish the scale
ofthegalaxiesbyafactoroftwo,sothatthegascontentwas ofdepartureofthegaseous haloes aroundgalaxies from the
likely toolow. Usingasuiteofcosmological hydrodynamics non-feedback cosmic-web prediction. Previously we found
simulations with a variety of wind feedback models, both models without feedback matched the observed absorp-
withandwithoutAGNfeedback,Suresh et al.(2015)found tion signatures of the “mesogalactic medium” at distances
some of the models match the cumulative covering fraction exceeding 500 − 800 comoving kpc around galaxies with
of optically thick absorbers within the virial radius of star- halo masses up to 1012 M⊙ at redshifts 2 < z < 2.5
forminggalaxies,butthepredictedvaluesareabouthalfthe (Meiksin et al.2015);thesimulationvolumesweretoosmall
measuredvaluewithintwicethevirialradius.Inananalysis tomakefirmstatementsforlargermasshaloes.Bycontrast,
of the EAGLE simulations (Schayeet al. 2015), for which clear excesses in the data compared with the non-feedback
a different supernova-driven wind feedback model was im- modelswerefoundwithintheinner100−300comovingkpc
plemented and which include AGN heating, Rahmati et al. (Meiksin et al. 2015). Using the Sherwood simulation suite
(2015)matchedthecoveringfractionprofileofsystemswith (Bolton et al. 2016), we reinvestigate the signatures of hy-
(cid:13)c 2016RAS,MNRAS000,1–10
Gas around galaxies III: hydrogen absorption 3
drogen absorption around galaxies and QSOs. We include linear proportionality of the diffusesoft x-ray luminosity of
simulations both without and with feedback, using a differ- galaxies with thestarformation rate(Mineo et al.2012),is
ent feedback model compared with previous simulations in matched in an energy-drivenwind model for an asymptotic
the context of CGM measurements to further test the sen- wind velocity that scales approximately as M˙∗1/6 (Meiksin
sitivity of the absorption signatures to the feedback model. 2016). For a star-formation rate scaling approximately like
Thesenewlargersimulationsalsosubstantiallyimprovethe thestellarmass(e.g.Elbaz et al.2007)andtheratioofstel-
precisionofthepredictions,particularlyfortheraremassive larmasstohalomassscalingapproximatelyasthehalomass
haloes exceeding 1012 M⊙, enabling us to quantify the dif- for halo masses 1011 < Mh < 1012M⊙ (e.g. Behroozi et al.
ferencesbetweentheabsorptionsignaturesbelowandabove 2013), the scaling required by the x-ray measurements cor-
this mass threshold. responds at least approximately to a wind velocity scaling
This paper is organized as follows. In Sec. 2 we sum- likethehaloescapevelocity.Themass-loadingfactorforthe
marise the numerical simulations used in this work. The wind model in the simulations has typical values of 1−10.
resulting HI absorption signatures are presented in Sec. 3. This is consistent with the values found for the FIRE sim-
We summarize and discuss our conclusions in Sec. 4. Sim- ulations (Muratov et al. 2015). By contrast, the mass load-
ulation convergence tests are presented in an Appendix. ing factor was less than unity for the galaxy in Shen et al.
Unless stated otherwise, all results are for a flat ΛCDM (2013).Massloadingfactorshavenotbeenpublishedforthe
universe with the cosmological parameters Ω = 0.308, EAGLE simulations.
m
Ωbh2 = 0.0222 and h = H0/100 km s−1 Mpc−1 = 0.678, Otherfeedback mechanismsare considered in theliter-
representingthepresent-daytotalmassdensity,baryonden- atureof galaxy formation simulations. Murray et al. (2005)
sity and Hubble constant, respectively. The power spec- consider winds driven by momentum-deposition from radi-
trum has spectral index ns = 0.961, and is normalized to ation pressure or supernovae. Keller et al. (2014) use a su-
σ8 = 0.829, consistent with the Cosmic Microwave Back- perbubble feedback model for which mass loading includes
grounddatafromPlanck(Planck Collaboration et al.2014). a contribution from thermal evaporation. Another possibly
Alldistancesarecomovingunlessstatedotherwise;‘cMpc’is importantwinddriveriscosmic-raypressure(Ipavich1975),
used to designate comoving megaparsecs, and ‘ckpc’to des- whichhasrecently beenusedin cosmological galaxy forma-
ignate comoving kiloparsecs. tion simulations (Salem et al. 2014; Pakmor et al. 2016).
The computational methods are described in detail in
Bolton et al. (2016). The runs used in this paper are sum-
marizedinTable1.Exceptforthewindsimulation,thecom-
2 NUMERICAL SIMULATIONS
putationsusethe“quickLyα”method(qLyα),forwhichall
The numerical simulations are performed using a modified gas cooler than 105 K and with an overdensity exceeding
version of the parallel Tree-PM SPH code P-GADGET-3, an 1000 is converted into collisionless “star”particles (without
updatedversionofthepubliclyavailablecodeGADGET-2(last feedback).Thisis acomputational trick usedin IGMsimu-
described by Springel 2005). Both simulations without and lationstosignificantlyspeedupthecomputation(Viel et al.
with galactic winds were performed. Because the computa- 2004), and is not meant to represent the stellar content
tionaldemandsforgeneratingagalacticwindfromfirstprin- of real galaxies. A detailed comparison with similar sim-
ciples are well beyond current resources, all wind feedback ulations using the finite-difference code Enzo (Bryan et al.
