The evolution of Balmer jump selected galaxies in the ALHAMBRA survey

9.4 MB·

by  P. Troncoso Iribarren

Astronomy&Astrophysicsmanuscriptno.bjg_arxiv cESO2016
(cid:13)
January22,2016
The evolution of Balmer jump selected galaxies in the ALHAMBRA
⋆
survey.
P.TroncosoIribarren1,2,L.Infante1,2,N.Padilla1,2,I.Lacerna1,2,5,S.Garcia1,2,A.Orsi3,A.MuñozArancibia1,14,
J.Moustakas4,D.Cristóbal-Hornillos3,M.Moles3,6,A.Fernández-Soto9,10,V.J.Martínez9,11,12,M.Cerviño6,7,8
E.J.Alfaro6,B.Ascaso6,13,P.Arnalte-Mur11,L.Nieves-Seoane10,11 andN.Benítez6
1 InstitutodeAstrofísica,UniversidadCatólicadeChile,Av.VicunaMackenna4860,782-0436Macul,Santiago,Chile
6 2 CentrodeAstro-Ingeniería,UniversidadCatólicadeChile,Av.VicunaMackenna4860,782-0436Macul,Santiago,Chile.
1 3 CentrodeEstudiosdeFísicadelCosmosdeAragón(CEFCA),PlazadeSanJuan1,Teruel,44001,Spain.
0 4 DepartmentofPhysicsandAstronomy,SienaCollege,515LoudonRoad,Loudonville,NY12211,USA.
2 5 MaxPlanckInstituteforAstronomy,Königstuhl17,69117,Heidelberg,Germany.
6 InstitutodeAstrofs´icadeAndalucía(IAA-CSIC),GlorietadelaAstronomía,18008-Granada,Spain.
n
7 InstitutodeAstrofísicadeCanarias,VíaLácteas/n,38200,LaLaguna,Tenerife,Spain.
a
J 8 DepartamentodeAstrofísica,FacultaddeFísica,UniversidaddeLaLaguna,38206LaLaguna,Spain.
9 UnidadAsociadaObservatorioAstronómico(IFCA-UV),E-46980,Paterna,Spain.
0
10 InstitutodeFísicadeCantabria(CSIC-UC),E-39005Santander,Spain.
2
11 ObservatoriAstronòmic,UniversitatdeValència,C/CatedràticJoséBeltrán2,E-46980,Paterna,Spain.

12 Departamentd’AstronomiaiAstrofísica,UniversitatdeValència,E-46100,Burjassot,Spain.
]
A 13 GEPI,ObservatoiredeParis,CNRS,UniversitéParisDiderot,61,Avenuedel’Observatoire75014,Paris,France.
14 InstitutodeFísicayAstronomía,UniversidaddeValparaíso,Avda.GranBretana1111,Valparaíso,Chile.
G
ReceivedOctober,2015;accepted
.
h
p
- ABSTRACT
o
r Context.Samplesofstar-forminggalaxiesatdifferentredshiftshavebeentraditionallyselectedviacolortechniques.TheALHAM-
t
s BRAsurveywasdesignedtoperformauniformcosmictomographyoftheUniverse,andwewillexploititheretotracetheevolution
a ofthesegalaxies.
[ Aims.OurobjectiveistousethehomogeneousopticalcoverageoftheALHAMBRAfiltersystemtoselectsamplesofstar-forming

galaxiesatdifferentepochsoftheUniverseandstudytheirproperties.
1
Methods.Wepresentanewcolor-selectiontechnique,basedontheBruzual&Charlot2003modelsconvolvedwiththeALHAMBRA
v
bands, and the redshifted position of the Balmer jump to select star-forming galaxies in the redshift range 0.5 < z < 1.5. These
3
galaxiesaredubbedBalmerjumpGalaxies(BJGs).WeapplytheiSEDfitBayesianapproachtofiteachdetailedSEDanddetermine
2
star-formationrate(SFR),stellarmass,ageandabsolutemagnitudes.Themassofthehaloeswherethesesamplesresidearefound
4
viaaclusteringanalysis.
5
Results.Five volume-limited BJG sub-samples with different mean redshifts are found to reside in haloes of median masses
0 1012.5 0.2M slightlyincreasingtowardz=0.5.Thisincrementissimilartonumericalsimulationsresultswhichsuggeststhatwear∼e
1. tracin±gthe⊙evolutionofanevolvingpopulationofhaloesastheygrowtoreachamassof 1012.7±0.1atz=0.5.Thelikelyprogenitors
∼
0 ofoursamplesatz 3areLymanBreakGalaxies,whichatz 2wouldevolveintostar-formingBzKgalaxies,andtheirdescendants
∼ ∼
6 inthelocalUniverseareellipticalgalaxies.Hence,thisallowsustofollowtheputativeevolutionoftheSFR,stellarmassandageof
1 thesegalaxies.
: Conclusions.Fromz 1.0toz 0.5,thestellarmassof thevolumelimited BJGsamplesnearlydoesnotchange withredshift,
v ∼ ∼
suggestingthatmajormergersplayaminorroleontheevolutionofthesegalaxies.TheSFRevolutionaccountsforthesmallvariations
i
X ofstellarmass,suggestingthatstarformationandpossibleminormergersarethemainchannelsofmassassembly.
r Keywords. extragalacticastronomy–galaxyevolution–starformation
a
1. Introduction asgalaxyscalingrelationsthatmightchangeornotwithredshift
(Mannuccietal. 2010; Elbazetal. 2011; Bouwensetal.
Most of the recent observational efforts to understand galaxy
2014; Troncosoetal. 2014), in high or low density environ-
evolution have been focused on determining the history
ments, in extreme physical conditions (starburst, AGN galax-
of cosmic star formation, gas density evolution, metallic-
ies), and in spatially resolved data due to internalvariationsof
ity evolution and mass growth of the Universe (Daddietal.
the galaxy properties (Sanchezetal. 2013). In parallel, theo-
2004; Mannuccietal. 2010; Madau&Dickinson 2014;
retical works and simulations have tried to explain the phys-
Tomczaketal. 2014; Bouwensetal. 2015).Thesemultiwave-
ical mechanisms that reproduce the measured global proper-
length observationalconstraintshave been usually summarized
ties (Daddietal. 2010; Davéetal. 2011; Lillyetal. 2013;
Lagosetal. 2014; Padillaetal. 2014). Despite these efforts,
Sendoffprintrequeststo:P.TroncosoIribarren,[email protected]
the completeness and cleanness of the sample are still chal-
⋆ BasedondataobtainedattheCalarAltoObservatory
Articlenumber,page1of14
A&Aproofs:manuscriptno.bjg_arxiv
lengingissues that dependon the sample selection-method,in- Table1.NameandeffectivewavelengthofeachALHAMBRAfilter.
