**VolumeTitle** ASPConferenceSeries,Vol.**VolumeNumber** **Author** (cid:13)c**CopyrightYear**AstronomicalSocietyofthePacific VAST-a real-timepipeline fordetecting radio transients and variablesonthe AustralianSKA Pathfinder (ASKAP)telescope 2 1 0 JayBanyer1,2,TaraMurphy1,2,3,andtheVASTCollaboration 2 1Sydney Institute forAstronomy(SIfA),SchoolofPhysics,TheUniversity of n Sydney, NSW2006,Australia a J 2ARCCentreofExcellence forAll-skyAstrophysics (CAASTRO),The 5 University ofSydney, NSW2006,Australia 1 3SchoolofInformation Technologies, TheUniversity ofSydney, NSW2006 ] M Australia I . Abstract. TheAustralianSKAPathfinder(ASKAP)isanextgenerationradiotele- h scope currently under construction in Western Australia. The fast survey speed and p widefieldofviewmakeitanidealinstrumentforblindtransientssearches.TheASKAP - o Variable and Slow Transients (VAST) survey is a one of the major science programs r planned for ASKAP. The scientific goals of VAST include the detection and charac- t s terisationofawiderangeoftransientandvariablephenomena,fromgamma-rayburst a afterglowstoextremescatteringevents,ontimescalesof5secondsorlonger. [ We describe the data and processing challenges involved in running the VAST 1 real-timetransientdetectionpipeline.ASKAPwillproduce2.5GBofvisibilitydataper v second,transformedintoone8GBimagecubeevery5seconds.Eachcubewillcontain 0 approximately twenty 100 megapixel images with 100s of radio sources detected in 3 eachepoch. TheVASTpipelinewillmeasureandmonitorallofthesesources,detect 1 variablesandtransientsandgeneratealertsusingtheVOEventframework. 3 ThegoaloftheVASTDesignStudyistodevelopaprototypepipelinetoestablish . 1 anddemonstratethe functionalityof thefinalASKAP pipeline. We givean overview 0 oftheprototypepipeline’sfunctionality,technicalimplementationandcurrentstatus. 2 1 : v i X 1. TheAustralian SKAPathfinder(ASKAP) r a The Australian SKA Pathfinder (ASKAP, Johnston etal. 2008) is a powerful radio telescope currently under construction inWestern Australia. Itisa36dish radio inter- ferometer with exceptional survey speed: it can image the entire southern sky to high sensitivity intwonights. Thefastsurveyspeedandwidefieldofviewmakeitanideal instrument forblindtransients searches. ASKAP is located at the Murchison Radio-astronomy Observatory (MRO), ap- proximately 315km north east of Geraldton, Western Australia. The MRO is also the location of the Murchison Widefield Array (MWA) telescope and is the candidate site forAustraliaandNewZealand’sbidtohosttheSquareKilometreArray(SKA). A 6 antenna array called BETA is currently being installed and will be commis- sioned during 2012. ASKAP is scheduled to begin scientific observations with a 6 to 12antennaarrayduring2013. 1 2 JayBanyer,TaraMurphy,andtheVASTCollaboration 2. TheVASTSurvey ASKAP will give us an unprecedented opportunity to investigate the transient sky at radio wavelengths. Its wide-field survey capabilities will enable the discovery and in- vestigation of variable and transient phenomena from the local to the cosmological, including flare stars, intermittent pulsars, X-ray binaries, magnetars, extreme scat- tering events, intra-day variables, radio supernovae and orphan afterglows of gamma ray bursts. In addition ASKAP probes unexplored regions of phase space where new classesoftransientsources maybedetected. The VAST survey (Murphy&Chatterjee 2009) is one of ten major science pro- grams planned for ASKAP. It will involve several survey regimes including one that images most of the visible sky every night for two years. This will produce measure- mentsofonemillionsourcesoverhundreds ofepochs. 3. TransientPipelineFunctionality The VAST transient pipeline will detect transients and variables by analysing the im- ages produced by the ASKAP imaging pipeline. The VAST pipeline operates by ex- tracting and measuring sources in images, constructing and analysing light curves and generating notifications of detections. Thisapproach isbased on techniques described byBannister etal.(2011). Thefunctionality oftheVASTpipeline isillustrated inFig- ure1anddescribed inTable1. TheVASTcollaboration isalsoinvestigating thetechniques oftransient detection usingimagesubtraction andvisibility domaintransient detection (Trottetal.2011). Figure1. IllustrationofVASTpipelinefunctionality 4. CapacityChallenges Thetraditionalmodelforradioastronomyistotakethedatafromthetelescopeandpro- cess itoff-site some timelater. ASKAPproduces too much data to store and transport VAST-detectingradiotransientsandvariableswithASKAP 3 Table1. DescriptionofVASTpipelinefunctionality Function Description Sourcefinding Blind source finding is performed on every image. This produces a list of detected sources with position, flux (brightness) andsize. Sourceassociation Eachdetectedsourceisassociatedwithonefromthemas- terlistbyfindingthemastersourceatthesamepositionas thedetection. Ifnomastersourceisfoundatthatposition thenanewmastersourceisadded. Imagequalitycontrol The image is assessed for quality by comparing the mea- sured parameters of the detected sources to a list of well-characterised sourcesusingatechniquedescribed by Bannister etal.