Time delay of SBS 0909+532

0.42 MB·English

by  A. Ullan

Astronomy&Astrophysicsmanuscriptno.camawi February5,2008
(DOI:willbeinsertedbyhandlater)
Time delay of SBS 0909+532
A.Ulla´n1,L.J.Goicoechea1,A.P.Zheleznyak2,E.Koptelova3,V.V.Bruevich3,T.
Akhunov4,andO.Burkhonov4
6
0
0 1 DepartamentodeF´ısicaModerna, UniversidaddeCantabria,Avda.deLosCastross/n,
2
39005Santander,Spain

n
e-mail:[email protected], [email protected]
a
J
2 InstituteofAstronomyofKharkov NationalUniversity, Sumskaya35,61022 Kharkov,

0
2 Ukraine

e-mail:[email protected]
1
v 3 SternbergAstronomicalInstitute,Universitetskipr.13,119992Moscow,Russia
3
7 e-mail:[email protected], [email protected]
4 4 UlugBegAstronomicalInstituteofUzbekAcademyofScience,Astronomicheskaya.Str.
1
0 33,700052Tashkent,RepublicofUzbekistan
6
e-mail:[email protected], [email protected]
0
/
h
p AcceptedJanuary12,2006
-
o
r
t Abstract. Thetimedelaysbetweenthecomponentsofalensedquasararebasictoolsto
s
a analyzetheexpansionoftheUniverseandthestructureofthemainlensgalaxyhalo.Inthis
:
v paper,wefocusonthevariabilityandtimedelayofthedoublesystemSBS0909+532A,B
i
X
as well as the time behaviour of the field stars. We use VR optical observations of SBS
r
a 0909+532A,Bandthefieldstarsin2003.TheframesweretakenatCalarAlto,Maidanak
andWiseobservatories,andtheVRlightcurvesofthefieldstarsandquasarcomponentsare
derivedfromapertureandpoint–spreadfunctionfittingmethods.WemeasuretheR–band
timedelay of the system fromthe χ2 and dispersion techniques and 1000 synthetic light
curvesbasedontheobservedrecords.Onenearbyfieldstar(SBS0909+532c)isfoundto
bevariable,andtheothertwonearbyfieldstarsarenon–variablesources.Withrespectto
thequasarcomponents,theR–bandrecordsseemmorereliableandaremoredenselypop-
ulatedthantheV–bandones.TheobservedR–bandfluctuationspermitapre–conditioned
measurement of the time delay. From the χ2 minimization, if we assume that the quasar
emissionisobservedfirstinBandafterwardsinA(inagreementwithbasicobservations
of the systemand thecorresponding predictions), weobtain ∆τ =− 45+1 days (95%
BA −11
confidenceinterval).Thedispersiontechniqueleadstoasimilardelayrange.Aby–product
oftheanalysisisthedeterminationofatotallycorrectedfluxratiointheRband(corrected
bythetimedelayandthecontamination duetothegalaxy light).Our 95%measurement
2 A.Ulla´netal.:TimedelayofSBS0909+532
∆m =m (t+∆τ )−m (t)=0.575±0.014magisinexcellentagreementwithprevious
BA B BA A
resultsfromcontaminatedfluxesatthesametimeofobservation.
Keywords. Gravitationallensing–Quasars:general–Quasars:SBS0909+532–Stars:
variables:general
1. Introduction
The system SBS 0909+532 was discovered by Stepanyan et al. (1991). Some years later,
a collaboration between the Hamburger Sternwarte and the Harvard–Smithsonian Center for
Astrophysics resolved the system into a pair of quasars (A and B) with a direct R–band flux
ratio(atthesametimeofobservation)∆m=m −m =0.58magandaseparationofabout1′.′1
B A
(Kochaneketal.1997).ThedirectR–bandfluxratiowasnotconsistentwiththedirectfluxratios
at other wavelengths: ∆m = 0.31 mag in the I band and ∆m = 1.29 mag in the B band. From
observationswiththe4.2mWilliamHerschelTelescope,aSpanishcollaborationgotspectrafor
eachcomponentofthesystem.Thedatashowedthatthesystemconsistsoftwoquasarswiththe
sameredshift(z =1.377)andidenticalspectraldistribution,supportingthegravitationallensin-
s
terpretationofSBS0909+532(Oscozetal.1997).Oscozetal.(1997)detectedaMgdoubletin
absorptionatthesameredshift(z =0.83)inbothcomponents,andtheysuggestedthattheab-
abs
sorptionfeatureswereassociatedwiththephotometricallyunidentifiedlensinggalaxy.Through
a singular isothermal sphere (SIS) lens model, the authors also inferred the first constraint on
thetimedelaybetweenthecomponents:|∆τ | ≤140days,where∆τ isthedelayofBwith
BA BA
respecttoAandtheHubbleconstantisassumedtobeH =70kms−1Mpc−1.