struments limit, and telescope time. The aforementioned is-
sues make the comparison between observational and theoret- Name λeff Name λeff Name λeff
ical works even more difficult. For example, Campbelletal. [nm] [nm] [nm]
(2014) compared the stellar mass of GALFORM galaxies pre- U 365.5 U 396.5 B 427.5
1 2 3
dicted by the model with the ones obtained via the fit of their
B 458.5 B 489.5 B 520.5
predicted broad-band colors. They find that, for an individ- 4 5 6
ual galaxy,both quantitiesdiffer, hence the clustering of mass- B7 551.5 R8 582.5 R9 613.5
selected samples can be affected by systematic biases. There- R 644.5 R 675.5 R 706.5
10 11 12
fore, mass-selected samples might provide erroneous conclu- I 737.5 I 768.5 I 799.5
13 14 15
sions regardingtheir progenitorsand descendants. Besides, the
I 830.5 z 861.5 z 892.5
evolutionofscalingrelationsisconstrainedwithobservationsof 16 17 18
galaxysamples,selectedwithluminosityorstellar-massthresh- z19 923.5 z20 954.5
olds, located at different redshifts, which does not necessarily
Notes:Cols.1,3,and5indicatetheadoptedfilternamethatwill
constitute causally connected populations (i.e. do not follow a
progenitor-to-descendantrelationship).Clusteringselectedsam- beusedthroughoutthepaper;Cols.2,4,and6showtheeffective
plesovercomethisissuebecause,inahierarchicalclusteringsce- wavelengthofeachALHAMBRAband.
nario, a correlation analysis allows us to estimate the bias and
hence statistically determine the progenitors and descendants
galaxyredshiftandselectthesampleaccordingtocertainprob-
of galaxy samples. The bias parametermeasures the clustering
abilitythresholddefinedbytheauthors.Therefore,bydefinition
difference between the galaxy spatial distribution and underly-
thismethodselectsacleanbutnotacompletesample.
ing dark-matter distribution. Thus, it relates the typical mass
Inthiswork,weaimtodevelopatechnique,basepurelyon
of haloes hosting the galaxies (Sheth,Mo&Tormen 2001).
photometric data, to select star-forming galaxies. This kind of
Hence,measuringitingalaxysamplesatdifferentredshifts,de-
selectionallowsustodirectlycomparewiththepreviouslymen-
termineswhetherwearefollowingtheevolutionofbaryonicpro-
tioned works that also use a dropout technique to select their
cessesoccurringonhaloesofsimilarmassesornot.Thisfactis
BzK, LBG, etc. samples. We use the uniform separation be-
ofextremeimportancebecauseonceitisdeterminedwecanuse
tweentwocontiguousmedium-bandsoftheALHAMBRA sur-
themultiwavelengthdatatostudytheevolutionofthebaryonic
veytoregistertheredshiftedpositionofthe3646Å Balmerjump
processesatcertainhalomass,establishingadirectlinkbetween
withintheopticaldomain,allowingustoselectgalaxysamples
observationsandgalaxyformationmodels. Padillaetal. (2010)
intheredshiftrange0.5<z<2.
selectedearly-typegalaxiesaccordingtotheirclusteringandlu-
We base our two-color selection technique on the
minosity function in the MUSYC survey. So far, no study that
Bruzual&Charlot (2003) models and apply it to the GOLD
selects star-forming galaxies according their clustering and lu-
ALHAMBRA catalogs(Molinoetal. 2014). In the following,
minosityfunctionhasbeenreported.
the galaxies selected by this method are dubbed Balmer jump
Galaxies (BJGs) and their physical properties are investigated.
Star-forming galaxies are of particular interest, because
Based on a clustering study, we find the progenitors and de-
they allow us to study the mechanisms that switch on/off the
scendantsofgalaxiesinthesesamples,allowingustostudythe
star formation and its evolution with redshift. Considering the
evolutionoftheSEDfitderivedpropertiesasafunctionofred-
lack of wide spectroscopic surveys, in the sense of wave-
shift on haloes of certain mass. This paper is organizedas fol-
length coverage and surveyed area, the majority of the star-
lows: in section §2, we summarize the ALHAMBRA observa-
forming galaxy samples have been chosen using two-color se-
tionsandintroducethenomenclatureoftheALHAMBRAfilter
lection techniques. The so-called “dropout” technique is based
system used throughout the paper. In section §3, the selection
on recording the difference between two distinct parts of the
method is described and attested with the Bruzual&Charlot
spectrum, which generate a break on it (e.g. the 912Å Ly-
(2003)models.TheBJGsamplesaredefinedhere.Insection§4,
man break, the 3646Å Balmer jump). This difference is strong eachgalaxySEDismodeledusingiSEDfit(Moustakasetal.
enoughthatithasbeenmeasuredinbroadbands,selectingstar- 2013)andthephysicalpropertiesofeachsamplearecharacter-
forming galaxies at early periods of the Universe (z > 1.4), izedasawhole.Insection§5,theclusteringpropertiesarecal-
such as the BzK, BX, BM, DOGs and LBGs (Steideletal. culated.Insection§6,themainresultsarediscussed.Finally,in
1996; Daddietal. 2004; Steideletal. 2004; Deyetal. 2008; section §7 our conclusionsare exposed. Throughoutthe paper,
Infanteetal. 2015). Several authors have measured the bias weuseastandardflatcosmologywithH =100hkms 1Mpc 1,
0 − −
(Gawiseretal. 2007; Blancetal. 2008; Guaitaetal. 2010) Ω (z = 0) = 0.3,Ω (z= 0)= 0.7,σ = 0.824 0.029,andthe
m Λ 8
determiningthemassofthehalowhereeachgalaxysamplere- magnitudesareexpressedintheABsystem. ±
sides and connecting the progenitors and descendants of these
galaxysamples.Otherworksselectedthesamplestrustingfully
on their photometric-redshift and the physical properties de- 2. Data:theALHAMBRA survey
termined via SED-fitting (Tomczaketal. 2014), far-IR lumi-
The Advanced Large Homogeneous Area Medium-Band Red-
nosity (Rodighieroetal. 2011) or Bayesian approaches such
shift Astronomical (ALHAMBRA1) survey provides a kind
asiSEDfit(Moustakasetal. 2013).Recently, Viironenetal.
of cosmic tomography for the evolution of the contents of
(2015)implementedamethodintheALHAMBRAsurveytose-
the Universe over most of the cosmic history. Benitezetal.
lectgalaxysamplesusingtheprobabilitydistributionofthepho-
(2009)haveespeciallydesignedanewopticalphotometricsys-
tometric redshift (zPDF). The quality of the detailed SED dis-
tem for the ALHAMBRA survey, which maximizes the num-
tribution, provided by the medium bands of the ALHAMBRA
ber of objects with an accurate classification by the Spectral
survey, allowed them to perform an accurate statistical analy-
sis. Indeed, they include our lack of knowledge on the precise 1 http://www.alhambrasurvey.com
Articlenumber,page2of14
P.TroncosoIribarrenetal.:TheevolutionofBalmerjumpselectedgalaxiesintheALHAMBRAsurvey.