modelsrelyon asubgridapproximation scheme.Previously 2014),withnogasremovalorfeedbackimplemented,shows
(Meiksin et al.2015),wehadusedastar-formationandfeed- that, while the quick Lyα approximation reduces the cir-
back model with a constant wind velocity for all haloes cumgalactic gas density within the virial radius of haloes
(Springel & Hernquist2003).Suchmodels, however,donot compared with Enzo, the integrated absorption signatures
wellreproducetheluminosityfunctionofgalaxies.Thewind are unaffected along baselines comparable to those used in
simulationusedhereisbasedonthestar-formationandwind measurements (Meiksin et al. 2015), even for lines of sight
model of Puchwein & Springel (PS13; 2013). This scheme passing within the virial radius. This was shown to be be-
is built on a subgrid model that allows for star formation cause the integrated absorption signals over baselines ex-
inamultiphaseinterstellarmedium,with thefeedbackpro- ceeding ∼ 1000km s−1 are dominated by gas outside the
vided by stochastic kinetic energy injection by assigning to virial radius for non-feedback models even for impact pa-
particlesenteringthewindnearthecentreofagalaxyakick rameters smaller than the virial radius. Weaccordingly use
velocity that scales with the escape velocity from the halo. the quick Lyα simulations to model the absorption signa-
Themassloadingfactor,definedastheratioofthemassflow turesin theabsence of feedback.
inthewindtothestarformationrate,isproportionaltothe Tointerpretthelarge-scalemesogalacticabsorptionsig-
inverse square of the escape velocity, thus using a constant nal through the extended galactic haloes outside the cir-
amount of energy per unit mass of stars formed for driving cumgalacticregion,wealsocomputesimpleone-dimensional
thewind.Thisformhasbeenfoundtorecovertheluminosity spherically symmetric collapse models. We use the com-
function and luminosity-metallicity relation of Local Group binedgravitating shell modeland hydrodynamicsfinitedif-
satellite galaxies (Okamoto et al. 2010). It also reproduces ference code described in Meiksin (1994), modified to allow
thestellarmassfunctionofgalaxiesoverredshiftsz<2,the foracosmologywithacosmologicalconstant.Torecoverthe
observed gas to stellar mass ratios and specific star forma- asymptotic mean profile at large distances from the centre,
tionratesasafunctionofstellarmass(Puchwein & Springel we use the mean profile derived by Bardeen et al. (1986)
2013), and a variety of other galaxy properties over a wide as the initial density profile. Since the gas is smooth, it
range of redshifts (Vogelsberger et al. 2013). It receives ad- is necessary to adopt a method to represent the density
ditional support from x-ray measurements. The measured fluctuations underlying the Lyα forest, without which the
(cid:13)c 2016RAS,MNRAS000,1–10
4 Avery Meiksin, James S. Bolton, Ewald Puchwein
Table1.Summaryofthesimulationsperformedinthiswork.Thecolumns,fromlefttoright,listthesimulationname,thecomovingbox
size,thedarkmatterparticlemass,thebaryonparticlemass,thecomovinggravitationalsofteninglengthandthemethodofgasremoval/
windfeedback.
Name Boxsize Mdm Mgas lsoft gasremoval/wind
[h−1 cMpc] [h−1 M⊙] [h−1 M⊙] [h−1 ckpc]
40-2048 40 5.37×105 9.97×104 0.78 qLyα
80-2048 80 4.30×106 7.97×105 1.56 qLyα
40-1024 40 4.30×106 7.97×105 1.56 qLyα
40-1024-ps13 40 4.30×106 7.97×105 1.56 PS13
20-512 20 4.30×106 7.97×105 1.56 qLyα
important effects of line blanketing will not be recovered. integrated across velocity windows. Previously we found
We do so by implementing the lognormal density fluctua- this provided an adequate description of the effect of over-
tion scheme of Bi (1993) and Bi & Davidsen (1997) along densestructuresnearthehaloesontheabsorptionstatistics
projected one-dimensional lines of sight through the spher- (Meiksin et al. 2015). For a halo at measured velocity v
halo
ically symmetric gas distribution. Details are described in along a line of sight to a background QSO, the equivalent
Chongchitnan & Meiksin (2014). To describe the fluctua- width at theLyman-αwavelength λ is computed as
α
tions around a halo, the density field is normalized to the
λ vhalo+∆v/2
local mean density in thehalo. w(b⊥,∆v)= cα Zvhalo−∆v/2 dvh1−e−τα(b⊥,v)i, (1)
Forboththefullcosmologicalsimulationandthespher-
over a velocity window of width ∆v centred on the posi-
icallysymmetriccode,atomicradiativecoolingfrom hydro-
gen and helium are included, along with photoionization tion of the halo, displaced transversely by an amount b⊥.
The equivalent width is then used to define the fractional
heating. Metal-line cooling is not included. Metal-line cool-
change δ in thetransmitted flux compared with themean
ing negligibly affects the large-scale IGM, but may play an F
intergalactic value for the corresponding velocity window
importantrolewithingalactichaloesfollowing thedistribu-
(Prochaska et al. 2013)
tionofmetalsbysupernovaefeedback.Thegasisphotoion-
iLzyedαabbysoarbmeedtaflguaxlalecvtiecliuslsterat-ivnioolredterbatockmgraotuchndt.heTmheeamsueraen- δF(b⊥,∆v)= w(b∆⊥,λ∆−v)w−wIGM. (2)
IGM
ments of Beckeret al. (2013). Radiative transfer of contin- Here ∆λ=λ ∆v/c and w =∆λ[1−exp(−τ )], where
α IGM eff
uum radiation is not included in the computations. Radia-
τ is the Lyman-αeffectiveoptical depth of theIGM.
