Closing in on a Short-Hard Burst Progenitor Constraints from Early-Time Optical Imaging and

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

ClosinginonaShort-HardBurstProgenitor:ConstraintsfromEarly-TimeOpticalImagingandSpectroscopyofaPossibleHostGalaxyofGRB050509bJ.S.Bloom1,J.X.Prochaska2,D.Pooley1,3,C.H.Blake4,R.J.Foley1,S.Jha1,E.Ramirez-Ruiz5,2,3,J.Granot6,5,A.V.Filippenko1,S.Sigurdsson7,A.J.Barth8,H.-W.Chen9,M.C.Cooper1,E.E.Falco4,R.R.Gal10,B.F.Gerke11,M.D.Gladders12,J.E.Greene4,J.Hennanwi1,13,L.C.Ho12,K.Hurley14,B.P.Koester15,W.Li1,L.Lubin10,J.Newman16,13,D.A.Perley1,G.K.Squires17,andW.M.Wood-Vasey4

arXiv:astro-ph/0505480v2 5 Sep 200512DepartmentofAstronomy,601CampbellHall,UniversityofCalifornia,Berkeley,CA94720-3411.UniversityofCaliforniaObservatories/LickObservatory,UniversityofCalifornia,SantaCruz,CA95064.34ChandraFellow.Harvard-SmithsonianCenterforAstrophysics,60GardenStreet,Cambridge,MA02138.56InstituteforAdvancedStudy,OldenLane,Princeton,NJ08540.KIPAC,StanfordUniversity,P.O.Box20450,MailStop29,Stanford,CA94309.7DepartmentofAstronomy&Astrophysics,525DaveyLaboratory,PennsylvaniaStateUniversity,UniversityPark,PA16802.8DepartmentofPhysics&Astronomy,4129FrederickReinesHall,UniversityofCalifornia,Irvine,CA92697-4575.9CenterforSpaceResearch,MassachusettsInstituteofTechnology,Cambridge,MA02139-430710DepartmentofPhysics,OneShieldsAve.,UniversityofCalifornia,Davis,CA95616-8677.11DepartmentofPhysics,366LeConteHall,UniversityofCalifornia,Berkeley,CA94720-7300.12CarnegieObservatories,813SantaBarbaraStreet,Pasadena,CA91101.13HubbleFellow.14UCBerkeley,SpaceSciencesLaboratory,7GaussWay,Berkeley,CA94720-7450.

15DepartmentofPhysics,UniversityofMichigan,AnnArbor,MI48109-1090.

16InstituteforNuclearandParticleAstrophysics,LawrenceBerkeleyNationalLaboratory,Berkeley,CA

94720.

17SpitzerScienceCenter,CaliforniaInstituteofTechnology314-6,Pasadena,CA91125.

ABSTRACT

Thelocalizationoftheshort-duration,hard-spectrumgamma-rayburstGRB050509bbythe

Swiftsatellitewasawatershedevent.NeverbeforehadamemberofthismysterioussubclassofclassicGRBsbeenrapidlyandpreciselypositionedinaskyaccessibletothebevyofground-basedfollow-upfacilities.ThankstothenearlyimmediaterelayoftheGRBpositionbySwift,webeganimagingtheGRB eld8minutesaftertheburstandcontinuedforthefollowing8days.ThoughtheSwiftX-rayTelescope(XRT)discoveredanX-rayafterglowofGRB050509b,the rsteverof

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

ashort-hardburst,noconvincingoptical/infraredcandidateaftergloworsupernovawasfound

fortheobject.Wepresentare-analysisoftheXRTafterglowand ndanabsolutepositionof

R.A.=12h36m13.s59,Decl.=+28 59′04′′.9(J2000),witha1σuncertaintyof3′′.68inR.A.,3′′.52

inDecl.;thisisabout4′′tothewestoftheXRTpositionreportedpreviously.Closetothis

positionisabrightellipticalgalaxywithredshiftz=0.2248±0.0002,about1′fromthecenter

ofarichclusterofgalaxies.Thisclusterhasdetectabledi useemission,withatemperatureof

.36kT=5.25+3 1.68keV.Wealso ndseveral(～11)muchfaintergalaxiesconsistentwiththeXRT

positionfromdeepKeckimagingandhaveobtainedGeminispectraofseveralofthesesources.

Neverthelessweargue,basedonpositionalcoincidences,thattheGRBandthebrightelliptical

arelikelytobephysicallyrelated.Wethushavediscoveredevidencethatsupportsthenotion

thatatleastsomeshort-duration,hard-spectraGRBsareatcosmologicaldistances.

Wealsoexploretheconnectionofthepropertiesoftheburstandtheafterglow, ndingthat

GRB050509bwasunderluminousinbothoftheserelativetolong-durationGRBs.However,we

alsodemonstratethattheratiooftheblast-waveenergytotheγ-rayenergyisconsistentwith

thatoflong-durationGRBs.Thissuggestsacomparablyhighe ciencyofγ-rayconversionas

inlongGRBsasmightbeexpectedifthesameemissionmechanismisatworkinshortandlong

GRBs.Basedonthisanalysis,onthelocationoftheGRB(40±13kpcfromabrightgalaxy),on

thegalaxytype(elliptical),andthelackofacoincidentsupernova,wesuggestthatthereisnow

observationalconsistencywiththehypothesisthatshort-hardburstsariseduringthemergerofa

compactbinary(twoneutronstars,oraneutronstarandablackhole).Inthiscontext,welimit

thepropertiesofaLi-Paczy´nski”mini-supernova”thatispredictedtoariseon～daytimescales.

Otherprogenitormodelsarestillviable,andadditionalrapidlylocalizedburstsfromtheSwift

missionwillundoubtedlyhelptofurtherclarifytheprogenitorpicture.

Subjectheadings:gammarays:bursts,gamma-raybursts:individual:050509b

1.Introduction

Thedistributioninduration(Mazetsetal.1981;Norrisetal.1984)andhardness(Kouveliotouetal.1993)revealsevidencefortwodistinctpopulationsofclassicgamma-raybursts(GRBs):long-durationbursts,withtypicaldurationsaround30sandpeakenergiesat～200keV,andtheminorityshort-durationbursts,withdurationsofafewhundredmilliseconds(ms)andharderspectra.Despiteremarkableprogressinunderstandingthenatureandprogenitorsoflong-durationGRBs,comparativelylittlehasbeenlearnedabouttheoriginofshort-hardbursts,primarilybecauseveryfewsuchburstshavehadrapidandpreciselocalizations.

Themodeledburstingrateatredshiftz=0oflong-softburstsoutnumbersshort-hardburstsbyaboutafactorof3.5intheBATSEcatalog(Schmidt2001);thisassumesthesameburstingrateasafunctionofredshiftanddoesnotincludethee ectofbeaming,which,ifdi erentforlongandshortbursts,wouldimplythattheintrinsicrelativeratesdi erfromthoseobserved.WhileanumberofburstshavebeentriangulatedthroughtheInterplanetaryNetwork(seeHurleyetal.2005b)onroughlyday-longtimescales,therehasonlybeenonepreciselylocalizedshort-hardburstrelayedtogroundobserversinlessthan1hr(GRB050202/Swift:Tuelleretal.2005)1;owingtoitsproximitytotheSunattimeoflocalization,sparsegroundbasedfollowup

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

wasundertaken.IncludingGRB050509b,thiscorrespondstoaratioof1:18forshort-hardtolong-softburstdetectionswithSwift,muchsmallerthantheBATSEresult.2

Aswithlong-durationbursts,thedistributionofshortburstsappearsverynearlyisotropic(Kouveliotouetal.1993;Briggsetal.1996),andtheirbrightnessdistribution(<V/Vmax>≈0.35)isconsistentwithbeingacosmologicalpopulation.Still,thereisnostrongevidencetosupporttheideathatshortburstsarepreferentiallyseenfromz<～0.37richAbellclusters(Hurleyetal.1997),noraretheyclearlyconnectedwithstarformationwithin～100Mpc(Nakaretal.2005).

Withoutpreciseandrapidlocalizations,thepopulationstatisticsdonotprovideastrongconstraintontheshort-burstprogenitors.Still,ithasbeenlargelyreckonedthattheleadingcandidatesforshortburstsarethemergerofaneutronstarbinary(NS–NS;Blinnikovetal.1984;Paczy´nski1986,1991;Narayanetal.1992;Katz&Canel1996;Ru ert&Janka1999;Rosswog&Ramirez-Ruiz2002;Rosswogetal.2003)orablackhole–neutronstarbinary(BH–NS;Lattimer&Schramm1976;Eichleretal.1989;Mochkovitchetal.1993;Kluzniak&Lee1998;Bethe&Brown1999;Pophametal.1999;Fryeretal.1999).Thesesystemsholdseveralparticularattractions.First,althoughuncertain,theestimatedrateofmergers(between1.5–20per106yrpergalaxy;Belczynskietal.2002;Sipior&Sigurdsson2002;Rosswogetal.2003)iscomparabletotheshort-burstrate(Schmidt2001).Second,thedynamicaltimescaleofsuchmergersisseveralmillisecondsandthesound-crossingtimesareofordertenmilliseconds,comparabletotheshortestobservedbursts(Miller2005).Third,compactmergersystemsarelikelytocontainenoughmass-energyinatransienttorustopowershort-burst uencesaswouldbeobservedifatcosmologicaldistances(Rosswogetal.2003;Leeetal.2004;Rosswog2005).Thetypicaldynamicaltimescaleinsuchbinariesimmediatelypriortocoalescence(ms)ismuchshorterthantheobservedburstduration,andsoitrequiresthecentralenginetoevolveintoacon gurationthatisstable,whileretainingasu cientamountofenergytopowertheburst(Leeetal.2004).