0
Inrecentyears,Lubinetal.(2000)indicatedthepossiblenatureofthemaindeflector(early–
typegalaxy)andconfirmeditsredshift(z =0.830).Leha´retal. (2000)reportedona program
d
includingHubbleSpaceTelescope(HST)observationsofSBS0909+532.Theydiscoveredthe
mainlensgalaxybetweenthecomponents,whichhasalargeeffectiveradius,withacorrespond-
ingly low surface brightness. This lens galaxy is closer to the brightest component(A), which
isnotincontradictionwithSIS–likelensmodelswhenthefartherandfaintercomponent(B)is
strongeraffectedbydustextinction(seebelow).Thecolorsofthelensareconsistentwiththose
ofanearly–typegalaxyatredshift0.83.Assumingasingularisothermalellipsoid(SIE)model,
Leha´retal.predictedatimedelay∆τ intherange[−10,−87]days(H =70kms−1Mpc−1).
BA 0
Ata givenemissiontime,the sign”−”meansthatthecorrespondingsignalis observedfirstin
BandlaterinA.TheCOSMOGRAILcollaborationprovidedthedistributionofpredictedtime
delaysofthesystem(Sahaetal.2005).Intheirhistogram(Fig.10ofSahaetal.),therearetwo
features: the main feature is an asymmetric peak around − 80 days and the secondary one is
another asymmetric peak around − 45 days. Therefore, if the COSMOGRAIL predictions are
right,thetimedelayisveryprobablyof2–3months(componentBleadingcomponentA),but
we cannot rule out a delay of about one and a half months. On the other hand, the flux ratio
Sendoffprintrequeststo:A.Ulla´n
A.Ulla´netal.:TimedelayofSBS0909+532 3
anomalypointed by Kochaneket al. (1997)was confirmedand accurately studied by Motta et
al. (2002)and Mediavilla et al. (2005),who reportedthe existence of differentialextinctionin
themainlensgalaxy.Chartas(2000)andPageetal.(2004)alsostudiedthesystemintheX–ray
domain.
TimedelaysarebasictoolstodiscussthepresentexpansionrateoftheUniverseandthestruc-
ture of the main lens galaxy haloes (e.g., Refsdal 1964; Kochanek, Schneider & Wambsganss
2004), so that variability studies are crucial. While some time delays have been measured
fromradiolightcurves(PKS1830−211:Lovelletal.1998;Q0957+561:Haarsmaetal. 1999;
B0218+357: Biggs et al. 1999; B1600+434: Koopmans et al. 2000; B1422+231: Patnaik &
Narasimha 2001; B1608+656:Fassnacht et al. 2002) or X–ray variability (e.g., Q2237+0305:
Dai et al. 2003), an important set of delays are based on optical monitoring of gravitation-
allylensedquasars.OpticalframestakenatApachePointObservatory,FredLawrenceWhipple
ObservatoryandTeideObservatorywereusedtoestimatea14–monthdelayforthedoublesys-
temQ0957+561(e.g.,Peltetal.1996;Kundic´etal.1997;Serra–Ricartetal.1999;Ovaldsenet
al.2003).Althoughthetimedelayofthisfirstmultiplequasarhasbeenconfirmedthroughinde-
pendentobservations,the measurementis only5% accurate,or equivalently,there is an uncer-
taintyofabout20days(Goicoechea2002).TheTel–AvivUniversity(TAU)grouphaverecently
determinedthetimedelaybetweenthetwocomponentsofHE1104−1805(Ofek&Maoz2003).