eff
tivetransfernormallyhaslittleeffectonthetemperatureof
IntheAppendixwepresentasuiteofconvergencetests
absorption systems optically thick to hydrogen and helium
ofthecosmological simulations,varyingthemassresolution
ionizing photons because their density is sufficiently high
and box size. In brief, we find that a comoving box size
thatthermalbalanceisachieved.Itwill,however,affectthe of 40h−1 Mpc with a dark matter particle mass resolution
amount of hydrogen absorption in optically thick systems. 4.30×106h−1M⊙ (10243 particles) yieldsadequateconver-
Because the large velocity shearing of absorption systems
gence.Furtherconvergencetestsoftheabsorptionsignatures
withinthevirialradiusincreasestheequivalentwidthsofthe
are described in Meiksin et al. (2015).
systems(Meiksin et al.2015),especiallyinthepresenceofa
wind,itisimportanttoincluderadiativetransfereffectsfor
predicting the amount of absorption within the virial radii
3 RESULTS
ofthehaloes. Ideally,self-consistent radiativehydrodynam-
ics should be implemented, but this is beyond the scope of 3.1 Deviation of Lyα absorption from the mean
this work. Instead we post-process the ionization fractions IGM
using the approximate treatment of Rahmati et al. (2013).
The left-hand panel of Fig. 1 shows the mean spherically
The effect the radiative transfer correction has on the ab-
sorption signature is described in the Appendix. The full averagedgasoverdensityaroundhaloesofmass1012 M⊙ at
redshift z =2.4. The profilesare constructed byspherically
effects including radiative hydrodynamics is still a largely
averaging the gas distribution around each halo in radial
unexplored topic.
bins,andthenaveragingtheresultingsphericalprofilesover
WeextractspectraoftheLyαforestfollowingthestan- the haloes in a narrow halo mass bin centred at 1012M⊙.
dard procedure, as summarized in Meiksin et al. (2015). A Results are shown for two simulations in a 40h−1 Mpc co-
fullVoigt profileisused throughouttheanalyses. Tomatch movingbox,onewithadarkmatterparticlemassresolution
the resolution of the observations, we smooth the resulting of4.30×106h−1M⊙ (40-1024) andtheotherwitharesolu-
spectra with a Gaussian of FWHM 125km s−1 for com- tionof5.37×105h−1M⊙ (40-2048).Thedensityprofilesfor
parison with Prochaska et al. (2013) and FWHM 8km s−1 the models agree well, and largely match the predictions of
for comparison with Rakicet al. (2012). We extract spec- thesphericalcollapsemodel.DeviationsfromtheBBKSlin-
tra for a range of impact parameters around haloes identi- eartheoryprofilebecomepronouncedonlywithin theinner
fiedinthesimulationsusingthefriend-of-friendsalgorithm, 5 cMpc.
as described in Bolton et al. (2016). We base our compari- Theabsorptionexcessδ relativetothemeanIGM,av-
F
son primarily on the Lyman-α absorption equivalent width eraged overa spectral window of width ∆v=2000km s−1,
(cid:13)c 2016RAS,MNRAS000,1–10
Gas around galaxies III: hydrogen absorption 5
no wind wind (z = 2.4)
100
10
1
0 2 4 6 8 10
Figure1.Left-handpanel:Radial-averagedgasoverdensityprofilesaroundhaloeswithmassMh=1012 M⊙atredshiftz=2.4fornon-
windsimulations with 20483 dark matter particles (40-2048; solidblue) and 10243 darkmatter particles (40-1024; short-dashed cyan),
bothinabox40h−1 comovingMpconaside.Alsoshownaretheoverdensityprofilesfor1Dsphericalinfall(long-dashedmagenta)and
theBBKSlinearoverdensity(dottedblack).Right-handpanel:FractionalabsorptionexcessδF relativetothemeanIGMabsorption,for
aspectralwindow∆v=2000kms−1 acrossthehalosystemicvelocitiesforthe40h−1 Mpccomovingboxsimulationswithandwithout
awind,shownagainstprojectedimpactparameter b⊥ forhalomasseslog10(Mh/M⊙)=11.1(triangles; blue),11.5(squares;magenta)
and12.0−12.6(circles;black)atredshiftz=2.4.Forclarityofpresentation, thepointsforthelowestandhighesthalomassbinshave
been interpolated to slightly offset values of b⊥. The halo velocities include a random component drawn from a Gaussian distribution
with standard deviation σH = 520kms−1. The curves (green) correspond to a 1D spherical infall model around haloes with masses
log10(Mh/M⊙)=11,11.5and12(seetext).
is shown in the right-hand panel of Fig. 1 for the non- averageenhancementinabsorptionwithinthevelocitywin-
wind simulation 40-1024 and the wind simulation 40-1024- dow from the chance interception of other haloes clustered
ps13 at z = 2.4. A random line-of-sight halo velocity withthecentralhalo.Estimatingthecorrectionanalytically
is allowed for, drawn from a Gaussian distribution with is not straightforward as it involves modelling the cluster-
dispersion σ = 520km s−1, typical of measurements ing of haloes in an overdense region. Instead we allow for a
H
(Prochaska et al. 2013). Results are shown as a function small additional amount of absorption common to all halo
of projected line-of-sight comoving impact parameter b⊥ masses, normalizing to theabsorption in thesimulations at
for haloes in mass bins centred at log10(Mh/M⊙) = 11.1 7 cMpc. Allowing for the correction, the spherically sym-
and 11.5 of width ∆log10(Mh/M⊙) = 0.2, and for haloes metric model predictions agree closely with the non-wind
with masses 12.0 < log10(Mh/M⊙) = 12.6. The amount numerical simulation from asymptotically large intergalac-
of absorption increases weakly with halo mass. While the tic scales to within thevirial radii of the haloes.
