Java Analysis Studio and Java Analysis Studio and the hep.lcd class librarythe hep.lcd class library

Mike Ronan - LBNL

Joanne Bogart, Gary Bower, Tony Johnson - SLAC

Nick Sinev - Oregon

Don Benton - U Penn

Physics and Experiments with Future Linear e+e- CollidersThursday, April 29 , 1999

ContentsContents

The hep.lcd framework for LC physics The hep.lcd framework for LC physics studiesstudies• OverviewOverview• Fast MCFast MC• Tracking ReconstructionTracking Reconstruction• Cluster FindingCluster Finding

Java Analysis Studio Java Analysis Studio Distributed Physics AnalysisDistributed Physics Analysis Performance ComparisonPerformance Comparison Conclustions - How to try it out! Conclustions - How to try it out!

LC Detector studies in US LC Detector studies in US Goals:Goals:

• Detailed Study of physics processes Detailed Study of physics processes in a variety of possible LC Detectors.in a variety of possible LC Detectors.

– Reference Small and Large detectorsReference Small and Large detectors

• Full simulation with GISMO Full simulation with GISMO – Switch to Geant4, when readySwitch to Geant4, when ready

• Analysis using Analysis using – Java & Java Analysis StudioJava & Java Analysis Studio– C++ & RootC++ & Root

• Software RequirementsSoftware Requirements– Flexibly handle different detector Flexibly handle different detector

geometries and technologiesgeometries and technologies– Rapid development of variety of Rapid development of variety of

reconstruction and analysis reconstruction and analysis algorithms algorithms

Java package hep.lcdJava package hep.lcd Reconstruction ProcessorsReconstruction Processors

• Track finder + track fitter Track finder + track fitter • Several clustering Several clustering

algorithmsalgorithms Parameterized MC Parameterized MC

ProcessorsProcessors• Can read generator output Can read generator output

(StdHEP) or Gismo output(StdHEP) or Gismo output• Track and Cluster smearingTrack and Cluster smearing

Analysis UtilitiesAnalysis Utilities• Event Shape + Thrust utilities Event Shape + Thrust utilities • Jet finders [Jade, Durham]Jet finders [Jade, Durham]• HistogramingHistograming

Event DisplayEvent Display• Simple 2D Event display Simple 2D Event display

currentlycurrently

FrameworkFramework• Driver frameworkDriver framework

– interactively controlinteractively control– calling of processorscalling of processors– debugging/histogramingdebugging/histograming

• Parameter (Constant) accessParameter (Constant) access– driven by detector geometrydriven by detector geometry

• MC event input (StdHEP MC event input (StdHEP format)format)

• IO system based on Java IOIO system based on Java IO– random access files allows random access files allows

efficient access to subset of efficient access to subset of datadata

• Can be run inside JAS or Can be run inside JAS or standalonestandalone

Track finding and fitting Track finding and fitting packagepackage

Track finding Track finding package derived package derived from TPC, Babarfrom TPC, Babar

Track fitting Track fitting based on SLD based on SLD Weight Matrix fit Weight Matrix fit algorithmalgorithm• Tracking Tracking

chamber or chamber or tracking+vertex tracking+vertex detector fit detector fit possiblepossible

Clustering + FastMCClustering + FastMC

Three Clustering AlgorithmsThree Clustering Algorithms implemented:implemented:• Simple Cluster Finder Simple Cluster Finder (contiguous energy](contiguous energy]

• Cluster Cheater Cluster Cheater [perfect clustering using MC info][perfect clustering using MC info]• JRBCluster - configurable cluster finderJRBCluster - configurable cluster finder

FastMCFastMC • Simple parameterized MC Simple parameterized MC

– Allows analysis directly from generator output without using Allows analysis directly from generator output without using full Gismo simulationfull Gismo simulation

• Produces same event format as GismoProduces same event format as Gismo– same analysis can be run with FastMC or Gismo.same analysis can be run with FastMC or Gismo.

Track-Cluster Association Track-Cluster Association being Developedbeing Developed

Physics UtilitiesPhysics Utilities

Physics UtilitiesPhysics Utilities• 4-vector, 3-vector classes4-vector, 3-vector classes• Event shape/Thrust finderEvent shape/Thrust finder• Jet FinderJet Finder

– Jade and Durham algorithms implementedJade and Durham algorithms implemented– Extensible to allow implementation of other algorithmsExtensible to allow implementation of other algorithms

HistogramingHistograming (from Java Analysis Studio) (from Java Analysis Studio) Event DisplayEvent Display

• Suitable for debugging reconstruction and analysisSuitable for debugging reconstruction and analysis• Plan to use Wired for full 3D support in futurePlan to use Wired for full 3D support in future

Event DisplayEvent Display

Event DisplayEvent Display

Event DisplayEvent Display

Event DisplayEvent Display

Java Analysis StudioJava Analysis Studio

Set of experiment independent analysis tools Set of experiment independent analysis tools for event oriented (High Energy Physics) data for event oriented (High Energy Physics) data • Data Access classes provide access to many common HEP Data Access classes provide access to many common HEP

data formatsdata formats• Histogram/Scatterplot Accumulation + Manipulation ClassesHistogram/Scatterplot Accumulation + Manipulation Classes• Plot Display classesPlot Display classes• Lightweight framework for users to create physics analysis Lightweight framework for users to create physics analysis

applications in Java.applications in Java.

