Data Analysis within an AcquRoot Framework - Nuclear Physics ...

Data Analysis within an AcquRoot FrameworkJ.R.M. AnnandDepartment of Physics and AstronomyUniversity of GlasgowAcquRoot Version 4v2July 2, 2008AbstractACQU is a set of software tools for the acquisition and analysis of data from sub-atomicphysics experiments. This document describes the C++ version of the data analyser which isbased on the CERN toolkit ROOT. The structure of AcquRoot is shown, along with workinginstances of analysis classes and methods for conguration. A companion package MCGeneratorfor Monte Carlo simulation of data is also described.1

Contents1 Introduction 42 Software Structure of ACQU 42.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 AcquRoot Software Structure 63.1 The main program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63.2 The ROOT Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63.3 The TAcquRoot interface to ACQU Data . . . . . . . . . . . . . . . . . . . . . 83.3.1 Mk1Loop and OineLoop . . . . . . . . . . . . . . . . . . . . . . . . . . 83.4 The Threaded Data Server . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83.4.1 Ring Buer TA2RingBuer . . . . . . . . . . . . . . . . . . . . . . . . . 104 The Analyser Structure 104.0.2 TA2DataManager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104.0.3 TA2HistManager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134.0.4 TA2Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144.0.5 TA2Apparatus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164.0.6 TA2Detector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 Setup Technicalities 175.1 Installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175.1.1 Core Acqu Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . 175.1.2 User Acqu Installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195.2 Parameter Conguration Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205.2.1 Base TAcquRoot conguration le . . . . . . . . . . . . . . . . . . . . . 205.2.2 Conguration le for TA2Analysis or derived class . . . . . . . . . . . . 215.2.3 TA2DataServer conguration le . . . . . . . . . . . . . . . . . . . . . . 225.2.4 Apparatus conguration les . . . . . . . . . . . . . . . . . . . . . . . . 245.2.5 Detector conguration les . . . . . . . . . . . . . . . . . . . . . . . . . 245.2.6 Physics conguration le . . . . . . . . . . . . . . . . . . . . . . . . . . 255.3 Data Filtering and Display . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265.3.1 LoadVariable a exible means to access addresses . . . . . . . . . . . . . 275.3.2 Data Cut Specication . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295.4 Display Specication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315.4.1 Histograms set up by TA2HistManager::ParseDisplay . . . . . . . . . . 315.4.2 Special histogram setup in TA2Analysis . . . . . . . . . . . . . . . . . . 336 Specimen Analysis Classes 346.1 Detector Classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346.2 Apparatus Classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346.2.1 TA2CrystalBall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346.2.2 TA2KensTagger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346.3 Physics Classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346.3.1 TA2PhotoPhysics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347 Obsolete Companion Software 358 AcquMC a Monte Carlo reaction-kinematics generator 368.1 TMCGenerator event generation . . . . . . . . . . . . . . . . . . . . . . . . . . 378.1.1 TMCFoamGenerator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378.1.2 TMCFoamInt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378.2 TMCParticle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 398.2.1 TMCResonance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408.2.2 TMCFoamParticle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408.2.3 TMCdS5MDMParticle . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418.3 Conguration le format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418.4 Output formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432

8.4.1 Mkin output format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438.4.2 AcquMC output format . . . . . . . . . . . . . . . . . . . . . . . . . . . 449 Analysis of Monte Carlo generated data 459.0.3 Features of the GEANT-3 based simulation . . . . . . . . . . . . . . . . 4510 DAQ Software 4710.1 TDAQexperiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4710.1.1 Conguration les for TDAQexperiment . . . . . . . . . . . . . . . . . . 4710.2 TDAQmodule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4810.2.1 Conguration les for TDAQmodule . . . . . . . . . . . . . . . . . . . . 4810.2.2 Supported hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50A Applications 52A.1 Cluster Determination Algorithms for Segmented Calorimeters . . . . . . . . . 52A.1.1 The basic cluster algorithm. . . . . . . . . . . . . . . . . . . . . . . . . . 52A.1.2 The extended cluster algorithm. . . . . . . . . . . . . . . . . . . . . . . 52A.1.3 The split-o search algorithm. . . . . . . . . . . . . . . . . . . . . . . . 53A.1.4 Conguration le options for TA2ClusterDetector . . . . . . . . . . . . 53A.1.5 Preliminary Cluster studies using Monte Carlo generated events. . . . . 543

1 IntroductionACQU is mainly, but not exclusively, used within the Mainz A2 collaboration. From the outsetin 1988 the desire has been to keep all parts as exible and general purpose as possible. With alittle thought this can generally be achieved without incurring any signicant ineciencies in thedata readout and analysis software. Up to the start of Crystal Ball experiments at MAMI thetraditional form of ACQU has been used, which is written in C and C++, with a little FORTRANto interface to HBOOK. The latest and probably last version of this software is 3v4. Starting at3v5 a new data analysis program (AcquRoot) has been written to replace the traditional Sortprogram. With the implementation of threaded multi-tasking, AcquRoot has developed to theextent that it has, on its own, more-or-less full ACQU functionality and the potential to do muchmore.This manual describes the AcquRoot (AR) analysis software (including AcquMC and AcquDAQcomponents), as traditional ACQU is covered by a separate manual. As the name implies, theAR analyser is built on the framework of ROOT, the CERN C++ based suite of software toolsand libraries. This has similar functionality to its predecessor, the FORTRAN-based PAW andHBOOK, although the programing languages are quite dierent. ROOT, based on object-orientedsources, has broader range of functionality and better performance. ROOT has a further advantagethat its macro language (essentially a script processor) is in nearly all respects the same as C++.Novice analysers are encouraged to learn the basics of Linux/UNIX operating systems and to readthe ROOT user guide. These aspects are not covered in this manual.2 Software Structure of ACQU2.1 OverviewFig. 1 shows a basic layout schematic of the multi-threaded AR, with the threaded tasks withinAR taking over from traditional ACQU separate-program components. This avoids the need forcommunication via shared memory as the threads share a common data store. It is colour-codedas follows.• Red: main programs.• Dark Pink: embedded threads in main programs.• Pale Pink: embedded procedures in main programs or threads.• Pale blue: text parameter les which are read in by the main programs at the initialisationstage. These, amongst other things specify entities to display and the desired analysisprocedures.• Green: I/O to/from disk, tape or network.AR has threads which supercede the old ACQU tasks• DataServer: input of non-root-format data from experiment, e.g.storage.via net or from mass• Analysis: supercedes the old sort program.• DataStore: output of reduced (ltered data) in native or root formats. This function hasbeen absorbed into the analysis thread.4

• Control: root command-line processing for display etc.• DAQ-Control: this is under development. The DAQ software will be made ROOT compatible.As of AR-3v11 the Analysis, Control and DataServer threads are in operation.ACQU−ROOT Data Analysis & StorageJ.R.M. Annandupdate 12th Sept 2003, 19th March 20047th Jan 2005NetworkonlineACQU−ROOTC++ Analyser/ControllerMULTI−THREAD OperationAnalysis ThreadTAcquRootOnlineData InputThe AnalyserTA2AnalysisBase CLASSControl ThreadOfflineData InputROOT "shell"full ROOT capabilitySpectrum Display etc.DAQ Control ThreadGUI basedCommand Interfaceto VMEbus CPUData Server ThreadTake over Ext. Server TaskMultiple input data streamsTA2DataServerCRYSTAL−BALLMonte CarloGEANT−3 FortranHBOOK NTUPLE FileROOTFileROOTFilePARTIALLYANALYSEDDATAAnalysis Config.text fileNetworkonlineDisk/Tape file"online"..... .......Figure 1: Schematic of the AcquRoot analysis scheme.5

3 AcquRoot Software StructureAR started as a component of ACQU and originally was intended as an alternative to the Sortprogram. However its scope has broadened to the extent that it has taken over all ACQU taskscompletely.AR is coded in C++, sticking to ANSI convention where possible, and makes extensive use of thefacilities oered by ROOT. Most of the AR classes are derived from ROOT TNamed (a TObjectwith a character-string Name) and may be manipulated directly using the ROOT macro interpreterCINT.Source code les may be found in subdirectories of $acqu_sys, but navigation of this softwareis facilitated by the html documentation, which is generated from the primary .cc and .h les usingthe appropriate make command, and can be found at directory $acqu_sys/htmldoc. On Glasgowsystems the html les have links to the main ROOT documentation directory /cern/root/doc/htmldoc,this directory being specied in the .rootrc setup le. The usefulness of the html les is of courseimproved if the source code is well commented, using the conventions laid out in the ROOT THtmlClass documentation. The conguration of environmental variables such as $acqu_sys is coveredin Sec.5.1.3.1 The main programAR may be invoked from any ROOT session through dynamic loading of the AR shared library andsubsequent manipulation from the ROOT shell. However for convenience it is also compiled as anexecutable AcquRoot which can be run without explicit invocation of ROOT. The main programcalls AR Class TA2Control which inherits from TRint and thus provides a ROOT commandshell whence any ROOT command or operation may be carried out. It also creates instances ofTAcquRoot and TA2Analysis (or an inherited user-dened analysis Class) calls their congurationmethods and starts data processing.Batch processing may be performed by supplying the appropriate options at the command line:AcquRoot --offline --batch Config-File-Name >&!Message.log3.2 The ROOT InterfaceAR is incorporated into the ROOT Class hieracrchy (Fig. 2) through the Class TA2System,inheriting from TNamed, which is a TObject with a name and a title. AR Classes thus gainaccess to ROOT methods, and may be manipulated interactively from the ROOT command shell(using the CINT interpreter) while AR is running. Thus for example kinematic cuts on data couldbe altered without restarting AR.TA2System has some general-purpose methods for• File handling, via an interface to the UNIX system, with basic methods for reading andparsing lines of text.• Error handling with variable grades of severity of response, up to exit AR. The default is toprint an error message, set a ag (protected parameter fIsError of Class TA2System) andproceed. The ag may be interrogated at a later stage, e.g. by the PostInitialise() virtualmethod of TA2System.• Conguration of the Class, either by command-line input or by lines read from a parameterle, held in directory $acqu/data. Methods FileCong() and CommandCong(), along withvirtual method SetCong() handle these tasks.6

3.3 The TAcquRoot interface to ACQU DataThe TAcquRoot Class inherits from TA2System (Fig. 2) and, in common with all AR classes,conguration is performed by the TA2System methods described in Sec. 3.2.Data processing is handled by the method Run(), which is started as a separate thread (by methodStart() ) so that the output of the sorting may be examined via the ROOT shell while processingcontinues. The ROOT shell may in principle be used to alter analysis parameters while sortingproceeds, e.g. through a SetXXX() function of a particular analysis class. Run operates in threepossible modes:1. Online mode (the default mode). Here TAcquRoot links to the data-server thread embodiedin AR class TA2DataServer (3.4) using the procedure LinkDataServer( TA2DataServer*ds). Data buers in either traditional (Mk1) or new (Mk2) ACQU format are read andtransferred to a buer by the server, which TAcquRoot processes using method DataLoop()for TA2DataServer linkage. This method operate on the data buer (typically 32 kbyte long)calling Mk1EventLoop() to process each sequential event, which in turn calls the Process()method of main analysis class which inherits from TA2Analysis (4). Formats other thanMk1 (e.g. TAPS) can be handled by the TA2DataServer if an appropriate format class isprovided (see 3.4).2. Oine mode (selected by command line option --offline). Here the data-server thread isnot required and the OineLoop() method of TAcquRoot reads events from a Tree le andcalls the Process() method of the TA2Analysis as above. TA2DataServer is not used in thiscase. Oine is usually used to process pre-ltered data (by previous passes of AcquRoot)or GEANT generated data.3. Batch mode (selected by command line option --batch). The control thread is not startedand display output and log les may optionally be sent to a directory dierent from $acqu.This is useful for batch farm operations where the batch control process does not have writeaccess to a private $acqu directory. Batch mode may be used with either Online or Oinedata-input modes.3.3.1 Mk1Loop and OineLoopThese methods perform the same function of taking input data and where appropriate passing itto the analyser. Mk1Loop works with data provided by TA2DataServer while OineLoop readsthe data itself directly from a ROOT Tree le.3.4 The Threaded Data ServerThe threaded data server TA2DataServer provides all of the functionality of the traditionalDataServer program and in addition allows the simultaneous processing of multiple input datastreams. In traditional ACQU it is assumed that for multi-node DAQ systems, one of the DAQnodes is an event builder which reads/assembles sub-events and sends a single data stream to theanalyser. This scheme can be inecient if the front-end DAQ event builder is not suciently powerfuland may be unfeasible if linkage of non-compatible DAQ systems is required. For exampleat Mainz the TAPS array, which plugs the forward hole in Crystal Ball acceptance, has its ownDAQ system which is not easily married to ACQU-DAQ.Thus the option to perform event building on the data storage and analysis computer systemhas been built into AR (Fig.3). In the case of Mainz tagged-photon experiments the data fromTAPS and the Crystal Ball are sent on separate bre-optic lines to the online analysis (a highperformance Linux PC), where TA2DataServer performs the following:8

Figure 3: Threaded data server TA2DataServer• Read all streams asynchronously via classes TA2NetSource (network data transfer) or TA2FileSource(data from disk/tape le), both is which inherited from TA2DataSource. The TA2DataSourceread loops are run as separate threads. Each source feeds data into a buer (Sec.3.4.1), whereit is retrieved by TA2DataServer.• Identify and extract simultaneous sub-events from all streams. Data is assumed by defaultto be in ACQU (Mk1) format, but other other formats are possible. The basis of theformat-recognition software is class TA2DataFormat and specic formats are handled byTA2Mk1Format (Mk1) and TA2TAPSFormat (TAPS).• In all formats it is assumed that each sub-event incorporates an identication stamp (hardwaregeneratedevent number) which is used check the synchronisation of the sub-events. An erroris generated when these do not match.• If the event matching is correct then the contents of each sub-event are merged into a singleevent which is then written to an output buer. This buer is optionally written to disk andalso passed to TAcquRoot (Sec.3.3) etc. for processing.9

Single input, where TA2DataServer operates in a similar way to the old DataServer, or multipleinput data sources may be chosen via conguration le (Fig.13).3.4.1 Ring Buer TA2RingBuerTA2RingBuer is a component of the data server. It is a linked list of data buers, with thelast in the list linked back to the rst. The intermediate data buers shown in Fig.3 are of typeTA2RingBuer.4 The Analyser StructureData analysis is performed by three Classes (Fig. 4), all of which inherit from TA2DataManager(Sec.4.0.2), which provides methods for passing data between Classes, and TA2HistManager(Sec.4.0.3) which provides a general purpose histogrammer of results using the ROOT TH1F/TH2F/TH3FClasses. All three Classes have the capability to:1. Read in conguration data from le or command line using TA2System methods (Sec.3.2)2. Communicate with TAcquRoot and other data-analysis Classes.3. Histogram the results of the local calculations.In principle TA2DataManager does not restrict the system to a 3-tier structure (Fig.4) so thatsub-detectors or even sub-sub-detectors would be possible, but this would be over complicatedand detector classes in any case are geared for segmentation.4.0.2 TA2DataManagerThis Class can be envisaged as the provider of the arrowed data links shown in Fig. 4, i.e. it isthe class which glues the structure together. Analysis, apparatus, detector and physics classesall inherit from this one and are able to communicate with each other via its methods.Protected class variables include pointers to ADC and scaler information, as well as pointers to theparent class and the list of child classes. Methods CreateChild() and AddChild() create instancesof child classes, while general purpose Decode() and Reconstruct() call the specialist analysisfunctions for each child on the list.TA2DataManager is responsible for setting up lists of data ltering conditions through the methodsParseCuts() and SetCut(). Three linked lists (type TA2MultiCut*) of data cuts (type TA2Cut)are used to maintain:1. TA2MultiCut* fDataCuts: list of cuts applied globally to the output of the decode orreconstruct methods of the particular analysis class.2. TA2MultiCut* fHistCuts: list of cuts applied ONLY to specic histograms.3. TA2MultiCut* fMiscCuts: list of wild-card cuts which may be used at the analysers discretion,e.g. to determine the species of particle.4. TA2MultiCut* fCompoundCut: a means of chaining simple cuts to form a compound cut10