Mergersofsuchcompactremnantsarebynomeanstheonlypossiblechanneltoproduceshortbursts.Evaporatingprimordialblackholesmayproduceshort(<100ms)GRBs(Clineetal.1999),thoughbasicenergeticsargumentssuggestthatitwouldbedi culttoseesuchsourcesfromdistanceswellbeyondtheGalaxy.Therecentdiscoveryofamega arefromSGR1806 20(Mereghettietal.2005;Hurleyetal.2005a;Palmeretal.2005;Terasawaetal.2005)ledtoplausiblesuggestionsthatasubstantialfraction(≈40%)ofshortburstscouldbeproducedbyextragalacticmagnetars(Hurleyetal.2005a).However,positional(Palmeretal.2005;Nakaretal.2005)andspectral(Lazzatietal.2005)argumentshaveledotherworkerstosuggestthatatmostafewpercentoftheBATSEcatalogcouldconsistofshort-burstmagnetars.Notethatnotallcompactmergerscreatefertileconditions(atransienttorusaroundaBH)formakingashortburst(e.g.,Janka&Ru ert2001;Rosswogetal.2004).Thedurationoftheburstinacompactbinarymergerisdeterminedbytheviscoustimescaleoftheaccretinggas,whichissigni cantlylongerthanthedynamicaltimescale,thusaccountingnaturallyforthelargedi erencebetweenthedurationsofburstsandtheirfastvariability(Leeetal.2004).Inthecollapsarscenarioforlong-durationbursts,ontheotherhand,theburstdurationisgivenbythefall-backtimeofthegas(Woosley1993;MacFadyen&Woosley1999),whichistypicallygreaterthanafewseconds.However,amodi edcollapsarscenarioinwhichtheburstdurationisdeterminednotbyfall-backbutratherbythedynamicaltimescalesassociatedwiththeexpandingout ow

table.html.

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

mightstillmeettheconstraintsofshortGRBs(Woosley2001).

ThetheoreticalpredictionsfortheafterglowsofshortGRBshavebeenconsideredbyPanaitescuetal.(2001).Sincethepeak uxofthepromptemissioniscomparableforshortandlongGRBs,iftheirdistancescalesaresimilartheisotropicequivalentenergyoutputingammarays(Eγ,iso)wouldbeproportionaltothedurationoftheGRB,whichis～10 100timeslargerforlongGRBs.Ifthee ciencyforproducingthegammaraysiscomparable,thentheisotropicequivalentkineticenergyintheafterglowshock(Ek,iso)wouldhaveasimilarratiobetweenlongandshortGRBs.ThiswouldimplytheafterglowofshortGRBstobeonaverage～10 40timesdimmerin uxthanthatoflongGRBs.TheafterglowsofshortGRBswouldbeevenmuchdimmerthanthisiftheyencounteramuchsmallerexternaldensitycomparedtolongGRBs;thisistheexpectationfromshort-burstsfrombinarymergersoutsideofthehostgalaxy.Panaitescuetal.(2001)arguedthatalowexternaldensitywouldnota ecttheX-rayband,asthelatterwasassumedtolieabovethecooling-breakfrequency,νc.We ndthatforaverylowexternaldensitytheelectroncoolingbecomesveryslowsothatνccanlieabovetheX-raybandforthe rstfewdays,thusreducingtheX-ray uxcomparedtothatforahigherexternaldensitytypicaloftheinterstellarmedium(ISM)foundnearstar-formingregionsoflong-durationGRBs.

Todatethedeepestearly-timeobservations( t<～1hr)yieldedupperlimitsVlim≈14magfromthe0.3mROTSE-Iexperiment(Kehoeetal.2001).Hurleyetal.(2002)compileddeepernon-detectionsatopticalandradiowavelengthsattimesfromdaystoweeksafterfourshortbursts,withthefaintestnon-detectionofR≈22.3magat t=20hr(seealsoGandol etal.2000).Clearly,deepandearlyobservationsinsearchofashort-burstafterglowwouldrequirearapidlocalizationtoanuncertaintycomparabletothe eldofviewofmeter-class(andlarger)telescopes.

GRB050509b(Gehrelsetal.2005)triggeredtheBATcoded-maskimageron-boardSwifton9May200504:00:19.23(UTdatesandtimesareusedthroughoutthispaper;Hurkettetal.2005).ThepositionofGRB050509b,withanuncertaintyof4′radius,wasrelayedtothegroundwithinafewseconds.TheinitiallocalizationwaslaterrevisedtoapositionR.A.=12h36m18s,Decl.=+28 59′28′′,witha95%con denceerrorradiusof2.8′(Barthelmyetal.2005a).Barthelmyetal.(2005a)describetheburstasasingle-peakedsourcewithdurationof～30ms,peak uxof2100countss 1(15–350keV),andahardnessratioconsistentwiththatoftheshort-hardpopulation.At06:29:23,afadingX-raysourcewasreportedwitha6′′localization(Kenneaetal.2005)andlaterupdatedtoan8′′uncertaintyradiusatpositionR.A.=12h36m13.9s,Decl.=+28 59′01′′(Roletal.2005).

GRB050509bthusrepresentsthe rstshort-hardburstlocalizedinrealtimetoapositionsuitableforimmediatefollow-upobservationsfromasuiteofground-basedfacilities.Inthispaperwedescribetheresultsofourobservationsofthe eldofGRB050509bandwhatbearingthesedatahaveonthenatureofshortburstsandthephysicsofshort-burstafterglows.In§2wedescribeimagingandspectroscopyofthe eld.OuranalysisoftheX-rayafterglowofGRB050509bisgivenin§3,leadingtoalocalizationnearanellipticalgalaxy(§4).In§5wepresentaspectrumofthatgalaxy,itsredshift,andinferredproperties.Wethenargue,onstatisticalgrounds,foraplausibleassociationofthisgalaxyandtheGRB.Wedemonstratein§6howGRB050509bappearstobeasubluminousburstrelativetolong-durationGRBs,butwitharatioofblast-waveenergytogamma-rayenergythatisconsistentwiththelong-durationpopulation.Intheremainingsectionswedescribenewconstraintsonthenatureofshort-burstprogenitors.Throughout,weassumeaconcordancecosmologywithH0=70kms 1Mpc 1, Λ=0.7,and m=0.3.Alloftheresultspresentedherein,thoughgenerallyconsistentwithourpreviousresultsinGCNCirculars,supersedethem.

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

2.ObservationsandReduction

Initially,severalgroupsreported(Ryko etal.2005;Ugarteetal.2005;Bloometal.2005a;Torii2005)nonewoptical/infraredsourcethatwasconsistentwiththeXRTpositionofGRB050509b(Kenneaetal.2005).At07:21:27wehighlightedtheproximityoftheXRTtoabrightredgalaxy(hereafterG1=2MASXJ12361286+2858580)andsuggestedaplausiblephysicalassociation(Bloometal.2005a)basedonitspresumedmembershipinaz≈0.22cluster(Barthelmyetal.2005a).WelaterreportedthedeterminationoftheredshiftinProchaskaetal.(2005)andProchaskaetal.(2005).At08:44:13wenotedthepresenceofafaint,compactsource(hereafterS1;see2)intheoutskirtsofG1,whichwedeemedaplausiblecandidatecounterpart(Bloometal.2005b).Averysimilarsuggestionwasmadeat09:36:49byCenkoetal.(2005c);inaddition,theynotedapparentvariabilityofthecandidate(laterretractingthevariabilityclaiminCenkoetal.2005b)anddetectionofthreeotherfaintsources(S2–S4)consistentwiththeXRTposition(seealsoCenkoetal.2005a).Twoadditionalsources(S5andS6)intheXRTlocationweresubsequentlynotedfromVeryLargeTelescope(VLT)imagingbyHjorthetal.(2005),followedbyanother5sources(J1–J5)reportedbyBloometal.(2005).Noradioemission(Parkinson2005b;vanderHorstetal.2005)orGeV/TeVemission(Parkinson2005a)isconsistentwiththeXRTerrorlocalization.Belowwediscusstheobservations,andfurtherinterpretation,leadingtothesereports.

2.1.OpticalandInfraredImaging

Weobservedthe eldofGRB050509bonMay9withtheWIYN3.5mtelescopeandtheOPTICCCDimagerwitha9.6′×9.6′ eldofviewandaplatescaleof0.14′′/pixel.Underpoor(～2′′)seeingconditions,twoexposurestotaling360swereobtainedinthei′bandbeginningat04.344hr.Inaddition,weobtained2400sofintegrationinther′bandunderimprovedseeingconditions(～1′′)beginningat06.088hr.

Thedatawerereducedintheusualmannerusing at- eldsfromboththeilluminateddomeandthetwilightsky.TheastrometricsolutionstotheindividualimageswerecalculatedbycomparisontotheUSNOB-1.0catalogwitharoot-mean-square(rms)residualof0.1′′.Thephotometriczero-pointsoftheimageswerecalculatedbycomparisontomorethan50starsintheSloanDigitalSkySurvey(SDSS)pho-tometryprovidedbyEisensteinetal.(2005).Thezero-pointsoftheWIYNimagesareuncertainataboutthe3%level.Limitingmagnitudeswereestimatedfromthehistogramof uxesin104seeing-matchedaperturesplacedrandomlywithinthe eld.Thedispersion(σ)ofaGaussian ttedtothisdistributionwasusedtoestimatethe5σlimiting uxineachimage,whichwasconvertedtoamagnitudeusingtheknownzero-point.

ThebrightgalaxyG1tothewestoftheXRTpositioncontaminatesasigni cantportionofthe8′′radiusXRTerrorcircle.Weusedgal t(Pengetal.2002)to tasmoothS´ersicpro letothisgalaxyinordertoremovemostofthecontaminantlightpriortoexaminingtheXRTerrorcircle.Aseriesof1000seeing-matchedaperturesplacedrandomlywithintheXRTerrorcircleidenti ednonewsources.ThefaintgalaxyS1wasdetectedatthe>5σlevelinourdeeperr′images.

Near-infraredimageswereobtainedwiththe1.3mPAIRITELintheJ,H,andKsbands(seeBlakeetal.2005).Observationsconsistedofa1130sintegrationcomprisedof7.8sditheredexposuresbeginningat04.1375hr.Thesedatawerereducedbymedian-combiningsetsofindividualexposureswithinamoving5-minutewindow.Theresultingmedianwasusedtosubtractthebrightskyfromtheindividualimages.Finally,alloftheindividualimageswerecombinedtomakehigh-resolutionmosaicsusingamodi edversionofdrizzle(Fruchter&Hook1997).Zero-pointsweredetermined2MASSstarsinthe eld.UpperlimitsintheJ,H,KsmosaicswereestimatedusingthesametechniqueasfortheWIYNdata.TheWIYNand

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

PAIRITELupperlimits,aswellaslimitsreportedintheliterature,areshowninFigure1.