TheTAUdelayofHE1104−1805disagreeswiththeearlierestimationbyGil–Merino,Wisotzki
&Wambsganss(2002),butitisinexcellentagreementwiththedeterminationbyWyrzykowski
etal.(2003).Schechteretal.(1997)measuredtwodelaysforthequadruplyimagedquasarPG
1115+080.TheBelgian–Nordiccollaborationcarriedoutaveryintenseactivityduringthepast
fiveyears.Theyparticipatedinseveralmonitoringprojectsandmeasuredseveraltimedelaysat
opticalwavelengths:B1600+434(Burudetal.2000),HE2149−2745(Burudetal.2002a),RXJ
0911.4+0551(Hjorth et al. 2002), SBS 1520+530 (Burud et al. 2002b) and FBQ 0951+2635
(Jakobssonetal.2005).Theformalaccuraciesofthese5estimationsrangefrom5to25%(the
1σ error bars vary from 4 to 24 days). Kochanek et al. (2005) also measured the time delays
betweenthecomponentsofthequadruplequasarHE0435−1223.
The aim of this paper is to present VR observations of SBS 0909+532in 2003 conducted
by the University of Cantabria (UC, Spain), the Institute of Astronomy of Kharkov National
University(IAKhNU,Ukraine),theSternbergAstronomicalInstitute(SAI,Russia)andtheUlug
BegAstronomicalInstituteofUzbekAcademyofScience(UBAI,Uzbekistan).Wealsopresent
TAUobservationsofthefieldstarsin2003,whichhavebeenkindlymadeavailabletous.This
new optical monitoring campaign was carried out at the Calar Alto Observatory (Spain), the
Maidanak Observatory (Uzbekistan) and the Wise Observatory (Israel), and the frames were
takenwith the1.5mSpanishtelescope,the1.5mAZT−22telescopeatMt.Maidanakandthe
WiseObservatory1mtelescope(Section2).InSection3,wedescribethemethodologytoobtain
the fluxesof the quasarcomponentsandthe field stars. The VR lightcurvesare also shownin
Section 3. Section 4 is devoted to the time delay estimation from the light curves of A and B
4 A.Ulla´netal.:TimedelayofSBS0909+532
s1
N
s2
E
b
c
GL
a
x
Fig.1.CalarAltoimageofSBS0909+532,showingaFOVof∼ 7′×7′.Thefieldcontainsthe
gravitationallylensedquasar(”GL”)andsixbrightandnon–saturatedstars(”a–c”,”s1–s2”and
”x”).Thenearby(”a–c”)andrelativelyfar(”s1–s2”)fieldstarswereintroducedbyKochaneket
al. (1997)and Nakos et al. (2003),respectively.A sixth field star (”x”) is also included in the
FOV.
(quasar components) in the R band. Finally, in Section 5 we summarize our conclusions and
discussthefeasibilityofanaccuratedeterminationofthecosmicexpansionrateandthesurface
densityinthemainlensinggalaxy.
2. Observations
We havethreedifferentsets offramesforSBS 0909+532.Thefirstsetofopticalframescover
the period between 2003 March 4 and June 2, and they are part of a UC project to test the
feasibilityofquasarmonitoringprogramsthrough1−2mtelescopesinSpain(Ulla´n2005).These
observationsweremadewiththe1.52mSpanishtelescopeatCalarAltoObservatory(EOCA),
Almeria, Spain (see Ziad et al. 2005 for a site–testing on Calar Alto). The EOCA is equipped
with a Tektronics 1024×1024 CCD detector, which has pixels with a physical size of 24 µm,
givinga0.4arcsecpixel−1angularscale.Thegainis6.55e−/ADUandthereadoutnoiseis6.384
e−. During this first monitoring,exposuresin the V and R Johnson–Cousinsfilters were taken
everynightwhenclear,whatmakesatotalof20observingnights.Badweatherin2003March
andAprilpreventedusfromachievingaverydensesampling.Foreachmonitoringnightwehave
A.Ulla´netal.:TimedelayofSBS0909+532 5
Fig.2.Maidanakzoomed–inimageofSBS0909+532(about8′′onaside).Wechooseoneofthe
framestaken with the best seeing conditions, and then we expandthe portion occupiedby the
twoquasarcomponentsA(thebrightestobject)andB(thefaintestobject),i.e.,theGLinFig.1.