Figure 4: The 3-tier structure of the AR analyser. A central analyser TA2Analysis (red), hasTA2Apparatus-type children (blue). The specic apparati themselves have TA2Detector-typechildren (blue-grey).Application of cuts (and other analysis procedures) requires a method to access the variables ofa specic instance of an analysis class. This is done in a similar way to ROOT where specicinstances of classes may be accessed though a Name character string and indeed most AR classesinherit from TNamed. Method LoadVariable() implements the storage of an associative structwhich contains a pointer to a variable, a character string by which the variable is known and aninteger which species the type of the variable. Possible permutations are:1. Double_t (64-bit oat) single-value scaler variableLoadVariable("Mtot",&fMtot,EDSingleX);where variable fMtot is of type Double_t2. Double_t (64-bit oat) array of single-value variablesLoadVariable("Energy",fEnergy,EDSingleX);where variable fEnergy is of type Double_t*11

fUniqueIDfBitsfgDtorOnlyfgObjectStatkCanDeletekMustCleanupkObjInCanvas@~TObjectMakeZombieDoErrorTObjectTObjectoperator=AppendPadBrowseClassNameClearCloneCompareCopyDeleteDistancetoPrimitiveDrawDrawClassDrawCloneDumpExecuteExecuteExecuteEventFindObjectFindObjectGetDrawOptionGetUniqueIDGetNameGetIconNameGetOptionTObjectkIsReferencedkHasUUIDkCannotPickkZombiekBitMaskkSingleKeykNoContextMenu kOverwritekInvalidObject kWriteDeletekIsOnHeap fgIsAkNotDeletedGetObjectInfoGetTitleHandleTimerHashInheritsFromInheritsFromInspectIsFolderIsEqualIsSortableIsOnHeapIsZombieNotifylsPaintPopPrintReadRecursiveRemoveSaveAsSavePrimitiveSetDrawOptionSetUniqueIDUseCurrentStyleWriteWriteoperator newoperator new@[@]operator newoperator new@[@]operator deleteoperator delete@[@] operator=operator delete Clearoperator delete@[@] CloneSetBitSetBitResetBitTestBitTestBitsInvertBitInfoWarningErrorSysErrorFatalAbstractMethodMayNotUseGetDtorOnlySetDtorOnlyGetObjectStatSetObjectStatClassClass_NameIsAShowMembersfNamefTitlefgIsATNamed@~TNamedTNamedTNamedTNamedTNamedCompareCopyFillBufferGetNameGetTitleHashIsSortableSetNameSetNameTitleSetTitlelsPrintSizeofClassClass_NameIsAShowMembersTA2SystemfCmdList[5]fInputFilefLogFilefLogStreamfIsInitfIsErrorfIsConfigPassfgIsA@~TA2SystemBaseInitPrintErrorPrintMessageMap2KeyMap2KeyMap2StringGetMapKeyGetMapStringAddCmdListFileConfigCommandConfigParseMiscSetConfigPostInitBuildNameCheckNameGetInputFileGetLogFileGetLogStreamSetLogStreamSetLogFileIsInitIsErrorClassClass_NameIsAShowMembersfRandomfRandSeedfParentfChildrenfVariablesfTreeFilefTreefBranchfNbranchfDataCutsfHistCutsfMiscCutsfCompoundCutfRateMonitorfBitPatternfPatternHitsfNPatternHits@~TA2DataManagerPostInitGetN2VGetVarTypeIsIntCreateChildLoadVariableParseCutResetCutPostInitialiseAddChildAddChildDecodeDecodeSavedReconstructCleanupUpdatePeriodPeriodicParsePeriodParseEndOfFileParseFinishSetEndOfFileSetFinishEndFileFinishMacroExeSetSaveEventTA2DataManagerfADCfADCdefinedfMultifScalerfScalerBufffScalerSumfEventfMaxADCfMaxScalerfScalerIndexfScalerCurrfScalerAccfNPeriodfMaxPeriodfPeriodCmdfEndOfFileCmdfFinishCmdInitSaveTreeInitSaveTreeSaveEventSaveEventCloseEventGetChildTypeGetChildGetGrandChildGetParentGetChildrenGetDataCutsGetHistCutsGetMiscCutsGetADCGetADCdefinedGetScalerGetScalerSumGetEventGetEventGetMaxADCGetMaxScalerGetScalerIndexGetScalerCurrGetScalerAccGetScalerErrorGetMultiGetMultifTreeFileNamefIsRawDecodefIsDecodefIsReconstructfIsRawDecodeOKfIsDecodeOKfIsReconstructOKfIsCutfIsBitPatternfIsRateMonitorfIsEndOfFilefIsFinishfIsSaveEventkValidChildfgIsAGetRateMonitorGetBitPatternGetPatternHitsGetNPatternHitsGetTreeFileGetTreeGetNPeriodGetMaxPeriodGetPeriodCmdGetEndOfFileCmdGetFinishCmdIsRawDecodeIsDecodeIsReconstructIsRawDecodeOKIsDecodeOKIsReconstructOKIsBitPatternIsRateMonitorIsEndOfFileIsFinishIsSaveEventClassClass_NameIsAShowMembersFigure 5: TA2DataManager inheritance scheme. This plot, auto-generated by THtml, lists classvariables and procedures.3. Double_t (64-bit oat) multi-value variableLoadVariable("Mmiss",fMmiss,EDMultiX);where variable fMmiss is of type Double_t*. Multi-value variables usually have a nonconstantmultiplicity event-by-event, with the end-of-buer delimited by the marker EBuer-End. If histogramed, all multiple values are lled into the same histogram.4. Int_t (32-bit integer) versions of items 1 3 above may be specied by substituting EI forED, so that for example EDSingleX → EISingleX.The associative struct is wrapped in a ROOT TObject in the form of class TA2_N2V so that itmay be added to the fVariables linked list of TA2DataManager. From here it may be accessedusing the name string with the GetN2V(char*) method, or the address of the variable may bereturned using GetVar(char*), which in addition pulls out the pointer from TA2_N2V. In caseswhere the TA2DataManager is a parent (e.g. TA2Apparatus) or a grandparent (TA2Analysis),GetN2V is able to decend through the child/child-of-child lists in search of the variable name.Where the variable is stored in a child class the supplied name would a compound of the nameof the child class and the name of the variable. For example at physics or analysis level access of12

the TimeOR variable of detector class NaI a part of apparatus CB would be performed usingthe string CB_NaI_TimeOR. Obviously the placement of the '_' character within a variable orclass name should be avoided as it will confuse GetN2V().Further information on the use of LoadVariable() and the setting up of data cuts is given inSec.5.3.2.4.0.3 TA2HistManagerThis class inherits from TA2DataManager (Fig.5) and contains methods ParseDisplay(), Setup1D(),SetupMulti1D() and Setup2D() to specify and initialise the histograming of analysis class variables.fNamefTitlefgIsATNamed@~TNamedTNamedTNamedTNamedTNamedoperator=ClearCloneCompareCopyFillBufferGetNameGetTitleHashIsSortableSetNameSetNameTitleSetTitlelsPrintSizeofClassClass_NameIsAShowMembersTA2SystemfCmdList[5]fInputFilefLogFilefLogStreamfIsInitfIsErrorfIsConfigPassfgIsA@~TA2SystemBaseInitPrintErrorPrintMessageMap2KeyMap2KeyMap2StringGetMapKeyGetMapStringAddCmdListFileConfigCommandConfigParseMiscSetConfigPostInitBuildNameCheckNameGetInputFileGetLogFileGetLogStreamSetLogStreamSetLogFileIsInitIsErrorClassClass_NameIsAShowMembersfRandomfRandSeedfParentfChildrenfVariablesfTreeFilefTreefBranchfNbranchfDataCutsfHistCutsfMiscCutsfCompoundCutfRateMonitorfBitPatternfPatternHitsfNPatternHitsTA2DataManagerfADCfADCdefinedfMultifScalerfScalerBufffScalerSumfEventfMaxADCfMaxScalerfScalerIndexfScalerCurrfScalerAccfNPeriodfMaxPeriodfPeriodCmdfEndOfFileCmdfFinishCmd@~TA2DataManager InitSaveTreePostInitInitSaveTreeGetN2VSaveEventGetVarType SaveEventIsIntCloseEventCreateChild GetChildTypeLoadVariable GetChildParseCutGetGrandChildResetCutGetParentPostInitialise GetChildrenAddChildGetDataCutsAddChildGetHistCutsDecodeGetMiscCutsDecodeSaved GetADCReconstruct GetADCdefinedCleanupGetScalerUpdatePeriod GetScalerSumPeriodicGetEventParsePeriod GetEventParseEndOfFile GetMaxADCParseFinish GetMaxScalerSetEndOfFile GetScalerIndexSetFinishGetScalerCurrEndFileGetScalerAccFinishGetScalerErrorMacroExeGetMultiSetSaveEvent GetMultifTreeFileNamefIsRawDecodefIsDecodefIsReconstructfIsRawDecodeOKfIsDecodeOKfIsReconstructOKfIsCutfIsBitPatternfIsRateMonitorfIsEndOfFilefIsFinishfIsSaveEventkValidChildfgIsAGetRateMonitorGetBitPatternGetPatternHitsGetNPatternHitsGetTreeFileGetTreeGetNPeriodGetMaxPeriodGetPeriodCmdGetEndOfFileCmdGetFinishCmdIsRawDecodeIsDecodeIsReconstructIsRawDecodeOKIsDecodeOKIsReconstructOKIsBitPatternIsRateMonitorIsEndOfFileIsFinishIsSaveEventClassClass_NameIsAShowMembersTA2HistManagerfHistListf1Dhistf2Dhistf3DhistfN1DhistfN2DhistfN3DhistfIsDisplayfgIsA@~TA2HistManagerDecodeDecodeSavedReconstructSetConfigParseDisplayFillHistFillAllHistHistZeroHistZeroAllSaveHistListHistH2DIsDisplayGetN1DhistGetN2DhistGetN3DhistGet1DhistGet2DhistGet3DhistClassClass_NameIsAShowMembersFigure 6: TA2HistManager inheritance scheme. This plot, auto-generated by THtml lists classvariables and procedures.This class, which is part of any analysis class, maintains lists of 1D (TList* f1Dhist), 2D (TList*f2Dhist) and 3D (TList* f3Dhist) histograms of the decoded or reconstructed local variables.Histograms are based on the TH1F, TH2F and TH3F ROOT Classes with outer, templatedwrapper Classes to handle specics:13

1. TA2H1S: 1D spectrum for a single-valued parameter, which may be of any type (integer,oat etc.), but is cast to (Double_t*) when the histogram is lled.2. TA2H1M: 1D spectrum for a multi-valued parameter. This is contained in a buer (theaddress of the buer is stored at construct). The buer is scanned and each value is histogrameduntil the sequence is terminated by a EBuerEnd delimiter.3. TA2H1MF: 1D spectrum of signal shape from a sampling ADC. Otherwise similar to TA2H1M.4. TA2H1MS: 1D spectrum of consecutive scaler counts. Operationally similar to TA2H1MF.5. TA2H2SS: 2D spectrum with single-valued x and y parameters.6. TA2H2MS: 2D spectrum with multi-valued x and single-valued y parameters.7. TA2H2SM: 2D spectrum with single-valued x and multi-valued y parameters.8. TA2H2MS: 2D spectrum with multi-valued x and y parameters.9. TA2H3S: 3D spectrum with single-valued x, y and z parameters.10. TA2H3M: 3D spectrum with multi-valued x, y and z parametersNote that the TA2H are passed the address(es) of the variable(s) to be histogramed on construction.The addresses are retrieved using the LoadVariable() and GetN2V() methods described inSec.4.0.2 and so the variable naming convention is the same as for data-ltering cuts.The method FillHist() calls the templated Fill() routine specic to type for each TA2H found inthe lists of spectra. Method Hist() plots a spectrum, List1D, List2D() output the contents ofthe 1D and 2D histogram lists, ZeroHist(), ZeroAll() zero the histogram contents and SaveHist()writes all listed histograms of a particular class to a ROOT le.4.0.4 TA2AnalysisTA2Analysis, which inherits from TA2HistManager (Fig.7) provides the core of the data analysissystem (Fig.4), with all sorting taking place within the Process() method.14

TA2SystemfCmdList[5]fInputFilefLogFilefLogStreamfIsInitfIsErrorfIsConfigPassfgIsA@~TA2SystemBaseInitPrintErrorPrintMessageMap2KeyMap2KeyMap2StringGetMapKeyGetMapStringAddCmdListFileConfigCommandConfigParseMiscSetConfigPostInitBuildNameCheckNameGetInputFileGetLogFileGetLogStreamSetLogStreamSetLogFileIsInitIsErrorClassClass_NameIsAShowMembersfRandomfRandSeedfParentfChildrenfVariablesfTreeFilefTreefBranchfNbranchfDataCutsfHistCutsfMiscCutsfCompoundCutfRateMonitorfBitPatternfPatternHitsfNPatternHitsTA2DataManagerfADCfADCdefinedfMultifScalerfScalerBufffScalerSumfEventfMaxADCfMaxScalerfScalerIndexfScalerCurrfScalerAccfNPeriodfMaxPeriodfPeriodCmdfEndOfFileCmdfFinishCmd@~TA2DataManager InitSaveTreePostInitInitSaveTreeGetN2VSaveEventGetVarType SaveEventIsIntCloseEventCreateChild GetChildTypeLoadVariable GetChildParseCut GetGrandChildResetCut GetParentPostInitialise GetChildrenAddChild GetDataCutsAddChild GetHistCutsDecodeGetMiscCutsDecodeSaved GetADCReconstruct GetADCdefinedCleanupGetScalerUpdatePeriod GetScalerSumPeriodicGetEventParsePeriod GetEventParseEndOfFile GetMaxADCParseFinish GetMaxScalerSetEndOfFile GetScalerIndexSetFinish GetScalerCurrEndFileGetScalerAccFinishGetScalerErrorMacroExe GetMultiSetSaveEvent GetMultifTreeFileNamefIsRawDecodefIsDecodefIsReconstructfIsRawDecodeOKfIsDecodeOKfIsReconstructOKfIsCutfIsBitPatternfIsRateMonitorfIsEndOfFilefIsFinishfIsSaveEventkValidChildfgIsAGetRateMonitorGetBitPatternGetPatternHitsGetNPatternHitsGetTreeFileGetTreeGetNPeriodGetMaxPeriodGetPeriodCmdGetEndOfFileCmdGetFinishCmdIsRawDecodeIsDecodeIsReconstructIsRawDecodeOKIsDecodeOKIsReconstructOKIsBitPatternIsRateMonitorIsEndOfFileIsFinishIsSaveEventClassClass_NameIsAShowMembersTA2HistManagerfHistListf1Dhistf2Dhistf3DhistfN1DhistfN2DhistfN3DhistfIsDisplayfgIsA@~TA2HistManagerDecodeDecodeSavedReconstructSetConfigParseDisplayFillHistFillAllHistHistZeroHistZeroAllSaveHistListHistH2DIsDisplayGetN1DhistGetN2DhistGetN3DhistGet1DhistGet2DhistGet3DhistClassClass_NameIsAShowMembersfRawDatafNDAQEventfNhitsfNEventfNEvAnalysedfNOnlineEventfNHardwareErrorfHardwareErrorfFlash@~TA2AnalysisTA2AnalysisSetConfigProcessMProcessCleanupCreateChildParseDisplayLoadVariableParseADCPostInitialiseScalersZeroSumScalersRawDecodeEndFileFinishInitSaveTreeChangeTreeFileInitTreeFileNameSaveHistTA2AnalysisfMultifPhysicsfPrevFileNamefParticleIDfIsPhysicsfIsProcessCompletefgIsAGetNhitsGetFlashGetMultiGetMultiGetPhysicsIsPhysicsGetNEventGetNEvAnalysedGetNOnlineEventGetNDAQEventGetParticleIDSetParticleIDIsProcessCompleteClassClass_NameIsAShowMembersStreamerFigure 7: TA2Analysis inherits from TA2HistManager, TA2DataManager and TA2System. Thisplot, auto-generated by THtml, lists class variables and procedures.The event processing loop shown in Fig.8 is the most important procedure of TA2Analysis.15

Figure 8: TA2Analysis::Process() ow diagram4.0.5 TA2ApparatusTA2SystemfCmdList[5]fInputFilefLogFilefLogStreamfIsInitfIsErrorfIsConfigPassfgIsA@~TA2SystemBaseInitPrintErrorPrintMessageMap2KeyTA2DataManagerfRandomfRandSeedfParentfChildrenfVariablesfTreeFilefTreefBranchfNbranchfDataCutsfHistCutsfMiscCutsfCompoundCutfRateMonitorfBitPatternfADCfADCdefinedfMultifScalerfScalerBufffScalerSumfEventfMaxADCfMaxScalerfScalerIndexfScalerCurrfScalerAccfNPeriodfMaxPeriodfPeriodCmdfTreeFileNamefIsRawDecodefIsDecodefIsReconstructfIsRawDecodeOKfIsDecodeOKfIsReconstructOKfIsCutfIsBitPatternfIsRateMonitorfIsEndOfFilefIsFinishfIsSaveEventkValidChildfgIsATA2HistManagerfHistListf1Dhistf2Dhistf3DhistfN1DhistfN2DhistfN3DhistfIsDisplayfgIsA@~TA2HistManagerDecodeDecodeSavedReconstructTA2ApparatusfP4fPDGtypefPDG_IDfP4totfParticleIDfMtotfMinvfEtotfMaxParticlefNparticlefROOToutputfPCutfNCutfNSectorfPCutStartfNSectorCutfgIsA@~TA2ApparatusSetConfigLoadVariablePostInitCreateChildGetNparticleGetPCutGetPCutGetNCutGetNSector16

4.0.6 TA2DetectorTA2SystemfCmdList[5]fInputFilefLogFilefLogStreamfIsInitfIsErrorfIsConfigPassfgIsA@~TA2SystemBaseInitPrintErrorPrintMessageMap2KeyMap2KeyMap2StringGetMapKeyGetMapStringAddCmdListFileConfigCommandConfigParseMiscSetConfigPostInitBuildNameCheckNameGetInputFileGetLogFileGetLogStreamSetLogStreamSetLogFileIsInitIsErrorClassClass_NameIsAShowMembersfRandomfRandSeedfParentfChildrenfVariablesfTreeFilefTreefBranchfNbranchfDataCutsfHistCutsfMiscCutsfCompoundCutfRateMonitorfBitPatternfPatternHitsfNPatternHitsTA2DataManagerfADCfADCdefinedfMultifScalerfScalerBufffScalerSumfEventfMaxADCfMaxScalerfScalerIndexfScalerCurrfScalerAccfNPeriodfMaxPeriodfPeriodCmdfEndOfFileCmdfFinishCmd@~TA2DataManager InitSaveTreePostInitInitSaveTreeGetN2VSaveEventGetVarType SaveEventIsIntCloseEventCreateChild GetChildTypeLoadVariable GetChildParseCut GetGrandChildResetCut GetParentPostInitialise GetChildrenAddChild GetDataCutsAddChild GetHistCutsDecodeGetMiscCutsDecodeSaved GetADCReconstruct GetADCdefinedCleanupGetScalerUpdatePeriod GetScalerSumPeriodicGetEventParsePeriod GetEventParseEndOfFile GetMaxADCParseFinish GetMaxScalerSetEndOfFile GetScalerIndexSetFinish GetScalerCurrEndFileGetScalerAccFinishGetScalerErrorMacroExe GetMultiSetSaveEvent GetMultifTreeFileNamefIsRawDecodefIsDecodefIsReconstructfIsRawDecodeOKfIsDecodeOKfIsReconstructOKfIsCutfIsBitPatternfIsRateMonitorfIsEndOfFilefIsFinishfIsSaveEventkValidChildfgIsAGetRateMonitorGetBitPatternGetPatternHitsGetNPatternHitsGetTreeFileGetTreeGetNPeriodGetMaxPeriodGetPeriodCmdGetEndOfFileCmdGetFinishCmdIsRawDecodeIsDecodeIsReconstructIsRawDecodeOKIsDecodeOKIsReconstructOKIsBitPatternIsRateMonitorIsEndOfFileIsFinishIsSaveEventClassClass_NameIsAShowMembersTA2HistManagerfHistListf1Dhistf2Dhistf3DhistfN1DhistfN2DhistfN3DhistfIsDisplayfgIsA@~TA2HistManagerDecodeDecodeSavedReconstructSetConfigParseDisplayFillHistFillAllHistHistZeroHistZeroAllSaveHistListHistH2DIsDisplayGetN1DhistGetN2DhistGetN3DhistGet1DhistGet2DhistGet3DhistClassClass_NameIsAShowMembersfElementfEnergyfEnergyORfTimefTimeORfPositionfMeanPosfShiftOpfShiftValuefNShiftOpfHitsfRawTimeHitsfRawEnergyHitsfDetectorCommfDetectorHistfTotalEnergy@~TA2DetectorDecodeBasicSetConfigPostInitDecodeDecodeSavedReconstructCleanupSaveDecodedReadDecodedSetupArraysDeleteArraysCreateChildLoadVariableParseShiftApplyShiftGetElementGetElementGetEnergyGetEnergyORGetEnergyGetTimeGetTimeORGetTimeGetPositionGetPositionGetMeanPosTA2DetectorfEnergyScalefTimeOffsetfNhitsfNADChitsfNTDChitsfNelementfNelemfMaxHitsfIsECalibfIsScalerfIsEnergyfIsTimefIsPosfIsRawHitsfgIsAGetShiftOpGetShiftValueGetNShiftOpGetHitsGetHitsGetRawTimeHitsGetRawEnergyHitsGetTotalEnergyGetEnergyScaleGetTimeOffsetGetNhitsGetNADChitsGetNTDChitsGetNelementGetNelemGetMaxHitsIsEcalibIsScalerIsEnergyIsTimeIsPosIsRawHitsClassClass_NameIsAShowMembersFigure 10: TA2Detector inherits from TA2HistManager, TA2DataManager and TA2System. Thisplot, auto-generated by THtml, lists class variables and procedures.5 Setup Technicalities5.1 InstallationAny user account on a Linux-based system may be set up for AR use. Installation procedures forboth core and user acqu distributions are given in the following subsections.5.1.1 Core Acqu DistributionThis contains the kernel of AcquRoot in shared libraries and header les. When installed ona commonly accessible disk, these may be linked by multiple users to provide core AcquRootfunctionality to personal analysis implementations. The following assumes that core software is17