Welaterimagedthe eldofGRB050509bwiththeKeckI10mtelescopeandtheLRIS-Binstrument(Okeetal.1995)usingthedichroicD560(50%transmissionpointat5696 A)withGandR lters.Startingat11.25May2005,beginningnearastronomicaltwilight,wetook5ditheredimagesineachbandforatotalof1660sand1620sexposuresinGandR,respectively.Thedatawerereducedintheusualmannerandcombined,weightedbyexposuretimes.

On17May2005,08:05.5,8.17daftertheGRB,weobtaineddeepR(Ellis)(whichissimilartoHarrisR;Baconetal.2003)imagingontheEchelleteSpectrographandImager(ESI;Sheinisetal.2002)ontheKeckII10mtelescope.InthepresenceofbrightglarefromtheMoon,wecombinedseveralreducedimagesforane ectiveexposuretimeof960s.SincethereisanegligiblysmallcolorterminconvertingHarrisRtoRc,3wefoundazero-pointrelativetotheLRISRimage.TherearenonewsourcestoRc≈25.0mag(5σ),norsigni cantvariationsofthefaintsourcesintheXRTerrorcircle.

3.TheX-rayEmission

TheSwiftXRT(Burrowsetal.2000)beganobservationsofGRB050509bon2005May9at04:00:56,approximately61saftertheBATtrigger.Theobservationsconsistedofelevenblocks,eachabout2.5ksinduration(exceptthe rstobservationof1.6ksandthelastobservationof1.8ks),spreadoveraperiodof～21hr.TheXRToperatedinanumberofdi erentmodesthroughouttheobservations.Themostcommon(32.3ksofexposure)andmostusefulmodeforthisobjectwasthe“PhotonCounting”mode,whichretainsthefullimagingandspectroscopicresolutionoftheinstrument.Theimagesare480×480pixels,withascaleof2′′.36perpixel.TheXRTpoint-spreadfunctionisenergydependent,withahalf-powerdiameterof′′18at1.5keV.Theenergyresolutionisalsoafunctionofenergy,varyingfromabout50eVat0.1keVtoabout190eVat10keV.

The rstPhotonCountingobservationbeganat04:01:20(Kenneaetal.2005)andlasted1640s.AsnotedinKenneaetal.(2005)andRoletal.(2005),afaintX-raysourceisdetectedinthis rstoftheelevenobservations,butitfadedquicklybelowthebackground.WehaveobtainedtheXRTdatafromtheSwiftarchive,andhaveanalyzedthemtodeterminethepositionofthisX-rayafterglowcandidateaswellastoexamineitsvariability.Webrie yreviewthedatareduction,andthenwediscussthelocalizationoftheafterglowcandidateandattempttoquantifythedecay.

3.1.SwiftDataReduction

UsingtheLevel1datafromtheSwiftarchive,weranthexrtpipelinescriptpackagedwiththeHEAsoft6.0softwaresuppliedbytheNASAHighEnergyAstrophysicsScienceArchiveResearchCenter4.Weusedthedefaultgradeselection(grades0to12)andscreeningparameterstoproduceaLevel2event lere-calibratedaccordingtothemostcurrent(asof2005May15)calibration lesintheSwiftdatabase5.Toproduceimagesforsourcedetection,weusedthexselectsoftware(alsopartofHEAsoft6.0),witha lterto

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

includeonlycountsinPIchannels30–1000(correspondingtophotonenergiesof0.3–10keV).ThePIchanneltophotonenergyconversionwasaccomplishedwiththeredistribution leswxpc0to12

2seriesstartingat4pixels(4,5.657,8,11.314,16),andthesigni cancethresholdwassetat4×10 6,correspondingtoa～1falsepositivedetectionofapointsourceintheimage.Wedetect22compactsourcesintheentire32.3ksdataset.

Tostudythepropertiesoftheafterglowcandidate,weextractedthealltheeventswithinanareaofradius10pixelsaroundthenominalwavdetectposition.Inthe rstobservationof1.6ks,thereare14countsinthisregion.Whenexaminingaplotofthecumulativedistributionversustime,wenoticedthatthemajorityofthecountsfromthisregionoccurredinthe rst300s.Wethereforefurtherinvestigatedthisbriefinterval.

Inthe rst300softhe rstPhotonCountingobservation,theXRTdetected92countsontheentirechip,with73ofthemoutsideofthe22sourceregions.Withinany10-pixelradiussourceregion,wethereforeexpectanaverageof0.1backgroundcounts.Wedetect9countsinthisregionoftheX-rayafterglow,ingthemeanlocationofjustthese9counts,wecanobtainarelativelyuncontaminatedestimateofthesourceposition.Wecalculatethe68%√con denceintervalineachdirectionasTσj/

isdistributedbytheNationalOpticalAstronomyObservatory,whichisoperatedbytheAssociationofUniversitiesforResearchinAstronomy,Inc.,undercooperativeagreementwiththeNationalScienceFoundation.6IRAF

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

ForeachXRTsourceotherthantheafterglowcandidate,Figure3plotstheo setbetweentheXRTpositionandthepositionoftheclosestopticalsource.TwoXRTsourceshadtwoopticalsourceswithin5′′;forthese,theclosestopticalsourceisrepresentedbydashedlinesandthenextclosestbydottedlines.Thereisanobviouslocusarounda4′′.5di erenceinR.A.,suggestingthattheseXRTsourcesareassociatedwiththecorrespondingnearestopticalsources.Atadetectionsensitivityaround10 14ergcm 2s 1,itisnotsurprisingto ndsomanyopticalcounterpartsinthemoderatelydeepBokimage.InaChandra/SubarustudyoftheR.A.=13hrXMM/ROSAT eld,McHardyetal.(2003) ndunambiguousopticalcounterparts¯=20.7,for61ofthe66X-raysourcesabove10 14ergcm 2s 1.ThemeanRmagnitudeofthesesourcesisRandthefaintestcounterpartisatR=24.4mag.

Usingthe14sourcesintheabovelocus(excludingthetwosourceswithmultiplepossiblecounterparts),wederiveano setbetweentheXRTframetotheopticalframeof4′′.49±0′′.72WinR.A.and0′′.42±0′′.30SinDecl.OurbestestimateforthelocationoftheX-rayafterglowisthereforeR.A.=12h36m13s59,Decl.=+28 59′04′′.9(J2000);thisis4.1′′westand3.9′′northoftherevisedXRTpositionreportedinRoletal.(2005).TheuncertaintyinourpositionisacombinationofthestatisticaluncertaintyoftheXRTlocalization(3′′.6inR.A.,3′′.5inDecl.)andtheuncertaintyinshiftingtheXRTframetotheICRS(0′′.76inR.A.,0′′.40inDecl.).

TheastrometryinouroriginalreportsfromWIYNandKeckimagingwerebasedonaframeofapprox-imately10starsinthe2MASScatalog.ThereleaseoftheSDSSdataandcalibrationsofthis eldallowustoimprovetheastrometrictietotheICRS.We ttheKeck/LRISG-bandimageto91sourcesincommonwiththeSDSSobjectcatalogwithathird-orderpolynomialsolutionusingIRAF/CCMAP.TheuncertaintyintheastrometrictietoSDSS,baseduponresidualsfromthe t,isσ(R.A.)=0.134′′andσ(Decl.)=0.153′′.Assuminga75masastrometricuncertaintyintheSDSSastrometriccalibrationtotheICRS(Pieretal.2003),weestimatetheabsoluteuncertaintyintheKeck-ICRStieisσ(R.A.)=0.154′′andσ(Decl.)=0.171′′.

TheXRTlocationis11.2′′±3.6′′(or40±13kpcinprojection)fromG1aswe rstnotedinBloometal.(2005a).Spectroscopyofthissourcerevealsthatitisindeedanearly-typegalaxy(see§5)andisamemberofaclusterNSCJ123610+285901atz≈0.22(Galetal.2003;Barthelmyetal.2005a).NearthelocationoftherevisedXRTerrorcircle,we nd～11faintsources(allofwhichweorothershavereportedpreviously;seeabove).Figure2showstheKeckGandRimageswithidenti edsourcelabeled.Table2givestheastrometricpositionsandmagnitudesofthesources.

3.3.X-rayAfterglowDecay

Weexaminethe rst1.6ksblockofobservationstocharacterizethetemporalpropertiesoftheX-rayafterglow.AKolmogorov-Smirno (K-S)testonthearrivaltimesofthe14photonsgivesaprobabilityof0.06%thattheycomefromasourcewithconstantcountrate.ThenextstepinmodelcomplexityisoneinwhichtheX-raycountrateRXinthisregionhasaconstantcomponent(duetothebackgroundanddi useclusteremission)plusacomponentwithapower-lawdependenceontime(duetothefadingafterglow).OurmodelisthusRX(t)=A(t t0) α+B,whereBistheconstant(backgroundpluscluster)countrate,t0isthetimeoftheBATtrigger,andAisanormalizationchosensuchthatthemodelpreservesthedetected uxoverthe1.6ksunderconsideration.WedetermineBtobe0.00107counts 1fromthelaterobservations.

Weconsideredarangeofαfrom0to4andcomputedtheK-Sprobabilityoftheobserveddatacomingfromthemodelforeachvalueofα.TheK-Sprobabilitywashighest(97.8%)atα≈1.3.Forα 1(0.77)andα 1.7(2.1),theK-Sprobabilitydroppedbelow32%(5%).Forα=1.3,thenormalization

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

Ais22counts 1.Wecantranslatethistoanenergy- uxnormalizationbydeterminingtheconversionfromcountstoergcm 2.Weconsideronlythe rst300sofdataforthisdeterminationinordertoreducecontaminationfromthebackground.Foreachofthe9counts,weknowitsenergyaswellasthee ectiveareaoftheXRTatthatenergy.Theaverageis3.14×10 11ergcm 2count 1.OurmodelfortheX-ray ux(0.3–10keV)oftheafterglowonlyisthenFX(t)=A′(t t0) α.Forα=1.3,thenormalizationisA′=6.9×10 10ergcm 2s 1.Forexample,theX-ray uxoftheafterglowatt=200saftertheBATtriggerforα=1,1.3,and1.7isFX(200)=5.8,7.0,and6.5×10 13ergcm 2s 1,respectively.