To avoidthepixellationeffectwhena subframeis expanded,the brightnessdistributionisalso
smoothed.
three consecutiveframeson each filter, i.e.,three 300s exposuresin the V passbandand three
180sexposuresintheRpassband.Thosewerethemaximumexposuretimestoavoidsaturation
ofselectedstarsinthefield.InFigure1weshowatypicalframe.Inthistypicalexposure,half
adozenbrightandnon–saturatedstarswerefittedwithinthefieldofview(FOV).Followingthe
notationof Kochaneketal. (1997),theFOV includedthe gravitationallylensedquasar(”GL”)
andnearbyfieldstars”a”(South),”b”(North)and”c”(West).TheFOValsoincludedtwostars
thatwereintroducedbyNakosetal.(2003)andwerelabelledas”s1”and”s2”.Thesetwostars
areplacedrelativelyfarfromthegravitationallenssystem,andtheyappearclosetotheNorth–
Westedgeoftheframe(seeFig.1).Asixthstar(”x”)appearsclosetotheSouth–Westedgeof
thetypicalframe.
ThesecondsetofobservationsincludeframesinFebruary2003aswellasduringApril–May
andOctober–November2003.Thetotalnumberofnightsis18.Inthissecondprogramtheim-
agesweretakenwiththe1.5mAZT−22telescopeatMaidanakObservatory(Uzbekistan),with
neardiffraction–limitedopticsandcarefulthermo–stabilization,whichallowforhigh–angular–
resolution imaging. The AZT−22 telescope has a LN–cooled (liquid nitrogen cooled) CCD–
camera,SITe-005CCD,manufacturedinCopenhagen(Denmark).Forthiscamera,theimaging
areaissplitinto2000×800pixels,wherethepixelsizeis15µmandtheintrinsicangularscale
is0.26arcsecpixel−1.TheframesweretakenintheRBesselfilter,whichcorrespondsapprox-
imatelytotheRJohnson–Cousinspassband.Thepoortrackingsystemofthistelescopeallows
onlyexposuresupto3minutes.Toobtainsufficientlyhighphotometricaccuracy,wetooksev-
eral frames each observation night. With respect to the rectangular FOV of the telescope, the
North/Southcoveragewas 2.5 timessmaller than the East/West one, so the ”s1”, ”s2” and ”x”
starswerenotincludedwithintheFOV. Figure2 showsazoomed–inimagemadefromoneof
the best frames in terms of seeing. There are two close quasar components, but the very faint
galaxyisnotapparent.TheobservationsatMt.MaidanakarepartofIAKhNU,SAIandUBAI
projectstofollowupthevariabilityofgravitationallylensedquasars.
6 A.Ulla´netal.:TimedelayofSBS0909+532
Observatory(Telescope) Frames/night(Filter) ObservationPeriods
CalarAlto(1.5m) 3×300s(V)+3×180s(R) March–June
Maidanak(1.5m) (3–11)×180s(R) February,April–May,October–November
Wise(1.0m) 1×420s(R) 22unevenlydistributednights
Table1.ObservationsofSBS0909+532in2003.
For the past six years the TAU group have been monitoring several gravitationally lensed
quasars with the Wise Observatory 1 m telescope. The targets are mainly monitored in the
Johnson–Cousins R–band, and the frames are obtained with a cryogenically cooled Tektronix
1024×1024–pixelback–illuminatedCCD. The angular scale is 0.7 arcsec per pixel. This pixel
scale and the median seeing (FWHM) of about 2′′ do not allow resolving most of the lensed
objects,e.g.,SBS0909+532.However,theframesofSBS0909+532in2003arecharacterized
bywideFOVs,whichincorporatethe”a–c”,”s1–s2”and”x”stars.Thisfactpermitstododiffer-
entialphotometrybetweenseveralpairsoffieldstars,andthus,totestthereliabilityoftheCalar
AltoandMaidanakrecords.