Env. Variable Typical Value Env. Variable Typical ValueROOTSYS /cern/root/dev acquversion 4v1ROOT /home/jrma/cern/root acqudate 7th Jul 2007LD_LIBRARY_PATH .:/cern/root/dev/lib:... acqu_site jrma1ACQU_SYSROOT /home/acqu/4v1 acqu_sys /home/acqu/4v1/acquACQUROOT /home/jrma acqu /home/jrma/acquCCCOMP g++ acqu_debugger dddCPUTYPE __LINUX acqu_debug -gacqu_usesubdirsYESTable 1: The ACQU UNIX environmental variables for user jrma linking to AcquRoot 4v1 installedon user acqu. Note that from 4v0 onwards code is written entirely in C++ and old environmentalvariables relating to other languages are no longer used.installed in a dedicated user account. This account, called here acqu, needs no special systempriveleges, but its directory system should be readable by other users.1. Login as acqu. Copy or download the ACQU distribution tar le to the home directory. Thename (e.g. 4v1.25.08.07.tar.gz) species the version (i.e. version 3, subversion 17) and date(i.e. 1st December 2004).2. Unpack the distribution. From the home directory...tar zxvf 4v1.25.08.07.tar.gzThis should create a directory called 4v1.25.08.07. For setup scripts to work properly thefollowing symbolic link should by created...ln -s 4v1.25.08.07 4v1Its often convenient to make a further symbolic link...ln -s 4v1 devThe next item assumes this has been done3. Basic set up of the acqu account (skip this item if the set up has been performed already).Two rc les .acqurc, .rootrc should reside on the home directory. From the home directory...cp dev/acqu/User/.acqurc .cp dev/acqu/User/.rootrc .Edit .acqurc. Check the setup of environment variables ACQU_SYSROOT, ACQUROOTand ROOTSYS and if necessary change.If running on systems where processors with dierent architecture may be used (e.g. if yourdesktop is an I386 and your batch farm uses X64) the .acqurc le should contain the linesetenv acqu_usesubdirs YESREAD THE COMMENTS in .acqurc for guidance.Edit .rootrc. Usually only the line specifying the directory where ROOT html documentationresides requires to be changed. The can also be set to a web address.Create or edit .tcshrc (or .cshrc) the setup script for the shell tcsh. Suitable lines to pastein for automatic ROOT and ACQU set up can be found in ~/dev/acqu/User/.tcshrc.4. Test the automatic set up (skip if performed already). Logout and then login again as acqu.Some messages should appear regarding executing the acqu_setup script. If any error isreported while the tcsh shell executes the .tcshrc initial script, its possible that the ACQUenvironment has not been initialised fully. The shell command printenv will print out alldened environmental variables. The ones congured by ACQU (there will be others) shouldlook similar to the typical values in Table 1.5. Build ACQU...cd $acqu18

make ultracleanmakeThis compiles and links the entire suite of ACQU programs including the html documentation.To compile and link AcquRoot only...make AcquRoot5.1.2 User Acqu InstallationDetermine the account holding the core ACQU distribution, which the user needs to access (readonly).This will usually be called something similar to acqu, acqusys or a2acqu. Listing thehome directory of this account should reveal directories, the working versions of which have asymbolic link pointer, e.g. directory 4v1.25.08.07 (version 4v1 at date 25th August 2007) pointedto by link 4v1. To set up normal-user AcquRoot data analysis...1. Unless otherwise directed select the highest version/subversion and execute the setup script:source ~acqu/4v1/acqu/acqu_setupThis line should be inserted into the setup script for the tcsh shell setup script usually called~/.tcshrc or ~/.cshrc as outlined in item 3 of Sec.5.1.1.2. If this is a completely new start then setup a private acqu directory:cd ~; tar ztvf $acqu_sys/User/UserAcqu.tar.gzThis merely lists the contents of the archive so that the le and directory specication maybe checked. To actually install:tar zxvf $acqu_sys/User/UserAcqu.tar.gzThis will create an acqu directory which contains bin, obj, lib, root/src and root/macrossubdirectories.3. If this is a completely new start copy over two rc les to your home directory (~):cd; cp $acqu_sys/User/.acqurc .cp $acqu_sys/User/.rootrc .These les may require editing to conform with the peculiarities of the local Linux system.Where the user wishes to retain the functionality of an existing .rootrc le the setup willneed to be merged, possibly requiring expert advice. However in many cases the .rootrcchanges are minimal, relating to the html documentation directory only.4. Look closely at the lines in .acqurcsetenv ACQU_SYSROOT /home/acqu/$acquversionsetenv CERNROOT /cern/devThis assumes /home/acqu is the acqu account home directory and /cern/dev the placewhere cernlib (the old F77 based CERN tools) software resides (if its installed).setenv acqu_debugger dddsetenv acqu_debug -gThis chooses the ddd GUI interface to the gdb debugger and sets the ag to compile indebug mode. To compile in standard mode comment out the latter line. The directoryspecications are highly system dependent.5. When satised that the rc les are correctly congured exit the shell and open a new shell(i.e. a start a new xterm of the equivalent). The acqu_setup script should execute andprint a banner with version and date. Now perform the following:cd $acqu; make ultraclean; makeThis will compile the sample user source code which is provided with the distribution indirectory $acqu/root/src and if successful write an executable AcquRoot to the $acqu/bindirectory and a private ROOT library $acqu/lib/libUserRoot.so.19

6. To run AcquRoot...cd $acqu; AcquRoot CB.OfflineCB.Oine is the name of the base le which congures an analysis of CB data from disk.Unless the AcquRoot parameter les are correctly congured this will very likely fail. Con-guration is described in Sec.5.25.2 Parameter Conguration FilesThe structure of the tree of parameter les to set up the AR analyser is basically similar tothat in Fig.4. The parameter les are human-readable text les, created using a standard editor.Comment lines are designated by the '#' character....please read the comments/explanations inthose displayed in subsequent gures.5.2.1 Base TAcquRoot conguration leThis supplies information to setup:1. The tree structure of ROOT les for oine analysis.2. The analysis class and analysis-class conguration le to use.3. The data server conguration le to use for online runningThe typed instructionAcquRoot CB.Offlinecauses AR to use the le $acqu/data/CB.Oine (Fig.11) as the base conguration le, i.e. theconguration le for the class TAcquRoot. If the le name is omitted then AR will try to open$acqu/data/ROOTsetup.dat. Any base le may be made the default by making a symbolic link.cd $acqu/data; ln -s CB.Offline ROOTsetup.datThe full set of cong commands provided by TAcquRoot are given in Table 2 and an example oftheir use in a cong. le is shown in Fig.11.Keyword Parameters FunctionName: char, int, char Output tree name, #branch, proc-typeBranch: char, int Name & buer size (byte) of branchDirectory: char Name of output-le directoryAnalysis: char Name of analysis class to useAnalysisSetup: char Name of analysis cong leServerSetup: char Name of data server cong leSetADC: int, int, int Manual ADC/scaler #adc, #scaler, bits/adcEventSize: int ONLY used if no Name: Branch: etc.EventIndex: int Manual setup of adc index of event ID stampTreeFile: char Name of input le for oine analysisScalerNoClear: int, int... Fragmented scaler read: block indicesLocal-DAQ: Char_t*, .... DAQ(s) on same node as AR, DAQ cong le name(s)Table 2: TAcquRoot conguration commands20

Wednesday December 15, 2004 /home/jrma/acqu/data/CB.OfflineDec 15, 04 12:42 CB.OfflinePage 1/2##−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−##−−Author JRM Annand 29th Apr 2003##−−Update JRM Annand... 1st Dec 2004##−−Description## *** Acqu++ Root ***## Online/Offline Analysis of Sub−Atomic Physics Experimental Data#### Parameter File: CB.Offline## Master parameter file for Acqu−Root analysis. Expected to reside in## directory $acqu/data/## To use invoke...## AcquRoot CB.Offline##−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−## Input data setup for Offline analysis...this info is ignored## if no offline connection is used, ie can be left in.## Tree name and # branches are used if storing filtered data in root tree## format or reading data ( −−offline ) from a chain of tree files.## Process Type can be "Raw", "MC" or "Physics".## For a standard Raw analysis 3 branches would normally be used.## For MC input the number of branches depends on the format of the saved data## Physics mode isn’t implmented yet## Ensure the buffer size is sufficient for each branch.## RawEvent_Hits: 32768 is the data record length...should be OK## RawEvent_Scalers: 4 * (number of scalers + 1) is sufficient## Mk1_Header: 4 * (number of ADCs + 1) is sufficient## NB with TAPS the number of ADCs = 29001## Tree Name # branches Process TypeName: RawEvent_AR_Save 3 Raw## Branch Name Buffer sizeBranch: RawEvent_Hits 32768Branch: RawEvent_Scalers 4192Branch: Mk1_Header 120000#### List of Tree files which will form the input TChain for −−offline runningTreeFile: /scratch/jrma/CB_3557.rd0TreeFile: /scratch/jrma/CB_3558.rd0#### Directory to save any created ROOT files (include final / )Directory: /scratch/jrma/#### Manual setup of ADC, Scaler, bits−in−ADC numbers## Usually needed only where experiment header information is not given## e.g. if the input tree file is generated by GEANT## #ADC #Scaler #bits#SetADC: 30000 492 12#### End of Offline analysis input specification##−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−## Needed for offline analysis if event marker ID hasn’t been saved in header## IDEventIndex: 400#### Normally don’t attempt to set eventsize manually##EventSize: 32768#### Analysis mode.## User−defined analysis classAnalysis: TA2UserAnalysis## Analysis setup fileAnalysisSetup: CBanalysis.dat#### Online running...setup of data serverFigure 11: AR sample base conguration le. It species the user analysis class TA2UserAnalysis,congured using the le CBanalysis.dat and the internal data server, congured using CBserver-Setup.datDec 15, 04 12:42 CB.OfflinePage 2/2## This is ignored for offline running and may be left inServerSetup: CBserver.OfflinePrinted by Red Hat Linux User5.2.2 Conguration le for TA2Analysis or derived classA sample analysis conguration le CBanalysis.dat is shown in Fig.12. This species:• handlers for multi-hit or ash ADC's (standard single-value ADC's do not require any inputhere)• the optional apparatus classes to load, and their conguration les. Apparatus-level reconstructionis also enabled.• the optional physics class to load and the conguration le.• the display of some important raw ADC's and scalers21

Tuesday December 14, 2004 /home/jrma/acqu/data/CBanalysis.dat### Stuff to monitor time dependence of counting rates and the hit pattern## coming from the trigger box....the parameter is the setup file nameRateMonitor: Trigger.RateBitPattern: Trigger.Pat#### Apparatus setup## Raw analysis may be performed without Apparati## Name Class Setup fileApparatus: CB TA2Calorimeter CB.datApparatus: TAPS TA2Calorimeter TAPS.datApparatus: TAGG TA2KensTagger KensTagger.datApparatus: LinPol TA2LinearPol LinearPolapparatus.dat#### Turn on apparatus−level reconstructionReconstruct−Analysis:#### Specify physics analysisPhysics−Analysis: Meson TA2PhotoPhysics CBphysics.dat#### Do some post initialisation...usually necessaryInitialise:#### Setup of extra procedures to run at periodic intervals## The procedures are usually held in macro script files## Period...every N events## N Macro File Macro ProcedurePeriod: 100000 PeriodMacro.C PeriodMacro()####−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−##−−Author JRM Annand 29th Apr 2003##−−Update JRM Annand... 1st Dec 2004##−−Description## *** Acqu++ Root ***## Online/Offline Analysis of Sub−Atomic Physics Experimental Data#### Parameter File: CBanalysis.dat (specified in CB.Offline)## Setup basic parameters of analysis to be used## This one sets up an analysis of Crystal Ball data####−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−#### The following give sample specifications for## handlers for flash or multihit ADCs or TDCs#### Catch−TDC, where a reference TDC channel has to be specified## The reference TDC channel contains the trigger time which is subtracted## from the other specified channels## CB TDC’s## #hits start stop referenceMultiADC: 3 2001 2767 2000MultiADC: 3 4000 4511 2000#### SG ADC read out in 3−sum mode (#hits is always 3)## DC # channels summed for the pedestal (before signal leading edge)## signal # channels summed for the signal region of the pulse## tail # channels summed for the tail region (after signal trailing edge)## CB S−ADC’s# #hits start stop DC signal tailMultiADC: 3 3000 3735 30 30 30## MWPC strip S−ADC’s (switched to single−value read)#MultiADC: 3 5000 5319 30 30 30Dec 14, 04 13:21 CBanalysis.datPage 2/2## EndFile...each time an input data file is closed#EndFile: EndFileMacro.C EndFileMacro()#### Finish...when all input data files processed#Finish: FinishMacro.C FinishMacro()#### Save all accepted events (ie those which pass all data cuts) to a## ROOT file. Data saved in RAW format (ie ADC indices and values)#Save−Event:#### Histogram ALL defined single value ADC’s#Display: All## Histogram ADCs 100 − 131#Display: ADC 100 131#### Display of multihit stuff etc.## CB CATCH TDC’s, 1st hit#Display: MultiHit 2001 2007 1000 −1000 0 0#Display: MultiHit 2008 2031 1000 −3000 −2000 0#### CB CATCH TDC’s, 2nd hit#Display: MultiHit 2032 2720 500 −2600 400 0#Display: MultiHit 4000 4383 500 −4000 2000 0#### CB SADC’s pedestal#Display: MultiHit 3000 3720 100 6000 7000 0## CB SADC’s signal#Display: MultiHit 3000 3720 1000 0000 4000 1## CB SADC’s tail#Display: MultiHit 3000 3720 500 −800 200 2#### Display blocks of scalersDisplay: Scaler 0 11Display: SumScaler 0 11Display: Scaler 12 363Display: SumScaler 12 363Display: Scaler 364 491#### Time dependence of individual scaler rates and ratiosDisplay: Rate 0 2500Display: Rate 1 2500Display: Rate 2 2500Display: Rate 3 2500Display: Rate 4 2500Display: Rate 5 2500#### Trigger pattern, and e−beam polarisation bitsDisplay: Pattern 0 16 0 16Display: Pattern 1 8 0 8#### CB−TAPS synchronisation monitorsDisplay: 2D 20401 200 0 4000 2001M0 200 −600 −400Display: 2D 20401 200 0 4000 2004M0 200 −600 −400Display: 2D 20401 200 0 4000 2005M0 200 −600 −400#### END OF FILEFigure 12: AR sample analysis conguration le.The full list of conguration commands for TA2Analysis is given in Table 3Keyword Parameters FunctionApparatus: char, char, char Name, class-name and cong le of an apparatusReconstruct-Analysis:Dec 14, 04 13:21 CBanalysis.datPage 1/2Enable apparatus reconstruct proceduresPhysics-Analysis char, char, char Name, class-name and cong le of physicsDisplay: see Fig.12 Set up histrgramsInitialise:Some further setting after all parms read inMultiADC: see Fig.12 Handler for multi-value ADCsFlashADC: see Fig.12 Handler for sampling ADCsBitPattern: char Hit pattern-unit decode, name of cong leRateMonitor: char Scaler time dependences, name of cong lePeriod: int,char,char period(# events), macro le, macro commandEndFile: char, char macro le, macro commandFinish: char, char macro le, macro commandSave-Event:Save ltered data to ROOT tree lePrinted by ACQU data analysisTable 3: TA2Analysis conguration commands5.2.3 TA2DataServer conguration leThe cong. le to use to setup the internal DataServer is specied in the base cong. le (Fig.11)and the server cong. le itself, CBserver.Oine, is listed at the left of Fig.13. It species thatdata is received in a single stream from a disk le. An alternative server le MergeServer.Onlinespecies that data is received from 2 network stream sockets. These 2 streams are stored directly22

to disk les for diagnostic purposes, the streams are merged and the merged data is also writtento a disk le.Wednesday December 15, 2004 /home/jrma/acqu/data/CBserver.Offline, /home/jrma/acqu/data/MergeServer.OnlineDec 14, 04 13:25 CBserver.OfflinePage 1/1##−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−##−−Author JRM Annand 13th Jan 2004##−−Update JRM Annand... 1st Dec 2004##−−Description## *** Acqu++ Root ***## Online/Offline Analysis of Sub−Atomic Physics Experimental Data#### Parameter File: CBserver.Offline## Configuration file for DataServer (data input for analysis)####−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−## Number of Input streams (sources of data), Record length to write data## and switch 0/1 = no/yes to control data write## Number streams Record Len Save DataInput−Streams: 1 32768 0#### Optional directory spec. If not specified File−Names must contain the full## directory pathFile−Directory: /scratch1/acqu/#### Specify each input stream## From disk file, ACQU (Mk1) format, save any output to file (save=1)## Input data record length, Output data record length#### Source Format Save? RecLen Buffers Swap MarkIDStream−Spec: File Mk1 0 32768 32 0 400#### Now the list of data files to analyse## File Name 1st rec Last rec#File−Name: CB_1867.dat 0 0#File−Name: CB_929.dat 0 0#File−Name: CB_1980.dat 0 0#File−Name: CB_1763.dat 0 0#File−Name: CB_2749.dat 0 0File−Name: CB_3557.dat 0 0File−Name: CB_3626.dat 0 0#### ENDDec 15, 04 20:25 MergeServer.OnlinePage 1/1##−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−##−−Author JRM Annand 13th Jan 2004##−−Update JRM Annand... 1st Dec 2004##−−Description## *** Acqu++ Root ***## Online/Offline Analysis of Sub−Atomic Physics Experimental Data#### Parameter File: MergeServer.Online## Default master parameter file for DataServer (data input for analysis)## Online data taking, merged ACQU & TAPS####−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−## Number of Input streams (sources of data), Record length to write data## and switch 0/1 = no/yes to control data write## Number streams Record Len Save DataInput−Streams: 2 32768 1#### Optional directory spec. If not specified File−Names must contain the full## directory pathFile−Directory: scratch/#### Specify each network−socket input stream## Mk1(ACQU) data, node slarti,IP−port 3010, 512 elem buffer, event stamp 400## TAPS data, node taps02,IP−port 4097, 32 elem buffer, event stamp 29000## Individual data stream save is turned on...usually for diagnostic only## Source Format Save? Node Port Buffers MkID SwapStream−Spec: Network Mk1 1 slarti 3010 512 400File−Name: CB_Mk1 0 0Stream−Spec: Network TAPS 1 taps02 4097 32 29000File−Name: CB_TAPS 0 0##Figure 13: Data Server conguration lePrinted by ACQU data analysisThe following set up commands are recognised by TA2DataServerKeyword Parameters FunctionInput-Streams int, int, int #-streams, record-length, save(0/1=no/yes)Stream-Spec: see Fig.13 parameters for data streamFile-Name: char Input data le nameFile-Directory: char Directory where data les resideTable 4: TA2DataServercap:TA2Analysis-conguration-commands conguration commands23

