Full Event Reconstruction in

Java

CHEP '04 Interlaken, Switzerland

Sep. 30, 2004

Norman Graf (SLAC)for the LCD development team

(“From 0 to analysis in 15 minutes”)

International Linear Collider International Linear Collider (ILC)(ILC)Detectors designed to exploit the physics Detectors designed to exploit the physics

discovery potential of ediscovery potential of e++ee-- collisions at collisions at s~1TeV.s~1TeV.

Will perform precision measurements of Will perform precision measurements of complex final states.complex final states.

Require:Require: Exceptional momentum resolution. Exceptional momentum resolution. Excellent vertexing capabilities.Excellent vertexing capabilities. “ “Particle Flow” calorimetry.Particle Flow” calorimetry. Hermeticity.Hermeticity.

Mission StatementMission Statement Provide full simulation capabilities for ILC physics program:Provide full simulation capabilities for ILC physics program:

Physics simulations.Physics simulations. Detector designs.Detector designs. Machine-Detector interface & backgrounds.Machine-Detector interface & backgrounds.

Need flexibility for:Need flexibility for: New detector geometries/technologies.New detector geometries/technologies. Innovative reconstruction algorithms.Innovative reconstruction algorithms.

Limited resources demand efficient solutions, focused effort.Limited resources demand efficient solutions, focused effort. Desire a common simulation environment used in ILC

studies which allows sharing of detectors, algorithms, and code.

The system should be flexible, powerful, yet simple to install, maintain and use.

Metric of performance:Ease with which physicist having an idea can implement and test its

effect!

Most important tool of all!Most important tool of all!

Why Java?Why Java? Java is a pure Object Oriented Language.Java is a pure Object Oriented Language.

Simpler to learn and use than C++ :Simpler to learn and use than C++ : Language design emphasizes ease-of-use over Language design emphasizes ease-of-use over

performance.performance. Garbage collector takes care of freeing unused objects.Garbage collector takes care of freeing unused objects. New language, no historical baggage.New language, no historical baggage. Very powerful standard library.Very powerful standard library.

– Large number of open-source libraries including libraries for Large number of open-source libraries including libraries for scientific computing.scientific computing.

Platform independent, compile once just runs Platform independent, compile once just runs everywhere (Linux, Windows, Mac OSX, Solaris, …).everywhere (Linux, Windows, Mac OSX, Solaris, …).

Physicist gets to concentrate on writing clean OO Physicist gets to concentrate on writing clean OO code to perform analysis tasks.code to perform analysis tasks.

Not understanding core dumps and learning difference Not understanding core dumps and learning difference between a pointer and a reference.between a pointer and a reference.

Performance of Java code is close to that of C++.Performance of Java code is close to that of C++. In real life maybe 20-30% overhead typical.In real life maybe 20-30% overhead typical.

Why Java?Why Java? Several very Several very

powerful, powerful, freefree, IDE’s , IDE’s now available:now available: e.g. Netbeans, Eclipse.e.g. Netbeans, Eclipse. Support editing, code Support editing, code

completion, GUI completion, GUI building, debugging, building, debugging, performance profiling, performance profiling, refactoring, CVS, etc…refactoring, CVS, etc…

Multiplatform.Multiplatform. Mantra “I prefer Mantra “I prefer

EMACS” is becoming EMACS” is becoming hard to justify!hard to justify!

Why a full reconstruction Why a full reconstruction now?now?

ILC ILC LEP LEP Much higher luminosity, energy, precision.Much higher luminosity, energy, precision. More complex events (not on a resonance).More complex events (not on a resonance).

ILC ILC LHC LHC Detector R&D focused on orthogonal issues.Detector R&D focused on orthogonal issues.

e.g. rad hardness not an issue, tracker material is!e.g. rad hardness not an issue, tracker material is! Desired precision and resolutions require Desired precision and resolutions require

a new detector paradigm, intimately a new detector paradigm, intimately connected with the reconstruction: connected with the reconstruction: “Particle Flow”.“Particle Flow”.