The pre–processing of the images included the usual bias subtraction, flat fielding using
sky flats, sky subtraction and cosmic ray removalby using the Image Reduction and Analysis
Facility (IRAF) and Munich Image Data Analysis System (MIDAS) environments. Some de-
tailsaboutthewholeobservationalcampaignareincludedinTable1(observatories,telescopes,
frames/night,filtersandobservationperiods).
3. PhotometryandVRlightcurves
Duetothesmallangularseparationbetweenthetwolensedcomponents,about1′.′1(Kochanek
etal. 1997),the photometryof SBS 0909+532is a difficulttask. This task is also complicated
by the presence of the main lensing galaxy between the components, which could make the
computation of individual fluxes even harder. In general, aperture photometry does not work,
so we must look for better approaches.An initial issue is to decide about the inclusion or not
inclusionofaphotometricmodelforthelensinggalaxy.Inprinciple,whencomputingthefluxes
of SBS 0909+532 we may use a galaxy model derived from the HST images of the system.
The galaxy model could also be inferred from the best images in terms of seeing. Once the
relevantinformationonthegalaxyisknown,wewouldapplyaPSFfittingmethodtoalloptical
images,settingthegalaxypropertiestothosederivedfromtheHSTorthebest–qualityimages,
and allowing the remaining parameters to vary (e.g., McLeod et al. 1998; Ulla´n et al. 2003).
Magain,Courbin&Sohy(1998)alsopresentedanalternativetask(deconvolution)thatcombines
all the frames obtained at different epochs to determine the numerical light distribution of the
lensing galaxy as well as the positions of the point–like sources (quasar components), since
these parametersdo notvary with time. The flux of the point–likesourcesare allowed to vary
fromimagetoimage,whichproducesthelightcurves.However,theseandotherprocedureshave
A.Ulla´netal.:TimedelayofSBS0909+532 7
areasonablelimitation:theyonlyworkwellwhenthegalaxylighthasasignificantcontribution
to the crowdedregionsin the individualframes. For a veryfaint galaxyin a standard(i.e., not
superb)frame,thereisconfusionbetweengalaxysignalandnoise,sotheuseofagivengalaxy
modelcouldleadtobiasedfluxesofthecomponents.Thebiaseswilldependonthequalityofthe
image (seeing, signal–to–noiseratio, etc), which must produceartificial variabilitysuperposed
totherealone.Ontheotherhand,theuseofadirectPSFfittingmethod(neglectingthegalaxy
brightness)leadsto contaminatedfluxesofthe components.Butif the galaxyisveryfaint,the
contaminations will be small. Moreover, the variation of the quasar fluxes, seeing conditions,
etc, will cause fluctuations in the contaminations, which are expected to be below the typical
contaminationlevels.Forstandardframesofaquasarlensedbyaveryfaintextendedobject,it
isreallydifficulttochoosebetweenbothapproaches(withandwithoutgalaxy).