wind simulation produces a greater amount of absorption
for all impact parameters, δ is enhanced by the wind by
F
3.2 Comparison with observations
anamount exceeding0.05 onlywithin theinner50 ckpcfor
log10(Mh/M⊙) = 11.1, 70 ckpc for log10(Mh/M⊙) = 11.5 Weusetheintegratedabsorptionexcessoverbroadvelocity
and200ckpcfor12.0<log10(Mh/M⊙)<12.6.Forcompar- windows to quantify the absorption around galaxy haloes.
ison, the corresponding virial radii are 160 ckpc, 220 ckpc Previously we found this provided an adequate description
and320−510ckpc,respectively,sothatthislevelofenhance- of the effect of overdense structures near the haloes on the
mentoccursataboutone-thirdthevirialradii.Asshownin absorption statistics (Meiksin et al. 2015).
the Appendix, when the impact parameters are rescaled to InFig.2wecomparethepredictionsofmodelswithout
thevirialradiusforeachhalomass,boththeprofileswithout andwith windsto theδ measurements surroundingQSOs
F
wind feedback and with are each nearly universal, indepen- reportedbyProchaska et al.(2013)atz ≈2.4,withaveloc-
dent of halo mass. itywindow∆v=2000km s−1.Arandomcomponentdrawn
The spherically symmetric collapse models similarly fromaGaussiandistributionwithσ =520km s−1 isadded
H
predict an increase in δ with halo mass, as shown by the tothehalovelocitiesinthesimulationstomatchthetypical
F
(green) curves in Fig. 1 (right-hand panel). The spherically errors in the measured halo redshifts. The mean QSO halo
symmetric solution is for an isolated halo. It neglects the mass at this epoch is (1−3)×1012M⊙ (Trainor & Steidel
(cid:13)c 2016RAS,MNRAS000,1–10
6 Avery Meiksin, James S. Bolton, Ewald Puchwein
Steidel et al. (2010)
Rakic et al. (2012)
Figure2.AbsorptionsignaturearoundQSOs.Thefractionalab- Figure 3. Absorption signature around star-forming galaxies.
sorptionexcessδF relativetothemeanIGMabsorptionisshown The rest-frame equivalent width (˚A) is shown within a veloc-
for a spectral window ∆v = 2000kms−1 across the halo sys- ity window ∆v = 1000kms−1 centred on the halo centre-of-
temicvelocitiesforthemodelwithoutawind(80-2048;solidsym- massvelocityasafunctionoftheline-of-sightimpactparameter
bols) and with a wind (40-1024-ps13; open symbols) at redshift b⊥ forhalo masses log10(Mh/M⊙)=11.7 (triangles; blue), 11.9
z=2.4,forhalomasses log10(Mh/M⊙)=11.9(triangles; blue), (squares; magenta) and 12.1 (circles; black) at redshift z = 2.4
12.1(squares;magenta)and12.2−12.8(circles;black).Thesolid for models without wind feedback (80-2048; solid symbols) and
lineshowsthepredictionforsimplesphericalaccretion.Thedata with(40-1024-ps13;opensymbols).Thesolidlineshowsthepre-
points(largegreenfilledsquares)areforabsorptionaroundQSOs diction for simple spherical accretion. The data are for star-
atz≈2.4takenfromProchaskaetal.(2013),withtheerrorbars forminggalaxiesfromSteideletal.(2010)(redfilledcircles)and
showingtheerrorsinthemean. Rakicetal.(2012)(largegreenfilledsquares).
2012;Whiteet al.2012;Eftekharzadeh et al.2015).Outside widths measured around star-forming galaxies at z ≈ 2.4
the virial radii (corresponding to 300−590 comoving kpc within 1000km s−1 wide velocity windows centred on the
forthehalomassesshown),themeasurementsagreeclosely systemic velocities of the galaxies using the data from
with the simulation predictions. The simulations slightly Steidelet al. (2010) (red filled circles) and Rakic et al.
over-predict the amount of absorption near 1−2 comov- (2012) (large green filled squares). A random component
ingMpc.Thismayindicatetoolowaphotoionizationback- drawn from a Gaussian distribution with σ =130km s−1
H
ground was used, but we conservatively maintain the level is added to the halo velocities in the simulations to
used in the simulations. The measured values also match match the typical errors in the measured halo redshifts.