Figure4shows,onacommonscale,X-raylightcurvesforanumberofGRBsandcore-collapsesuper-novae.ThelocationoftheX-raytransientassociatedwithGRB050509b,whenplacedatz=0.2248,isstriking.WhiletheslopeofthetransientagreeswellwiththoseoftypicalGRBs,its uxfallswellbelowthetypicallong-GRBrange.Infact,foranyreasonableredshift(i.e.,z 3–5),GRB050509bwouldstillbesigni cantlyunderluminousinitsX-rayafterglowwhencomparedtothoseoflong-durationGRBs(seeFigure4).Fortheassumedredshiftofthetentativehostgalaxy,theextrapolatedX-rayluminosityatafewdays,whichisalsoconsistentwiththeChandraupperlimit(Pateletal.2005),isclosetothoseseenintypicalcore-collapsesupernovae.

3.4.Di useGalaxyClusterEmission

Weusedwavdetecttosearchforlarge-scalestructuresinthefull32.3ksXRTdataset.Thepixelscalessearchedwere(20,28.28,40,56.57,80).Thecenterofthedi useemissionpresumablyassociatedwiththegalaxyclusterhadawavdetect-determinedpositionof12h36m18s26,+28 59′06′′.7.Figure5showsanadaptivelysmoothedimage(usingtheCIAOtoolcsmooth)oftheXRTdatawiththeclustercenterandGRBindicated.Thecolorsrepresentthe0.3–10keVcountdensity.Contoursaredrawnat0.00449,0.00646,0.00934,0.0136,0.0197,and0.0273countarcsec 2.Astheimageshows,thewavdetect-determinedpositionofthedi useemissionisabout14′′tothewestand4′′southofthepeakofthedi useemission,whichisat12h36m19s33,+28 59′10′′.8(J2000).[Notethattheopticalclustercenteris12h36m10+28 59′00.9′′(J2000)asde nedbythecenterofthegalaxyoverdensity;thisisabout125′′eastand10′′southofthepeakofthedi useX-rayemission.]Wethus ndthattheXRTafterglowpositionis75′′west,6′′northoftheclustercenter,asde nedbythepeakofthedi useX-rayemission,about270kpcinprojection.

Weextractaspectrumfromaregionof110′′inradiuscenteredonthewavdetectposition.Weuseasimilar-sizedregioninasource-freeareatoextractaspectrumforbackgroundsubtraction.Werequiretheclusterspectrumtocontainatleast20countsperbin,andweconsidertherange0.3–10keV.We tthebackground-subtractedclusterspectruminXspecv12.2(Arnaud1996)withaMEKAL(warmplasma)modelabsorbedbyaGalacticcolumndensityof1.52×1020cm 2(Dickey&Lockman1990).WesettheMEKALredshiftatz=0.2248andthemetallicityat[Fe/H]=0.26(Mushotzky&Loewenstein1997)and

.36allowthetemperatureandnormalizationtovary.Thebest ttemperatureiskT=5.25+3 1.68keV,which

givesχ2/dof=22.4/20.

4.AssociatingGRB050509bwithG1

WearenowinapositiontoexplorethepossibleassociationofGRB050509bwiththeclusterandwiththenearbyellipticalgalaxyG1.FocusingontheBATlocalizationalone,we rstconsidertheprobabilitythatarandompositionintheskywouldbeinarichclusterofgalaxies(hereweneglectthee ectsoflensing,

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

expectedtobesmall;forexample,Grossman&Nowak1994).AreasonableestimateofthecoveringfractionofclustersontheskyisgivenbythetheDPOSSNorthernSkyOpticalClusterSurvey(Galetal.2003).Althoughthissurveyisnotverydeep(zlim～0.3),low-redshiftclustersshoulddominatetheskydensity.Galetal.(2003) ndacoveringfractionof～0.03assumingatypicalclusterradiusof1Mpc,whichsuggestsachancealignmentisimprobablebutnotimpossible.Moreover,theXRTlocalizationofGRB050509btowithin45′′ofthecenterofsuchaclusterwouldoccurbychancewithaprobabilityofjust～7×10 4,butitisdi culttoestimateaposteriorihowlargeadistancefromaclustercenteronewouldhaveconsidered“signi cant.”

Whilethegasfromtheclusterenvironmentmayenhancetheprobabilityoflocalizingshort-burstafter-glows(seebelow),ourexpectationisthatshortGRBprogenitorsarecausedbythedeathofstarsofsomesort,withtheburstratedeterminedbyprocessesonscalessigni cantlysmallerthanclusterlengths.Tothisend,weshouldconsiderthechanceprobabilityoftheGRBeventoccurringatcloseimpactparametertoagalaxysimilartoG1.AsreportedbyEisensteinetal.(2005),thegalaxyG1hasaPetrosianr′magnitudeof17.18±0.02magbasedonimagingbytheSDSS.TheskydensityofgalaxieswithcomparableapparentmagnitudebrighterthanG1is～40persquaredegree(Blantonetal.2003b).Therefore,theprobabilityofaneventrandomlyoccurringwithin20′′(abouttwicetheobservedo set)ofthisbrightgalaxyis～5×10 3.Weconsiderthisaconservativeestimatebecausethisprobabilitymakesnoreferencetothegalaxyredshift,type,size,orage(whichareconsistentwithaprioridiscussionsofshort-bursthostgalaxies;e.g.,Bagotetal.1998andBloometal.1999).

IfonearguesthatGRB050509bisindeedphysicallyassociatedwiththisbright,low-redshiftellipticalgalaxy,onemustconsiderwhytheseveralotherwell-localizedshortburstshavenotshownsimilarassoci-ations.The rstpossibility,thattheshortburstsarisefromamorelocalpopulation(assuggestedbythemagnetar arefrom27December2004;Hurleyetal.2005a)andGRB050509bmustthereforearisefromadi erentpopulation,wasdiscountedforfourofthebest-localizedshortbursts(Nakaretal.2005).AnotherpossibilityisthatGRB050509bwassigni cantlycloserthantheotherwell-localizedshortbursts.Astrongtestofthishypothesisistodetermineifothershortburstsareassociatedwithmoredistantclustersorin-trinsicallybright,massivegalaxies(e.g.,throughadeepimagingcampaign).Thethirdpossibilityissimplythatshort-burstprogenitorsneednotalwaysariseinsuchgalaxies.Infact,fortheNS–NShypothesiswewouldexpectmergersingalaxiesspanningawiderangeofHubbletypes.AdelayedBH–NSmergerisalsopossible,butlesslikelyifGRB050509bisassociatedwithG1:statistically,thedistributionoftimescalesforBH–NScoalescenceisasbroadasthatforNS–NScoalescence,albeitquitemodeldependent(Belczynskietal.2002;Sipior&Sigurdsson2002),butthesystemickickvelocityisexpectedtobesystematicallylowerbyafactorofafewinmosttheoreticalmodelsofformationofBH–NSbinaries(kickvelocityisroughlyinverselyproportionaltomass,somoremassivebinariesreceivelesskick).Moreover,largervelocitykicksgenerallyleadtoshortermergertimes.Thusthe～40kpco settendstofavorNS–NSoverBH–NSmergers.

Whilethesepossibleassociationsaretantalizing,aposterioristatisticsareverysuspect.HadtheGRBbeennearabrightspiralgalaxy,wemighthavemadesimilarclaimsbasedonchanceprobabilities.Nevertheless,itremainsthecasethatmanyworkershadpredictedthedistinctpossibilitythatawell-localizedcompactmergerand/orshortburstcouldbenearanellipticalgalaxy(Bagotetal.1998;Bloometal.1999;Panchenkoetal.1999)(seealsoLivioetal.1998),andsowesuggestthattheseargumentsmightreasonablyre ectatrueassociation.Moreover,thepossibleassociationwithanearly-typehoststandsinstarkcontrasttoresultsfromlong-durationGRBs(Bloometal.2002;LeFloc’hetal.2003;Djorgovskietal.2003),andisreminiscent(e.g.,Bloometal.2002;Dar2004)ofthedichotomybetweencore-collapseandthermonuclear(TypeIa)supernovae.OnlyalargersampleofshortGRBswillprovidetrulycompellingevidenceforsuch

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

aparallel.Still,basedontheargumentsabove,weproceedbyacceptingthehypothesisthatGRB050509bisphysicallyassociatedwithG1.

5.APutativeHostGalaxyatz=0.2248

WeobtainedaspectrumofG1withDEIMOS(Faberetal.2003)ontheKeckII10mtelescopeunderphotometricconditions.Thedatawereacquiredinaseriesoftwoexposuresstartingat07.47hronthenightoftheburst.Theinstrumentalsetupincludedthe600linemm 1gratingblazedat7500 Aandcenteredat7200 A,theGG455order-blocking lter,andstandardCCDbinning.Thissetupgivesnearlycontinuouswavelengthcoverageintherange4500–9000 A.Weobservedthegalaxythougha1.1′′×20′′slitatskypositionangle90 andanairmassof1.0.ThissetupyieldsaFWHMresolutionof～5 A(i.e.,σ=100

1kms).ThedatawerereducedandcalibratedwiththeDEEPspectroscopicpipelineforDEIMOSdata(Cooperetal.2006).Wavelengthcalibrationand at- eldingwereperformedusingspectraofXe-Ne-Kr-Arandquartzlamps(respectively)obtainedthatnight.