Aim for very tight loopAim for very tight loop design design simulate simulate reconstruct reconstruct analyze analyze

Particle Flow MotivationParticle Flow Motivation Measure momenta of charged tracks in the

tracker with superb resolution. Measure photons in highly segmented EM

calorimeter with reasonable resolution. Remaining neutral hadrons measured in hadron

calorimeter.

2confusion

2neut.had.

2photons

2charged

2jet

had. neut.photonschargedjet

EEEE

EEEE

22confusionconfusion is the hardest term, cannot be is the hardest term, cannot be

correctly simulated with fast 4-vector smearing. correctly simulated with fast 4-vector smearing. Needs detailed calorimeter shower simulation Needs detailed calorimeter shower simulation and reconstruction to understand.and reconstruction to understand.

Imaging CalorimeterImaging CalorimeterEMCal: dense, small Moliere radiushighly segmented:laterally (~30-40 layers)longitudinally (~5x5 mm2)

HadCal: emphasize segmentation & granularity over intrinsic resolution.Digital readout?

ILC Detector SimulationsILC Detector Simulations1.1. Use simplified volumes to investigate Use simplified volumes to investigate

larger number of designs, allowing larger number of designs, allowing dynamic topologies using single dynamic topologies using single executable.executable.

2.2. Describe “optimal” detectors in great Describe “optimal” detectors in great detail.detail.

TPC Tracker, Si Disks, CCD VTXTPC Tracker, Si Disks, CCD VTX All Si Tracker, CCD VTXAll Si Tracker, CCD VTX

Java Reconstruction/Analysis Java Reconstruction/Analysis OverviewOverview Runs standalone or inside Java Analysis Studio (JAS3).Runs standalone or inside Java Analysis Studio (JAS3). Fast MC provides target (“best you can do”).Fast MC provides target (“best you can do”).

Smeared tracks and calorimeter clusters, fiducial acceptance.Smeared tracks and calorimeter clusters, fiducial acceptance. Beam & Physics background overlays at hit level.Beam & Physics background overlays at hit level. Hit digitization allows detector readout schemes to be Hit digitization allows detector readout schemes to be

varied, provides more realistic hits (merged, ghost, varied, provides more realistic hits (merged, ghost, pospos ). ). Full Reconstruction: Full Reconstruction:

ab initioab initio track finding and fitting. track finding and fitting. calorimeter clustering.calorimeter clustering. Individual Particle reconstruction (cluster-track association).Individual Particle reconstruction (cluster-track association).

Java Neural Net available for training & use.Java Neural Net available for training & use. Physics Tools (Vertex Finding, Jet Finding, Flavor Tagging).Physics Tools (Vertex Finding, Jet Finding, Flavor Tagging). Analysis Tools (including LCD-specific WIRED event Analysis Tools (including LCD-specific WIRED event

display).display).

Event Reconstruction FlowEvent Reconstruction Flow

Sequential R

ecord Loop

LC

Event

G4 Detector Simulation InputSimTrackerHits, SimCalorimeterHits, MCParticles

Detector Digitization/ Hit FindingCCD Pixels, Si strips*, TPC waveform*

Tracker hit clustering, uncertainty assignment

Track Finding/FittingStandalone 3D trackfinding (e.g. TPC, CCD)

Track merging, fitting

Calorimeter Clustering“Nearest” Neighbor, MST*, Fuzzy*Cluster Particle ID (e.g. , , h±, h0)

Reconstructed ParticleTrack-Cluster association,

Jet-Finding, Flavor TaggingZVTop vertexing, NN Flavor Tagging (b, c, uds)*

Reconstruction OutputLCIO (Tracks, Clusters, ReconstructedParticles…)

AIDA (tuples, histograms)

Reconstructors implemented as Drivers with hooks, e.g.: add(Processor p) processEvent(EventData evt) beforeFirstEvent() afterLastEvent() setDetector(Detector det) …Drivers implement Processor interface, so can be nested.

Event contains some predefined collection hooks, evt.getCalorimeterHits()Arbitrary objects can be added to/retrieved from Event evt.put(“myStuff”, myStuff)

BckgndEvents

+

Detector Hit DigitizationDetector Hit Digitization SimCalorimeter quantized into cells in Geant4.SimCalorimeter quantized into cells in Geant4.