5.2.4 Apparatus conguration lesApparatus conguration les will depend on the specic coding of the SetCong() method ofthe particular apparatus class, but in general this will call the TA2Apparatus method as well assupplying its own particular commands. Conventionally the local command list of the derivedapparatus class is searched rst for a match with the input command string and if not found thenthe command list of TA2Apparatus is searched. The le listed in Fig.14 has been used to set upthe user-dened TA2Calorimeter class but all of the lines shown are quite general. A summary ofTA2Apparatus commands is given in Table 5The gure shows how to specify...• Names, Class Names and Setup les for all detectors which are part of the class. Theapparatus contains TA2CalArray, TA2PlasticPID and TA2CylMWPC types of detectors• Some post-initialisation. This normally calls the TA2Apparatus method and may add someextras.• Procedure to execute periodically (every N events)• Filtering cuts to apply to the data, described in Sec.5.3.2.• Histograms to be made of variables, described in Sec. ??Keyword Parameters FunctionDetector: char, char, char Name, class name, cong leData-Cut: see Fig.14 Specify data cutReconstruct:Enable reconstruct methodInitialise:Further initialisationDisplay: see Fig.14 Specify histogramsParticleID: int init. PDG data base (max # particles)ParticleID-Cut int int setup particle-ID cuts from misc listMisc:NOT IMPLEMENTEDPeriod: int, char, char Period(events), Macro le, Macro commandEndFile: char, char End of File procedure (Macro le, Macro command)Finish: char, char End of Run procedure (Macro le, Macro command)Save-Event:Not yet implementedTable 5: TA2Apparatus conguration commands5.2.5 Detector conguration lesAs with apparati (Sec.5.2.4) detector conguration les depend on the specic coding of theSetCong() method of the detector class, but user detector classes generally call the TA2Detectormethod if the input command is not found in their own particular command list. The le listedin Fig.15 is used to set up a TA2ClusterDetector class which is derived from TA2Detector andillustrates cong. commands or both base and inherited class.The le supplies:• ADC, TDC and Position calibration information for each element of the array of detectorsnumbering 720 in this case.• Time-walk specication for TDC correction.24

Tuesday December 14, 2004 /home/jrma/acqu/data/CB.datDec 14, 04 13:44 CB.datPage 1/2##−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−##−−Author JRM Annand 29th Apr 2003##−−Update JRM Annand...##−−Description## *** Acqu++ Root ***## Online/Offline Analysis of Sub−Atomic Physics Experimental Data#### Parameter File: CBapparatus.dat (specified in ROOTanalysis.dat)## or equivalent## Setup analysis of the Crystal Ball apparatus## This one sets up an analysis of the Crystal_Ball NaI(Tl) array## and the inner Particle ID Detector####−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−#### Specify list of detectors which constitute the Crystal Ball apparatus## Name Class Name Setup fileDetector: NaI TA2CalArray CB_NaI_June.datDetector: PID TA2PlasticPID CB_PID_June3.datDetector: MWPC TA2CylMWPC MWPC.16.10.04#### Data reconstruction method for Crystal BallReconstruct:#### Do some post−initialisation.## Should usually be done before display is setupInitialise:#### Periodic procedure to execute every N * NA events## NA is the period specified in the analysis (see CBanalysis.dat).## Thus if NA = 100000, CBPeriodMacro() will execute every 200000 events## If no analysis periodic specified, no apparatus periodic will execute## N Macro File Macro ProcedurePeriod: 2 CBPeriodMacro.C CBPeriodMacro()#### Cut specification## NB 2D cuts, ie a general polygon on the X−Y plane not yet implemented## Simple rectangle cuts may be made by an AND of X and Y conditions## 1/2D Name Link Scope Low/High Limits#Data−Cut: 1D MtotC AND compound−data#Data−Cut: 1D Mtot OR element 100 180#Data−Cut: 1D Mtot OR element 500 600#Data−Cut: 1D PID_Nhits AND data 1 16#Data−Cut: 1D NaI_ClNhits AND data 1 16#### Histograms of apparatus−level variables## histogram name should be preceeded by 1D or 2D (for dimension of spectrum)## Name chan/low/high min/max condition(opt)#Display: 1D Nhit 16 0 16#Display: 1D Mtot 800 0 800 0 0 MtotC#Display: 1D Minv 500 0 500 0 7Display: 1D NphotMinv 800 0 800 0 1#Display: 1D NaI_ClNhitsOR 16 0 16###Display: 1D PID_Energy 100 0 20 0 23#### X−Name X−ch/low/high/id Y−Name Y−ch/low/high/id Condition(opt)#Display: 2D Mtot 100 0 800 −1 NphotMinv 100 0 800 1#Display: 2D Mtot 100 0 800 −l NphotMinv 100 0 800 2#Display: 2D Mtot 100 0 800 −1 Nhit 16 0 16 0###Display: 2D NaI_EnergyOR 100 0 800 −1 PID_Energy 100 0 20 0Dec 14, 04 13:44 CB.datPage 2/2##Display: 2D PID_Hits 24 0 24 −1 NaI_ClPhi 180 −180 180 −1Display: 2D MWPC_PhiWire0 180 0 6.5 −1 NaI_ClPhi 180 −180 180 −1Display: 2D MWPC_PhiSWDiff0 180 −0.1 0.1 −1 NaI_ClPhi 180 −180 180 −1Display: 2D MWPC_PhiWire0 180 0 6.5 −1 PID_Hits 24 0 24 −1Display: 2D MWPC_ZIntersect0 100 −200 200 −1 NaI_ClTheta 180 0 180 −1Display: 2D MWPC_ZIntersect1 100 −300 300 −1 NaI_ClTheta 180 0 180 −1Display: 2D MWPC_TrackTheta 180 0 3.15 −1 NaI_ClTheta 180 0 180 −1Display: 2D MWPC_TrackPhi 180 −3.14 3.14 −1 NaI_ClPhi 180 −180 180 −1Display: 2D MWPC_PhiDiff 180 −0.5 0.5 −1 NaI_ClPhi 180 −180 180 −1##Display 1D MWPC_Xvertex 1000 −100 100Display 1D MWPC_Yvertex 1000 −100 100Display 1D MWPC_Zvertex 200 −500 500Display 1D MWPC_PsXvertex 1000 −100 100Display 1D MWPC_PsYvertex 1000 −100 100Display 1D MWPC_PsZvertex 1000 −500 500Display: 1D MWPC_ZIntersect0 500 −220 220Display: 1D MWPC_PhiSWDiff0 100 −0.2 0.2Display: 1D MWPC_PhiWire0 360 0 6.4Display: 1D MWPC_ZIntersect1 500 −320 320Display: 1D MWPC_PhiSWDiff1 100 −0.2 0.2Display: 1D MWPC_PhiWire1 360 0 6.4Display: 1D MWPC_PhiDiff 100 −0.5 0.5Display: 2D MWPC_Xvertex 100 −50 50 −1 MWPC_Yvertex 100 −50 50 −1Display: 2D MWPC_Zvertex 150 −300 300 −1 MWPC_Rvertex 40 0 40 −1Display: 2D MWPC_PsXvertex 100 −50 50 −1 MWPC_PsYvertex 100 −50 50 −1Display: 2D MWPC_PsZvertex 150 −300 300 −1 MWPC_PsRvertex 40 0 40 −1Display: 2D MWPC_ZIntersect0 100 −220 220 −1 MWPC_PhiWire0 180 −0.1 6.4 −1Display: 2D MWPC_ZIntersect0 100 −220 220 −1 MWPC_ZIntersect1 100 −320 320 −1Display: 2D MWPC_ZIntersect0 100 −220 220 −1 MWPC_PhiSWDiff0 100 −0.1 0.1 −1Display: 2D MWPC_ZIntersect1 100 −320 320 −1 MWPC_PhiWire1 180 −0.1 6.4 −1Display: 2D MWPC_ZIntersect1 100 −320 320 −1 MWPC_PhiSWDiff1 180 −0.1 0.1 −1Display: 2D MWPC_PhiSWDiff0 100 −0.1 0.1 −1 MWPC_PhiSWDiff1 100 −0.1 0.1 −1Display: 2D MWPC_PhiWire0 180 0.0 6.4 −1 MWPC_PhiSWDiff0 100 −0.1 0.1 −1Display: 2D MWPC_PhiWire1 180 0.0 6.4 −1 MWPC_PhiSWDiff1 100 −0.1 0.1 −1Display: 2D MWPC_ZIntersect0 100 −220 220 −1 MWPC_PhiDiff 50 −0.1 0.1 −1Display: 2D MWPC_PhiWire0 180 0.0 6.4 −1 MWPC_PhiDiff 50 −0.1 0.1 −1Display: 2D MWPC_PhiWire1 180 0.0 6.4 −1 MWPC_PhiDiff 50 −0.1 0.1 −1Figure 14: Crystal Ball apparatus conguration lePrinted by ACQU data analysis• Nearest neighbour information for each element of the array.• Some postinitialisation, which is highly detector specic.• Filtering cuts to apply to the data, described in Sec.5.3.2.• Histograms to be made of variables, described in Sec. ??A summary of all recognised cong. commands for TA2Detector is given in Table 6 and extensionsto the command list provided by various derived detector types are given in Tables 7, 8 and 95.2.6 Physics conguration leTA2Physics and its inherited classes combine the information from the apparati into physicalquantities such as missing mass or missing momentum. The actual physics class to use is specied25

Dec 08, 05 19:33 CB_NaI.dat Page 1/2##−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−##−−Author JRM Annand 29th Apr 2003##−−Update JRM Annand... 1st Dec 2004##−−Description## *** Acqu++ Root ***## Online/Offline Analysis of Sub−Atomic Physics Experimental Data#### Parameter File: CB_NaI.dat. Setup of decoding of NaI information##−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−## Total # Elements in array, Pulse−Height info? Time info? Position info?## Info switch is off/on = 0/1## Total Energy? Time? Position?Size: 720 1 1 1#### Save separate ADC and TDC hits arrays...no energy/time cuts appliedRawHits:#### Get global coordinate−shift variables from input line## A shift operation is characterised by a 2−char name and value## Up to 12 operations may input and the associated corrinate## transforms are performed in order they appear on the line## Allowed operations## Translate in X,Y or Z.....TX,TY,TZ by assocoaied value## Rotate about X,Y or Z.....RX,RY,RZ by associated value in deg.## Invert X,Y,Z..............IX,IY,IZ (associated value ignored)## Rotate about mean pos## (Z only for now) .....MX,MY,MZ by associated value in deg.## This hypothetical set of shifts has NOT been applied#Shift−Coord: IX 0.0 RX 20.0 TZ 60.0#### Calibration factors for each crystal...one line for each crystal## ADC# : index of pulse−height ADC including any multi−hit spec (M1 here)## low,high: energy cuts for valid hit (in MeV)## ped : the pedestal(zero pulse height) channel (NOT MeV)## MeV/chan: conversion gain in these units## TDC# : index for time ADC including any multi−hit spec (M0 here)## l,high : time cuts for valid hit (in ns)## off : time offset (nominal time zero) channel## ns/ch : conversion gain in these units## Pos : coordinates of the detector. Here x,y,z mid front face of NaI## Optionally after each Element: line a time−walk correction may be specified## Thresh : the threshold in MeV of the leading edge discriminator## RiseTime: the rise time in ns of the signal from the detector element#### ADC# low high ped MeV/ch TDC# l,high off ns/ch PosElement: 3015M1 0.5 400. 0.00 0.0625 2032M0 0 200 −2200 0.117 3.62 45.21 2.63## Thresh RiseTimeTimewalk: 1.5 90.Element: 3014M1 0.5 400. 0.00 0.0677 2033M0 0 200 −2200 0.117 7.32 44.51 5.32Timewalk: 1.5 90.## ....## There would now follow a further 718 "Element:" and optionally## "Timewalk:" lines so that calibration factors may be given for all## 720 elements of the detector#### Specification of adjacent detectors for Clustering## Only if the detector is a TA2ClusterDetector## Max # clusters to analyse Cluster threshold Energy (MeV)Max−Cluster: 8 25#### Turn on search for any connected hits outside of near−neighbour list## hits−size−fact min−dist max−distIterate−Neighbours: 3 4.5 5.5#### Search for clusters which could be split offs from the main cluster## Thresh−SplitOff(MeV) Split−Off Opening angle (deg.)#Split−Off: 30.0 5.0#### Nearest neighbours for each element (see above) of array## Parameter N specifies the number of neighbours (adjacent elements)## This is followed by a list of the indices of the adjacent elements.## 1st in the list is the index of the "central element"## N List indices of neighboursNext−Neighbour: 12 0 1 144 576 145 577 432 579 2 3 146 288Next−Neighbour: 13 1 0 2 3 5 6 4 146 145 144 576 577 579## ....## There would now follow a further 718 "Next−Neighbour:"## lines so that adjacent elements may be given for all## 720 elements of the detector#### Do some further array initialisation BEFORE display specificationInitialise:#### Periodic procedure to execute every N * NAp * NA events## NAp is the (CB) apparatus period, NA is the analysis period## If NA = 100000, NAp = 2; NaIPeriodMacro() will execute every 400000 events## If no analysis/apparatus periodic, no detector periodic will execute## N Macro File Macro ProcedurePeriod: 2 NaIPeriodMacro.C NaIPeriodMacro()#### Cut specification. Syntax the same for all analysis classes## see apparatus file CB.dat## 1/2D Name Link Scope Low/High Limits#Data−Cut: 1D CBCross OR compound−data#Data−Cut: 1D ClTheta NAND element 50 65#Data−Cut: 1D ClPhi NAND element 50 65#Data−Cut: 1D CBBox OR compound−data#Data−Cut: 1D ClTheta NOR element 120 140#Data−Cut: 1D ClPhi NOR element −150 −120#Data−Cut: 1D ClNhits AND data 6 6#### Histograming of basic detector stuff## NaI Crystal hit multiplicity, 64 channels, range 0−64## Name ch/low/high imin/imaxDec 08, 05 19:33 CB_NaI.dat Page 2/2Display: 1D Nhits 64 0 64Display: 1D NADChits 64 0 64Display: 1D NTDChits 64 0 64#### NaI hit frequency distribution, 720 elements, range 0 −− 720Display: 1D Hits 720 0 720Display: 1D ADCHits 720 0 720Display: 1D TDCHits 720 0 720#### Total Energy deposited in the array, 400 channels, range 0 −− 800 MeVDisplay: 1D Etot 400 0 800#### Energy/time for individual crystals, 200 channels, range 0−400 MeV, 0−200 ns## Make histogram for each element (0−719) of the CB#Display: 1D Energy 200 0 400 0 719#Display: 1D Time 200 0 200 0 719Display: 1D EnergyOR 200 0 400Display: 1D TimeOR 200 0 200#### Cluster Histograms## Cluster energy OR, 200 chan, 0−800 MeVDisplay: 1D ClEnergyOR 200 0 800#### Hit pattern within each cluster, 32 chan, 0−32, 720 histograms#Display: 1D ClHits 32 0 32 0 719#### # hits in each cluster#Display: 1D ClMulti 32 0 32 0 719#### # clusters detected per eventDisplay: 1D ClNhits 16 0 16Display: 1D ClNhitsOR 16 0 16#### Distribution of cluster hits...index of the central element plottedDisplay: 1D ClTotHits 720 0 720#### 2D diagnostic plots. id = −1 means not specified## X−Name X−ch/low/high/id Y−Name Y ch/low/high/id Condition(opt)Display: 2D ClTheta 180 0 180 −1 ClPhi 180 −180 180 −1Display: 2D EnergyOR 100 0 400 −1 Hits 720 0 720 −1Display: 2D TimeOR 180 0 180 −1 Hits 720 0 720 −1Printed by ACQU administrationThursday December 08, 2005 CB_NaI.dat1/1Figure 15: TA2ClusterDetector conguration le. Most of the Element: and Next-Neighbour:lines (720 lines of each command) have been removed for brevity and clarity.in the analysis le (Fig.12) as is the physics conguration le. User-dened physics methods willvary considerably, but all will need target nuclei specication as in the physics conguration leshown in Fig.16. This le also has some wild-card setup, which user physics classes can interpretas they see t, and of course specication of cuts and histograms. The syntax for the latterfollows the standard rules as TA2Physics is a type of TA2Apparatus. Indeed TA2Physics calls thestandard TA2Apparatus::SetCong() method and so has the commands given in Table 5.A summary of TA2Physics-specic commands is given in Table 10. The coding of TA2Physics ispreliminary and suggestions of general-purpose utilities to augment the bare bones are welcome.5.3 Data Filtering and DisplayAny data analysis system requires that chosen variables may be tested, histogramed and displayed.This may be hard-wired into code, but in AcquRoot a more exible approach is taken.26