Thesoftwareprovidesatwo-dimensional,sky-subtractedimageofthespectrumacrosstwoCCDsoftheDEIMOSmosaic.Unfortunately,theCCDthatincludesthebluestdatahasapairofblockedcolumnswhichlienearthecenterofthegalaxypro le.Therefore,weextractedtheone-dimensional(1D)spectrumonthisCCDusingoptimalextractiontechniquesassumingaGaussianpro lewithσ=9.2pixels(i.e.,1.1′′).FortheotherCCD,weextracteda1Dspectrumbyadoptinga26-pixel(3′′)boxcaraperture.Finally,weprocessedandcalibratedaspectrophotometricstandardstar(BD+28 4211)observedattheendofthisnight.Aftercomparingitsobserved ux(indigitalnumbers)againsttheSTISCALSPECcalibration7,wecalculatedasensitivityfunctionwhichcouldbeappliedtoourgalaxyspectra.

SpectroscopicobservationsofG1,S1,S2,and2unidenti edsourceswereobtainedwiththeGMOSspectrometer(Hooketal.2004)ontheGeminiNorth8mtelescopebeginningat10.27May2005underphotometricconditions.Weuseda0.75′′slit,aR400gratingblazedat7640 A,GG455order-blocking lter,andsetthecentralwavelengthto6500 A.Theairmasswaslow(1.0–1.1),sothee ectsofatmosphericdispersionwerenegligible(Filippenko1982).StandardCCDprocessingandspectrumextractionwereac-complishedwithIRAFusinga1.74′′apertureforthesesourcesofinterest(S1,S2).ThedatawereextractedusingtheoptimalalgorithmofHorne(1986).Low-orderpolynomial ingtechniquesdiscussedinWade&Horne(1988)andMathesonetal.(2000),weemployedIRAFandourownIDLroutinesto ux-calibratethedataandtoremovetelluriclinesusingthewell-exposedcontinuaofthespectrophotometricstandardEG–131(Bessell1999).

Figure6presentsthe1D ux-calibratedspectrumofG1againstavacuum,heliocentric-correctedwave-lengtharray.Thedottedlinetracesa1σerrorarraybasedonPoissoncountingstatistics.Wehavemarkedanumberofdetectedabsorption-linefeaturesandalsotheexpectedpositionforseveralstrongtransitionsfrequentlyobservedinemission-linegalaxies(e.g.,Hα,[OIII]).Wehave tadouble-Gaussianpro letoCaIIH&Kandmeasurez=0.2248±0.0002.ThisisconsistentwiththeredshiftinferredphotometricallyforthisclusterfromDPOSS(Galetal.2003).Atthisredshift,theluminositydistanceis1117.4Mpc,and1′′correspondsto3.61kpcinprojection.

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

Toestimatethevelocitydispersionofthegalaxy,wehavecomparedthespectrumagainstatemplatespectrumofHD72324(e.g.,Kelsonetal.2000)smoothedbyawiderangeofσ.ThebestmatchtotheabsorptionlinesofG1withλrest=4000–5300 Aisσ=275±40kms 1.Accountingfortheinstrumentalresolution,wederivealight-weightedvelocitydispersionforthisgalaxyof260±40kms 1.

ThespectralfeaturesevidentinFigure6aretypicalofearly-typegalaxies.Thespectraltypeandvelocitydispersionindicateamassiveellipticalgalaxywithnoapparentongoingstarformation.Aquantitativelimittothecurrentstar-formationrate(SFR)canbeinferredfromtheupperlimittotheHαluminosityofthisgalaxy.Theemission-line uxina10 Awindow( v≈300kms 1)centeredattheexpectedwavelengthofHαhasa3σupperlimitof1.2×10 16ergs 1cm 2.Adoptingthecurrentconcordancecosmology,wederiveanHαluminosityLHα<1.2×1040ergs ingtheempiricalrelationbetweenSFRandLHα(Kennicut1998),the3σupperlimittothecurrentSFRis0.1M⊙yr 1.

Amorphological ttoWIYNI-bandimagingusinggal t(Pengetal.2002)showsgoodagreementwithadeVaucouleurspro le(S´ersicindex=4),withaχ2/dof=1.22.Thee ectiveradiusisRe=0.96′′=

3.47kpc.Thegalaxyhasanaxisratioof0.81withthesemimajoraxisalignedalongapositionangleeastofnorthat～90 .Therewaslittleimprovementinχ2/dofbyaddingmorecomplicatedmorphologiesorlettingtheS´ersicindexvary.

ThecoincidenceofapointsourceatradiowavelengthswiththeopticalcenterofG1mightsuggestthepresenceofalow-levelactivegalacticnucleusdespitethelackoftelltalefeaturesobservedintheG1opticalspectra.Moreover,inspectionofarchivalimagesofthisgalaxyfromtheNear-EarthAsteroidTrackingProgram(Pravdoetal.1999)on9April2002,20April2002,3May2002,22March2003,and8April2003revealsnoapparentvariabilityoftheopticallightfromG1.However,radioemissionwithoutcorrespondingopticalemissionisnotuncommoningiantellipticalgalaxiesharboringmildlyactivenuclei(Ho1999).TheradioemissioninG1isunlikelytobeassociatedwithstarformation,giventhelowSFRdeducedabove.

Thepropertiesofthisprobablehostgalaxycontrastsigni cantlywiththosemeasuredforthegalaxyhostsoflong-durationGRBs.First,mosthostsoflong-durationGRBsexhibitemission-linefeaturesin-dicativeofhighSFRs(e.g.,Djorgovskietal.2003).Second,theabsoluteK-bandluminosityofthisgalaxy(≈1.6×1011L⊙)exceedsthatofallpreviouslyidenti edGRBhostgalaxies(Charyetal.2002).Third,theimpactparameteroftheGRB(asde nedbythe90%XRTerrorcircle)islargerthanthatofallpreviouslyassociatedGRB-hostgalaxypairs(long-bursto sets<～10kpc;Bloometal.2002).

5.1.S1,S2:FaintBlueGalaxiesinaHigh-RedshiftGroup?

TheGemini/GMOSspectraofS1andS2arefeaturelessandblue.ExaminingtheregionsofthespectrumwhereHαorHβwouldlieifattheredshiftofthecluster,wedetectnomeasurableemission.Assuming,forthemoment,thatthesourcesareattheclusterredshiftof0.22,weputa3σupperlimitontheHαluminosityofLHα<1.5×1039ergs 1andLHα<1.4×1039ergs 1forS1andS2,ing

theequationfromKennicut(1998)relatingtheHαluminositytotheSFR,we ndthattheupperlimitsfortheunextinguishedSFR,assumingthatS1andS2areclustermembers,are～1.1×10 2M⊙yr 1forthegalaxies.IfS1andS2areclustermembers,thentheyarenotformingstars,whichwouldseemtocon ictwiththeirbluecolors.

Amorelikelyscenario,alsomentionedbyCenkoetal.(2005a),isthatS1andS2arebothbackgroundgalaxies.AlthoughtheGeminispectrarangefrom4600 Ato8600 A,thedatahavepoorsignal-to-noiseratio

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

bluewardof5200 A.Nevertheless,thespectralslopeiswellconstrainedanditsuggeststhatthesegalaxiesareformingstars(i.e.,thecontinuumisrelativelyblue).Thelackofcorrespondingemissionlines(Hα,Hβ,

[OIII]λ5007,[OII]λλ3727)fallinginourspectralwindowthereforesuggeststhatS1andS2havez>～1.3.Additionalspectroscopywillberequiredtocon rmourhypothesisthatS1andS2arefaintbluegalaxymembersofasmallgroupatmoderateredshift.

6.TheoreticalInterpretation

The uenceofthepromptgamma-rayemissionmeasuredbytheSwiftBATisf=(2.3±0.9)×10ergcm 2(Barthelmyetal.2005b),whichattheredshiftofthetentativehostimpliesanisotropicequivalentenergyoutputofEγ,iso=(2.7±1)×1048erg.SinceνFνisstillrisingroughlyasν0.5inthe15–150keVSwiftrange,thetotal uencecouldbe 3timeslargerifthepeakenergyEp 1–2MeV.Figure9showstheisotropicequivalentluminosityofGRBX-rayafterglowsscaledtot=10hraftertheburst(inthecosmologicalrestframeofthesource),LX(10hr),asafunctionoftheirisotropicgamma-rayenergyrelease,Eγ,iso,forGRB050509btogetherwithasampleoflongGRBs.LX(10hr)forGRB050509bisestimatedbyextrapolatingthe uxmeasuredbytheSwiftXRTusingthebest- tpower-lawdecayindexofα=1.3,whichisalsoconsistentwiththeChandraupperlimit. 8

Alinearrelation,LX(10hr)∝Eγ,iso,seemstobebroadlyconsistentwiththedata,probablysuggestingaroughlyuniversale ciencyforconvertingkineticenergyintogammaraysinthepromptemissionforbothshortandlongGRBs.This“universal”e ciencyisalsolikelytobehigh(i.e.,theremainingkineticenergyiscomparableto,orevensmallerthan,thatwhichwasdissipatedandradiatedinthepromptemission).Ifthisisthecase,thewell-knowne ciencyproblemforlongGRBsalsopersistsforshortGRBs.

TheX-rayluminosityat10hrisusedasanapproximateestimatorfortheenergyintheafterglowshock,since(a)at10hrtheX-raybandistypicallyabovebothνmandνcsothatthe uxhasaveryweakdependenceon B[tothepowerof(p 2)/4]andnodependenceontheexternaldensity,bothofwhichhaverelativelylargeuncertainties(Freedman&Waxman2001;Piranetal.2001;Bergeretal.2003);and(b)at10hrtheLorentzfactoroftheafterglowshockissu cientlysmall(Γ≈10)sothatalargefractionofthejetisvisible(outtoanangleof～Γ 1≈0.1radaroundthelineofsight)andlocalinhomogeneitiesonsmallangularscalesareaveragedout.Furthermore,thefactthattheratioofLX(10hr)andEγ,isoisfairlyconstantformostGRBssuggeststhatbothcanserveasreasonablemeasuresoftheisotropicequivalentenergycontentoftheejectedout ow.ApossiblecaveattotheabovestatementarisesiftheobserverisinfactnotwithintheapertureoftheGRBjet(asissuggestedtobethecaseinbothX-ray ashesandX-rayrichGRBs;Granotetal.2005).InthiscaseEγ,isocanbesigni cantlysmallerthantheisotropicequivalentkineticenergyintheafterglowshock,whichisbetterre ectedbyLX(10hr).ThisislikelytobethereasonwhyGRB031203isabovethecorrelationshowninFigure9(Ramirez-Ruizetal.2005).Ano -axisinterpretationforGRB050509b,ontheotherhand,isunlikelysinceitsX-rayafterglowlightcurvewasobservedtodecayfromaveryearlyepoch(att≈102s).ThisisalsoconsistentwiththefactthatGRB050509bfallsclosetothecorrelation.