Study readout segmentation by ganging cells.Study readout segmentation by ganging cells. SimTrackerHit has full MC info: x, t, dE/dx…SimTrackerHit has full MC info: x, t, dE/dx…

Digitized at reco level: e.g. strip pitch, pixel size, Digitized at reco level: e.g. strip pitch, pixel size, charge sharing, electronic noise, …charge sharing, electronic noise, …

Provides more realistic simulation of effects of Provides more realistic simulation of effects of backgrounds, efficiencies and noise.backgrounds, efficiencies and noise. Nearby hits merged, ghosts simulated.Nearby hits merged, ghosts simulated. Hit-specific measurement uncertainties.Hit-specific measurement uncertainties.

Can add Random Hits (salt & pepper Can add Random Hits (salt & pepper backgrounds).backgrounds).

Efficiencies simulated by dropping hits, channels.Efficiencies simulated by dropping hits, channels.

CCD DigitizationCCD Digitization VXD hits from simulated events, finds charge VXD hits from simulated events, finds charge

deposited in each pixel, adds electronics noise deposited in each pixel, adds electronics noise and digitizes signal. and digitizes signal.

Finds CCD clusters, splitting if necessary.Finds CCD clusters, splitting if necessary. Coordinates of found centers are used to Coordinates of found centers are used to

replace TrackerHits in the events. Further replace TrackerHits in the events. Further event processing (track finding, fitting, and so event processing (track finding, fitting, and so on) proceeds the same way as it was before.on) proceeds the same way as it was before.

Can set CCD parameters (like thickness, Can set CCD parameters (like thickness, depleted layer depth, epitaxial layer thickness depleted layer depth, epitaxial layer thickness and so on), electronics parameters (noise, and so on), electronics parameters (noise, ADC conversion scale, pixel and cluster ADC conversion scale, pixel and cluster thresholds), processing parameters (like thresholds), processing parameters (like cluster center calculation method).cluster center calculation method).

Stereo Angle

Wedges

Hexagons

Tiling Forward DisksTiling Forward Disks

Track Finding/FittingTrack Finding/Fitting Full pattern recognition in 3D detectors.Full pattern recognition in 3D detectors. Extrapolations of found tracks to 2D det’s.Extrapolations of found tracks to 2D det’s. Tuned for Large and Silicon Detectors.Tuned for Large and Silicon Detectors.

Working on generalizing this.*Working on generalizing this.* Track Fitters:Track Fitters:

SLD Weight Matrix Fitter.SLD Weight Matrix Fitter. Single Detector or Combined fit (e.g. VTX+TPC).Single Detector or Combined fit (e.g. VTX+TPC). Kalman Fitter*Kalman Fitter*

MIP Tracking in Cal (MIP Tracking in Cal (’s, track starting).’s, track starting).

Calorimeter ClusteringCalorimeter Clustering Cluster Interface defined, several clustering Cluster Interface defined, several clustering

algorithms currently implemented.algorithms currently implemented. MC “Cheater”, “Nearest” Neighbor, Fixed-cone, MC “Cheater”, “Nearest” Neighbor, Fixed-cone,

… … Cluster QA package developed.Cluster QA package developed. Cluster Refinements:Cluster Refinements:

Combining HAD + EM clusters.Combining HAD + EM clusters. Endcap + Barrel overlap region.Endcap + Barrel overlap region. Identify cluster as originating from charged or Identify cluster as originating from charged or

neutral hadron, EM.neutral hadron, EM.

Reconstructed Particle/ Jet Reconstructed Particle/ Jet FindingFinding Particle Flow algorithms being developed with Particle Flow algorithms being developed with

minimal coupling to specific detector designs.minimal coupling to specific detector designs. Photon and muon reconstruction fairly mature.Photon and muon reconstruction fairly mature. Emphasis on track-following for charged Emphasis on track-following for charged

hadrons.hadrons. MIP reconstruction quite promising. MIP reconstruction quite promising.

Canonical data samples identified and will be Canonical data samples identified and will be used to characterize detector response.used to characterize detector response.