Keyword Parameters FunctionSize: int, int, int, int # elements, yes/no energy, time, positionElement: see Fig.15 Calibration parameters for element of a detecorAllElement: DIAGNOSTIC ONLY Give all elements same calibrationInitialise:Further initialisationDisplay: see Fig.15 Histogram cong.Data-Cut: see Fig.15 Data-ltering cuts cong.TimeWalk: double, double Walk correction to TDC. Discr thresh, rise timeAllTimeWalk: DIAGNOSTIC ONLY All elements same walk correctionRawHits:Save ADC/TDC hit patterns (no cuts)BitPattern: char Hit pattern-unit decode, name of cong lePeriod: int, char, char Period(events), Macro le, Macro commandEndFile: char, char End of File procedure (Macro le, Macro command)Finish: char, char End of Run procedure (Macro le, Macro command)Energy-Scale: double Global scaling for fEnergy array (scale factor)Energy-Random: int Smearing of energies (random generator seed)Shift-Coord: see Fig.15 Global transforms of detector coordinatesTable 6: TA2Detector conguration commandsKeyword Parameter FunctionMax-Cluster: int, double Max # clusters, cluster energy thresholdNext-Neighbour: see Fig.15 Specify adjacent detectors for cluster ndingAll-Neighbour: DIAGNOSTIC ONLY All elements same neighboursSplit-O: double, double Split threshold (MeV), max main-split angle (deg)Iterate-Neighbours: int, double, double Array size factor, new neighbour min, max distanceTable 7: TA2ClusterDetector conguration commandsKeyword Parameter FunctionBar: int # long scintillator barsBar-Size: int, int, double, double, double bar parametersTable 8: TA2LongScint conguration commands. The detector is assumed to consist of a long barof plastic scintillator read at both ends by PMT.Keyword Parameter FunctionNumber-Layers: int # active layers in detectorNumber-Chambers: int # active chambers in detectorLayers-In-Chamber: see MWPC.16.10.04 Indices of layes in chamberChamber-Parameter: see MWPC.16.10.04 Size and position of chamberPlane-Wire:Not yet implementedCyl-Wire: see MWPC.16.10.04 Cylindrical array of wiresPlane-Drift:Not yet implementedPlane-Strip:Not yet implementedCyl-Strip: see MWPC.16.10.04 Helical wound cathode stripsTable 9: TA2WireChamber conguration commands. Cong. le MWPC.16.10.04, part of theUserData bundle, shows how a TA2CylMWPC is set up.5.3.1 LoadVariable a exible means to access addressesTA2DataManager::LoadVariable() is used to associate a name (character string) with the addressof a particular variable and the associative list is then stored in the private data banks of any27

Tuesday December 14, 2004 /home/jrma/acqu/data/CBphysics.dat##−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−##−−Author JRM Annand 29th Apr 2003##−−Update JRM Annand... 1st Dec 2004##−−Description## *** Acqu++ Root ***## Online/Offline Analysis of Sub−Atomic Physics Experimental Data#### Parameter File: CBphysics.dat## Last stage in event−by−event analysis####−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−#### Target parameters (LH2)## x,y,z Mass(MeV)Target: 0. 0. 0. 938.271## Would be CD#Target: 0. 0. 0. 1875.613 11188.174#### Wild−card stuff for fitting combinations of particles to a reaction## Not fully implementedPermutations: 1#### Particle data group database, max # particlesParticleID: 16#### Do some post−initialisation.## Should usually be done before display is setupInitialise:## Cut specification...see apparatus setup, same syntax applies# 1/2D Name Link Scope Low/High Limits#Data−Cut: 1D MtotC OR compound−hist#Data−Cut: 1D Mtot OR element 100 200#Data−Cut: 1D Mtot OR element 400 600## Histograms of physics−level variables# histogram name should be preceeded by 1D or 2D (for dimension of spectrum)# Name chan/low/high min/max condition(opt)#Display: 1D Mtot 800 0 800 0 0 MtotC#Display: 1D Minv 500 0 500 0 7#Display: 1D Mmiss 500 0 2000#Display: 1D DpnMmiss 1000 0 1000#Display: 1D DpnMmiss 1000 0000 2000#Display: 1D DpnAngle 180 0 180#Display: 1D DpnAngleCM 180 0 180#Display: 1D PhiDiff 180 0 360#Display: 1D ThetaDiff 180 0 180#Display: 1D ThetaDiffCM 180 0 180#Display: 1D ThetaP 180 0 180#Display: 1D ThetaN 180 0 180#Display: 1D ThetaPCM 180 0 180#Display: 1D ThetaNCM 180 0 180Display: 1D CBTAPS2PhotonMinv 200 0 800#Display: 1D Pmiss 500 0 500 0 7#Display: 1D Ptheta 500 0 500 0 7#Display: 1D Pphi 500 0 500 0 7## X−Name X−ch/low/high/id Y−Name Y−ch/low/high/id Cond(opt)#Display: 2D Mtot 100 0 800 −1 NphotMinv 100 0 800 1#Display: 2D Mtot 100 0 800 −l NphotMinv 100 0 800 2#Display: 2D Mtot 100 0 800 −1 Nhit 16 0 16 0#Display: 2D CB_NaI_TimeOR 100 0 200 −1 TAPS_BaF2_TimeOR 100 0 200 −1## Any wild−card stuff here#Misc: Not yet implemented#Display: 2D CB_NaI_EnergyOR 100 0 200 −1 TAPS_BaF2_EnergyOR 100 0 200 −1Display: 2D CB_NphotMinv 100 0 800 1 Mmiss 100 0 2000 0Display: 2D CB_NphotMinv 100 0 800 5 Mmiss 100 0 2000 0#Display: 2D DpnAngle 180 0 180 0 TAGG_Ladder_Hits 176 0 352 0#Display: 2D DpnAngle 180 −180 180 0 PhiDiff 180 0 360 0#Display: 2D DpnAngleCM 180 −180 180 0 PhiDiff 180 0 360 0#Display: 2D ThetaDiff 180 −180 180 0 PhiDiff 180 0 360 0#Display: 2D ThetaDiffCM 180 −180 180 0 PhiDiff 180 0 360 0#Display: 2D ThetaDiffCM 180 −180 180 0 DpnMmiss 100 000 1000 0#Display: 2D ThetaDiff 180 −180 180 0 DpnMmiss 100 000 1000 0#Display: 2D ThetaP 180 0 180 0 ThetaPCM 180 0 180 0#Display: 2D ThetaDiff 180 −180 180 0 ThetaDiffCM 180 −180 180 0Figure 16: Sample conguration le for a user, inherited physics class of TA2PhysicsDec 14, 04 16:18 CBphysics.datPage 1/2Dec 14, 04 16:18 CBphysics.datPage 2/2Printed by ACQU data analysisKeyword Parameters FunctionTarget: see Fig.16 Target x,y,z, atomic massesPermutations: int,int,int,int,int,int,int,int Wild-card integer inputTable 10: TA2Physics conguration commandsanalysis class which inherits from TA2DataManager. The variable's address may then be retrievedfor setting up histrograms or data cuts by supplying the name. The basic method isTA2DataManager::LoadVariable(const char* name, T* var, Int_t type);The templated variable T* var is pointer to (address o) an Int_t or Double_t and Int_t typetells TA2DataManager what type of variable to expect. It may have 4 dierent values...1. EISingleX for a single value Int_t (4-byte integer)2. EIMultiX for a multiple value Int_t3. EDSingleX for a single value Double_t (8-byte oat)28

4. EDMultiX for a multiple value Double_tMultiple value variables are held in a buer and it is the start address of the buer which issupplied. Usually the number of values held in the buer will change from one event to the nextand so the end of buer is marked with the special value EBuerEnd. AcquRoot histogram andcut managers know to look for this delimiter if the supplied type is EIMultiX or EDMultiX. Atypical instance of a EDMultiX is the Time-OR of the focal plane detectors of the A2 Tagger.Usually a set of LoadVariable statements are bundled together in a single method as in the followingextracted from TA2ClusterDetector.ccvoid TA2ClusterDetector::LoadVariable( ){ // Input name - variable pointer associations for any subsequent// cut or histogram setup// LoadVariable( "name", pointer-to-variable, type-spec );// NB scaler variable pointers need the preceeding &// array variable pointers do not.// type-spec ED prefix for a Double_t variable// EI prefix for an Int_t variable// type-spec SingleX for a single-valued variable// MultiX for a multi-valued variableTA2DataManager::LoadVariable("ClNhits", &fNCluster, EISingleX);TA2DataManager::LoadVariable("ClTotHits", fClustHit, EIMultiX);TA2DataManager::LoadVariable("ClTheta", fTheta, EDMultiX);TA2DataManager::LoadVariable("ClPhi", fPhi, EDMultiX);TA2DataManager::LoadVariable("ClNhitsOR", fNClustHitOR, EIMultiX);TA2DataManager::LoadVariable("ClEnergyOR",fClEnergyOR, EDMultiX);TA2Detector::LoadVariable();}5.3.2 Data Cut SpecicationAt the present stage of development data ltering cuts may be applied to computed (i.e. producedin a decode or reconstruct procedure) variables only. The set up of cuts is performed bythe TA2DataManager::ParseCut( char* ) method (Sec.4.0.2) which is thus automatically part ofanalysis, apparatus, detector and physics classes.Instances of cut specication may be found in the descriptions of conguration les in Sec.5.2, butthe syntax is covered here in detail.1. Cut-spec. lines are agged by the keyword Data-Cut:. Multiple Data-Cut: lines may bespecied in which case a chain of logical evaluations is carried out.2. After the keyword the line should contain at least 4 (often more) string parameters... dimension,cutX, link, scope.3. dimension = 1D cut in 1 dimension; = 2D cut in 2 dimension.4. cutX is the name of the variable to be tested (the X variable if working in 2D)...see Sec.5.3.15. link species how multiple conditions are joined logically. With two Boolean variables toevaluate a chain of cut-results: A (always initialised to 1 = TRUE before starting evaluation)which is the standing value of the chain, and B which is the value of the particular elementof the chain. The 4 allowed values of link and their counterpart Boolean operations are asfollows...(a) AND A = A.B(b) OR A = A + B29

(c) NAND A = A.B(d) NOR A = A + B(e) If an AND operation gives 0 (=FALSE) then evaluation of the logical chain is stoppedand 0 is returned as the combined Boolean value. If an OR operation gives 1 then chainevaluation stops and 1 is returned.6. scope may have 7 dierent values(a) histogram the cut applies to a histogram only and has to be specied in the histogramsetup command line (Sec.5.4)(b) data the cut applies to the entire data stream. If the test of a particular cut returnsFALSE, then no further analysis of the event takes place. Note that cut evaluation takesplace just before histogramming and after the variables of a particular analysis classhave been determined, e.g. in TA2Detector::Decode() or TA2Apparatus::Reconstruct().Thus a cut at apparatus level will have no eect on detector histograms, while a detectorcut will aect apparatus histograms.(c) misc the cut is added to the wild-card list, to be employed at the discretion of theuser. By default misc cuts do nothing, but for example they may be used to identifyparticles. The TA2Apparatus::SortParticleIDCuts() method sorts the cuts containedin the fMiscCuts list for use in Reconstruct() procedures. However to do this correctlythe cut options must be set correctly.(d) compound-hist start a compound-logic cut for histogram(s) only. Here the name givenin cutX above should be some unique string which has not already been used as avariable name (Sec. 5.3.1).(e) compound-data is similar to compound-hist except it acts on the entire data stream.(f) compound-misc is the misc equivalent of compound-data(g) element species a logical building block of a compound-hist, compound-data orcompound-misc which must have been started before the element is introduced. Elementswill be added to the compound cut until a Data-Cut: line with scope ≠elementis encountered. Within a compound cut evaluation of the internal logical chain proceedsas in item 5 above.1D cut additional parametersWhere scope = histogram, data, misc or element a further 2 Double_t numbers low,high and one optional Int_t ind must be supplied on the Data-Cut: input line. The standard 1Dwindow cut applied to a variable with value v returns TRUE if low ≤ v ≤ high. Parameter indis supplied if the variable is a member of a xed length array (N.B. NOT a multi-value variabledescribed in Sec.5.3.1). For example the line in a TA2Detector congurationData-Cut:1D Energy AND data 20 200 238 (opt)Would specify that the Energy of element 238 of the detector array must be within the range20-200 MeV to pass the cut test.Optionally up to 8 numbers may be appended at the end of the line. These are wild-card parameterswhich are stored in the cut's private (read-only) data. They have no bearing on the operationof the cut returning TRUE/FALSE but in the case of misc cuts can be used to determine howthe cut is used2D cut additional parametersData-Cut: 2D Echarged AND misc DeltaE p0 PIDdE.07.04.05.root 0 0 0 2212Where scope = histogram, data, misc or element (misc in the case above) a further 3 parametersare mandatory, followed by options30

1. DeltaE is the name of the Y-variable.2. p0 is the name of the 2D polygon which species the cut limits.3. PIDdE.07.04.05.root is the name of the root le which contains the polygon in the formof the TCutG ROOT class. Many polygons may be held in a single ROOT le.4. 0 0 are optional indices for non scaler variables (ix followed by iy). In the case above bothEcharged (X) and DeltaE (Y) variables are arrays so the 0 0 species the 0th element ofeach.5. 0 2212 are 2 (of up-to-eight) optional parameters which are appended to the line and whichare stored in the cut's private, read-only data bank. Normally these would only be used formisc cuts. In the present case the 0 refers to sector 0 of the detector system (ix, iy = 0 inthe previous item also refer to the same sector but this is not always the case). 2212 is thePDG code for a proton. Thus if this misc cut is used for particle ID, a cut return of TRUEsays that the particle is a proton.5.4 Display SpecicationInstances of display specication may be found in the conguration le descriptions of Sec.5.2. Acommon setup procedure is provided in TA2HistManager::ParseDisplay() which recognises setuplines beginning with the keyword Display:.5.4.1 Histograms set up by TA2HistManager::ParseDisplayThis method is used by Detector, Apparatus and Physics classes. A setup command for a 1D plotwould typically look likeDisplay: 1D Energy 1000 10. 800. 0 299 Cut1D WeightWhere the parameters are explained as follows:1D histograms1. 1D species a one-dimensional histogram. 2D is also possible.2. Energy is the name (see Sec.5.3.1) of the variable to histogram3. 1000 10. 800. specify a 1000-channel histogram covering the range of values 10.0 to 800.04. 0 299 apply in the case where Energy is a xed-length array of values, e.g. the Energies ofeach element of a composite detector. These Int_t values specify the range of indices forwhich a histogram should be created, in this case creating 300 histograms to cover Energy[0]- Emergy[299]. If Energy is a single-valued these indices may be omitted or -1 -1 entered.5. Cut1D is an optional parameter which species the name of a cut which should be applied tothis histogram only. Omit if no cut is desired. The cut of scope histogram or compound-hist(Sec. 5.3.2) should have been initialised previously.6. Weight is an optional parameter, the name of a variable (see Sec.5.3.1) which species anon-unity increment value for the histogram. If its not given, normal unity increments areimplemented. The weight variable can be any Double_t class variable (as with Energyabove), and it is the users responsibility to set this variable at each event.31

The names of the histograms resulting from this line would be:ClassName_Energy0.Cut1D.Weight ...ClassName_Energy299.Cut1D.WeightClassName is the name of the analysis class which owns the histogram. For a scaler variable(item 4) the indices are omitted. Cut and Weight names are appended only if the cut or weight isspecied and exists.2D histogramsTwo dimensional histograms would typically be set up using a command line likeDisplay: 2D Etot 100 10. 800. -1 Time 100 0. 200. 55 Cut2D Weightwith the parameters explained as follows1. 2D to distinguish from a 1D plot.2. Etot the name of the X variable.3. 100 10. 800. The X-axis has 100 bins and values ranging from 10. to 80.4. -1 This value says that Etot is a scaler variable. In the case that Etot is an array, an integer≥ 0 species the array index. Multiple indices are not allowed5. Time is the name of the Y-variable6. 100 0. 200. The Y-axis has 100 bins and values ranging from 0.0 to 200.7. 55 Time is an array so plot Time[55] here.8. Cut2D is an optional name of a cut to be applied to this histogram only.9. Weight is an optional name of a parameter to specify increment weightThe name of the histogram resulting from this line would be:CLassName_Time55_v_Etot.Cut2D.WeightClassName is the name of the analysis class which owns the histogram. For scaler variables(items 4,7) the indices are omitted. Cut and Weight names are appended only if the cut or weightis specied and exists.3D histogramsThree dimensional histograms would typically be set up using a command line likeDisplay: 3D Etot 100 10. 800. -1 Time 100 0. 200. 55 Efrac 20 0.0 1.0 22 Cut3DWeightwith the parameters explained as follows1. 3D to distinguish from a 1D plot.2. Etot the name of the X variable.3. 100 10. 800. The X-axis has 100 bins and values ranging from 10. to 80.4. -1 This value says that Etot is a scaler variable. In the case that Etot is an array, an integer≥ 0 species the array index. Multiple indices are not allowed for 3D32

5. Time is the name of the Y-variable6. 100 0. 200. The Y-axis has 100 bins and values ranging from 0.0 to 200.7. 55 Time is an array so plot Time[55] here.8. Efrac is the name of the Z-variable9. 20 0.0 1.0 The Z-axis has 20 bins and values ranging from 0.0 to 1.010. 22 Efrac is an array so plot Efrac[22].11. Cut3D is an optional name of a cut to be applied to this histogram only.12. Weight is an optional name of a parameter to specify increment weightThe name of the histogram resulting from this line would be:CLassName_Efrac22_v_Time55_v_Etot.Cut3D.WeightClassName is the name of the analysis class which owns the histogram. For a scaler variables(items 4,7,10) the indices are omitted. Cut and Weight names are appended only if the cut orweight is specied and exists.5.4.2 Special histogram setup in TA2AnalysisTA2Analysis has a private ParseDisplay() method as some of the display related to ADC or scaleroutput is slightly non-standard although the keyword is still Display:. After the keyword comes aplot delimiter and associated parameters which are explained (and exemplied) as follows.• All Display all single value ADCsDisplay: All• ADC Display a range (ADC100 - ADC131) of single value ADCsDisplay: ADC 100 131• 2D Plot 1 ADC value against anotherDisplay: 2D 20401 200 0 4000 2001M0 200 -600 -400Plot ADC20401 X-axis 200 bins range 0 4000 against multihit ADC2001 (0th hit) Y-axis200 bins range -600 to -400• Scaler Display last-read-buer values of scalers 0 - 11Display: Scaler 0 11• SumScaler Display accumulated values of scalers 0 - 11Display: SumScaler 0 11• Rate Display Rate 0 in multiscaler mode (equivalent to time dependence). Allocate 2500bins to plot. Assumes a TA2RateMonitor has been set up previously.Display: Rate 0 2500• Pattern Display hit pattern 0. Allocate 16 bins covering the range 0 - 16. Assumes aTA2BitPattern has been set up previously.Display: Pattern 0 16 0 16• ADChits Display number of hits in each single value ADC. Here hit from the 1st 4096 ADCsare plottedDisplay ADChits 4096 0 409633