Energy Distribution (SED) and photometric redshift. It em- early release of the first ALHAMBRA field, they determineda
ploys 20 contiguous, equal-width 310 Å, medium-bandcov- changeintheslopeoftheK-bandnumbercountsinthemagni-
ering the wavelength range from 3∼500 Å to 9700 Å, plus the tude range 18.0 < KVega < 20.0 and that the data is complete
standard JHK near-infrared bands. Molesetal. (2008), and until KVEGA = 20 or KAB = 21.8. We have checked the K-
AparicioVillegasetal. (2010) presentedanextensivedescrip- bandnumberscountsofthe48individualpointingthatcompose
tion of the survey and filter transmission curves. The observa- theALHAMBRAGOLDcatalog.Everysinglepointingtendsto
tionsweretakenintheCalar AltoObservatory(CAHA,Spain) fallsat KAB 22.Hereafter,weusethe KAB=22limittodefine
∼
with the 3.5-m telescope using the two wide-field imagers in ourcompletesamples.
the optical (LAICA) and NIR (Omega-2000). The total sur-
veyedarea is 2.8deg2 distributedin eightfieldsthat overlapar-
3.1.Selectionofstar-forminggalaxies
eas of other surveys such as SDSS, COSMOS, DEEP-2, and
HDF-N. The typical seeing of the optical images is 1.1arcsec,
We use the models of Bruzual&Charlot (2003) convolved
while for the NIR images is 1.0arcsec. For details about the
with the ALHAMBRA filter system to define a two-color cri-
survey and data release, please refer to Molinoetal. (2014),
teria to select star-forminggalaxiesanalogouslyto the workof
and Cristóbal-Hornillosetal. (2009). In thiswork, we use the
Daddietal. (2004).TheycreatedtheBzKcolor-selectiontech-
public GOLD catalogs 2, which contains data of seven, out of
nique to cull star-forming galaxies at z>1.4. This technique is
the eight, ALHAMBRA fields. The magnitude limits of these
based on the Balmer jump, which is an indicatorof recent star
catalogues are m 25 for the four blue bands and from
AB formation.Theyusedthemodelsof Bruzual&Charlot (2003)
h i ∼
m 24.7 mag to 23.4 mag for the redder ones. The NIR
AB toidentifytheredshiftedBalmerjumpinthez-band,creatingthe
h i ∼
limitsatABmagnitudesare J 24mag, H 23mag,K 22
colorselectioncriteriaBzK (z K) (B z) = 0.2;where
≈ ≈ ≈ AB AB
mag.Inthefollowing,thefilternomenclaturepresentedinTable ≡ − − − −
BzK 0.2 selects star-forming galaxies at z > 1.4. Our ap-
1 is used and the ALH-4 and ALH-7 fields are excluded from ≥ −
proachisanaloguetotheBzK method,inthesensethatwealso
ouranalysis.Previousauthorshaveshownthatoverdensitiesre-
use the Bruzual&Charlot (2003) models, whereas we record
sideinthesefieldsandtheymightaltertheredshiftdistribution
redshiftedBalmerjumpsinvarious(diverse)medium-bandsse-
of the selected samplesas well as the clusteringmeasurements
lecting galaxies at different redshifts. In Fig.1, this situation is
(see section §5). Arnalte-Muretal. (2014) obtained the clus-
illustrated, the typical spectrum of a star-forming galaxy red-
teringpropertiesofALHAMBRAgalaxiesandstudiedthesam-
shiftedtoz 0.5,z 1.0andz 1.5isshown.TheALHAMBRA
ple variance using the seven independentALHAMBRA fields. ∼ ∼ ∼
bandsareoverplottedfromtheopticaltoNIRranges.Wechoose
Theyquantifiedtheimpactofindividualfieldsonthefinalclus-
theU -bandtosampletheregionbeforetheBalmerjumpinstead
2
tering measurements, using the Norbergetal. (2011) method,
oftheB-band.WeselecttheU bandbecauseitreachesahigher
2
they determined two “outliers regions”, i.e. ALH-4 and ALH-
completenesslevelwith respectto U1. Objects thatare not de-
7 fields. Besides, partof the ALH-4 field spatially corresponds
tected in the U band are also considered in our selection and
2
to the COSMOS field. Previous works have shown that this
its magnitude limit is used. Redder bands might sample weak
region is dominated by some large-scale structures (LSS), the
featuresbluewardsto the Balmerjumpoflow redshiftgalaxies
mostprominentarepeakingatz 0.7,and0.9(Scovilleetal.
z < 0.4. The K band is exactly the same used by Daddietal.
∼
2007).ThesestructureshaveX-raycounterpartsandaprobabil-
(2004).While,thez-bandisreplacedwiththevariableX -band,
n
ityhigherthan30%ofbeingLSS. Guzzoetal. (2007)studied
whichcoverstheopticalrangefrom613Å to954Å,i.e. X can
the clusters located at the center of the LSS (at z 0.7), while n
Finoguenovetal. (2007) found diffuse X-ray emi∼ssion in the be any ALHAMBRA filter from R9 to z20 (see Table 1). The
X -band samples the region directly red-wards to the Balmer
mostcompactstructures.ThereareotherLSSfoundintheCOS- n
jump,hencetheU X colorindicatesthegalaxyredshiftrange
MOSfield,buttheyfalloutoftheredshiftrangeofthesamples 2− n
depending on the selected n. On the other hand, the X K
studiedinthispaper. n −
color registers the duration of the star-formation age. The dif-
ferent panels in Fig.2 and Fig.3 show the theoretical evolution
3. Sampleselection (Bruzual&Charlot 2003)oftheU2XnK (Xn K) (U2 Xn)
≡ − − −
colorasafunctionofredshiftwith9<n<20(X =R ,....,z ).
n 9 20
To exclude the stars from the original ALHAMBRA Gold cat- Thered,greenandbluesolidlinesshowtheU X K-colorevo-
2 n
alog we use the stellarity index given by SExtractor (CLASS- lution of constant star-formation rate models for ages 0.2, 1,
STAR parameter C(K)) and the statistical star/galaxy separa- and 2 Gyr and reddening E(B V) = 0.3. The red, green and
tion (Molinoetal. 2014, Section 3.6) encoded in the variable bluedashedlinesshowtheU X−K-colorevolutionofpassively
2 n
Stellar-Flag (Salh) of the catalogs. Throughout the present pa- evolving models for formation redshift of zf = 2,3 and 6. In
per, we define as galaxies those ALHAMBRA sources with each panel, the star-forming models always lie in the region
C(K) ≤ 0.8 and Salh ≤ 0.5. The ALHAMBRA Gold catalog U2XnK > 0. Consequently, in order to select a sample at red-
isanF814W (i.e.almostanI-band)selectedcatalog.Thisband shifthigherthanz 0.5,a combinationthatinvolvesa X filter
n
was createdby theALHAMBRA teamas a linearcombination redderthanR must∼beused,asitisshowninthebottompanels
9
ofALHAMBRAbands(seeEq.5in Molinoetal. (2014)),and ofFig.2andFig.3.