MostoftheCalarAltoandMaidanakindividualframesofSBS0909+532donotshowevi-
dencesforagalaxybrightnessprofile.Thisfactisduetothefaintnessofthegalaxy,aswecorrob-
orateherebelow.Ifweconsideranhypotheticalastronomerthatneglectsthegalaxybrightness
anddoesdirectPSFfitting(withouttakingintoaccountthegalaxywhendoingthecomputation
ofthefluxes),itispossibletoattainaroughestimationofthemaximumcontaminationfromthe
galaxytotheclosestcomponentA(at0′.′4fromthecentreofthedeflector).Wetakeintoaccount
thepaperabout10lenssystemsbyLeha´retal. (2000),where,inTable3,we canfindthebest
availablephotometricandastrometric(HST)data ofSBS 0909+532.Theauthorswere ableto
tracethegalaxylightintheH passband,bymeasuringitspositionandbrightness.Ifweusethe
colorsinthesametable,weconcludethatm ∼19magandm ∼16magintheI band(near–
gal A
IR),andm >20.4magandm ∼16.7magintheV opticalband.Therefore,astheRfilteris
gal A
placedjustbetweentheIfilterandtheV one,wemayassumethatm −m ∼3.5magintheR
gal A
band.Thedifferenceof3.5magisconsistentwitharatiooffluxesF /F ofabout1/25.Thus,
gal A
inthecaseofQSO0957+561wefoundaR–bandratiooffluxesF /F ofabout1/2.5(Ulla´net
gal A
al.2003),andnowwehave F /F ∼1/25,whatexplainsourunsuccessfuleffortswhenmea-
gal A
suringthefluxofthelensgalaxyinstandardframes.Asaresultofthat,inanextremecase(when
directPSFfittingleadstoamagnitudem insteadofm ,i.e.,allthegalaxylightisincluded
A+gal A
intheprofileoftheAcomponent)wefindarelationship:m =m +F /F ,wherethetrue
A A+gal gal A
flux(inmagnitudes)m differsfromthecontaminatedfluxthroughdirectPSFfitting(m )in
A A+gal
aquantityF /F .ThismaximumcontaminationofAwouldbeonlyof40mmag,andthereal
gal A
contaminationof both componentswill be less than our upper limit. The artificial fluctuations
(causedbyvariablecontamination)willbeevensmallerthanthetypicalcontaminationlevels,so
weexpecttheywillnotplayanimportantroleinanalysesofquasarvariability(e.g.,timedelay
estimates).
In order to derive the light curves of the components A and B, we decide to use a direct
PSF fitting method and do not consider the galaxy brightness in the fits. The key idea of this
procedureis toobtainthe differentfluxeswe areinterestedin by usinga PSFthatcomesfrom
abrightstar inthefield commontoallframes.Thepoint–likeobjects(quasarcomponentsand
8 A.Ulla´netal.:TimedelayofSBS0909+532
Fig.3. (S/N) versus FWHM for the Calar Alto framesin the R filter (top panel)and the V
max
filter (bottompanel).The circlesandtrianglesrepresentthe frameswith goodand badpost–fit
residues,respectively(seemaintext).
stars)aremodelledbymeansoftheempiricalPSF.Hence,wedonotuseatheoreticalPSF(i.e.,
Gaussiandistribution,Lorentziandistribution,etc),butthetwo–dimensionalprofileofastarin
this field (a PSF star). Apart from a PSF star, we also need a reference star to do differential
photometryand to obtain relative fluxesm −m and m −m . The goodbehaviourof the
A ref B ref
referencestarisusuallycheckedbyusingacontrolstar,sothefluxesm −m areexpected
con ref
toagreewithaconstantlevel.Nevertheless,sincetheR–bandfluxratioisdiscussedinSection
5, we also want to obtaina roughestimation of the contaminationsfrom this directtechnique.
Withthisaim,adeconvolutiontechnique(Koptelovaetal.2005)isalsoappliedtoasetofframes
withgoodseeingandsignal.Theselectedframesarefittedtoamodelincludingthegalaxy,and
thus,weareabletoobtainafewcleanfluxesofcomponentsAandBandcomparethemwiththe
correspondingcontaminatedfluxes(throughadirectPSFfitting).Theaveragedcontaminations
areusedinSection5.
A.Ulla´netal.:TimedelayofSBS0909+532 9
Fig.4. Calar Alto light curves of the double quasar in the R filter. We use the ”b” star as the
reference object, because it is confirmed as a non–variable star in Fig. 7. The circles are the
fluxesy (y = m −m )andthesquaresarethefluxesy −0.45mag(y = m −m ).TheA
A A A b B B B b
lightcurvehasafinaldecline,whiletheBlightcurveendswitharise.
Fig.5.CalarAltolightcurvesofthedoublequasarintheV filter.We usethenon–variablestar
”b”asreferenceobject(seeFig.7).Thecirclesarethefluxesy andthesquaresarethefluxes
A
y −0.65mag.