• FlashShape Display the waveform given by sampling ADC 55.obtained automatically from the sampling ADC software handlerDisplay FlashShape 55Binning and range are• MultiHit Display a range of multihit ADCsDisplay: MultiHit 3000 3720 1000 0000 4000 1Display Make 1D histograms for multihit ADCs 3000 - 3720. Each plot has 1000 bins coveringthe range 0 - 4000. The nal 1 species that the 1st (starts with 0th) hit of the multi-blockis plotted.6 Specimen Analysis ClassesSome very simple examples of analysis classes are provided with the AR core distribution. Howeverany serious data analysis requires some esh on top of the bare bones. The $acqu_sys/Usersubdirectory contains 2 tar archives1. UserAcqu.tar.gz: a set of analysis classes which will perform analysis to physics level ofMainz Tagger/CB/TAPS data2. UserData.tar.gz: a set of conguration les to setup various types of analysis.These are intended as a guide to enable the user to build his/her personal analysis.6.1 Detector ClassesThe system provides an abstract base detector class TA2Detector, from which all detectors arebuilt.6.2 Apparatus ClassesThe system provides an abstract base apparatus class TA2Apparatus, from which all apparati arebuilt.6.2.1 TA2CrystalBall6.2.2 TA2KensTagger6.3 Physics ClassesThe system provides an abstract base physics class TA2Physics, from which all physics analysesare built, and which is itself derived from TA2Apparatus. Thus TA2Physics may be regarded asa super apparatus which assembles all apparatus information to give an overall view of a reactionprocess.6.3.1 TA2PhotoPhysics34

7 Obsolete Companion SoftwareThe original ACQU was coded before multi-thread functionality became standard on UNIX-likeoperating systems and so multi-tasking components of ACQU were separate programs which communicatedvia System-V semaphores and shared memories. The use of these separate tasks wasphased out in the evolution of AcquRoot v3 and is entirely absent from v4. Indeed all remnantsof the original ACQU have been removed from v4 and coding is now entirely in C++ within theframework of ROOT.35

8 AcquMC a Monte Carlo reaction-kinematics generatorAs part of the ongoing eort to convert the computer models of A2 detector systems fromHBOOK/GEANT-3 to ROOT/GEANT-4 standard, a C++, ROOT-compatible kinematics generatorAcquMC has been written. This program normally runs stand alone, but the classes whichit uses are ROOT/CINT compatible and may be used in any ROOT session if the appropriatelibraries are loaded.AcquMC inputs information from a conguration le, by default MCsetup.dat, which resides inthe same directory as the AcquRoot conguration les. Other conguration les may be speciedon the command line along with switches which control the loading of base classes, e.g.:AcquMC GP_PPi0_SAID.datWhich species the use of conguration le GP_PPi0_SAID.dat. The format of the congurationles is very similar to AcquRoot les, as input is handled by the SetCong() method of TA2System(Sec.3.2), from which the main generator classes are decended.A schematic diagram of the software structure of AcquMC is given in Fig.17. The main task ofevent generation is performed by TMCGenerator (or its decendent TMCFoamGenerator), whileparticle properties are handled by TMCParticle and its decendents. Random sampling is performedwith the ROOT class TRandom (and its variants) with non-uniform sampling distributionshandled via ROOT histogram class TH1D. This is the double-precision version of 1D histogramclass TH1 as all calculations are performed in double precision. Multi-dimensional random samplingis performed via the ROOT class TFoam, implemented in TMCFoamGenerator which inheritsfrom TMCGenerator.AcquMC C++ Event GenerationJ.R.M. Annand5th July 2007Kinematics Generator GP4(TMCParticle* Reaction)ReturnReturnNoTMCParticleXYesdecayproducts?YesGenerateDecaysDecay particles’decay listsLastin list?NoOptionTMCFoamParticleTMCParticle::GenDecay()2−bodyPhase Spaceor "Physics">2particles?N4−momentaCMBoostCM −> LABN−bodyPhase Spaceor "Physics"Configure AcquMCTA2System::FileConfigStart Event LoopGEvent()OptionalCall N−Dim TFoamSave sampled vectorBeam 4−MomentumBeam + Target= Reaction ParticleOptionalTMCFoamIntInitialise TFoamDensity() fn.Next EventParameter(text) FileConfigurationParameter(text) FilePhysics ModelTabulate on N−D GridN−Dim Interpolateto calculate TFoamvolume elementsRecursiveGP4Errors?ROOT TreeAcquMC FormatStore EventEither/ORROOT ntuplemkin−equivalentFigure 17: Flow chart of AcquMC36

8.1 TMCGenerator event generationAcquMC has a skeletal main program and the main work horse is the class TMCGenerator which isa direct decendent of TA2System. TMCGenerator contains the main event loop which calculatesbeam energy, direction and point of interaction in an extended target. The beam and target4-momenta are added and the resultant entity is treated as a particle using the TMCParticleclass (Sec.8.2). Thus the reaction is handled as a decay of the reaction particle and details ofthe reaction dynamics are stored within the associated class. The reaction nal-state productsare also handled by TMCParticle and decay of any unstable product is handled in a similar way.Multi-step decays are handled recursively and in principle there is no limit to the number of steps.The recursive chain terminates when a daughter particle reports back that it is stable.Usually the AcquMC main program runs the TMCFoamGenerator class (8.1.1) which can functionexactly as its parent TMCGenerator, but also has the capability of multidimensional randomsampling via the ROOT class TFoam.The inheritance scheme of the generator classes is given in Fig.18.8.1.1 TMCFoamGeneratorThis variant of TMCGenerator uses the ROOT class TFoam to sample an N-dimensional vectorfrom a foam of volume elements (bubbles) in the hyperspace. This foam is generated froman abritrary relationship, which for example could be a 5-fold dierential cross section whichdetermines completely a 3-body nal state in terms of a set of energies and angles. Where thisrelationship can not be specied by simple relations, it is tabulated on a grid of values andinterpolated at intermediate values. This is handled by the class TMCFoamInt (8.1.2) and itsdecendents.8.1.2 TMCFoamIntThis class is a decendent of the ROOT class TFoamIntegrand, which constructs the foam ofN-dimensional volume elements. TMCFoamInt has procedures to perform the following:• ReadData( Char_t* le )Read in data points from a le, tabulated on an N-dimensional grid, which species thekinematic relationships which TFoam has to reproduce.• LinInterpN( Double_t* x, Bool_t comp )Perform multi-linear interpolation on the grid of values (as read in above) to provide thevalue at the point x = (x 0 , x 1, . . . x n ). The switch-variable comp, if set to TRUE (defaultis FALSE), causes the interpolator to apply error compensation, equivalent to quadraticinterpolation.• Density( Int_t nDim, Double_t* x )This is the procedure used by TFoamIntegrand to construct the foam of volume elements.TFoamIntegrand supplies a vector x, sampled randomly in N-D space, where each elementis sampled on the interval {0,1}. The element values are scaled to the interval of physicalinterest, e.g. for polar angle one might scale x i = {0, 1} → {0, π}. The return is the valueof the N-dimensional relation, as for example calculated using the interpolator describedabove. Weighting factors calculated using a particular x i value may optionally be appliedto the return value. If x i is a polar angle then the weighting factor sin(x i ) can be applied toaccount for phase space.So far two specialist decendents of TMCFoamInt have been coded:37

fNamefTitlefgIsATNamed@~TNamedTNamedTNamedTNamedTNamedoperator=ClearCloneCompareCopyFillBufferGetNameGetTitleHashIsSortableSetNameSetNameTitleSetTitlelsPrintSizeofClassClass_NameIsAShowMembersTA2SystemfCmdList[5]fInputFilefLogFilefLogStreamfIsInitfIsErrorfIsConfigPassfgIsA@~TA2SystemBaseInitPrintErrorPrintMessageMap2KeyMap2KeyMap2StringGetMapKeyGetMapStringAddCmdListFileConfigCommandConfigParseMiscSetConfigPostInitBuildNameCheckNameGetInputFileGetLogFileGetLogStreamSetLogStreamSetLogFileIsInitIsErrorClassClass_NameIsAShowMembersfRandfSeedfNrunfPDGfTargetMassTMCGeneratorfTargetQFRecoilMassfTargetRadiusfTargetLengthfTargetCentrefVertexfBeamCentrefP4TargetfP4TargetQFfP4BeamfP4RfBeamfTargetQF@~TMCGeneratorTMCGeneratorFlushParseBeamParseTargetParseTargetQFParseParticleInitParticleDistPostInitInitTreeSaveEventCloseEventSetConfigGenerateGEventGP4InitNtupleSaveNtupleCloseNtupleGetRandGetSeedGetPDGGetTargetMassGetTargetRadiusGetTargetLengthGetTargetCentreGetVertexfReactionfNThrowfNParticlefResonancefReactionListfParticlefTreeFileNamefTreeFilefTreefNtuplefBranchfEventfIsTreeOutfIsNtupleOutfIsErrorfgIsAGetBeamCentreGetP4TargetGetP4BeamGetP4RGetBeamGetReactionGetNthrowGetNParticleGetResonanceGetReactionListGetParticleGetTreeFileNameGetTreeFileGetTreeGetBranchGetEventIsTreeOutIsNtupleOutIsErrorGetParticleClassClass_NameIsAShowMembersStreamerTMCFoamGeneratorfFoamListfFoamXfIntfXfXscalefScalefFoamInitOptfNDimfNDataBasefFnOptfFoamModelfgIsA@~TMCFoamGeneratorTMCFoamGeneratorFlushPostInitSetConfigReadDatabaseGEventGetFoamListGetFoamInitOptGetXGetXscaleGetScaleGetNDimGetNDataBaseGetFnOptClassClass_NameIsAShowMembersStreamerFigure 18: Inheritance scheme of TMCGenerator, and TMCFoamGenerator.TMCPhotoPSInt This has a specialised ReadData procedure to input SAID generated observablesfor pseudo-scalar meson photoproduction. For each incident energy and meson productionangle, 16 observables (cross section + the various single and double polarisation quantities) areinput. The data base is split into 16, one section for each of the observables. The Density procedureoperates in 3-D with vector elements x 0 ≡ E γ , x 1 ≡ θ π , x 2 ≡ φ π , but the observables areevaluated by 2-D interpolation in energy and polar angle. Following that the asymuthal dependenceis calculated explicitly using the observables and the polarisation states of beam, target andrecoil particle. Input of polarisation vectors is described in 8.3.38

TMCdS5MDMInt This is a specialist integrand class, designed to handle radiative, pseudoscalarmeson photoproduction. This has been measured to access magnetic dipole moments ofnucleon resonances. It has a modied ReadData procedure which allows input of more than oneobservable on the 5-D grid of dσ/dE γ ′dΩ γ ′dΩ π , where γ ′ is the radiated photon and π is thepseudo-scaler meson. If the overall dimension of TFoam is set to 6, then a block of dierentialcross section values, followed by a block of Σ asymmetry values, is expected as input. If thedimension is set to 5 then only dierential cross section is read in.5-D interpolation is performed to evaluate cross section (and Σ) for a given randomly sampledvector x, x 0 ≡ E γ x 1 ≡ E γ ′ x 2 ≡ θ γ ′ x 3 ≡ φ π x 4 ≡ θ π . If Σ is generated then x 5 ≡ φ π andthe Density procedure weights the cross section according to [1 − cos(2φ π )]. The weighted crosssection is then returned as the value of Density. Energy and phase-space weighting factors mayalso be applied as described in 8.1.2.Note that although TMCdS5MDMInt was designed for radiative meson photoproduction it can inprinciple be applied to any reaction producing a 3-body nal state.8.2 TMCParticleTObjectTNamedTMCParticleTMCFoamParticleTObjectTNamedTMCParticleTMCResonancefUniqueID kIsReferenced kZombiefNamefBitskHasUUID kBitMaskfTitlefgDtorOnly kCannotPick kSingleKey fgIsAfgObjectStat kNoContextMenu kOverwritekCanDelete kInvalidObject kWriteDelete@~TNamedkMustCleanup kIsOnHeap fgIsATNamedkObjInCanvas kNotDeletedTNamed@~TObject GetObjectInfo operator new@[@] TNamedMakeZombie GetTitle operator delete TNamedDoError HandleTimer operator delete@[@] operator=TObject Hashoperator delete ClearTObject InheritsFrom operator delete@[@] Cloneoperator= InheritsFrom SetBitCompareAppendPad Inspect SetBitCopyBrowse IsFolder ResetBitFillBufferClassName IsEqual TestBitGetNameClearIsSortable TestBitsGetTitleCloneIsOnHeap InvertBitHashCompare IsZombie InfoIsSortableCopyNotifyWarningSetNameDeletelsErrorSetNameTitleDistancetoPrimitive PaintSysErrorSetTitleDrawPopFatallsDrawClass PrintAbstractMethod PrintDrawClone ReadMayNotUse SizeofDumpRecursiveRemoveGetDtorOnlyClassExecute SaveAs SetDtorOnly Class_NameExecute SavePrimitive GetObjectStat IsAExecuteEvent SetDrawOption SetObjectStat ShowMembersFindObject SetUniqueID ClassFindObject UseCurrentStyle Class_NameGetDrawOption WriteIsAGetUniqueID WriteShowMembersGetName operator newfRandom fPhiDist fTcmfPDGfPhiLow fWtfParent fPhiHigh fWtMaxfP4fTDistfPDGindexfP4dfTLowfNdfP4Beam fTHighfDecayModefDecayList fPDistfIsDecayfThetaDist fPLowfIsTrackfThetaLow fPHighfIsResonancefThetaHigh fMassfgIsAfCosThDist fMassQFfCosThLow fMdfCosThHigh fMd2@~TMCParticle GetP4d GetTcmPdkGetP4d GetWtTMCParticle GetP4Beam GetWtMaxTMCParticle GetDecayList GetPDGindexTMCParticle GetThetaDist GetNdFlushGetThetaLow GetDecayModeInitGetThetaHigh SetDecayModePostInit GetCosThDist IsDecayGenDecay GetCosThLow IsTrackGen2Decay GetCosThHigh IsResonanceGenNDecay GetPhiDist SetDecayBoostLab GetPhiLow SetTrackFindDist GetPhiHigh GetDecaySetDist GetTDist ThetaSetDist GetTLow CosThetaSetDist GetTHigh PhiAddDecay GetPDist EnergyReplaceDecay GetPLow MomentumGenMCDist GetPHigh P4GetRandom GetMass ClassGetPDG GetMassQF Class_NamefXscalefScalefIThetafICosThetafIPhifIEnergyfIMomentumfgIsA@~TMCFoamParticleTMCFoamParticleTMCFoamParticleFlushInitGetXscaleGetScaleGetIThetaGetICosThetaGetIPhiGetIEnergyGetIMomentumSetFoamLinkThetaCosThetaPhiEnergyMomentumDistClassClass_NameIsAShowMembersStreamerStreamerNVirtualfUniqueID kIsReferenced kZombiefNamefBitskHasUUID kBitMaskfTitlefgDtorOnly kCannotPick kSingleKey fgIsAfgObjectStat kNoContextMenu kOverwritekCanDelete kInvalidObject kWriteDelete@~TNamedkMustCleanup kIsOnHeap fgIsATNamedkObjInCanvas kNotDeletedTNamed@~TObject GetObjectInfo operator new@[@] TNamedMakeZombie GetTitle operator delete TNamedDoError HandleTimer operator delete@[@] operator=TObject Hashoperator delete ClearTObject InheritsFrom operator delete@[@] Cloneoperator= InheritsFrom SetBitCompareAppendPad Inspect SetBitCopyBrowse IsFolder ResetBitFillBufferClassName IsEqual TestBitGetNameClearIsSortable TestBitsGetTitleCloneIsOnHeap InvertBitHashCompare IsZombie InfoIsSortableCopyNotifyWarningSetNameDeletelsErrorSetNameTitleDistancetoPrimitive PaintSysErrorSetTitleDrawPopFatallsDrawClass PrintAbstractMethod PrintDrawClone ReadMayNotUse SizeofDumpRecursiveRemoveGetDtorOnlyClassExecute SaveAs SetDtorOnly Class_NameExecute SavePrimitive GetObjectStat IsAExecuteEvent SetDrawOption SetObjectStat ShowMembersFindObject SetUniqueID ClassFindObject UseCurrentStyle Class_NameGetDrawOption WriteIsAGetUniqueID WriteShowMembersGetName operator newfRandom fPhiDist fTcmfPDGfPhiLow fWtfParent fPhiHigh fWtMaxfP4fTDistfPDGindexfP4dfTLowfNdfP4Beam fTHighfDecayModefDecayList fPDistfIsDecayfThetaDist fPLowfIsTrackfThetaLow fPHighfIsResonancefThetaHigh fMassfgIsAfCosThDist fMassQFfCosThLow fMdfCosThHigh fMd2@~TMCParticle GetP4d GetTcmPdkGetP4d GetWtTMCParticle GetP4Beam GetWtMaxTMCParticle GetDecayList GetPDGindexTMCParticle GetThetaDist GetNdFlushGetThetaLow GetDecayModeInitGetThetaHigh SetDecayModePostInit GetCosThDist IsDecayGenDecay GetCosThLow IsTrackGen2Decay GetCosThHigh IsResonanceGenNDecay GetPhiDist SetDecayBoostLab GetPhiLow SetTrackFindDist GetPhiHigh GetDecaySetDist GetTDist ThetaSetDist GetTLow CosThetaSetDist GetTHigh PhiAddDecay GetPDist EnergyReplaceDecay GetPLow MomentumGenMCDist GetPHigh P4GetRandom GetMass ClassGetPDG GetMassQF Class_NamefGammafMassDistfMLowfMHighfgIsA@~TMCResonanceTMCResonanceTMCResonanceTMCResonanceFindDistMassGetGammaGetMassDistGetMLowGetMHighClassClass_NameIsAShowMembersStreamerStreamerNVirtualGetIconNameoperator new@[@]SetParentSetQFIsAGetIconNameoperator new@[@]SetParentSetQFIsAGetOptionoperator newGetParentMassShowMembersGetOptionoperator newGetParentMassShowMembersGetP4SetMassStreamerGetP4SetMassStreamerSetP4GetMdStreamerNVirtualSetP4GetMdStreamerNVirtualSetP4GetMd2SetP4GetMd2Figure 19: Inheritance schemes of TMCParticle classesParticles are characterised within AcquMC by the TMCParticle class, which is a direct decendent ofTNamed and thus does not have any of the I/O functionality of TMCGenerator classes. Inheritanceschemes for TMCParticle and its decendents TMCResonance and TMCFoamParticle are given inFig.19. The main features of this class are described as follows:• Basic static properties of the particle may be obtained via the Particle Data Group (PDG)data base which may be passed, along with a particle descriptor index, when the class isconstructed. For example mass would be obtained through the instructionfMass = fPDG->GetParticle(ipdg)->Mass() ;If the employed root distribution has no PDG database then particle mass (the minimumnecessary information) can be specied at construct time.39