itwasusedfortheirsourcedetection.Objectswithfaintfeatures
in this band are not detected, it might affect the completeness
of the selected samples as will be discussed furtherin sections 3.2.BJGsamples
§4.3 and §6.1. This catalogue is complete up to F814W = 23
We use the condition U X K (U X ) (X K) >
(Molinoetal. 2014),henceweusethislimittobuildourcom- 2 n 2 n n
≡ − − −
0 and the homogeneous coverage in the optical range of
pletesample.Ontheotherhand,theNIRcompletenesshasbeen
the ALHAMBRA filter system, where n is ranging from 9
studiedpreviously(Cristóbal-Hornillosetal. 2009). Using the
to 20 (X = R ,..,z ), to select galaxies at redshift z >
n 9 20
2 http://cosmo.iaa.es/content/ALHAMBRA-Gold-catalog 0.5,0.6,0.7,0.8,0.85,0.95,1.05,1.1,1.2,1.25,1.3, respectively.
Articlenumber,page3of14
A&Aproofs:manuscriptno.bjg_arxiv
Table2.PropertiesoftheBJGselectedsamples.
Sample Initialset Cleansets N <z>
Name U X K U X K Two-color
2 n 2 n
BJG R n 10 5489 0.5-0.6
1 9
≥
BJG R n 11 5174 0.6-0.7
2 10
≥
BJG R n 12 4497 0.7-0.8
3 11
≥
BJG I n 13 4012 0.8-0.85
4 12
≥
BJG I n 14 3550 0.85-0.95
5 13
≥
BJG I n 15 2878 0.95-1.05
6 14
≥
BJG I n 16 2231 1.05-1.1
7 15
≥
Fig.1.ALHAMBRAfiltertransmissioncurvecoveringfromtheoptical
BJG I n 17 2325 1.1-1.2
toNIR.Fromtoptobottom,theredcurvesshowthetypicalredshifted 8 16 ≥
spectrum of a star-forming galaxy at z 1.5, z 1.0 and z 0.5. The BJG9 z17 n 18 2058 1.2-1.25
∼ ∼ ∼ ≥
greyshadedareasmarkthepositionoftheALHAMBRAfiltersR8,I14, BJG10 z18 n 19 2140 1.25-1.3
andz ,lighteningthepositionoftheBalmerjump.ToselecttheBJG ≥
20 BJG z n=20 3391 1.3-1.5
samples,weusethefilterdirectlyredwardstothegrayfiltertosample 11 19
theendingpositionoftheBalmerjump. Notes.Col.1,Nameoftheselectedsample;Col.2,Initialfilter
setusedforselection;Col.3,Secondfiltersetsusedtosubtract
higherredshiftgalaxiesfromtheinitialsample;Col.4,Number
IneachALHAMBRAfiltersetU X K,thiscolorconditionse-
2 n
ofgalaxiesselected.Col.5,Expectedredshiftrangeofthesam-
lects star-forming and passive galaxies in the wide mentioned
ple.
redshiftrange.Theoretically,these selectedpassivegalaxieslie
always at higher redshift (i.e., ∆z > 0.25) with respect to the
selected star-forming galaxies. In order to select galaxies in a
narrow redshift range, we use more than one color condition;
firstthe(U X ) (X K)>0conditionwithn= j,toselect the zBPZ distribution of the eleven BJG samples. For a bet-
2 n n
− − − tervisualizationofthe distributiontails, the upperpanelshows
all galaxies above certain redshift z and secondly we subtract
j
thehigher-redshiftsamples,selectedwithn j+1(passiveand the BPZ distribution of the BJG1,BJG3,BJG5,BJG7,BJG9,
≥ and BJG samples, while the lower panel presents the
star-forming),from the first galaxyselection. For example,our 11
BJG ,BJG ,BJG ,BJG ,andBJG samples.Theverticalcol-
lowest redshift sample was selected by imposing the condition 2 4 6 8 10
U R K = 0 (i.e.select galaxiesat z > 0.5)and subtractingthe oredlinesshowthemedianofthezBPZ distributionforthese-
2 9
lected samples. The median of zBPZ distribution, reported on
galaxysamplesselectedbyU X K > 0withn 10,whichse-
2 n
≥ Table 3, clearly agree with the median of the expectedredshift
lect all galaxies at z > 0.6. In this way, only the star-forming
range estimated via the two-color selection criteria that are re-
galaxiesin the redshiftrange 0.5 < z < 0.6 were selected (see
portedonTable2.
yellowshadedregioninFig.2andFig.3).Followingthismethod
andusingthehomogeneousseparationbetweeneachALHAM-
BRAband,weselectelevenstar-forminggalaxysamplespeak-
4.2.SEDfitting
ingatz 0.55,0.65,0.75,0.8,0.9,1.0,1.05,1.15,1,2,1.25,and
∼
z 1.4.Itbycullingthegalaxiesthatsatisfythestar-formingcri-
ForeachgalaxyoftheBJGsamples,wefitthe20opticalbands
∼
teriaU X K (U X ) (X K)>0,whereX =R ,...,z
2 n 2 n n n 9 19 plus the three NIR bands, using the Bayesian SED modeling
≡ − − −
and“cleaned”ofhigherredshiftgalaxiesbysubtractingthesam- code iSEDfit (Moustakasetal. 2013). Once we fix the red-
ples that satisfy U X K > 0, with n 10,...,20, respectively.
2 n shift to the best-fit value given by BPZ in the ALHAMBRA
≥
Table 2 summarizes the properties of the selected samples us- catalog(Molinoetal. 2014),iSEDfitcalculatesthemarginal-
ingthemethoddescribedabove.Weensurethatallsamplesare
ized posterior probability distributions for the physical param-
roughlyindependentofeachotherbyremovingthehighredshift
eters, in certain model space that is previously defined by the
samples,whoseare selectedusingthe X filterswithn j+1
n user.UsingaMonteCarlotechnique,wegenerate20,000model
≥
(see column 3 of Table 2), from the sample selected with the SEDswithdelayed star-formationhistoriesSFH te t/τ,where
−
filterX . ∼
n=j τ is the star formation timescale. The SEDs were computed
employing the Flexible Stellar Population Synthesis models
4. Physicalpropertiesofthe BJGsamples (FSPS, v 2.4; Conroy,Gunn&White 2009; Conroy&Gunn
2010) based on the miles (Sanchez-Blázquezetal. 2006) and
Inthissection,weaimtocharacterizeeachBJGsample,provid- Basel (Lejeuneetal. 1997, 1998; Westeraetal. 2002) stel-
ing a mean characteristic value of its redshift, absolute magni- lar libraries. We assume a Chabrier (2003) initial mass func-
tude,stellarmass,age,starformationrate,etc. tion from 0.1 100 M , and a time-dependent attenuation
curve of Char−lot&Fall⊙ (2000). We adopt uniform priors
on stellar metallicity Z/Z [0.04,1.0], galaxy age t
4.1.Photometricredshifts [0.01,age(z )]Gyr,rest-f⊙ram∈eV-bandattenuation A [0 ∈
BPZ V
∈ −
We use the Bayesian photometricredshift (zBPZ) publishedin 3]mag,andstarformationtimescaleτ [0.01,age(z )]Gyr,
BPZ
∈
theGoldALHAMBRAcatalogs(Molinoetal. 2014)toverify whereage(z )istheageoftheUniverseateachgalaxy’spho-
BPZ
our selection method. The BPZ code was optimized to deter- tometric redshift. Figures9 and 10 show the SEDs of a galaxy
minephotometricredshifts,fordetailsonthezBPZcalculations randomlypickedoutofeachBJGsample.Thefilledgreendots
see Molinoetal. (2014)and Benitezetal. (2009).Fig.4shows show the photometric data, the red line shows the model that
Articlenumber,page4of14
P.TroncosoIribarrenetal.:TheevolutionofBalmerjumpselectedgalaxiesintheALHAMBRAsurvey.