(2011). Sourcemonitoring Master sources known to be in the field that were not de- tected by thesource finder are measured byattempting to fitaGaussianatthepositionwherethesourceisexpected to be. If no source can be fit then an upper-limit on the fluxatthatposition istakeninstead. Lightcurvecreation All measurements for each source are collated into a ra- dio light curve -a collection of flux measurements for all epochs wherethesourcewasobserved. Detect and classify tran- Theupdatedlightcurvesarescannedtodetecttransientor sientsandvariables variable behaviour. Sources that exhibit transient or vari- able behaviour areclassified using amachine-learning al- gorithm. Thisattemptstodeterminewhatkindofphysical phenomenon iscausing thevariability. Notification Detectionsandclassificationsaresubjectedtoqualitycon- trolandcross-matching toexternalcatalogues. Thosethat appeargenuineareaddedtoanarchiveandthecommunity isalertedusingVOEventnotifications. sotheVASTtransient surveywillbeimplementedasanearreal-timepipeline running onthesamelargecomputingcluster astheASKAPimagingsystems. ASKAP’s imaging pipeline will produce two streams of images: one with inte- gration times of 5 seconds, the other with longer integration times depending on the observing mode e.g. 1 minute, 1 hour, 8 hours. Each 5 second image cube will con- tain approximately twenty ˜100 megapixel images: ˜5 frequency bands each with 4 polarisations. Eachcubewillbe˜8GBinsize. In the transient pipeline the 5 second image stream will require approximately 12,000 source measurements per second (approximately 100 sources per square de- gree, an image size of 30 square degrees, 20 images per cube, one cube every 5 sec- onds). Light curves for approximately 3,000 sources will change each second. Each source detection requires aconesearch toassociate thedetection withaknown source 4 JayBanyer,TaraMurphy,andtheVASTCollaboration position. 3,000 cone searches will be performed each second however most of these can be optimised by searching a cached list of the known sources in the current field. The cached list will be refreshed each time the telescope moves to a new field, which willhappennomorefrequently thaneveryminuteorso. The image stream with longer integrations (e.g. 1 hour) will produce images at a lower rate but each image will contain more sources due to higher sensitivity. The sourcemeasurementloadforthisstreamwillvarydepending ontheintegration time. 5. VASTPipelinePrototype A prototype transient pipeline has been developed as part of the VAST Design Study. It allows iterative development of the functional requirements of the final ASKAP pipeline. The prototype pipeline will also be used to perform transient detection on data from other telescopes, including: the Murchison Widefield Array (MWA), the Australia Telescope Compact Array (ATCA), the Very Large Array (VLA), and the SKAMolongloPrototype(SKAMP). TheprototypetakesaseriesofFITSimagesasinput,detectsandmeasuressources, performsimagequalitycontrol,constructslightcurvesandoutputsresultstoadatabase. Transientdetection, classification andVOalertsarebeingdeveloped. The prototype uses a custom source finder that is able to characterise multiple overlapping sources moreaccurately than other existing source finders (Hancock etal. inprep). Adynamicwebsiteprovidesauserinterfacetothedatabase. Itspurposeistohelp visualise anddiagnose thepipeline output easily andefficiently. Itcanbrowsethelists of images, master sources and detections, plot light curves, create postage stamps and perform cross-matching withexternalcatalogues. The prototype has already proven to be a useful tool for analysing datasets from radio telescopes even when not looking for transients due to its ability to measure and reportonallsources inallimageswithverylittleeffort. Thisassistswithchecking the imagequality, calibration andpositional accuracyofthedataset. 5.1. Languages,ToolsandLibraries The prototype pipeline is developed using the Python programming language and the Django web framework. These were chosen for their suitability for rapidly changing requirements during the VAST Design Study and the availability of astronomical li- braries. PostgreSQL is used for the SQL database with the Q3C plugin for optimised celestial coordinate searches (conesearches). References Bannister,K. W., Murphy,T., Gaensler,B. M., Hunstead,R.W., &Chatterjee, S.2011,MN- RAS,412,634 Johnston,S.,etal.2008,ExperimentalAstronomy,22,151 Murphy, T., & Chatterjee, S. 2009, VAST Memo 1, Tech. Rep. 1. URL http://www.physics.usyd.edu.au/sifa/vast/uploads/Main/vast_memo1.pdf Trott,C.M.,Wayth,R.B.,Macquart,J.-P.R.,&Tingay,S.J.2011,ApJ,731,81