• Details of the kinematics of the decay are implemented in the procedures Gen2Decay() andGenNDecay(). In the case that the particle is a reaction particle the decay is just thereaction process. The daughter particles participating in the decay process return kinematicquantities via the following functions:1. Double_t Theta() the polar angle2. Double_t CosTheta() cosine of the polar angle3. Double_t Phi() the asimuthal angle4. Double_t Energy() the kinetic energy of the particle5. Double_t Momentum() the momentum of the particle.6. TLorentzVector P4() samples the complete 4-momentum using the CosTheta(), Phi()and Momentum() functions above.By default these functions use uniform random distributions over a previously-input range ofvalues. However a non-uniform distribution may be specied for any particle via the P-Distr:directive which is described in Sec.8.3. Alternatively the kinematic quantity may be linked(8.3) to an element of the output vector from TFoam.• The 2-body decay algorithm can use either phase-space (the default) or non-uniform distribution(s)for any of the 5 possible kinematic variables. For the decay to 3 or more particlesphase space is assumed by default and the entry of non-uniform distributions will produceerroneous results. If non-phase-space ≥ 3-body kinematics are desired, this must be encodedin the GenNDecay() procedure of a specialist particle class decended from TMCParticle ofTMCFoamParticle (see 8.2.3).• A linked list of pointers to TMCParticle classes characterises the decay products of a particularparticle. For each event the generator loop GP4() decends through the lists of decayproducts, calling their GenDecay() procedures until a stable particle is reached.• A pointer is provided to the class which characterises its parent particle if any....the reactionparticle has no parent.8.2.1 TMCResonanceThis class is a derivative of TMCParticle (Fig.19) and diers in that the mass is not assumedconstant, but is sampled from a Breit-Wigner type of distribution when using the Mass() function.The standard TMCParticle::Mass() returns the constant fMass. The alternative functionGetMass() always returns the constant fMass regardless of particle class.8.2.2 TMCFoamParticleThis derivative of TMCParticle provides additional options for choosing the kinematics of particles,in particular to allow correlated sampling of variables for a single or multiple particles. By defaultTMCFoamParticle will behave exactly as TMCParticle. However it may be linked to the n-dimensional vector which is sampled in the GEvent() procedure of TMCFoamGenerator. Then-dimensional hyperspace, which is sampled, is initialised prior to the start of the event loop inTMCFoamGenerator (Fig.17).Any (or none) of the 5 possible kinematic variables may be tied to any element of the TFoamvector, e.g. if element x j (j = 0 − n − 1) corresponds to the energy of the particular particlethen the a call to the Energy() function returns x j . Linkage is specied by the Foam: directivedescribed in Sec.8.3.If no linkage to the TFoam vector is specied then Energy() and other kinematic quantities arereturned exactly as for TMCParticle.40

8.2.3 TMCdS5MDMParticleThis version of TMCFoamParticle has a specialist GenNDecay procedure. This takes the 5(or 6) dimensional vector output from TFoam sampled from a 5-fold dierential cross sectiondσ/dE γ ′dΩ γ ′dΩ π . This species completely a three-particle nal state (8.1.2). Its assumed thatthe vector elements are0 : E beam , 1 : E 1 , 2 : θ 1 , 3 : φ 1 , 4 : θ 2 , 5 : φ 2E 2 and E 3 , θ 3 , φ 3 are calculated from the above input. Supply of φ 2 is optional. If not supplied itis sampled randomly in one dimension.8.3 Conguration le formatAs with AR classes, initialisation is performed by the FileCong() procedure which reads lines froma text le and then calls SetCong() to parse the commands enumerated below. The commandline in type-face as it would appear in the conguration le is followed by a short explanation.Note that there must be a space after the colon in the key-word (the 1st word on the line).1. Foam-Model: MDMEvent-generation model for TFoam multi-dimensional sampling.(Char_t* = MDM) is for radiative pseudo-scalar meson productionA SAID option is also available. It takes ouput from the eponymous partial-wave analysisdatabase [4], e.g. for pion or eta photoproduction. If this directive is not supplied the defaultis to make no particular model assumption.2. Number-Throws: 200000 888888 10Number of events to run (Int_t =200000); random generator seed (Int_t=888888); numberof runs (Int_t=10). Each run would be of 200000 events.3. Beam: 22 0.0 0.0 -500.0Beam particle PDG index (Int_t=22); origin x,y,z coordinates (Double_t = 0.0 0.0 -500.0)in cm. Alternatively the particle mass in GeV may be given instead of the PDG index, butthe number must be an explicit oat with a decimal point, e.g. photon = 0.0, proton =0.938.4. Target: 0.938 2.5 5.0 0.0 0.0 0.0Target particle/nucleus mass in GeV (Double_t=0.938); Target radius and length in cm(Double_t=2.5 5.0); Target-centre x,y,z coordinates in cm (Double_t=0.0 0.0 0.0).5. QF-Target: 0.938 10.255The beam interacts with a quasi-free particle of mass in GeV (Double_t=0.938), whichis embedded within the target, which is 12 C. The mass of the recoiling system in GeV(Double_t=10.255) is here that of 11 B in its ground state. The QF-Target particle (name= TargetQF) accepts kinematic distribution input (see item 7).6. Particle: Pi0 111 1 0 Delta_1 NULLParticle name (Char_t*=Pi0); PDG index (Int_t=111); Decay to other particles switch 0/1= OFF/ON (Int_t=1 ON); Track particle switch (Int_t=0 OFF); Parent particle, if any,name (Char_t*=Delta_1); Particle Option (Char_t*=NULL). The name of each particlemust be unique in a particular run. As with Beam: above, a oating-point mass may begiven instead of the PDG index. Unstable particles should have decay set to 1 if one expectsto detect the daughter products. Setting track to 1 causes the particles 4-momentum tobe saved in the output ROOT le. If the particle is a decay product then the name of theparent particle must be given, otherwise enter NULL. Options reect non-standard modes ofsampling kinematic variables which are described below, but usually NULL (standard mode)is used.41

7. Resonance: S11 1.535 0.15Species a short lived particle whose mass has a signicant width. Name (Char_t*=S11);Resonant peak mass in GeV (Double_t=1.535); Resonance width in GeV (Double_t=0.15).This command merely enters resonance characteristics which may then be used in theParticle: specication (item 5), by substituting the resonance name for PDG index. If aparticle is also a resonance then its mass can be sampled from a Breit-Wigner distribution.A few resonances, e.g. P33(1232), are included in the PDG tables in which case PDG index(Int_t=2214 for P33(1232)) may be substituted for the resonant mass and resonant widthomitted.8. P-Distr: Beam Cos(Theta) TMath::Gaus(x,1.0,0.000001) 0.9999975 1.0 200Specify a non-uniform distribution for a kinematic variable of a particle of name (Char_t*= Beam);the variable name (Char_t*=Cos(Theta)); the distribution (Char_t*=TMath::Gaus(x,1.0,0.000001));the limits over which to sample this distribution (Double_t=0.9999975 1.0); the numberof bins in the TH1D used to perform the sampling (Int_t=200). If the desired distributiondoes not have a simple analytical form then the limits and TH1D parameters are omittedand the distribution string is then the name of a le, by default in the $acqu/data directory,which contains a table of x, f(x) values (1 line for each pair) which dene the distribution.The x values should ascend monotonically. The present line says that the cosine of the polarangle of the beam particle should follow a Gaussian centered on 1.0 with a σ of 0.000001in the range 0.9999975 to 1.0. Multiple P-Distr operations may be applied to a singleparticle, which may be any of the dened particlesThe recognised variable names are:Theta-Dist [polar angle], Cos(Theta) [cosine polar angle], Phi-Dist [asimuthal angle],T-Dist [kinetic energy], P-Dist [momentum].9. Tree-Output: /scratch/acqu/GP_PPi0C_200kT 2ROOT TTree output to le (Char_t*=/scratch/acqu/GP_PPi0C_200kT). The TTree has(Int_t=2) TBranch. Output format is given in Sec.8.4.10. Ntuple-Output: /scratch/acqu/GP_PPi0C_200kNROOT TNtuple (basic form of TTree) to le (Char_t*=/scratch/acqu/GP_PPi0C_200kN).Output format is given in Sec.8.4.The following commands are supported by the TMCFoamGenerator class in addition to thoseenumerated above.1. Foam-Distribution: 5 0 data/MDM_dSig5b.datEnter a distribution for input into TFoam. The N-dimensional distribution has dimension(Int_t=5), interpolation option (Int_t=0) and is read from le (Char_t*=data/MDM_dSig5b.dat).2. Foam-Option: 1000 200 8 1 2 25 1 1.1Parameters which control the initialisation of TFoam which are nCells (Int_t=1000); nSampl(Int_t=200); nBin (Int_t=8); OptRej (Int_t=1); OptDrive (Int_t=2); EvPerBin (Int_t=25);Chat (Int_t=1); MaxWtRej (Double_t=1.1). The parameters above represent the defaultwhich is applied if no explicit Foam-Option: command is supplied. See the ROOT referencesection on TFoam for the meaning of these variables. In general the default will work quitewell.3. Foam: Gamma_0 2 -1 3 1 -1Couple the kinematic variables of TMCParticle name (Char_t*=Gamma_0) to the TFoamgenerated N-dimensional vector. Five integers are input (Int_t=2 -1 3 1 -1), whichcorrespond to the 5 cases of kinematic variable recognised by TMCParticle1st:Theta-Dist, 2nd:Cos(Theta), 3rd:Phi-Dist, 4th:T-Dist, 5th:P-Dist. Each integerrepresents the corresponding element of the TFoam generated vector. A value of -1 species42

that there is no corresponding TFoam vector element and standard TMCParticle samplingshould be used (either uniform or weighted). The present line species the polar angleis TFoam element 2, the cosine polar angle element is unspecied, the asimuthal angle isTFoam element 3, the kinetic energy is TFoam element 1 and the momentum is unspecied.In the present case N=5 the range of valid values for the 5 integers is 0 to 4.4. Foam-Limits: 0 0.2 0.4Reset the limits to a kinematic variable which have been input via Foam-Distribution: wherethe variable identier is the TFoam vector index (Int_t=0); the lower and upper limits(Double_t=0.2 0.4). For example the data base input by Foam-Distribution: might spanan incident energy (index 0) of 0.15 to 1.5 GeV, while the current calculation is desired overthe range 0.2 to 0.4 GeV.5. Beam-Pol: 0.0 0.0 0.0Set the components of the beam polaristion vector (Double_t = 0.0 0.0 0.0).6. Target-Pol: 0.0 0.0 0.0Set the components of the target polaristion vector (Double_t = 0.0 0.0 0.0).7. Recoil-Pol: 0.0 0.0 0.0Set the components of the recoil-particle polaristion vector (Double_t = 0.0 0.0 0.0).8. Foam-Weight: Bremsstrahlung 0Foam-Weight: SinTheta 2The rst (Char_t*=Bremsstrahlung) is for use in extended energy range bremsstrahlungbeam experiments where the photon-energy dependence is given in the N-dimensional TFoamdistribution with the vector index of the incident photon (Int_t=0). The incident photonenergy dependence is weighted with a simple 1/E γ dependence.The second (Char_t*=SinTheta) applies a phase-space factor to a polar angle distributionspecied as index (Int_t=2)8.4 Output formatsData may be output in either of two data formats.1. A ROOT version of mkin format2. AcquMC format8.4.1 Mkin output formatThis output format mimics that of the F77 program mkin, an event generator which has been usedpreviously with the GEANT-3 based model of the Crystal Ball cbsim. In AcquMC the output iswritten as a ROOT (as opposed to hbook) ntuple. Each event contains the following.• V x , V y , V z the Cartesian coordinates of the reaction vertex.• p x , p y , p z , P tot , E to specify the 4-momentum of the beam particle. P tot =and p x = P x /P tot . E is the total energy.√P 2 x + P 2 y + P 2 z• This is followed by the 4-momentum specication of each particle which has been markedas trackable. This specication is identical to that of the beam particle.This format is compatible with the Geant-4 model of the Crystal Ball and TAPS43

8.4.2 AcquMC output formatThis output format used a ROOT TTree to save the reaction vertex and 4-momenta of the particlesproduced in a particular generated event. Each event contains:• V x , V y , V z the Cartesian coordinates of the reaction vertex.• P x , P y , P z , E to specify the 4-momentum of the beam particle. Symbols are as in Sec.8.4.1.• This is followed by the 4-momentum specication of each particle which has been markedas trackable. This specication is identical to that of the beam particle.The ROOT macro MCReplay(FileName.root) provides a diagnostic of the output, in le namedFileName.root generated by AcquMC. The le name requires full directory specication. MCReplaycreates histograms of• The vertex x, y, z coordinates and x − y, r − z 2D distributions.• The momentum balance, which is expected to be zero if all nal-state momentum is tracked.• The lab. x, y, z momentum components and energy of the beam particle.• The lab. polar and asimuthal angles of the beam particle.• The lab. x, y, z momentum components and energy of each tracked particle.• The lab. polar and asimuthal angles of each tracked particle44

9 Analysis of Monte Carlo generated dataThe analysis of Crystal-Ball Monte Carlo output may be done within the same framework (ROOT)as the real data analysis outlined above. The present GEANT-3 model of CB/TAPS will be withus for some time and it outputs data as an HBOOK ntuple, which may trivially be convertedto ROOT format using the h2root utility. This root le can then be run through AcquRoot inoine mode in a similar way to the analysis of reduced data. Creating and analysing MonteCarlo data is summarised as follows:1. Create an input data le for mkin or MCGenerator to characterise the reaction of interest.These are standard text les which may be created using an editor such as emacs.2. Run the reaction event generator mkin or MCGenerator. This will create an ntuple (or tree)le of N reaction events where N is specied from le or command line. Each event consists ofbeam 4-momentum, reaction vertex in the target and the 4-momenta of the reaction productparticles.3. Create an ***.cards data le for cbsim to further characterise the reaction of interest.This is also a standard text le.4. Run the reaction kinematics ntuple/tree le through the CB/TAPS simulation cbsim. Trackparticles through the detector systems and calculate and store energy deposits (and times)in detector elements. The output data is stored in an ntuple le.5. Convert cbsim output from hbook to root format using h2root.6. Run the cbsim output through AcquRoot.9.0.3 Features of the GEANT-3 based simulationIn cbsim, the GEANT-3 based model of CB and TAPS, the position of the TAPS elements is readin from the le taps.dat held in the cbsim directory. In early versions of this le the x coordinateis inverted with the result that 4-momenta derived from the MC generated events have their p xcomponent inverted. If the 1st line of taps.dat reads:0 -1 -1then all should be correct. Conversely if the 1st line of taps.dat reads:0 1 -1then the x coordinates are inverted. Monte Carlo les generated using the inverted TAPS xcoordinate may still be analysed correctly in AcquRoot using the global coordinate shift feature ofthe TA2Detector class. To apply this feature the conguration le for the BaF 2 array, commonlycalled BaF2_MC.dat for analysis of Monte Carlo events, should contain the line:Shift-Coord: IX 0.0which tells the class to invert the x coordinate when it stores the position of a detector element.Note that the IX operation is one of a possible 12 types of shift operation provided by TA2Detector.As there is considerable scope for confusion in this business it is prudent to have a kinematicsmonitor in the AcquRoot analysis. The following example used a phase-space generation of thereaction γ + p → p + η, η → γγ for E γ = 900 MeV. The monitor is just a comparison of missing4-momentum P miss , derived from the 2 photons detected in the Crystal Ball, with recoil proton4-momentum P p obtained from a hit in the TAPS array which has the appropriate time of ight.The results are plotted in Fig.20.Plotted are the polar and asimuthal angle dierences δθ = θ miss − θ p and δφ = φ miss − φ p . Aswould be expected inverting x has a negligible eect on the apparent tight correlation in θ, whilethe x invertion completely destroys the correlation in φ.45

TA2PhotoPhysicsMDM_Angle1PTA2PhotoPhysicsMDM_Angle1P45004000Correct TAPS x coordinateEntries 16948Mean 0.03425RMS 2.3434000Inverted TAPS x coordinateEntries 16806Mean -0.05458RMS 2.4753500350030003000250025002000200015001500100010005005000-20 -15 -10 -5 0 5 10 15 20δθ (deg)0-20 -15 -10 -5 0 5 10 15 20δθ (deg)TA2PhotoPhysicsMDM_Angle2PTA2PhotoPhysicsMDM_Angle2P900Correct TAPS x coordinateEntries 16948Mean 0.1477RMS 17.5670Inverted TAPS x coordinateEntries 16806Mean 2.03RMS 102.680060700600505004004030030200100200-150 -100 -50 0 50 100 150δφ (deg)-150 -100 -50 0 50 100 150δφ (deg)Figure 20: TAPS Monte Carlo kinematic monitor. δθ = θ miss − θ p , δφ = φ miss − φ p46

10 DAQ SoftwareAcquDAQ is structured on the model of an array of front-end processors, each reading data fromsome component of the detector system, and feeding that data to one or more central eventbuilders. Event-building software is embodied in the MergeBuers procedure of TA2DataServer(Sec.3.4), while the TDAQexperiment class of AcquDAQ constitutes a container for all the DAQcomponents which typically run on a front-end machine.Figure 21: TDAQexperiment schematic diagram10.1 TDAQexperimentFig.21 gives a schematic diagram of the tasks which are run within the TDAQexperiment envelope.The operational sequence of TDAQexperiment is:1. Congure the DAQ session using the TA2System::FileCong() procedure, as employed byAcquRoot classes. This inputs data from text conguration les in a local disk store (usually$acqu/data/). The data are described in Sec.10.1.1.2. Initialise the hardware according to the instructions input from data les3. Start the threads which run, in parallel, component processes of the DAQ system.4. Start the experimental-control process embodied in TDAQSupervise. External control commandsmay be received from either a local keyboard or via the network.10.1.1 Conguration les for TDAQexperimentAs with AR classes, initialisation is performed by the SetCong() procedure with the commandsitemised below. The command line in type-face as it would appear in the conguration le isfollowed by a short explanation. Note that there must be a space after the colon in the key-word.47

1. Description: Test of AcquDAQ 4v0A brief description of the experiment, which is written into the data header.2. File-Name: /scratch/acqu/AcquDAQ 32 50000 32768 Disk 32768 3010Data output specication: le name including directory spec. (/scratch/acqu/AcquDAQ);number of output buers to create (32); number of data records to write to single le (50000);size in bytes of a data record (32768); ID of IP-socket used to send data (3010).3. Event-Counters: 100 0Frequency of running Scaler readout (100) and slow control (0) procedures. Entering 0 turnsthis process OFF.4. Start-Run: 0Default run number to start DAQ session. This value is overriden by any value read fromthe le $acqu/Run.log (if it exists), which in turn is overriden by the optional run-numberparameter supplied with the Run command received via TDAQsupervisor.5. #Control: Local 0 0Control: ENet AcquCtrl 3020Specify the communication method for TDAQsuperviser. Here the commented-out localkeyboard input (Local) has been disabled and ethernet (ENet) input enabled from (AcquCtrl)6. Module: CBD_8210 cbd_0 cbd8210_0.dat 800000 0 0DAQ hardware component electronics, in this case a CAMAC parallel branch driver (CBD_8210),which is given device title (cbd_0), reads conguration data from le (cbd8210_0.dat),gives the hexadecimal base address of the module (800000) in this an A24 VMEbus address,(0) base index not used here, (0) number of channels not used here. The 1st 3 parametersare always supplied for a module. Starting with 800000 the parameters supplied for theCBD_8210 are VMEbus device dependent.7. IRQ-Ctrl: cbd_0Hardware module, having device title (cbd_0) which is used to ag an interrupt, used totrigger event-by-event readout.10.2 TDAQmoduleDAQ hardware is characterised by inherited classes of TDAQmodule (Fig.22), TDAQmodule beingpurely virtual, ie. an abstract framework on which real hardware characteristics are hung.10.2.1 Conguration les for TDAQmoduleThe following summarises the available options specied in module les (Sec.10.1.1 Item #6).Items 1-3 are completely general, 4 is VMEbus1. BaseSetup: 10000000 256 ADCThe module sits at hex base address 1000000 (0 species dened elsewhere), has 256 registers(ie allocate space for 256 register structures) and is of type ADC. Allowed type values are ADC,Scaler, SlowCtrl (slow control), P-Ctrl (primary DAQ controller), S-Ctrl (secondaryDAQ controller) and IRQ (trigger processor, e.g. control start of readout)2. Init:Run PostInit() procedure48