Tools work alone, in combination, or withinTools work alone, in combination, or within• Java Analysis Studio GUI which gives:Java Analysis Studio GUI which gives:

– Integrated editor and compiler– Efficient access to local and remote data– Extensibility via Plug-ins, Fitters, Functions etc

GUI makes getting started easyGUI makes getting started easy“Wizards” guide beginners“Wizards” guide beginners

Built in Editor and CompilerBuilt in Editor and Compilerfor writing analysis codefor writing analysis code

Histogram and Scatterplot Histogram and Scatterplot displaydisplayInteractive Fitting and RebinningInteractive Fitting and Rebinning

GUI can be extended to addGUI can be extended to addexperiment specific featuresexperiment specific features

Distributed Data Analysis Distributed Data Analysis with JASwith JAS

With many different simulated detectors and With many different simulated detectors and many physics processes, total MC data sample is many physics processes, total MC data sample is largelarge

JAS has built in support for efficient JAS has built in support for efficient distributed distributed physics analysisphysics analysis

LCD has set up central data repository at UPenn, LCD has set up central data repository at UPenn, accessible from anywhereaccessible from anywhere

Distributed Data Analysis Distributed Data Analysis with JASwith JAS

Java allows objects to move from client to server - even Java allows objects to move from client to server - even across different platformsacross different platforms

Since analysis Since analysis code is moved to datacode is moved to data - analysis is fast - analysis is fast Transparent to end user, who “feels” as if analysis is Transparent to end user, who “feels” as if analysis is

running locallyrunning locally

GUIDataAnalysis Engine

UsersJava Code

JavaCompiler +Debugger

ExperimentExtensions(Event Display)

TCP/IP Network

Padded Cell

Data

Is Java fast Enough for HEP Is Java fast Enough for HEP offline?offline?

Current (266Mhz PII, JDK 1.1.7)Current (266Mhz PII, JDK 1.1.7)– Clustering .6 secs/event Clustering .6 secs/event

– 13.5 Million Calorimeter Cells13.5 Million Calorimeter Cells

– Fast MC 6 ms/eventFast MC 6 ms/event– Track Finding + Fitting ~5secs/eventTrack Finding + Fitting ~5secs/event

• Very competitive with C++/Root implementation (where Very competitive with C++/Root implementation (where they exist)they exist)

Will get even better!!!Will get even better!!!– JDK 1.2, HotSpot - Run-time optimizationJDK 1.2, HotSpot - Run-time optimization

In In real lifereal life may be faster than C++ may be faster than C++– Better, cheaper performance analysis toolsBetter, cheaper performance analysis tools– Java encourages lightweight, module interfaces which Java encourages lightweight, module interfaces which

promote efficienct coding stylespromote efficienct coding styles

Try it out!Try it out!

Works on Windows Works on Windows (95/98/NT](95/98/NT] or Unix or Unix (Linux, (Linux, Solaris,...]Solaris,...]

Online tutorial availableOnline tutorial available• Suitable for complete beginners:Suitable for complete beginners:

– no knowledge of Java or JAS assumedno knowledge of Java or JAS assumed– starts with instructions on downloading and installingstarts with instructions on downloading and installing– Shows simple sample analysis jobsShows simple sample analysis jobs

• http://www-sldnt.slac.stanford.edu/jas/documentation/lcd/http://www-sldnt.slac.stanford.edu/jas/documentation/lcd/

JAS Home PageJAS Home Page• http://www-sldnt.slac.stanford.edu/jashttp://www-sldnt.slac.stanford.edu/jas

ConclusionConclusion

Use of Java + JAS looks very promisingUse of Java + JAS looks very promising• Have been able to develop complete framework + full Have been able to develop complete framework + full

reconstruction package in < 6 monthsreconstruction package in < 6 months• People have quickly learned and use it, and to contribute to People have quickly learned and use it, and to contribute to

the reconstruction packagethe reconstruction package• Performance looks goodPerformance looks good

FutureFuture• New version of JAS available this monthNew version of JAS available this month• Standard Java interface to Geant4?Standard Java interface to Geant4?• Continue development of reconstruction and FastMCContinue development of reconstruction and FastMC

– Direct speed comparison with C++ codeDirect speed comparison with C++ code