theexpectationforsphericalaccretionontoahalo.Thedata A spatial correlation analysis gives an estimated median
showexcessabsorptionoverthenon-windmodelpredictions halo mass for the galaxies in the survey of 1011.9±0.1M⊙
at a statistically significant level (more than 3σ) out to im- (Trainor & Steidel2012).Weshowresultsforhalomassbins
pact parameters of 300−700 comoving kpc, with the ex- centred at log10(Mh/M⊙) = 11.7, 11.9 and 12.1 of width
cessincreasingwithdecreasingimpactparameter.Thewind ∆log10(Mh/M⊙) = 0.2. Following the observational anal-
model shows excess absorption over the non-wind model ysis procedure of Rakic et al. (2012) used to match their
within the virial radii for the most massive haloes, but the newer data to the continuum level of the earlier data of
amount still falls short of the measured values. The central Steidelet al. (2010), the simulated spectra have been ad-
twodatapointsaremoredominated byhaloeswith z≃2.0 justed by applying a renormalization factor of 0.804 to the
(Rahmati et al. 2015). The values for δ in the simulation fluxes.Outsidethe virial radii (250−350 comoving kpcfor
F
at z = 2.0 agree within the errors with those at z = 2.4 the halo masses shown), the predicted equivalent widths of
oraresomewhatsmaller.Theresultssuggestthatwhilethis allthemodelsagreewell withthedatawithinthemeasure-
particularwindmodelperformsbetterthantheconstantve- menterrors. Theyalso agree with theprediction forspheri-
locitywindmodelsstudiedpreviously(Meiksin et al.2015), calinfallwithoutfeedback.Excessabsorptionoverthenon-
it is still not able to account fully for the high amounts of wind model predictions is detected at a statistically signif-
measuredabsorption.Alternatively,itmaybeanindication icant level (more than 4σ) only within the inner b⊥ < 140
of the need to includeAGN feedback. comovingkpcbinaroundthegalaxies.Thewindmodelnow
In Fig. 3, we compare with the absorption equivalent booststheamountofabsorptiontomatchthemeasuredlev-
(cid:13)c 2016RAS,MNRAS000,1–10
Gas around galaxies III: hydrogen absorption 7
no wind wind
no wind wind
Figure4.CoveringfractionaroundQSOs.Thecoveringfraction Figure 5. Covering fraction around star-forming galaxies. The
of HI gas with column density log10NHI > 17.3 is shown for coveringfractionofHIgaswithcolumndensitylog10NHI>16.0
a spectral window ∆v = 3000kms−1 across the halo systemic is shown for a spectral window ∆v = 600kms−1 across the
velocities forthe model without awind(80-2048; solidsymbols) halo systemic velocities for the model without a wind (80-2048;
and with a wind (40-1024-ps13; open symbols) at z = 2.4. The solidsymbols)andwithawind(40-1024-ps13; opensymbols)at
datapoints (largegreenfilledsquares)arefromProchaskaetal. z = 2.4. The data are from Rudieetal. (2012) for star-forming
(2013) for absorption around QSOs at z ∼ 2.4, with the error galaxies(greenhistogram).
barsshowingtheerrorsinthemean.
4 DISCUSSION AND CONCLUSIONS
We use the Sherwood simulations, a suite of cosmological
els.Thiscontrastswiththeconstantvelocitywindmodel,for N-body+hydrodynamics IGM simulations, to predict ab-
which toolittle absorption wasfound (Meiksin et al. 2015). sorption signatures around QSOs and star-forming galax-
We have also checked against HI covering fractions ies. Both models without and with feedback in the form
around QSOs and star-forming galaxies. Fig. 4 shows the of supernovae-driven winds are included. No metals are in-
fraction of lines of sight through haloes with masses char- cluded. In this paper, we use an energy-driven feedback
acteristic of QSOsthatcontain absorption systems with HI modelthatscalesthewindvelocitywiththeescapevelocity
columndensitiesexceeding1017.3cm−2 atz=2.4.Thepre- from the halo and varies the amount of mass loading as its
dicted covering fractions at z = 2.0 agree within the errors inverse square, using a constant amount of mechanical en-
withthoseatz=2.4oraresomewhatsmaller.Aswefound ergy per unit mass in stars formed to drive the wind. This
previously (Meiksin et al. 2015), the covering fraction con- contrastswithourprevioustreatmentwhichinvokedawind
tinues to fall short of the measurements of Prochaska et al. with constant velocity and constant mass loading indepen-
(2013),evenforthenewsupernovae-drivenwindmodelused dent of halo mass.
here. Part of the discrepancy, such as near 1−2 comoving A comoving box size of 40h−1Mpc with 10243 dark
Mpc, may be attributed to unresolved absorption systems matter and gas particles each is shown to provide a well-
(Prochaska et al.2013),butnotthesteeprisetowardssmall convergedestimateoftheamountofabsorption,whenthere
impact parameters. By contrast, Fig. 5 shows that the new aresufficientnumbersofhaloes.Toproduceaccuratestatis-
windmodelnowrecoverstheHIcoveringfractionofsystems tics for QSO haloes, however, a larger box size is preferred.
around haloes with masses typical of star-forming galaxies, We use a comoving box size of 80h−1Mpc with 20483 dark
as measured by Rudieet al. (2012) for absorption HI col- matterandgasparticleseachforcomparisonofthenon-wind
umndensitiesexceeding1016cm−2.Rudieet al.(2012)also simulationswithallobservations.Becauseofthegreaterex-
report cumulativecovering fractions of 20 per cent.for sys- pense of the wind simulations, a 40h−1Mpc comoving box
temswithcolumndensitiesintherange1017.2−1020.3cm−2 wasused.Anapproximateanalyticformulationisusedtoes-
for impact parameters within both the virial radius and timatetheeffectsofradiativetransferontheabsorptionsig-
twicethevirialradius.Wefind18percent.and8percent., naturesinapost-processingstep.Formodelswithoutwinds,
respectively. While the cumulative covering fraction within the radiative transfer corrections negligibly affect the ab-
the virial radius is comparable to the measured value, the sorptionsignature,buttheeffectsarenon-negligiblewhena
fraction within twice thevirial radius is somewhat low. wind is included. The radiative transfer correction was ap-
(cid:13)c 2016RAS,MNRAS000,1–10
8 Avery Meiksin, James S. Bolton, Ewald Puchwein
plied in all cases for consistency. Radiative hydrodynamics in Europe (PRACE) 8th Call. This work also made use of
effects are not taken intoaccount. the DiRAC High Performance Computing System (HPCS)
Excellentagreementisfoundbetweenthemeasuredand andtheCOSMOSshared memory service at theUniversity
predicted mesogalactic neutral hydrogen absorption signa- of Cambridge. These are operated on behalf of the STFC
turesaroundstar-forming galaxies outsidethevirialradius. DiRAC HPC facility. This equipment is funded by BIS
Bothline-of-sightabsorptionmeasurementsandneutralhy- National E-infrastructure capital grant ST/J005673/1 and
drogencoveringfractionsarereproduced.Thisdemonstrates STFCgrantsST/H008586/1, ST/K00333X/1. WethankV.
thatthereisnoneedfortheintroductionofadditionalphys- Springel for making P-GADGET-3 available. JSB acknowl-
icalassumptionstodescribethegasdensityandpeculiarve- edges the support of a Royal Society University Research
locity fields for gas on these scales. Good agreement is also Fellowship. EPacknowledges supportbytheKavliFounda-
found well outside the virial radius for the QSO data. A tion.
small excess in the predicted amount of absorption may be
aresultofanunder-estimatedphotoionizingradiationfield.