Fig.2.Criteriatoselectstar-forminggalaxies.Thesolidanddashedlinesshowthecolorevolutionofstar-formingandpassivegalaxies,respec-
tively.Thered,green,bluecolorofeachlineindicatestheformationredshiftofthegalaxyz = 2,3,6,respectively.Thehorizontaldashedline
f
indicatesthecolorcutU X K (U X ) (X K)=0.Ateachpanel,thestar-forminggalaxiesarelocatedabovethisthreshold.Theyellow
2 n ≡ 2− n − n−
shadedregionsmarktheredshiftrangeoftheselectedBJGsamplesineachcolorcombination.Dashedverticallinesindicatetheredshift0.5and
1.5.
minimizes the χ2 using iSEDfit, the black squares mark the 4.3.ComparisonbetweenBJGsamples
convolution between this model and the ALHAMBRA filters.
In this subsection, we aim to compare the properties, derived
The blue shading shows the universe of models, generated by
from SED fitting, of the BJG samples for galaxies of similar
iSEDfitusing the previously described priors, scaled by their
reducedχ2.Thecolorbarindicatesthereducedχ2scale. K-bandabsolutemagnitude,i.e.similarstellarmass.Therefore,
inadditiontorestrictingthe BJGsampleswithinthemagnitude
InTable3,foreachBJGsample,themedianvalueofthesum survey limit (see section §3) we also apply an absolute mag-
oftheposteriorprobabilitydistributionsforsomephysicalprop- nitudelimitforallsamples.Inthefollowing,wesearchforthe
ertiesarereported.Theuncertaintiesindicatethe1-σconfidence mostappropriateabsolutemagnitudelimitthatallowsustobuild
level,to accountforasymmetricdistributionswedeterminethe thesesamples.Fig.5showstheK-bandabsolutemagnitude,ob-
percentiles16and84. tainedviaiSEDfit,asafunctionofBPZ redshift.Thered,or-
Articlenumber,page5of14
A&Aproofs:manuscriptno.bjg_arxiv
Fig.3.Criteriatoselectstar-forminggalaxies.Thesolidanddashedlinesshowthecolorevolutionofstar-formingandpassivegalaxies,respec-
tively.Thered,green,bluecolorofeachlineindicatestheformationredshiftofthegalaxyz = 2,3,6,respectively.Thehorizontaldashedline
f
indicatesthecolorcutU X K (U X ) (X K)=0.Ateachpanel,thestar-forminggalaxiesarelocatedabovethisthreshold.Theyellow
2 n ≡ 2− n − n−
shadedregionsmarktheredshiftrangeoftheselectedBJGsamplesineachcolorcombination.Dashedverticallinesindicatetheredshift0.5and
1.5.
ange, green, cyan, light blue and blue dots indicate the galax- K = 21.2,whichcorrespondtotheabsolutelimitwherethe
abs
−
ies in the BJG , BJG , BJG , BJG , BJG , and BJG sam- BJG sample at z 1 is complete and comparable,in terms of
1 4 6 7 9 11 5
∼
ples with apparent magnitude brighter than the magnitude sur- K luminosity,totheotherlowerredshift BJG samples.In
abs n<6
vey limit K = 22, respectively. We can note that partly due to Fig.5 the black solid line shows the absolute magnitude limit
Malquist bias, the farthest sample BJG is roughly complete K = 21.2.Wehavechosenthez 1limitbecausetheAL-
11 abs
− ∼
onlyuntil K = 22.5(bluedots).However,bychoosingthis HAMBRA Gold data is an F814W (i.e. almost an I-band) se-
abs
−
bright absolute limit to compare all BJG samples we restrict lected catalog and thus less sensitive to galaxies at z 1 with
≥
our study only to the most massive and bright objects as well a pronouncedBalmer jump. For galaxies at z 1, the spectral
≥
asenormouslyreducethestatisticsforeachsample,speciallyat region directly bluewards of the Balmer jump is barely or not
lowredshift.Hence,wedecidetostudyallobjectsbrighterthan detectedintheF814W becausetheBalmerjumpfallsintheI
13
Articlenumber,page6of14
P.TroncosoIribarrenetal.:TheevolutionofBalmerjumpselectedgalaxiesintheALHAMBRAsurvey.
Table3.MedianphysicalpropertiesoftheBJGsamples.
Name N zBPZ <M > SFR Age
M∗ M yr 1 [Gyr]
−
⊙ ⊙
BJG 5489 0.52+0.16 10.15+0.11 0.08+0.22 5.52+1.62
1 0.14 0.13 0.23 1.87
BJG 5174 0.64−+0.11 10.25−+0.12 0.27−+0.23 5.13−+1.50
2 0.18 0.13 0.22 1.76
BJG 4497 0.74−+0.10 10.30−+0.13 0.38−+0.23 4.83−+1.40
3 0.17 0.14 0.24 1.68
BJG 4012 0.81−+0.12 10.37−+0.13 0.42−+0.25 4.67−+1.29
4 0.14 0.15 0.26 1.62
BJG 3550 0.90−+0.12 10.36−+0.14 0.49−+0.24 4.37−+1.24
5 0.18 0.16 0.25 1.54
BJG 2878 0.98−+0.12 10.41−+0.15 0.52−+0.25 4.12−+1.15
6 0.19 0.17 0.27 1.46
BJG 2231 1.07−+0.12 10.46−+0.15 0.63−+0.25 3.83−+1.09
7 0.20 0.17 0.26 1.40
BJG 2325 1.15−+0.12 10.45−+0.16 0.69−+0.26 3.63−+1.05
8 0.21 0.19 0.27 1.36
BJG 2058 1.24−+0.14 10.45−+0.18 0.72−+0.25 3.41−+1.01
9 0.25 0.20 0.25 1.32
BJG 2140 1.29−+0.21 10.38−+0.19 0.72−+0.24 3.16−+1.05
10 0.33 0.23 0.22 1.29
BJG 3391 1.33−+0.28 10.34−+0.20 0.72−+0.23 2.99−+1.05
11 0.42 0.24 0.21 1.27
− − − −
Notes. Col. 1, Name of the selected sample; Col. 2, number
of galaxies selected; Col. 3, BPZ photometric redshift; Col. 4,
logarithmofthestellarmass(ChabrierIMF);Col.5,logarithm
ofthestarformationrate;Col.6,galaxyage.