Theaboveargumentssuggestthattheenergyintheout owejectedbyGRB050509bwas～Eγ,iso≈10erg,ifitwasspherical.Ontheotherhand,ifitwascollimatedintoanarrowjetofhalf-openingangle

2θ0,thenthetrueenergywouldbesmallerbyafactoroffb=(1 cosθ0)≈θ0/2.Sinceasigni canto -axis

viewingangleisnotlikely,thetrueenergyprobablydoesnotexceedEγ,iso.AhigherredshiftwouldincreaseEγ,isoandwithittheestimatefortheenergyreleaseinthisevent;however,itwouldstillremainsigni cantly48.5

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

lessenergeticthantypicallongGRBs(seeFigure9).

AsalsoarguedbyLeeetal.(2005),thefactthattheX-rayafterglowluminosityofGRB050509bismuchsmallerthanthatoflongGRBsisprobablybecausetheeventwassub-energetic,ratherthanduetodi erencesonthevaluesoftheexternaldensityorthemicrophysicalparameters.ThisisillustratedinFigure8bya ttothecurrentlyavailableafterglowdatausingparametervaluesthataretypicalforlongGRBs,exceptfortheisotropicequivalentenergyintheafterglowshock,Ek,iso,whichisheretakentobeequaltoEγ,isoassumingz=0.2248.Otherparametervaluescouldalsogiveareasonabledescriptionoftherathersparsedata.InTable3wedemonstrateafewdi erentsetsofparametersthat ttheafterglowdata.Again,wereferthereadertoLeeetal.(2005)foramoredetaileddescription.Regardlessoftheredshift,itwillbeverydi culttodetecttheafterglowintheradio,sincethemaximal uxdensity(giventheobservationalconstraints)isunlikelytoexceed～15µJy.

IfshortGRBsoccursigni cantlyoutsideoftheirhostgalaxies,asmaybecommonforbinarymergers(Tutukov&Yungelson1994;Bloometal.1999;Fryeretal.1999;Buliketal.1999;Belczy´nskietal.2000),thenonemightexpecttheexternaldensityencounteredbytheafterglowshockofsomeGRBstobeverylow,typicaloftheintergalacticmedium(IGM),nIGM≈10 6.5(1+z)3cm 3.Thismayhelpexplainwhysomeshortburstscouldhaveveryfaintafterglows.SinceGRB050509bhappenedtooccurnearthecenterofagalaxyclusterwheretheexternaldensityisrelativelyhigh,itsX-rayafterglowwasrelativelybrighter.IfindeedGRB050509bisassociatedwiththegalaxyclusteratz≈0.22,thenonemightexpecttheexternaldensitytobeintermediatebetweentheIGMandISM:theGRBis～76′′fromthecenteroftheclusterasdeterminedbytheX-rayposition(§3.4),correspondingto～270kpcinprojectionandwellwithinthedi useemissionfromthehotintraclustermediumgas(whichextendstoaradiusof～1Mpc).ThissuggestsanambientdensitynearthepositionoftheGRBofn≈10 3–10 2cm 3,thoughthisestimateisuncertainbecausethespacepositionoftheburstrelativetotheclustercenterandtheintraclustermedium(ICM)densitypro learenotknownprecisely.

7.Discussion

ThelackofastrongafterglowsignaturesetsGRB050509bapartfrommostotherGRBs.8Asacom-parison,thelow-redshiftlong-softburst(GRB030329,z=0.1685;Greineretal.2003),ifplacedattheredshiftofG1,wouldhavebeenR≈14magatt=8000s;thisisapproximate10magbrighterthanthedetectionlimitsfoundherein.Evenatz=3,theopticalafterglowofaGRB030329-likeburstshouldhavebeendetectedatearlytimes(neglectingthee ectsofdustextinction).Ournon-detections(R>～24mag)ofvariabilityat1.3hrinthewhatwouldbetherestframeatz=1ismorethan3.5magnitudesdeeperthanthefaintestopticaltranisentfoundforalongGRB(021211;z=1.0;see g.2ofFoxetal.2003).

Thelackofdetectableoptical/infraredafterglowisnotsurprisingongroundsrelatedtotheprogenitorsandtoGRBafterglowtheory.First,sincetheluminosityoflong-wavelengthafterglowsscaleswiththesquarerootoftheambientdensity(Begelmanetal.1993;M´esz´arosetal.1998),eventsthatoccurintheISMorIGMshouldbeintrinsicallyfainter(atoptical/infraredwavelengths)thanthoseoccurringinthecircumburstenvironmentsofcollapsars(seePanaitescuetal.2001).Second,basedon<V/Vmax>studies,theisotropic-equivalentpeakluminosity(Lp(γ))ofshortburstsissimilartothatoflongbursts(Schmidt

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

2001),implyingthatthetotalenergyoutput(Eγ,iso≈Lp(γ)/η×duration,withηastheconversione ciencytogammarays)isatleastanorderofmagnitudesmallerforshortbursts.AsarguedbyPanaitescuetal.(2001),sinceafterglowbrightnessscaleswithEγ,iso(1 η),short-burstafterglowswouldbesystematicallyfaint.

NowthatthereisadetectedX-rayafterglowweareinapositiontodirectlytestthefaintnessclaim,byinferringthegamma-rayenergyreleaseandX-rayafterglowluminosity(aproxyforthekineticenergyintheblastwave).FromFigure9itisclearthatthisratioforGRB050509bissimilartothatfoundinlong-durationGRBs.Thisisastrikingobservationalbridgetolong-durationburstsandsuggestsacommonphysicalmechanismsforpromptanddelayed(afterglow)emissionforbothlong-durationandshort-durationGRBs,eventhoughtheirprogenitorsareprobablydi erent.

Atentativedetectionofanafterglowsignalbyaddinguptheemissionof76shortBATSEburstswasreportedbyLazzatietal.(2001)(seealsoConnaughton2002).Thesignalpeakedatt≈30safterthebursttriggerwitharelatively atνFν≈5×10 10ergcm 2s 1.ThiswouldcorrespondtoanX-ray uxinthe0.2–10keVrange,ofFX≈2×10 9ergcm 2s 1.TheX-ray uxoftheafterglowofGRB050509bisbestconstrainedaroundt≈200s,andisfoundtobeFX≈6.5×10 13ergcm 2s 1.Extrapolatingthis uxtot≈30swithapower-lawindexintherangeinferredfromthedata,1.0 α 1.7,givesa uxthatislowerthantheonefoundbyLazzatietal.byafactorof～(1 5)×102.ThismightsuggestthateitherthepossibledetectionbyLazzatietal.(2001)wasnotstatisticallysigni cant,ortheX-rayafterglowofGRB050509bisunderluminouscomparedtotheaveragevalueforshortGRBsbyatleasttwoordersofmagnitude.

WithessentiallynoindicationofrecentstarformationinG1,massiveprogenitorstarsleadingtocol-lapsarscannotbepresentinG1.S1andS2,thebrightestandthird-brightestsourceswithintheXRTerrorcircle,havenoindicationofrecentstarformationiftheirredshiftsare 1.3(SFR<0.05M⊙year 1forz<0.3andSFR<1M⊙year 1forz<1.2).Thefainter(andblue)objectsdiscussedin§5.1arelikelytobebackgroundgalaxies.IftheoriginofGRB050509bisfromacollapsar,itislikelythatitsredshiftexceeds

1.3.

IfGRB050509bisabackgroundobjectatz 2,someprogenitorscenariosaredi culttoreconcile.Withanobserveddurationof～30msec,therest-framedurationwouldbeonlyabout10msec.ThisisimplausiblyshortforanNS–NSmerger,andmarginallypossibleforaBH–NSmergerifthecoalescenceisthroughunstablemasstransfer(Lee&Klu´zniak1999;Rosswog2005;Miller2005).Itishardtosimulta-neouslyaccommodatetheshortintrinsictimescaleandthehigherenergybudgetoftheburstwithinanycompactmergermodel,ifitisathighredshift.

IfshortGRBstracestarformationwithatimedelaythroughdoublecompactmergerswithcoalescencetimescalesof107–1010yr(asopposedtoprompttracersofstarformationaswiththecollapsarscenarioforlongGRBs;Bagotetal.1998andBloometal.1999),thenweexpectsomefraction(10–30%)ofshortGRBstobeseeninassociationwithearly-typegalaxiesingeneralandclustersspeci cally(seeNutzmanetal.2004forratedensityinthelocaluniverse).Thisissomewhatmodeldependent,sincethedistributionofcompactmergertimescalesispoorlyconstrainedbydata,butbroadlyconsistentwithbothobservedandmodeldistributions.

Acore-collapsesupernova(SN)producesnoelectromagneticradiationuntilitsenvelopeiscompletelyconsumedbytheexplosion(althoughseeKhokhlovetal.1999).Thisphaseends,however,withabrilliant ashofX-rayorextremeultravioletphotonsastheshockreachesthestellarsurface.The“breakout” ashisdelayedintime,andvastlyreducedinenergy,relativetotheneutrinotransientproducedbycorecollapse.