Systematic investigation of Systematic investigation of jet-jet mass as a function of BnRmaplq (B-field, Cal radius, Cal cell area, Cal longitudinal segmentation), material and readout technology employing a Particle Flow paradigm being undertaken.

Java, Python, Pnuts editorDataset, Program, AIDA Explorer

Scripting Console

JAS3 IDEJAS3 IDE

Tutorial: http://jas.freehep.org/jas3/Tutorial/index.htmlTutorial: http://jas.freehep.org/jas3/Tutorial/index.html

Event Browser

Histogram & ntuple analysis,

fitting

WIRED Event DisplayWIRED Event Display

Tree controls visibility. (HepRep2 support in WIRED3

is preliminary)

Picking supports viewing physics

attributes

Next Generation (org.lcsim) Next Generation (org.lcsim) GoalsGoals

Retain core functionality from hep.lcd package.Retain core functionality from hep.lcd package. Full suite of reconstruction and analysis tools available to Full suite of reconstruction and analysis tools available to

all LCIO users.all LCIO users. Update to use LCIO for I/O and as basis for Update to use LCIO for I/O and as basis for

simulation, raw data and reconstruction event simulation, raw data and reconstruction event formats.formats.

Update/simplify framework using experience from Update/simplify framework using experience from hep.lcd.hep.lcd.

Internationalization:Internationalization: Try to make package independent of detector & geometry Try to make package independent of detector & geometry

assumptions so can work with any detector.assumptions so can work with any detector. Read properties of detectors at runtime.Read properties of detectors at runtime.

Update to Java 1.5.Update to Java 1.5. Many improvements since hep.lcd framework was Many improvements since hep.lcd framework was

created.created. Revitalize work on reconstruction algorithms.Revitalize work on reconstruction algorithms.

ConclusionsConclusions ILC Detector design is an area of active ILC Detector design is an area of active

development with many design choices, development with many design choices, requiring close coupling of design, requiring close coupling of design, simulation, and reconstruction.simulation, and reconstruction.

Ease of use and speed of development are Ease of use and speed of development are essential for physicists conducting ILC essential for physicists conducting ILC studies part-time & in a heterogeneous studies part-time & in a heterogeneous environment. environment.

A fairly complete suite of simulation tools A fairly complete suite of simulation tools written in Java exists and is being written in Java exists and is being successfully used for these studies.successfully used for these studies.

LinksLinks linearcollider.orglinearcollider.org ILC Simulations: ILC Simulations: http://http://lcsim.orglcsim.org hep.lcd: hep.lcd:

http://www-sldnt.slac.stanford.edu/jas/Documentation/lcd/http://www-sldnt.slac.stanford.edu/jas/Documentation/lcd/ TutorialsTutorials: :

http://jas.freehep.org/jas3/Tutorial/index.htmlhttp://jas.freehep.org/jas3/Tutorial/index.html http://www-http://www-sldnt.slac.stanford.edu/snowmass/Welcome.htmlsldnt.slac.stanford.edu/snowmass/Welcome.html

JAS3: JAS3: http://jas.freehep.org/jas3http://jas.freehep.org/jas3 WIRED: WIRED: http://wired.freehep.org/http://wired.freehep.org/ HepRep: HepRep: http://heprep.freehep.org/http://heprep.freehep.org/ LCIO: LCIO: http://http://lcio.desy.delcio.desy.de Discussion Forums:Discussion Forums:

http://forum.linearcollider.orghttp://forum.linearcollider.org (linear collider specific) (linear collider specific) http://http://forum.freehep.orgforum.freehep.org// (general JAS, Wired etc) (general JAS, Wired etc)

Related TalksRelated Talks248 - LCIO - A persistency framework and

data model for the linear collider, Frank Gaede

395 - WIRED 4 - A generic Event Display plugin for JAS 3, Mark Dönszelmann

397 - AIDA, JAIDA and AIDAJNI: Data Analysis using interfaces, Victor Serbo

401 - Interactive Data Analysis on the Grid using Globus 3 and JAS3, Tony Johnson

402 - Writing Extension Modules (Plug-ins) for JAS3, Mark Dönszelmann