B
3.1.CalarAltoframesandlightcurves
We adopt a model of the system including two point–like sources and a constant background.
Themodelisfittedtoeachimagebyadjustingits7freeparameters(two–dimensionalpositions
ofAandB,instrumentalfluxesofbothcomponentsandbackground)tominimizethesumofthe
squareresiduals,asdescribedinMcLeodetal.(1998)andLeha´retal.(2000).Weusewindows
of64×64pixels.EachempiricalPSFisasubframeof64×64pixelsaroundthePSFstar(the”a”
starinFig.1),whilethelenssystemisanalyzedfromasubframeofthesamesize,butcentered
onthedoublequasar.Theinstrumentalfluxesofthe”b”,”c”,”s1”and”s2”starsarealsoinferred
from64×64pixelswindowscenteredonthem.We initiallyfocusonthenearbyfieldstars,and
take the ”b–c” stars as the control–referenceobjects. The ”a” objectis the brighteststar in the
”a–c”triangle,and”b”and”c”werespectroscopicallyidentifiedbyKochaneketal.(1997)and
10 A.Ulla´netal.:TimedelayofSBS0909+532
Fig.6. Calar Alto light curves of the stars in the R filter. The Calar Alto fluxes are checked
from six Wise frames. Filled and open symbols are associated with PSF fitting and aperture,
respectively.First,thenon–variablestar”b”(seeFig.7)istakenasthereferenceobject.Thetop
boxincorporatesthebehaviourofy +2.15mag(y =m −m ):CalarAlto(opensquares)and
a a a b
Wise(opentriangles).Thesecondbox(underthetopbox)containsthefluxesy (y =m −m ):
c c c b
CalarAlto(filledsquares)andWise(opencircles).Thethirdandfourthboxesincludethey +
s1
2.38mag(y =m −m )andy +0.29mag(y =m −m )records,respectively.Inthethird
s1 s1 b s2 s2 s2 b
box:CalarAlto(filledandopentriangles)andWise(openastroids),whereasinthefourthbox:
CalarAlto(filledandopencircles)andWise(openrhombuses).Second,weplotthem −m −
s2 s1
0.20magrecordsin the bottombox:Calar Alto (filledandopenstar symbols)andWise (open
crosses).
Zickgrafetal.(2003):”b”isaFGstar,whosespectrumincludestheG–bandandCaH–Klines,
and”c”isaM3star.TheR−I andB−Rcolorsofthebrightestcomponent(A)andthe”a”star
are similar,the colorsof the faintestcomponent(B) are close to the colorsof the ”b”star, and
the ”c” star hascolorsdifferentto those of thecomponentsandthe ”a–b”stars (see Table 1 of
Kochaneketal.1997).Ontheotherhand,aftercheckingthePSFsofthethreenearbyfieldstars
(”a–c”),wedonotfindsignificantdifferencesbetweenthem.Thissuggeststhattheglobalshape
ofthePSFsaroundthelenssystemdoesnotdependonthepositionandcolorofthepoint–like
objects,so the PSF ofthe ”a” star seemsto be a reliable tracerofthe PSF associatedwith any
point–likeobjectintheregionofinterest.
As afirstattemptforobtaininglightcurveswe usethe ”b”and”c” starsasthecontroland
referenceobjects,respectively.Unfortunately,wefindclearevidencesinfavourofvariabilityof
the”c”star,sincethethreecurvesm −m ,m −m andm −m haveasimilarglobalbehaviour.
A c B c b c
This fact forces us to rule out the ”c” star as a reference–controlobject and, thus, to take the
”a”and”b”nearbyfieldstarsasthecontrolandreferencepoint–likesources,respectively.Inthe
nextsubsection,weanalyzetheMaidanak–Wisefluxesm −m andshowthatbothstars(”a”and
a b
”b”)arenon–variableobjects.Thisresultpermitstoassurethegoodbehaviourof”b”.TheCalar
Altofluxesm −m arenotincludedintheanalysis,sincemostCalarAltodatadisagreewiththe
a b