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

However,itconveysusefulinformationabouttheexplosion.Shockbreakout asheswerepredictedbyColgate(1968)asasourcefor(thethenundetected)gamma-raybursts.TheexplosionofSN1987Astimulatedareanalysisofsupernovabreakout ashesbyEnsman&Burrows(1992)and,morerecently,byBlinnikovetal.(1998)andBlinnikovetal.(2000).Thesestudiesrepresentanincreaseinsophisticationtowardthefullnumericaltreatmentofthiscomplicated,radiation-hydrodynamicproblem.Inprinciple,theXRTdatacouldconstraintheexistenceofashockbreakoutproducedbybotharedsupergiantexplosionlikeSN1993J(VanDyketal.2002)andabluesupergiantexplosionanalogtoSN1987A,buttheX-rayluminosityissensitivetotheuncertaindistributionofextragalacticgascolumnandthespeci cXRTobservingepochs.

UsingourESIopticalimaging,wecanalsolimitthepresenceofbrighteningduetoasupernovaorsupernova-likeemissionat8.17daftertheGRBtoRc≈25.0mag.Anormal,unextinguishedTypeIa(thermonuclear)supernovaatz=0.22wouldhaveR≈22mag,around6.7dafterexplosion(t=8.17dintheobserver’sframe).AverysubluminousSNIalikeSN1991bg(Filippenkoetal.1992)wouldhaveR≈24mag,stillsomewhatbrighterthanourlimit.ExtinctionwouldobviouslymaketheSNfainter,buttheMilkyWaycontributionissmall(AV≈0.06mag;Schlegeletal.1998),andtheoutskirtsofanellipticalgalaxyinaclustershouldhaveessentiallynodust.Whilesomecore-collapsesupernovaecouldbeasfaintas(orfainterthan)ourlimit,thepresenceofsuchasupernovaintheoutskirtsofanellipticalgalaxywouldbetrulyextraordinary(seevandenBerghetal.2005).OthershavealsoreportednoevidenceforaSNatlater-times(Hjorthetal.2005;Bersieretal.2005).

Thelocationofthis(andfuture)shortburstprovidesausefuldiscriminantfordistinguishingbetweendi erentprogenitormodelsofshortbursts.Simplistically,wewouldexpectevaporatingblackholestooccurnearthecenterofdeeppotentialwells(asdiscussedinthecontextofGalacticBHs;Clineetal.1999);thus,theo setfromG1seemstodisfavorthishypothesis.Agiant arefromamagnetarwouldneedtohaveaisotropicluminosity(Lγ,iso)largerbyafactorof～103andanEγ,isolargerbyafactorof～102comparedthetheinitialspikeofthe27December2004giant arefromSGR1806-20(thedi erenceinthefactorbetweenthetwoquantitiesarisessinceGRB050509blastedonly～30ms,whichis～10timesshorterthantheinitialspikeofthegiant arefromSGR1806-20).Burstsfrommagnetarsmightbeexpectedfromlater-typegalaxiesthanG1whereneutronstarswouldbeformedcopiously:magnetic elddecaywouldcuttheactivelifetimeformega areactivityafter～104yr.

8.Conclusions

WehavemonitoredthelocationofGRB050509batopticalandinfraredwavelengthsfrom8minutesto8daysafterthetriggerandfoundnoindicationofvariabilityatthelocationofthefadingX-raysource,the rstsolidX-raydetectionofanafterglowofashort-hardburst.Nearthelocationofthissourceweandothershavefoundanapparentgroupoffaintbluegalaxiesatredshifts 1.3.Whileitisindeedplausiblethatthisshortburstarosefromaprogenitorconnectedwiththosegalaxies,wefound—basedonapositionalargument—plausibleevidencethattheprogenitorislikelyassociatedwithG1,abrightellipticalgalaxyatz=0.2248.Wehavearguedthattheobservations ndnaturalexplanationwithacompactmergersystemprogenitor.Ifso,thenshort-hardGRBsprovideabridgefromelectromagnetictogravitationalwaveastronomy:indeedhasGRB050509boccurredafactorof～3closerinluminositydistanceismighthaveproducedadetectablechirpsignalwiththenextgenerationLIGO-IIgravitationalwavefacility9.

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

Brighteningemissionfrommosttypesofsupernovaewouldhavebeenseeninourimaging,sothelackofsuchemissionappearsinconsistentwiththenotionthatshortburstsareduetocollapsarsorvariantsthereof.Ourafterglowmodelingisalsoconsistentwith,thoughdoesnotrequire,acircumburstmediumhavinglowerdensitythanthatinferredinlong-durationGRBs;iftrue,thiswouldsuggestthattheprogenitorproducesaGRBinanenvironmentthatisbaryonpoorcomparedtothatexpectedforcollapsars.Moreover,wehaveseennoevidenceforongoingstarformationintheputativehost,sotherearelikelynoremainingmassivestars.Giventheshortactivelifeofaneutronstarhavingahighmagnetic eld,thisalsodisfavorsthemagnetarhypothesis.

Thenon-detectionofbrighteningemissionmayplacelimitsonthepresenceofathermal“mini-supernova”fromnon-relativisticejectaofacompactmergersystem(Li&Paczy´nski1998;Rosswog&Ramirez-Ruiz2002).Inthisscenario,thesmalldensemass(mej)ingthescalingsofLi&Paczy´nski(1998)andcrudelyassumingthat10%ofthebolometriclightatpeakisradiatedintheRband,theR-bandbrightnessshouldpeakatobservertimet≈1.2(mej/0.01M⊙)1/2daftertheburst,withabsolutemagnitudeMR≈ 18.5 1.25log(mej/0.01M⊙)mag.AssumingthattheGRBdidindeedoriginatefromtheredshiftz=0.2248,uponinspectionofFigure1,withnon-detectionsatMR≈ 16magatt≈1d,wecanveryroughlyexcludemej>few×10 3M⊙.ThoughtheLi&Paczy´nski(1998)modelwasintendedasasimplisticsketchofthephenomenon,thislimitonmejissomewhatsurprisinggiventheamountofescapingnon-relativisticmaterialexpectedincompactmergers(Rosswog&Ramirez-Ruiz2002).Indeed,weconsiderthislackofa“mini-supernova”asweakevi-denceagainstaz=0.22originfromacompactmergersystem.Still,theselimitsaresubjecttoconsiderableuncertaintyinanumberofuncertainparametersofejecta.Forinstance,ifthevelocityoftheejectaweretobe～0.01cinsteadof0.3c(asassumedbyLi&Paczy´nski1998)thenthepeakofthethermalemissionwouldoccurafterabout1month,andwouldnothavebeendetectedwiththecurrentlimits.

WeconcludebyemphasizingthatintheNS–NSorBH–NSprogenitorhypothesisforshort-hardbursts,thehostsgalaxiesmaybearangeofHubbletypes(e.g.,Livioetal.1998).CompactmergersystemscoalesceinappreciableratesfromMyrtoGyrafterastarburst(e.g.,Fryeretal.1999;Bloometal.1999).Obviously,thelongerthetimesincethestarburst,thelargerthedistanceabinarysystemwilltravelbeforecoalescence.Aclearpredictionfromthismodelisthatasmoreshortburstsarelocalized,thoseassociatedwithlater-typegalaxiesofagivenmassshouldbepreferentiallyclosertothestar-formationcentersofthehost;thatis,weexpectamoreconcentrateddistributionaroundaspiralgalaxywiththesamemassasanearly-type.Ontheotherhand,dwarfstar-forminghostshaveshallowenoughpotentialsthatmergersystemsfromthesegalaxiescouldcoalesceatappreciabledistances(>～100kpc)evenshortlyafterstarburst.AsSwiftlocalizesmoreshort-hardbursts,weexpectthattheo setdistributionaroundgalaxieswillfurtherelucidatetheprogenitorquestion.

Wethanktheanonymousrefereefortheirinsightfulanddetailedcommentsthatleadtotheimprovementofthispaper.MuchofthisworkcouldnothavebeenundertakenwithouttheworkofS.BarthelmyrunningtheGCNCircularsandtheSwiftteamhavingdeployedsuchamarvelousexperiment.WethanktheGeminista fortheirexcellentworkinobservingGRB050509b,especiallyK.Roth.WethankR.Duncan,A.Loeb,P.Kumar,andE.Quataertforinformativeexchanges.WethankB.Simmonsforherassistancewithusinggal t.WealsothankM.Skrutskie,D.Starr,W.Peters,B.HutchinsfortheircontributionstoandassistancewithPAIRITEL.WeareindebtedtoG.WirthandR.ChornockfortheircontributionsinacquiringsomeoftheKeckdata.J.S.B.,J.X.P.,andH.-W.C.arepartiallysupportedbyNASA/SwiftgrantNNG05GF55G.TheworkofA.V.F.issupportedbyNSFgrantAST-0307894andNASA/Swiftgrant

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

NNG05GF35G;heisalsogratefulforaMillerResearchProfessorshipatU.C.Berkeley,duringwhichpartofthisworkwascompleted.S.S.issupportedbytheCenterforGravitationalWavePhysicsfundedbytheNSFundercooperativeagreementPHY01-14375,andbyNSFgrantPHY02-03046.D.P.andE.R.-R.gratefullyacknowledgesupportprovidedbyNASAthroughChandraPostdoctoralFellowshipgrantnumbersPF4-50035andPF3-40028(respectively),awardedbytheChandraX-rayCenter,whichisoperatedbytheSmithsonianAstrophysicalObservatoryforNASAundercontractNAS8-03060.S.J.gratefullyacknowledgessupportviaaMillerResearchFellowshipatU.C.Berkeley.J.N.acknowledgessupportfromNASAthroughHubbleFellowshipgrantHST-HF-01165.01awardedbytheSpaceTelescopeScienceInstitute,whichisoperatedbytheAssociationofUniversitiesforResearchinAstronomy,Inc.,undercontractNAS5-26555.TheresearchofJ.G.issupportedbytheUSDepartmentofEnergyundercontractnumberDE-AC03-76SF00515.ThispublicationmakesuseofdataproductsfromtheTwoMicronAllSkySurvey,whichisajointprojectoftheUniversityofMassachusettsandtheInfraredProcessingandAnalysisCenter/CaliforniaInstituteofTechnology,fundedbytheNationalAeronauticsandSpaceAdministrationandtheNationalScienceFoundation.SomeofthedatapresentedhereinwereobtainedattheW.M.KeckObservatory,whichisoperatedasascienti cpartnershipamongtheCaliforniaInstituteofTechnology,theUniversityofCalifornia,andNASA;theObservatorywasmadepossiblebythegenerous nancialsupportoftheW.M.KeckFoundation.TheanalysispipelineusedtoreducetheDEIMOSdatawasdevelopedatU.C.BerkeleywithsupportfromNSFgrantAST-0071048.BasedonobservationsobtainedattheGeminiObservatory,whichisoperatedbytheAssociationofUniversitiesforResearchinAstronomy,Inc.,underacooperativeagreementwiththeNSFonbehalfoftheGeminipartnership:theNationalScienceFoundation(UnitedStates),theParticlePhysicsandAstronomyResearchCouncil(UnitedKingdom),theNationalResearchCouncil(Canada),CONICYT(Chile),theAustralianResearchCouncil(Australia),CNPq(Brazil)andCONICET(Argentina).Finally,wewishtoextendspecialthankstothoseofHawaiianancestryonwhosesacredmountainweareprivilegedtobeguests.