Fig.4.Photometricredshiftdistributionoftheelevenselectedsamples.
For a better visualization of the distribution tails, the upper panel
shows the BPZ distribution of the BJG ,BJG ,BJG ,BJG ,BJG ,
1 3 5 7 9
and BJG samples, while the lower panel presents the
11
BJG ,BJG ,BJG ,BJG , and BJG samples. The vertical col-
2 4 6 8 10
ored lines show the median of the zBPZ distribution for the selected
samples.
andtheregiondirectlybluewardstoitfallsintheR band.Since
12
the F814W image detection is a linear combination involving
the R band, this kind of galaxies are barely or not detected
12
as sourcesin the finalcatalog. We expecta deeper selection of
galaxieswithapronouncedBalmerjumpatz <1.Saidthat,the
BJG samplesatz 6 1selectedfrom X bandsareoptimized
n 13 Fig.5.Absolutemagnitudeasafunctionofredshift.Thered,orange,
to be complete accordingto the F814W≤selection, whereas the
green,cyan,lightblue,andbluedotsindicatesthegalaxiesintheBJG ,
samplesatz 1areincomplete.Thelevelofincompletenessof 1
BJG , BJG , BJG , BJG ,and BJG samples withapparent magni-
the BJGn 6 s≥amples is difficult to quantify because there may tude4brighte6rthanK7 =22,9respectivel1y1.Theblacksolidlineshowsthe
be many ≥effects working together (e.g. the F814W is a linear absolutemagnitudecutK = 21.2.
combinationofALHAMBRAbands,undetectablebluewardsre- abs −
gionsoftheBalmerjumpduetointrinsicgalaxypropertiessuch
asredshift,dustexcess,etc.).Tominimizetheincompletenessof 5. Halomassesthroughclusteringanalysis
oursamplesinthecomparisonofthephysicalpropertiesatdif-
ferentredshifts,inthefollowingwere-definethefirstfive BJG Inhierarchicalclustering,structuresbuildupintimefromsmall
samplesselecting all galaxiesbrighterthan K = 21.2.Fur- densityfluctuations.Smallstructuresagglomeratetobuildlarge
abs
−
thermore,the BJG samples, with n 6, arenotconsideredin structures.Darkmatterhaloesarebiasedtracersoftheunderly-
n
≥
thefollowinganalysis.Table4presentsthepropertiesofthefi- ing matter density field. Massive haloes lie in higher and rarer
nal sample definition (BJGK , with n < 6), which takes into densitypeaksandaremoreclusteredthanlowermasshaloes.If
n
accounttheabsoluteluminositycut(K = 21.2).InFig.6,the agalaxypopulationishostedbyhaloesofagivenmass,thenthe
abs
−
comparison of the physical properties of galaxies with similar clusteringamplitudeof the galaxypopulation,comparedto the
K band luminosity is shown. Red squares show the properties expecteddarkmatterclusteringatthesameredshift,canbeused
−
oftheBJGK (withn < 6),thedashedlinesshowtheresultsof toderivethetypicalhalomasscorrespondingtothatgalaxypop-
n
vanDokkumetal. (2013) for galaxies of similar stellar mass. ulation.Thisisencapsulatedinthebiasparameter,b,definedas
Thecoloredareasindicatethedispersionofeachsample,while b2 = ξ (r)/ξ (r), where ξ and ξ are the 2-pointspatial
gal DM gal DM
the error bars show the median errorof the physicalproperties correlation function for galaxies and dark matter, respectively.
derivedviaiSEDfit.Insection§6.3theseresultsarediscussed. Thecorrelationfunctionatacertainredshift,ξ(r,z),canbechar-
Articlenumber,page7of14
A&Aproofs:manuscriptno.bjg_arxiv
Fig.6.EvolutionofthephysicalpropertiesoftheBJGK samples.Thepanelsshowthemedianoftheprobabilitydistributionof,fromuptodown
n<6
andlefttoright,thespecificstarformationrate,starformationrate,age, B bandluminosity,K bandluminosity,andstellarmass,respectively.
RedsquaresshowtheevolutionofBJGK galaxiesbrighterthanK = 2−1.2.Thecoloredarea−sindicatesthedispersionofeachsample,while
n<6 −
theerrorbarsshowthemedianerrorofthephysicalpropertiesdeterminedbyiSEDfit.DashedlinesshowthefittotheCANDELSanalysistaken
from vanDokkumetal. (2013)forgalaxiesofstellarmassessimilartotheMilkyWay.
Table4.MedianphysicalpropertiesoftheBJGKsamplesconsidering b2 σ2 (r ,z)/σ2 (z),whereσ (σ )istherootmean
theabsolutemagnitudelimitK= 21.2. ≃ 8,gal 0 8,DM 8,gal 8,DM
− square fluctuation amplitude in 8h 1 Mpc spheres of galaxies
−
(darkmatter).Thehalomassisobtainedfromthebiasbyusing
Name N zBPZ <M > SFR Age
M∗ M yr 1 [Gyr] the models of Sheth,Mo&Tormen (2001). We have chosen
− thisprocedureandthe8h 1Mpcscale,inordertodirectlycom-
⊙ ⊙ −
BJGK 3322 0.56+0.18 10.50+0.10 0.24+0.29 5.80+1.43 pareourresultswithpreviousworksthatconnectedprogenitors
1 0.14 0.11 0.30 1.87
BJGK 3778 0.67+−0.11 10.46+−0.11 0.40+−0.27 5.31+−1.37 anddescendantsofsamplesofstar-forminggalaxies(seeFig.7,
2 0.16 0.12 0.26 1.74 Adelbergeretal. 2005; Ouchietal. 2005; Kashikawaetal.
BJGK3 3576 0.75+−00..1102 10.45+−00..1123 0.46+−00..2266 4.92+−11..3617 2006; Gawiseretal. 2007; Hildebrandtetal. 2007;
BJGK 3480 0.83+−0.11 10.45+−0.13 0.49+−0.27 4.69+−1.24 Hayashietal. 2007; Blancetal. 2008; Hartleyetal. 2008;
4 0.09 0.14 0.28 1.60
BJGK 3150 0.92+−0.10 10.44+−0.14 0.55+−0.26 4.37+−1.19 Yoshidaetal. 2008; Guaitaetal. 2010; McCrackenetal.