Thismay reflectan under-estimateofthemetagalactic ion-
izing radiation background. It may also be an indication of
REFERENCES
atransverseproximityeffectresultingfromalocalcontribu-
tiontotheionizingradiationfromtheQSOsthemselves,for Bardeen J. M., Bond J. R., Kaiser N., Szalay A. S., 1986,
which there is some independent evidence (Jakobsen et al. ApJ,304, 15
2003; Syphers& Shull 2014). Outside the virial radius, the Becker G. D., Hewett P. C., Worseck G., Prochaska J. X.,
spherically averaged absorption signatures are still well de- 2013, MNRAS,430, 2067
scribedbysphericalinfallofintergalacticgasontothehaloes BehrooziP.S.,WechslerR.H.,ConroyC.,2013,ApJ,770,
for both star-forming galaxies and QSOs, as predicted for 57
the extended haloes of galaxies and QSOs forming at the Bi H., 1993, ApJ,405, 479
peaks of a primordial gaussian field of density fluctuations, Bi H., Davidsen A. F., 1997, ApJ,479, 523
allowing for secondary infall. Birnboim Y.,Dekel A., 2003, MNRAS,345, 349
By contrast, for impact parameters lying within the BoltonJ.S.,PuchweinE.,SijackiD.,HaehneltM.G.,Kim
virialradii,statisticallysignificantexcessabsorptionismea- T.-S., Meiksin A., Regan J. A., Viel M., 2016, ArXiv e-
sured for both star-forming galaxies and QSOs compared prints,1605.03462
with the non-feedback predictions. The excess absorption BryanG.L.,Norman M.L.,O’SheaB.W.,AbelT., Wise
extends to substantially larger distances, of 300−700 co- J. H., Turk M. J., Reynolds D. R., Collins D. C., Wang
moving kpc, around haloes exceeding log10(Mh/M⊙) =12, P., Skillman S. W., Smith B., + 2014, ApJS,211, 19
typical of those harbouring QSOs, than for the less mas- Chongchitnan S., Meiksin A., 2014, MNRAS,437, 3639
sivehaloesinwhichstar-forminggalaxiesusuallyreside,for Croom S. M., Smith R. J., Boyle B. J., Shanks T., Miller
which excess absorption is measured only within the inner L.,OutramP.J.,LoaringN.S.,2004,MNRAS,349,1397
∼150 comoving kpc. This suggests that the affected region DekelA.,ZolotovA.,TweedD.,CacciatoM.,CeverinoD.,
islargeraroundQSOseitherbecausethefeedbackimpacton Primack J. R., 2013, MNRAS,435, 999
thesurroundinggas dependson halo massbeyondthescal- EftekharzadehS.,MyersA.D.,WhiteM.,WeinbergD.H.,
ing with virial radius found for the feedback model used, Schneider D. P., Shen Y., Font-Ribera A., Ross N. P.,
or because of a difference in the feedback mechanisms for Paris I., StreblyanskaA., 2015, MNRAS,453, 2779
star-forming galaxies and QSOs. ElbazD.,DaddiE.,LeBorgneD.,DickinsonM.,Alexander
The supernovae-driven wind model used accounts for D.M.,CharyR.-R.,StarckJ.-L.,BrandtW.N.,Kitzbich-
boththeexcessabsorptionaroundstar-forminggalaxiesand ler M., MacDonald E., Nonino M., Popesso P., Stern D.,
the neutral hydrogen covering fraction, unlike our previ- Vanzella E., 2007, A&Ap,468, 33
ous wind model simulations. The agreement suggests nei- Faucher-Gigu`ere C.-A., Feldmann R., Quataert E., Kereˇs
ther AGN heating nor metal cooling are required to match D.,HopkinsP. F., Murray N., 2016, MNRAS,461, L32
theabsorptionsignaturesforstar-forminggalaxies,although Faucher-Gigu`ere C.-A., Hopkins P. F., Kereˇs D., Muratov
thisconclusionmaydependonthefeedbackmodelused.By A.L., Quataert E., Murray N., 2015, MNRAS,449, 987
contrast, even for themodel with feedback, thesimulations Fumagalli M., Hennawi J. F., Prochaska J. X., Kasen D.,
failtoaccountfortheamountsofexcessabsorptionandthe DekelA., Ceverino D., Primack J., 2014, ApJ, 780, 74
covering fractions of optically thick gas near QSOs, even Haas M. R., Schaye J., Booth C. M., Dalla Vecchia C.,
thoughthefeedbackmodelpredictstheboost inabsorption SpringelV.,TheunsT.,WiersmaR.P.C.,2013,MNRAS,
fromthewindshouldincreasewiththehalomassatafixed 435, 2931
impact parameter. Additional heating by black-hole accre- Hennawi J. F., Prochaska J. X., Burles S., Strauss M. A.,
tion may be required, although it may also indicate a need Richards G. T., Schlegel D. J., Fan X., Schneider D. P.,
for higher numerical resolution of thegas. Zakamska N. L., Oguri M., Gunn J. E., Lupton R. H.,
Brinkmann J., 2006, ApJ, 651, 61
Hopkins P. F., Kereˇs D., On˜orbe J., Faucher-Gigu`ere C.-
A.,QuataertE.,MurrayN.,BullockJ.S.,2014,MNRAS,
ACKNOWLEDGMENTS
445, 581