Figure 22: TDAQmodule schematic3. BaseIndex: 4096 128 16This command is used with ADCs and Scalers to specify the start index (4096), the numberof readout channels in the module (128) and the number of signicant bits (16) of eachdatum read from the module4. Register: 200 32 09 0843 0Setup a VMEbus module register at address addr (oset from the base address), width is thenumber of signicant bits (e.g. 8,16,32) in the register, am is the VMEbus address modiercode (usually 09), datum is an optional parameter which is the number written to theregister at initialisation, and the optional nreg species how many registers at consecutiveaddresses to setup identically. If datum is omitted no initialisation of the register is carriedout and if nreg is omitted a single register only is specied.5. CAMAC-CNAF: 0 0 0 1 IRQThis is the CAMAC variant of the Register: directive. It sets up function code andsubaddress(es) for the CAMAC module whose branch, crate and station (BCN) are givenin the main cong. le (Sec.10.1.1). The 5 parameters are datum (0), function code (0),1st sub-address (0), last sub-address (1), register type (IRQ). Allowed types are IRQ (eventby-eventreadout), Init (initialise) and Test. For write function codes (e.g. 16,17...) thevalue of datum is written. In the present case a read function (0,1,2....) is given and datumis ignored. Thus the module sub-addresses A0, A1 are read each event. Set 1st and lastsub-addresses equal for a single sub-address.Other directives have a more specialist module-specic functions which are detailed in Sec.10.2.2.49

10.2.2 Supported hardwareThis will be updated periodically as new hardware is included.CAEN V2718 PCI-to-VMEbus bridge This hardware connects the PCI bus of a standardPC to the VMEbus via optical bres. CAEN software from a provided library is used to initialisethe link and read/write VMEbus addresses. The module has full Address Modifer (AM) VMEbusaddressing capabilities. It is introduced to the DAQ via the TDAQexperiment line:Module: CAEN_V2718 v2718_0 v2718.dat 0 0The supplied parametersMainz KPh X86 based single board VMEbus computer This VMEbus slot-1 controleris produced by the Mainz Institut fuer KernPhysik and is based on an X86 CPU. It is introducedto the DAQ via the TDAQexperiment line:CES CBD 8210 CAMAC parallel branch driver This VMEbus module converts an A24VMEbus address to a CAMAC BCNAF command, which is geographical specifying Branch # (B),Crate # (C), Crate Station # (N), Module Subaddress # (A), Function Code (F). The followingsetup line species base address 800000 (hex) which corresponds to Branch 0.Module: CBD_8210 cbd_0 cbd8210_0.dat 800000 0 0Generic CAMAC module This may be used to initialise any CAMAC module which requiresonly generic CAMAC-CNAF: directives (Sec.10.2.1). The directive from the main cong. le isas follows:Module: CAMAC a2_0 a2ctrl_0.dat 1 30The module in this case is a type-A2 crate controller on crate 1 station 30. Its virtual stationis 30 which physically doesn't exist. File a2ctrl_0.dat contains the setup AF directives. It isassumed to lie on the branch specied by the last introduced VME-CAMAC interface, e.g. theCBD 8210 above.LeCroy 4508 memory lookup unit This 2×8-bit memory lookup unit is often used for triggercontrol in AcquDAQ. It has some specialist directives to setup the internal RAMS. It is speciedin the main cong. le by the following:Module: LRS_4508 mlu_0 lrs4508_0.dat 1 4which says it sits in crate 1 at station 4. The setup le lrs4508_0.dat contains, in addition togeneric directives, a specialist of the form:RAM: 0 ffff c0RAM: 0 03 f7The 3 parameters specify in order RAM # (0), which for the 4508 may be 0/1 for the top/bottomsections of the module; the hex input bit pattern (03) for which the valid range is 0-ff; thecorresponding hex output bit pattern (f7) for which the valid range is 0-ff. If the input bitpattern is set to ffff then all RAM locations are written in the above case with c0, ie its adefault output value, given in cases where an input pattern is not specied.50

LeCroy 2373 memory lookup unit This 16-bit memory lookup unit has some specialistdirectives to setup internal RAMS. It is specied in the main cong. le by the following:Module: LRS_2373 mlu_1 lrs2373_0.dat 1 5which says it sits in crate 1 at station 5. The setup le lrs2373_0.dat contains, in addition togeneric directives, specialists of the form:RAM: ffff c0RAM: 3bf 4fa0CCR: 60The RAM: directives are very similar to the LeCroy 4508 above except that there is no RAM 0/1.The CCR: parameter is set to hex 60 for pulsed mode and hex 40 for latched mode.Generic VMEbus module This may be used to initialise any VMEbus module which doesnot require specialist handling:Module: VMEbus disc_0 v895_0.dat e0000000 5 16In this case it is a CAEN V895, 16-channel leading-edge discriminator which sits at base addresshex e0000000.CAEN V792 32-channel QDC Module: CAEN_V792 v792_0 v792_0.dat ee000000 100 32Virtual Modules for DAQ-software testing Module: Virtual vADC_0 vADC_0.dat ADC 20032 12Module: Virtual vSCA_0 vSCA_0.dat Scaler 200 32 24Module: Virtual vIRQ_0 vIRQ_0.dat S-Ctrl 0 0 1651

AApplicationsA.1 Cluster Determination Algorithms for Segmented CalorimetersIn a segmented calorimeter, such as the Crystal Ball or TAPS arrays, an energetic particle willgenerally produce a shower of secondary particles which will re a group of (usually) adjacentelements of the array. This group is known as a cluster. Dierent types of particle and dierentenergies will produce cluster signatures with dierent spatial extents and dierent distributions ofenergy deposition. The event reconstruction software has to:1. identify which hits in the calorimeter array group together to form a cluster.2. calculate the total energy deposition within the cluster, which can be related to the energyof the incident particle.3. calculate a weighted mean position of the cluster hits, which can be related to the entranceposition of the incident particle.4. calculate diagnostic parameters which may be used to aid identication of the incidentparticle.In cases where a reaction produces multiple nal-state products, e.g. in kaon or eta productionwhere the meson decays could give 6 or 8 photons, identication of the individual clusters can bea non-trivial process. Within AR the detector class TA2ClusterDetector has a variety of methodsto identify and calculate the properties of clusters.A.1.1The basic cluster algorithm.The basic method to identify hit clusters is essentially that used by UCLA for analysis of CrystalBall experiments at BNL. The array of calorimeter hits for a particular event is searched for localmaxima in energy deposition. Starting with the array element which has the absolute maximumenergy, its nearest neighbours are scanned and if they contain energy they are added to the clusterand also crossed o the hits array to avoid double counting. Each element of the calorimeter hasan associated list of neighbour elements which are input from the detector conguration le. Thusno particular neighbour geometry has to be assumed a priori and the extent of the neighbourarray, e.g. one ring, two rings around the element, can be tuned by hand.After this cluster is determined, the remaining hits are re-scanned to nd the next maximum inenergy deposition and a similar neighbour scan performed to determine the next cluster. Thisprocedure continues until there are no hits left or the number of clusters exceeds a maximumnumber specied in the detector conguration le. To be saved as a cluster, the group of adjacenthits has to have a total energy E tot = ∑ i E i which exceeds a threshold E th specied in the detectorconguration le. The weighted mean position is given by r m = ∑ i r i√Ei / ∑ √ E i and variousquantities such as a the fraction of total energy contained in the central element of the cluster arealso saved.A.1.2The extended cluster algorithm.The basic algorithm does not always catch all of the hits in a cluster and so an optional further scanmay be performed to sweep up any hits which are connected to the cluster but are not includedin the neighbour array of the central element. This method was adapted from that used in TAPSanalysis with some modication to improve computational eciency.52

After a basic cluster is determined (Sec.A.1.1) the remaining hits array is scanned. The distanced ij = |r i − r j | is calculated for each cluster member i and each remaining hit j. If d min ≤ d ij ≤d max , where d min and d max are specied in the detector conguration le, then element j isadded to the cluster and crossed o the remaining-hits list. The procedure is then iterated overthe new cluster elements to search for any further connections and the iteration stops when nonew elements are added. Once this is performed E tot ,r m etc. are calculated as before.A.1.3The split-o search algorithm.Occasionally a shower produces a set of calorimeter hits which cannot be combined into onecluster using the methods outlined above. Characteristically a photon shower might produce amain cluster and a separate smaller cluster containing a relatively small amount of energy. Thelatter is known as a split-o and it is often, but by no means always, situated within a relativelysmall opening angle with respect to the main cluster. In this scheme clusters are determinedby either of the methods above (Sec.A.1.1,A.1.2), but in order to recognise split-os the clusterenergy threshold E sptot is set lower than normal, e.g. E sptot ∼ 5 MeV as opposed to say E tot ∼30 MeV for π 0 decay photons. The resulting clusters are then sorted and those whose energyexceeds E tot are labelled main clusters and anything remaining is labelled a potential split-o.The opening angles ϑ kl between all combinations of main k, and split-o clusters l are thencalculated and sorted. Those combinations having ϑ kl ≥ ϑ max , where ϑ max is read from thedetector conguration le, are rejected and any remaining k, l combinations are assigned on thepriority of minimum ϑ kl . Once this is done cluster l is merged with cluster k so that Etot k =Etot k + Etot l and rm (r k = √ )m√ k Ektot + rml Eltot /(√Ektot + √ )Etotl .A.1.4Conguration le options for TA2ClusterDetectorIn addition to the standard Element: and Next-Neighbour: lines which have to be entered foreach element of the calorimeter array, the TA2ClusterDetector class (and its parent TA2Detector)have a number of options which are important to determination of cluster parameters.1. The global energy scale factor. After TA2Detector converts from ADC channel to energyusing calibration parameters particular to the given detector element, the resultant energyis multiplied by the global scaling factor. By default the scale factor is set to 1.0. To use avalue of say 1.05 the detector conguration le would contain the line:Energy-Scale: 1.052. The maximum number of clusters to handle and cluster threshold energy. This requires aninput line of the form:Max-Cluster: 12 30.0which species that the class will handle up to 12 clusters and that each cluster must haveE tot ≥ 30 MeV.3. The option to turn on iterative searching for additional cluster members. This requires aninput line of the form:Iterate-Neighbours: 3 5.9 6.1The factor 3 causes the neighbour-list array for any particular element to be increased in sizeby a factor 3, e.g. if 12 neighbours are given on a Next-Neighbour: line then the size of thearray is increased to 36 to allow for potential extra neighbours. The iterative search checksthe array boundaries. Numbers 5.9, 6.1 are respectively d min , d max the distance limits usedto determine if an element is connected to the cluster.4. The option to turn on split-o searching. This requires an input line of the form:Split-Off: 5.0 20.053

This species that E spth, the initial low cluster threshold to catch split-os, is set to 5.0 MeVand that the maximum opening angle between main and split-o clusters ϑ max is 20.0 degrees.A.1.5Preliminary Cluster studies using Monte Carlo generated events.A study of the eects of dierent methods of cluster analysis was performed using events generatedby mkin and played through the cbsim Monte Carlo model. The simulation assumed amonoenergetic photon of energy 900 MeV incident on a 5 cm hydrogen target, producing anη anda recoil proton according to phase space. Both η → γγ and η → π 0 π 0 π 0 decays were considered,again with energy and angle distributions determined by phase space consideration only.The resulting ntuples containing 10 5 events were converted to root trees and played through theAR analyser. The signal produced by the nal-state photons and proton were analysed using thecluster-determination methods (Sec.A.1.1,A.1.2,A.1.3) of the AR class TA2ClusterDetector, forboth the Crystal Ball and TAPS, and TA2PhotoPhysics analysed the 4-momenta produced by theapparati.Note that the simulations were performed using the correct (not x-inverted) TAPS coordinatesystem (as input to cbsim) and also that this study is not intended as a denitive statement onhow to perform cluster analysis. It is merely a guide as to the type of eects or pitfalls which maybe encountered.Fig.23 shows proton missing mass distributions obtained when only the nal-state photons aredetected and plots labelled 1 and 2 refer respectively to η → γγ and η → π 0 π 0 π 0 nal statedecays. The sub-labels A,B,C,D are explained as follows:A: The standard cluster-determination algorithm of Sec.A.1.1 was used. The cluster energy thresholdwas 30 MeV.B: The extended cluster-determination algorithm of Sec.A.1.2 was used, also with a cluster energythreshold of 30 MeV.C: The standard algorithm was used, followed by a split-o search of Sec.A.1.3. The cluster energythreshold was 30 MeV and the split-o threshold was 5 MeV.D: This was similar to C, except that the extended algorithm was used before the split-o search.Note that the global energy scales of both TAPS and the Crystal Ball were set to 1.0 so that theabsolute degree of energy loss in cluster determination could be assessed.Case/Parameter 1A 1B 1C 1D 2A 2B 2C 2D〈M miss 〉 − M p (MeV) 31.3 25.3 26.2 23.9 33.7 29.5 29.2 27.9δ σ M miss (MeV) 19.1 20.3 20.3 20.6 18.1 18.7 18.7 19.0〈E miss 〉 (MeV) 75.3 68.4 69.0 66.6 76.9 74.5 71.2 72.3〈θ miss − θ p 〉 (deg) -0.15 0.03 0.02 0.10 -0.21 -0.11 -0.06 -0.06δ RMS (θ miss − θ p ) (deg) 2.3 2.3 2.4 2.4 2.2 2.2 2.2 2.2〈φ miss − φ p 〉 (deg) 0.18 0.15 0.11 0.09 0.04 -0.18 -0.06 -0.27δ RMS (φ miss − φ p ) (deg) 17.3 17.6 17.6 17.9 19.0 18.2 18.3 17.9% events with ηp ID 33.3 40.0 39.4 41.3 9.6 8.3 12.0 9.0% events with 2/6 γ 80.3 82.0 79.8 81.9 43.2 30.4 43.4 30.5% events with >2/6 γ 11.9 9.6 12.2 9.8 5.4 3.2 5.5 3.2Table 11: Summary of cluster-study results. M miss is as in Fig.23, E miss is the missing energy ifboth nal-state η and p are identied, θ and φ angles are as dened in Sec.9.0.3Table 11 summarises some results extracted from this exercise. If can be seen that the variousalgorithms have a minimal eect on angle determination and also on energy resolution.54

TA2PhotoPhysics5000400030001A: η→γγMDM_EmEtaPEntries 41744Mean 969.5RMS 19.68χ 2/ ndf542.8 / 31Constant 5168 ±31.3Mean 969.6 ±0.1Sigma 19.08 ±0.07TA2PhotoPhysics60005000400030001BMDM_EmEtaPEntries 50534Mean 963.4RMS 20.69χ 2/ ndf352.4 / 34Constant 5916 ±32.8Mean 963.6 ±0.1Sigma 20.3 ±0.120002000100010000800 850 900 950 1000 10500800 850 900 950 1000 1050TA2PhotoPhysics60005000400030001CMDM_EmEtaPEntries 49760Mean 964.3RMS 20.73χ 2/ ndf363 / 33Constant 5829 ±32.5Mean 964.5 ±0.1Sigma 20.28 ±0.07TA2PhotoPhysics60005000400030001DMDM_EmEtaPEntries 52294Mean 962.1RMS 21χ / ndf316 / 352Constant 6028 ±32.9Mean 962.2 ±0.1Sigma 20.64 ±0.0720002000100010000800 850 900 950 1000 10500800 850 900 950 1000 1050TA2PhotoPhysics160014001200100002A: η→3πMDM_EmEtaPEntries 12476Mean 972.7RMS 19.74χ / ndf178.2 / 302Constant 1622 ±18.7Mean 972 ±0.2Sigma 18.14 ±0.13TA2PhotoPhysics1400120010008002BMDM_EmEtaPEntries 10655Mean 968.5RMS 20.062χ / ndf119.9 / 30Constant 1347 ±16.6Mean 967.8 ±0.2Sigma 18.72 ±0.148006004006004002002000800 850 900 950 1000 10500800 850 900 950 1000 1050TA2PhotoPhysics2000180016001400120010002CMDM_EmEtaPEntries 15608Mean 968.1RMS 20.19χ / ndf193 / 312Constant 1971±20.3Mean 967.5 ±0.2Sigma 18.72 ±0.12TA2PhotoPhysics1400120010008002DMDM_EmEtaPEntries 11563Mean 966.8RMS 20.27χ / ndf135.4 / 302Constant 1444 ±17.2Mean 966.2 ±0.2Sigma 18.95 ±0.148006004006004002002000800 850 900 950 1000 10500800 850 900 950 1000 1050Figure 23: Proton missing mass from the reaction γ + p → p + η for E γ = 900 MeV. Events weregenerated by the mkin/cbsim Monte Carlo model. The solid curves represent Gaussian ts to themissing-mass distributions.In the case of η → γγ, detection of the relatively high energy photons is improved, both interms of energy determination and in eciency of identifying the correct number of photons, ifthe extended cluster algorithm is applied. Case 1B is in all respects better than 1A. Turning onthe split-o search (Case 1C,1D) does improve the energy balance slightly, but does nothing to55

suppress erroneous detection of more than 2 photons.In the case of η → 3π 0 the situation is less clear cut. Energy determination in Case 2B is betterthan 2A, but conversely the eciency of detecting 6 photons is signicantly poorer in 2B. Againthe split-o search (Case 2C, 2D) gives a small improvement in energy determination but seemsto have little of no eect in suppressing erroneous detection of more than 6 photons.Thus the overall message is choose the best horse for a particular course...there is currently nomagic bullet to tackle this problem globally.References[1] ACQU[2] ROOT Users Guide[3] A2 Geant-4 base simulation of the Crystal Ball & TAPS @ MAMI, D.Glazier.[4] SAID[5] MDM56