5 0.12 0.15 0.26 1.52 2010; Linetal. 2012). More detailed treatments, such as
− − − −
consideringbothhalotermsanddifferentscales,arebeyondthe
Notes. Col. 1, Name of the selected sample; Col. 2, number
scopeofthispaper.
of galaxies selected from the BJG samples after the K-band
absolute cut; Col. 3, BPZ photometric redshift; Col. 4, stellar
mass(ChabrierIMF);Col.5,starformationrate;Col.6,galaxy
5.1.Angularcorrelationandcorrelationlength
age.
Todeterminetheangularcorrelationfunction,weusetheLandy
andSzalay(1993)prescription
w(θ)=(N 2N +N )/N , (1)
acterizedwithapowerlawandacorrelationlength,r ,through gg gr rr rr
0 −
ξ(r,z) = (r/r0(z))γ,whereγisthepowerindex.Forsparsesam- where Ngg and Nrr are the number of pairs at a separation
plesand atsmall scales thisrepresentationis typicallyused. In θ of galaxies in the catalog and points in a random catalog
largersurveysandsimulationsthecorrelationfunctionisusually with the same layout as the galaxy sample, respectively. N
gr
modelledcombiningtermsforgalaxiesinthesamehaloandin is the cross number of points between the galaxy and random
differentones(Zehavietal. 2011; Contrerasetal. 2013). distributions. For our calculations we consider five ALHAM-
In the following, we measure the bias and masses of the BRA fields (see section §2) that encompass 36 pointing cata-
halos that host the BJGK samples. Specifically, for each logs. Errors were estimated by a jackknife method, which has
n<6
BJGK sample, we calculate the 2-point angular correlation been shown as a robust error estimator (Zehavietal. 2011;
n<6
function, and then, using Limbers’ de-projection, given a Cabréetal. 2007),althoughitcanoverestimatethevarianceon
redshiftdistributionandacosmologicalmodel,wecalculatethe smallscales(Norbergetal. 2009). We havecalculatedthe an-
correlationlength. The bias parameter followsfrom r through gular correlation function 36 times, each time eliminating one
0
Articlenumber,page8of14
P.TroncosoIribarrenetal.:TheevolutionofBalmerjumpselectedgalaxiesintheALHAMBRAsurvey.
Table5.ClusteringpropertiesoftheBJGKsamples.
Name N <z> <K > A r Bias logM
gal abs w 0 h
BPZ mag 10 3 [h 1Mpc] (r=8h 1Mpc) [h 1M ]
− − − −
⊙
BJGK 3332 0.56+0.2 -22.3 4.2 0.2 4.67 0.33 1.36 0.09 12.71+0.13
1 0.1 ± ± ± 0.15
BJGK 3778 0.67−+0.1 -22.3 4.3 0.5 4.58 0.39 1.41 0.11 12.66−+0.15
2 0.2 ± ± ± 0.17
BJGK 3576 0.75−+0.1 -22.4 4.0 0.2 3.79 0.25 1.24 0.07 12.27−+0.14
3 0.1 ± ± ± 0.16
BJGK 3480 0.83−+0.1 -22.4 4.2 0.1 4.15 0.23 1.40 0.07 12.45−+0.10
4 0.1 ± ± ± 0.12
BJGK 3150 0.92−+0.1 -22.5 3.7 0.2 4.01 0.23 1.40 0.07 12.37−+0.11
5 0.1 ± ± ± 0.12
− −
Notes.Col.1,Nameoftheselectedsample;Col.2,numberofgalaxiesselectedfromtheBJGsamplesaftertheK-bandabsolutecut
andimagemask;Col.3,BPZphotometricredshift;Col.4,mediumK-bandabsolutemagnitude;Col.5,amplitudeofcorrelation;
Col.6,correlationlength;Col.7,biasfactorcalculatedwiththevarianceat8h 1 Mpc.Col.8,logarithmofthehalomassinunits
−
of[h 1M ].
−
⊙
catalog out of the 36 available. The uncertaintyis estimated as the spatial correlation function by b2 ξ (r,z)/ξ (r,z) or
gal DM
≈
thevarianceofw(θ).We followthemethodof Infante (1994); b2 σ2 (r,z)/σ2 (r,z).Thenumeratorξ (r,z) J σ2 (r,z)
Quadrietal. (2007) to correct for the integral constraint. As ista≈kengaflrom PeDebMles (1980)(Eq.59.3),wghalere J ≈isa2pgaarlame-
2
mentionedabove,tocalculater0 weusethepowerlawapprox- ter defined in terms of γ as J2 = 72/[(3 γ)(4 γ)(6 γ)2γ].
imationξ(r,z) = (r(z)/r0(z))−γ thatdependsonthespatialsepa- We fix the spatial scale at 8 h−1 Mpc −such th−at ξgal(−8,z) =
rationandredshift.Likewise,the2-pointangularfunctionturns (r (z)/8h 1Mpc)γ.Theevolutionofthedarkmatterdensityvari-
out to be w(θ,z) = Aw(z)θ(1−γ), where Aw(z) is the angular am- an0ce in a−comovingsphere of radius 8h 1 Mpc is σ2 (8,z) =
plitude. By fitting the correlation function to this power law, − DM
σ D(z),whereD(z)isthelineargrowthfactoratredshiftz.The
8
between 0.005and 0.2 degree,the Aw(z) is inferred.The angu- biasmeasuredforeachBJGKsample,atthescaleof8h 1Mpc,
−
lar amplitudes are reported in Table 5. To calculate the spatial
isreportedinTable5.
fromtheangularfunction,weusetheLimber’s(1953)inversion,
The figure 7 shows the evolution of the bias factor as
whichrequiresaredshiftdistributionN(z),andassumeacosmo-
a function of redshift for the BJGK samples (red squares).
logicalmodel.Limber’sinversioninvolvessolvingthefollowing
The lines show the bias evolution for different halo masses
integral(Kovacetal. 2007),
(Sheth,Mo&Tormen 2001), whose are indicated below of
2 each line in the right side of the panel. Approximately, the
rγ = Aw(c/H0)[R N(z)dz] , (2) BJGK samples follows the bias evolution of haloes of masses
0 CγR F(z)D1θ−γ(z)N2(z)gd(z)dz ∼tio1n0(182.)5ho−f1MS⊙h.eFthin,Malloy,&wTeocramlceunlat(e2t0h0e1)hatlhoatmraeslsatuessintgheeqbuiaas-
where the cosmology plays the role in the Hubble parameter with the peak height, ν = δ (z)/σ(M,z), where δ is the crit-
sc sc
tHh0e,agndg(uz)lar=dia(m1et+erzd)2ispta1n+ceΩDmθz.+TΩheΛ[p(a1r+amz)e−te2r−C1γ],daenpdendins iHcaelreowveerdasesnusmityecδoscm=pu1t.e6d9.uTsihnegn,ththeeshpahleorimcaalsscoMllhapisseobmtaoidneeld.
on the power index such that Cγ = Γ(0.5)Γ[Γ0.[50(.γ5−γ]1)]. In order ltihsrtoeudgihntσa(bMle,4z.)Ienvfialgu.a8tetdheathtahloemreadssshoifftsthoefBthJeGBKJGsaKmpsalemspalseas
to be consistent internally and compare with other works
functionofredshiftisshown.Squaresshowthemedianvaluesof
(Adelbergeretal. 2005; Ouchietal. 2005; Kashikawaetal.
thehalomass,polygonsshowaconservativelimit,drawingthe
2006; Hayashietal. 2007; Hildebrandtetal. 2007;
masslimitstakingintoaccountthe1σdeviationoftheredshift
Gawiseretal. 2007; Blancetal. 2008; Hartleyetal. 2008;
andbias.TheseresultsarepresentedinTable5.