REFERENCES

Arnaud,K.A.1996,inASPConf.Ser.101:AstronomicalDataAnalysisSoftwareandSystemsV,17Bacon,D.J.,Massey,R.J.,Refregier,A.R.,andEllis,R.S.2003,MNRAS,344,673

Bagot,P.,PortegiesZwart,S.F.,andYungelson,L.R.1998,A&A,332,L57

Barthelmy,S.etal.2005a,GCNReport3385

—.2005b,GCNReport3385

Begelman,M.C.,M´esz´aros,P.,andRees,M.J.1993,MNRAS,265,L13

Belczy´nski,K.,Bulik,T.,andZbijewski,W.2000,A&A,355,479

Belczynski,K.,Kalogera,V.,andBulik,T.2002,ApJ,572,407

Berger,E.,Kulkarni,S.R.,andFrail,D.A.2003,ApJ,590,379

Bersier,D.,Fruchter,A.,Rhoads,J.,Levan,A.,andTanvir,N.2005,GCNCircular3521

Bessell,M.S.1999,PASP,111,1426

Bethe,H.A.andBrown,G.E.1999,ApJ,517,318

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

Blake,C.H.etal.2005,Nature,435,181

Blanton,M.R.etal.2003a,AJ,125,2348

—.2003b,ApJ,594,186

Blinnikov,S.,Lundqvist,P.,Bartunov,O.,Nomoto,K.,andIwamoto,K.2000,ApJ,532,1132

Blinnikov,S.I.,Eastman,R.,Bartunov,O.S.,Popolitov,V.A.,andWoosley,S.E.1998,ApJ,496,454Blinnikov,S.I.,Novikov,I.D.,Perevodchikova,T.V.,andPolnarev,A.G.1984,SovietAstronomyLetters,

10,177

Bloom,J.,Blake,C.,Prochaska,J.X.,Hennawi,J.,Gladders,M.,andKoester,B.2005a,GCNReport3386Bloom,J.,Blake,C.,Prochaska,J.X.,Hennawi,J.,Gladders,M.,Koester,B.,andChen,H.-W.2005b,

GCNReport3388

Bloom,J.S.etal.2005,GCNReport3417

Bloom,J.S.,Kulkarni,S.R.,andDjorgovski,S.G.2002,AJ,123,1111

Bloom,J.S.,Sigurdsson,S.,andPols,O.R.1999,MNRAS,305,763

Breeveld,A.etal.2005,GCNReport3397

Briggs,M.S.etal.1996,ApJ,459,40

Bulik,T.,Belczy´nski,K.,andZbijewski,W.1999,MNRAS,309,629

Burrows,D.N.etal.2000,inProc.SPIEX-RayandGamma-RayInstrumentationforAstronomyXI,ed.

K.A.Flanagan&O.H.Siegmund,Vol.4140,64–75

Cenko,S.B.,Soifer,B.T.,Bian,C.,Desai,V.,Kulkarni,S.R.,Berger,E.,Dey,A.,andJannuzi,B.T.

2005a,GCNReport3401

—.2005b,GCNReport3409

—.2005c,GCNReport3391

Chary,R.,Becklin,E.E.,andArmus,L.2002,ApJ,566,229

Cline,D.,Matthey,C.,andOtwinowski,S.1999,apj,827

Colgate,S.A.1968,CanadianJournalofPhysics,46,476

Connaughton,V.2002,ApJ,567,1028

Cooper,M.etal.2006,inpreparation

Dar,A.2004,astro-ph/0405386

Dickey,J.M.andLockman,F.J.1990,Ann.Rev.Astr.Ap.,28,215

Djorgovski,S.G.etal.2003,inDiscoveriesandResearchProspectsfrom6-to10-Meter-ClassTelescopes

II.,ed.P.Guhathakurta,Vol.4834,238–247

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

Eichler,D.,Livio,M.,Piran,T.,andSchramm,D.N.1989,Nature,340,126

Eisenstein,D.J.,Hogg,D.W.,andPadmanabhan,N.2005,GCNReport3418

Engelbracht,C.W.andEisenstein,D.J.2005,GCNReport3420

Ensman,L.andBurrows,A.1992,ApJ,393,742

Faber,S.M.etal.2003,inProceedingsoftheSPIE,ed.M.Iye&A.F.M.Moorwood,1657–1669Fenimore,E.E.etal.2004,GCNReport2735

Filippenko,A.V.1982,PASP,94,715

Filippenko,A.V.etal.1992,AJ,104,1543

Fox,D.W.etal.2003,apjl,586,L5

Freedman,D.L.andWaxman,E.2001,ApJ,547,922

Freeman,P.E.,Kashyap,V.,Rosner,R.,andLamb,D.Q.2002,ApJS,138,185

Frei,Z.andGunn,J.E.1994,AJ,108,1476

Fruchter,A.andHook,R.N.1997,inApplicationsofDigitalImageProcessingXX,Proc.SPIE,Vol.3164,

ed.A.Tescher(SPIE),120–125

Fryer,C.L.,Woosley,S.E.,andHartmann,D.H.1999,ApJ,526,152

Gal,R.R.,deCarvalho,R.R.,Lopes,P.A.A.,Djorgovski,S.G.,Brunner,R.J.,Mahabal,A.,and

Odewahn,S.C.2003,AJ,125,2064

Gandol ,G.etal.2000,inGammaRayBursts:5thHuntsvilleSymposium,ed.G.J.F.R.MarcKippen,

RobertS.Mallozzi,Vol.526(Meville,NewYork:AIP),23

Gehrels,N.etal.2005,astro-ph/0505630

Granot,J.,Ramirez-Ruiz,E.,andPerna,R.2005,astro-ph/0502300

Granot,J.andSari,R.2002,ApJ,568,820

Greiner,J.etal.2003,Nature,157

Grossman,S.A.andNowak,M.A.1994,ApJ,435,548

Hjorth,J.etal.2005,astro-ph/0506123

Hjorth,J.etal.2005,GCNReport3410

Ho,L.C.1999,ApJ,510,631

Hook,I.M.,Jørgensen,I.,Allington-Smith,J.R.,Davies,R.L.,Metcalfe,N.,Murowinski,R.G.,and

Crampton,D.2004,PASP,116,425

Horne,K.1986,PASP,98,609

Huang,K.Y.etal.2005,apjl,628,L93

The localization of the short-duration, hard-spectrum GRB 050509b was a watershed event. Thanks to the nearly immediate relay of the GRB position by Swift, we began imaging the GRB field 8 minutes after the burst and continued for the following 8 days. No

Hurkett,C.etal.2005,GCNReport3381

Hurley,K.etal.2002,ApJ,567,447

—.2005a,Nature,434,1098

Hurley,K.,Hartmann,D.,Kouveliotou,C.,Fishman,G.,Laros,J.,Cline,T.,andBoer,M.1997,ApJ

(Letters),479,L113

Hurley,K.etal.2005b,ApJS,156,217

Janka,H.-T.andRu ert,M.2001,astro-ph/0101357

Katz,J.I.andCanel,L.M.1996,ApJ,471,915

Kehoe,R.etal.2001,ApJ(Letters),554,L159

Kelson,D.D.,Illingworth,G.D.,vanDokkum,P.G.,andFranx,M.2000,ApJ,531,159

Kennea,J.A.etal.2005,GCNReport3383

Kennicut,R.C.1998,Ann.Rev.Astr.Ap.,36,131

Khokhlov,A.M.,H¨o ich,P.A.,Oran,E.S.,Wheeler,J.C.,Wang,L.,andChtchelkanova,A.Y.1999,ApJ

(Letters),524,L107

Kluzniak,W.andLee,W.H.1998,ApJ,494,L53

Kouveliotou,C.,Meegan,C.A.,Fishman,G.J.,Bhat,N.P.,Briggs,M.S.,Koshut,T.M.,Paciesas,W.S.,

andPendleton,G.N.1993,ApJ(Letters),413,101

Kouveliotou,C.etal.2004,ApJ,608,872

Lattimer,J.M.andSchramm,D.N.1976,ApJ,210,549

Lazzati,D.,Ghirlanda,G.,andGhisellini,G.2005,astro-ph/0504308

Lazzati,D.,Ramirez-Ruiz,E.,andGhisellini,G.2001,A&A,379,L39

LeFloc’h,E.etal.2003,A&A,400,499

Lee,W.H.andKlu´zniak,W.L .1999,MNRAS,308,780

Lee,W.H.,Ramirez-Ruiz,E.,andGranot,J.2005,astro-ph/0506104

Lee,W.H.,Ramirez-Ruiz,E.,andPage,D.2004,ApJ,608,L5

Li,L.-X.andPaczy´nski,B.1998,ApJ(Letters),507,L59

Livio,M.etal.1998,inGammaRayBursts:4thHuntsvilleSymposium,ed.C.A.Meegan,R.Preece,&

T.Koshut,Vol.428(Woodbury,NewYork:AIP)

MacFadyen,A.I.andWoosley,S.E.1999,ApJ,524,262

Matheson,T.etal.2000,AJ,120,1487

Mazets,E.P.etal.1981,Ap&SS,80,3

本文来源：https://www.bwwdw.com/article/wamj.html