This work used the Curie supercomputer at the Tr´e Grand Ipavich F. M., 1975, ApJ, 196, 107
CentredeCalcul(TGCC)inFrance,madeavailablethrough Jakobsen P., Jansen R. A., Wagner S., Reimers D., 2003,
time awarded by the Partnership for Advanced Computing A&Ap,397, 891
(cid:13)c 2016RAS,MNRAS000,1–10
Gas around galaxies III: hydrogen absorption 9
KauffmannG.,HeckmanT.M.,TremontiC.,Brinchmann N., Bogosavljevi´c M., Rudie G. C., Rakic O., 2010, ApJ,
J.,CharlotS.,WhiteS.D.M.,RidgwayS.E.,Brinkmann 717, 289
J., Fukugita M., Hall P. B., Ivezi´c Zˇ., Richards G. T., Suresh J., Bird S., Vogelsberger M., Genel S., Torrey P.,
SchneiderD.P., 2003, MNRAS,346, 1055 SijackiD.,SpringelV.,HernquistL.,2015,MNRAS,448,
Kay S. T., Pearce F. R., Frenk C. S., Jenkins A., 2002, 895
MNRAS,330, 113 SyphersD., ShullJ. M., 2014, ApJ, 784, 42
Keller B. W., Wadsley J., Benincasa S. M., Couchman Trainor R.F., Steidel C. C., 2012, ApJ, 752, 39
H.M. P., 2014, MNRAS,442, 3013 vandeVoortF.,SchayeJ.,BoothC.M.,DallaVecchiaC.,
Kereˇs D., Katz N., Weinberg D. H., Dav´e R., 2005, MN- 2011, MNRAS,415, 2782
RAS,363, 2 VielM.,Haehnelt M.G.,SpringelV.,2004, MNRAS,354,
Meiksin A.,1994, ApJ, 431, 109 684
Meiksin A.,2016, MNRAS,461, 2762 Vogelsberger M., Genel S., Sijacki D., Torrey P., Springel
MeiksinA.,BoltonJ.S.,TittleyE.R.,2015,MNRAS,453, V.,Hernquist L., 2013, MNRAS,436, 3031
899 WhiteM.,MyersA.D.,RossN.P.,SchlegelD.J.,Hennawi
Mineo S., Gilfanov M., Sunyaev R., 2012, MNRAS, 426, J. F., Shen Y., McGreer I., Strauss M. A., Bolton A. S.,
1870 Bovy J., 2012, MNRAS,424, 933
Muratov A. L., Kereˇs D., Faucher-Gigu`ere C.-A., Hopkins York D. G., Adelman J., Anderson Jr. J. E., et al. 2000,
P. F., Quataert E., Murray N.,2015, MNRAS,454, 2691 AJ,120, 1579
MurrayN.,QuataertE.,ThompsonT.A.,2005,ApJ,618,
569, arXiv:astro-ph/0406070
Okamoto T., Frenk C. S., Jenkins A., Theuns T., 2010,
MNRAS,406, 208 APPENDIX A: CONVERGENCE TESTS ON HI
STATISTICS
Pakmor R., Pfrommer C., Simpson C. M., Springel V.,
2016, ApJ, 824, L30 The convergence of the fractional absorption excess δ at
F
PlanckCollaborationAdeP.A.R.,AghanimN.,Armitage- z = 2.4 with mass resolution at a fixed comoving box size
CaplanC.,ArnaudM.,AshdownM.,Atrio-BarandelaF., of 40h−1Mpc is tested in Fig. A1 for a velocity window
Aumont J., Baccigalupi C., Banday A. J., et al. 2014, of width ∆v = 1000km s−1 and a halo redshift uncer-
A&Ap,571, A16 tainty σ = 130km s−1, similar to the of observations of
H
Prochaska J. X., Hennawi J. F., 2009, ApJ, 690, 1558 Rakicet al. (2012). A weak trend of increasing absorption
Prochaska J. X., Hennawi J. F., Lee K.-G., Cantalupo S., with halo mass is found for decreasing impact parameters.
BovyJ.,DjorgovskiS.G.,EllisonS.L.,LauM.W.,Mar- Foragivenmasshalo,thereisnosignificanttrendwithpar-
tinC. L.,MyersA.,RubinK.H.R.,SimcoeR.A.,2013, ticle mass, demonstrating the fluctuationsin thegas giving
ApJ,776, 136 rise to the absorption are well resolved. The absorption re-
Prochaska J. X., HennawiJ. F., Simcoe R. A., 2013, ApJ, sultsshown in thispaperuseamass resolution correspond-
762, L19 ing to a dark matter particle mass of 4.3×106h−1M⊙.
Puchwein E., Springel V., 2013, MNRAS,428, 2966 The convergence of the fractional absorption excess
Rahmati A., Pawlik A. H., Raicevi´c M., Schaye J., 2013, δ at z = 2.4 with comoving box size at a fixed mass
F
MNRAS,430, 2427 resolution, corresponding to a dark matter particle mass
RahmatiA.,SchayeJ.,BowerR.G., Crain R.A.,Furlong Mdm ≃ 4.3×106h−1M⊙, is tested in Fig. A2. The same
M., Schaller M., Theuns T., 2015, MNRAS,452, 2034 trend of increasing absorption with halo mass for decreas-
RakicO.,SchayeJ.,SteidelC.C.,RudieG.C.,2012,ApJ, ing impact parameter is recovered. For a given halo mass,
751, 94 there is no significant trend with simulation box size, al-
Rudie G. C., Steidel C. C., Trainor R. F., Rakic O., Bo- though convergence is slowest at small impact parameters
gosavljevi´c M., PettiniM., ReddyN.,ShapleyA.E., Erb for the smallest halo mass bin. The results show that the
D.K., Law D.R., 2012, ApJ,750, 67 large-scale power giving rise to fluctuations in the absorp-
Salem M., Bryan G. L., HummelsC., 2014, ApJ, 797, L18 tion has been well captured for the halo masses of interest.