Yoshidaetal. 2008; Guaitaetal. 2010; McCrackenetal.
2010; Linetal. 2012),thevalueofγwasfixedtothecanonical
γ = 1.8. This value is fully justified by experiments where γ
6. Discussion
wasleftfreeandisconsistentwiththeclusteringofluminosity-
selected ALHAMBRA samples (Arnalte-Muretal. 2014).
6.1.Selectiontechnique
F(z) accounts for the redshift evolution of the correlation
function,where F(z) = (1+z)−(3+ǫ), andwe use ǫ = −1.2 that The new two-color selection technique U2XnK based on AL-
corresponds to the value adopted for a constant clustering in HAMBRA medium-bands, allows us to extract eleven sam-
comovingcoordinates(Quadrietal. 2007).Forthecalculation ples of star-forming galaxies at z >0.5 in narrower redshift
of the correlation length error, the error contribution from the ranges with respect to previous studies (Daddietal. 2004;
amplitude of the angular correlation function, the Poissonian Steideletal. 2004; Adelbergeretal. 2005). This selection
errors ( √N(z)), and the effects of the photometric redshift method is based purely on the ALHAMBRA photometric data
∼
errorinshiftingandbroadeningN(z)aretakenintoaccount. andit doesnotdependson the modelsassumptions,methodol-
ogyortemplatesusedtodeterminethephotometricredshiftsor
properties derived from SED fit. We have validated this tech-
5.2.Biasandmassmeasurements
nique with the Bruzual&Charlot (2003) models as well as
In turn we estimate the bias parameter, b, from the correla- checked with other stellar population synthesis models giving
tion length. As pointed out above, b is related to r through consistentresults.
0
Articlenumber,page9of14
A&Aproofs:manuscriptno.bjg_arxiv
Fig.7.Evolutionofthebiasfactor,calculatedwiththevarianceat8h 1Mpc,asfunctionofredshiftforsamplesofstar-forminggalaxies.Red
−
diamondsindicatestheBJGK samples.Atz 2,thegrey,blue,red,green,andlightgreensquaresshowthesBzKsamplesof Hayashietal.
n<6 ∼
(2007); Blancetal. (2008); Hartleyetal. (2008); Linetal. (2012),and McCrackenetal. (2010),respectively.YellowsquaresshowtheBM,
BX selected galaxies of Adelbergeretal. (2005). At z 3, the blue, green and red squares indicates the LBGs selected by Yoshidaetal.
∼
(2008); Kashikawaetal. (2006); Hildebrandtetal. (2007); Ouchietal. (2005), respectively. Gray squares show the Ly-α emitters selected
by Guaitaetal. (2010); Gawiseretal. (2007).Thelinesshowthebiasevolutionfordifferenthalomasses(Sheth,Mo&Tormen 2001),whose
areindicatedbelowofeachlineintherightsideofthepanel.
The ALHAMBRA GOLD data is an F814W-selectedcata- thereforethenumberdensitiesmightbeunderestimatedaccord-
log,thusisless sensitiveto detecttheregionsbluewardsof the ingtoδ (BPZ)=0.014z(BPZ).
z
Balmerjumpofgalaxiesatz>1.TheR bandcontributesonly
12
to 10%of the final F814W detectionimage, while the I , and
13
According to this colour technique (visual inspection of
I bandscontribute18%each one.Hence, the detectionin the
14
Fig.2, and Fig.3), the redshift distributions should be narrower
F814W image becomesworse for galaxiesat z 1 with a pro-
≥ than in Fig. 4. The distributions can become wider if the er-
nouncedBalmerjump.
rors of the photometric redshifts are properly taken into ac-
Inordertoavoidanycontaminationoflow-redshiftgalaxies
count.ThemeanformalBPZerroroftheBJG andBJG sam-
inoursamples,weareinclinedtousethebluestbandavailable 1 11
ples are 0.02 and 0.03, corresponding to 20% and 30% of the
tosampletheregionbluewardsoftheBalmerjump.Hence,we
total expected width ( 0.1), respectively. We performed sim-
studiedthedetectionleveloftheU1andU2thatreaches97%and ulations considering G≤aussian redshift distributions filled ran-
99.7%, respectively. Since the U has a higher detection level
2 domlyusingthesameamountofgalaxiesfoundineachsample,
withrespecttoU ,wehavetailoredtheselectiontechniqueusing
1 the expected width (0.1) and the formal BPZ error δ (BPZ) =
theU band. z
2 0.014(1+z(BPZ)).Byperturbingthephotometricredshiftwith
Toselectsamplesatz<0.5,wetriedtousefiltersbluerthan
itscorrespondingerror,andchoosingrandomlyapositiveorneg-
theR9, whosecentralwavelengthislowerthan613.5Å.Never- ativevariancethewidthofthedistributionsincreasebyafactor
theless, the theoretical evolution of the color U2Xn<9K of pas- of 1.5 for the first BJG1 sample, and up to 2 times this value
siveandstar-forminggalaxies,basedonthe Bruzual&Charlot for the BJG sample, with increasing redshift. Clearly, higher
11
(2003) models, tend to occupy the same locus in the color- photometric redshift errors increase the distribution width. In
redshift(U2Xn<9K z)plane.Hence,itdoesnotallowtosepa- the AHAMBRA data, BPZ tends to decrease its precision at I
−
ratethestar-formingfromthepassivegalaxiesasclearasforthe <24.5AB(Molinoetal.2014).Theglobalredshiftdistribution
U2XnK selectionwithn>9(seeFig.2andFig.3). showsameanof<z>= 0.86atthismagnitudelimit.Hencethe
ThenumberdensityofeachBJGsampledecreaseswithred- distributionsat z 0.9 tend to broadeneven morebecause the
≥
shift,probablyduetothenatureoftheseobjectsoralsorelatedto photometric redshift errors are bigger than 0.014(1+ z). Con-
anon-consideredMalquist-biasselectioneffectintheU -band. sideringadistributionatz 1withanoriginalwidthof0.1and
2
∼
Besides, we verify our selection method with the BPZ photo- δ (BPZ)=0.03(1+z(BPZ)),thewidthincreaseuptothreetimes.
z
metric redshifts, whose uncertainty increase with redshift, and Outlayersdue to either photometricredshift or the color selec-
Articlenumber,page10of14