SchayeJ., Crain R. A., Bower R.G., Furlong M., Schaller The absorption results in the paper are based on a simula-
M., Theuns T., Dalla Vecchia C., Frenk C. S., McCarthy tion volume of comoving side 40h−1Mpc for models with a
I.G.,HellyJ.C.,JenkinsA.,Rosas-GuevaraY.M.,White wind and 80h−1Mpc for models without.
S.D. M., 2015, MNRAS,446, 521 We examine the possible role radiative transfer may
Schaye J., Dalla Vecchia C., Booth C. M., Wiersma play on the amount of absorption using the simplifying
R.P.C.,TheunsT.,HaasM.R.,BertoneS.,DuffyA.R., approximation of an attenuated radiation field within sys-
McCarthy I. G., van de Voort F., 2010, MNRAS, 402, tems sufficiently dense to be self-shielded from an external
1536 photoionizing radiation field. We adopt the semi-analytic
SchneiderD.P.,HallP.B.,RichardsG.T.,etal.2005,AJ, prescription of Rahmati et al. (2013), usinga characteristic
130, 367 self-shieldingtotalhydrogendensityof0.0064T0.17cm−3 for
4
Shen S., Madau P., Guedes J., Mayer L., Prochaska J. X., temperature T = T/104K. Fig. A3 shows the absorption
4
Wadsley J., 2013, ApJ, 765, 89 excesses in a velocity window of width ∆v = 2000km s−1,
Somerville R.S., Dav´eR., 2015, ARA&A,53, 51 allowing for a halo redshift uncertainty σ = 520km s−1,
H
Springel V., 2005, MNRAS,364, 1105 similar to the observations of Prochaska et al. (2013). For
Springel V., Hernquist L., 2003, MNRAS,339, 289 the non-wind models, we find the effect on the mean val-
SteidelC.C.,ErbD.K.,ShapleyA.E.,PettiniM.,Reddy ues of δ is under 5 per cent., and generally less than one
F
(cid:13)c 2016RAS,MNRAS000,1–10
10 Avery Meiksin, James S. Bolton, Ewald Puchwein
20 40
FigureA1.FractionalabsorptionexcessδF relativetothemean FigureA2.FractionalabsorptionexcessδF relativetothemean
IGMabsorptionforavelocitywindow∆v=1000kms−1 around IGMabsorption withina∆v=1000kms−1 window centred on
the halo centre-of-mass velocity. The data are shown as a func- the halo centre-of-mass velocity. The data are shown as a func-
tion of the line-of-sight impact parameter b⊥ for halo masses tion of the line-of-sight impact parameter b⊥ for halo masses
log10(Mh/M⊙) = 11.1 (blue triangles), log10(Mh/M⊙) = 11.5 log10(Mh/M⊙) = 11.1 (blue triangles), 11.5 (magenta squares)
(magenta squares) and log10(Mh/M⊙) = 11.9 (black circles) at and11.9(blackcircles)atredshiftz=2.4,forthenon-windsim-
z =2.4 for simulations in a comoving box size 40h−1 Mpc with ulations. No substantial differences are found between the box
darkmatterparticlemassresolutionsof3.44×107h−1M⊙(5123 sizes withinthe errors,except forthe smallesthalomass binfor
particles; starred symbols), 4.30×106h−1M⊙ (10243 particles; whichtheconvergenceissomewhatslower.Halovelocitiesinclude
opensymbols),and5.37×105h−1M⊙(20483particles;filledsym- a random component drawn from a Gaussian distribution with
bols).Nosystematicdifferenceswithresolutionarefoundwithin standarddeviationσH=130kms−1.
theerrors,althoughthelowestresolutionsimulationsshowasmall
tendency towards slightlylarger absorptiondeficits. Haloveloci-
ties include arandom component drawnfrom aGaussian distri- of halo mass when expressed as a function of the rescaled
butionwithstandarddeviationσH=130kms−1. impact parameters.
per cent. By contrast, the wind simulation shows a marked
increase in δ when the radiative transfer correction is in-
F
cluded. The difference increases with increasing halo mass
and decreasing impact parameter, reaching correction fac-
torsashighas10–20percent.Forconsistency,inthispaper
we includetheeffects of radiative transfer for all themodel
predictions of theintegrated amount of absorption.
For the wind model, which also creates stars following
amorephysicallymotivatedmodelthanthequickLyman-α
method does, we recomputed the halo centres and peculiar
velocities by centring on the minimum of the gravitational
potential ratherthan thecentre-of-mass. Theeffects on the
integrated absorption statistics and covering fractions were
negligible, well within the dispersion in the mean values,
except for a small boost in the HI covering fractions for
theQSOswithin theinnermost 800 comoving kpc, butstill
within thestatistical errors.
As shown in Fig. A4, rescaling the impact parameter
to the virial radius of the halo masses in each mass bin re-
sults in nearly universal profiles for both the cases without
andwithwindfeedback,independentofhalomasstowithin
the Poisson errors. The enhancement in the amount of ab-
sorption produced by wind feedback is nearly independent
(cid:13)c 2016RAS,MNRAS000,1–10