rarDatasets Class Reference

Dataset holder class. More...

#include <rarDatasets.hh>

Inheritance diagram for rarDatasets:

List of all members.

Public Member Functions
	rarDatasets ()
	Trivial ctor.
	rarDatasets (const char *configFile, const char *configSec, const char *actionSec)
	Default ctor.
virtual	~rarDatasets ()
virtual RooArgSet *	getPrimaryObs ()
	Return primary observables in dataset files.
virtual RooArgSet *	getAddOnCols ()
	Return addon columns for datasets.
virtual RooArgSet *	getFullFObs ()
	Return full observables.
virtual TString	getDSName (TString name)
	Create dataset name based on input string.
virtual RooDataSet *	getData (const char *name=0)
	Return the named dataset.
virtual TList *	getDatasetList ()
	Get dataset list.
virtual TString	ubStr (TString dsName, const char *ubStrVal=0)
	Return/set the unblind string for the dataset.
virtual Bool_t	isBlind (TString dsName)
	Check if the named dataset is blind or not.
Protected Member Functions
void	init ()
	Initial function called by ctor.
virtual void	tabulateDatasets (const char *dsName=0)
	Tabulate datasets wrt cats.
TString	getWeightVarName (TString datasetName)
	return the name of the weighted variable to be used in this dataset
Protected Attributes
TString	_actionSec
	Action config section name.
rarDatasetDef *	_dsd
	Dataset definition object.
TList	_dataSets
	Defined datasets.
RooArgSet *	_fullFObs
	Full set of fundamental observables.
RooArgSet	_UBs
	Unblind strings for datasets.
Private Member Functions
	rarDatasets (const rarDatasets &)
	ClassDef (rarDatasets, 0)

---

Detailed Description

Dataset holder class.

This class instantiates a rarDatasetDef class for dataset definition, reads in and holds all the datasets from ascii/root files. It also holds datasets derived from those primary datasets.

Config Directives:

See doc for dataset input section.

Definition at line 27 of file rarDatasets.hh.

---

Constructor & Destructor Documentation

rarDatasets::rarDatasets

(

)

Trivial ctor.

Usually the objects should be created using other ctors.

Definition at line 44 of file rarDatasets.cc.

References init().

  : rarConfig(),
    _dsd(0), _fullFObs(0)
{
  init();
}

rarDatasets::rarDatasets	(	const char *	configFile,
		const char *	configSec,
		const char *	actionSec
	)

Default ctor.

Parameters:

	configFile	The config file
	configSec	The config section
	actionSec	The action section

The default ctor initializes common data members, and because there is only one instantiation of this class, _configStr, name and title are set to constant values. It then calls init.

Definition at line 61 of file rarDatasets.cc.

References init().

  : rarConfig(configFile, configSec, "null", "Datasets", "Datasets"),
    _actionSec(actionSec), _dsd(0), _fullFObs(0)
{
  init();
}

rarDatasets::~rarDatasets

(

)

[virtual]

Definition at line 69 of file rarDatasets.cc.

{
  // _dataSets.Delete();
}

rarDatasets::rarDatasets

(

const rarDatasets &

)

[private]

---

Member Function Documentation

rarDatasets::ClassDef	(	rarDatasets	,
		0
	)			[private]

virtual RooArgSet* rarDatasets::getAddOnCols

(

)

[inline, virtual]

Return addon columns for datasets.

Returns:

The addon columns defined

Reimplemented from rarConfig.

Definition at line 40 of file rarDatasets.hh.

References _dsd, and rarDatasetDef::getAddOnCols().

Referenced by rarBasePdf::getAddOnCols().

{return _dsd->getAddOnCols();}

RooDataSet * rarDatasets::getData

(

const char *

name = 0

)

[virtual]

Return the named dataset.

Parameters:

name

The name of the dataset to return

Returns:

The returned dataset

It first checks if dataset named name exsits in _dataSets, if yes, it returns the dataset, if no, it parses name and if the number of tokens is greater than 0, it will check if there is dataset named as the first token, if yes, it will reduce that dataset according to the second token and return the reduced dataset. Otherwise, it returns null.

Reimplemented from rarConfig.

Definition at line 241 of file rarDatasets.cc.

References _dataSets, getDSName(), isBlind(), rarStrParser::nArgs(), tabulateDatasets(), and ubStr().

Referenced by rarMLFitter::doContourPlot(), rarBasePdf::doPdfPlot(), rarMLFitter::doProjPlot(), rarMLFitter::doScanPlot(), rarMLFitter::doSPlot(), rarMLFitter::doToyStudy(), rarMLFitter::generate(), rarMLPdf::init(), rarMLFitter::run(), rarBasePdf::setFitData(), and ubStr().

{
  RooDataSet *theData=(RooDataSet*)_dataSets.FindObject(getDSName(name));
  if (theData) return theData;
  // can not find the data set parser the name
  rarStrParser nameParser=name;
  if (nameParser.nArgs()<=0) return theData;
  theData=(RooDataSet*)_dataSets.FindObject(getDSName(nameParser[0]));
  if (!theData) return theData; //can not find any
  // reduce the dataset
  theData=(RooDataSet*)theData->reduce(nameParser[1]);
  theData->SetName(getDSName(name));
  _dataSets.Add(theData);
  // check if need to set ub bit
  if (!isBlind(getDSName(nameParser[0]))) {
    ubStr(getDSName(name), "Unblinded");
  }
  cout<<" New dataset created"<<endl;
  if (theData) {
    //theData->Print("v");
    theData->Print();
    tabulateDatasets(theData->GetName());
  }
  return theData;
}

virtual TList* rarDatasets::getDatasetList

(

)

[inline, virtual]

Get dataset list.

Returns:

the dataset list

Definition at line 51 of file rarDatasets.hh.

References _dataSets.

Referenced by rarMLFitter::doToyStudy(), and rarMLFitter::run().

{return &_dataSets;}

TString rarDatasets::getDSName

(

TString

name

)

[virtual]

Create dataset name based on input string.

Parameters:

name

The input name string

Returns:

The created dataset name

Definition at line 210 of file rarDatasets.cc.

Referenced by rarMLFitter::doToyStudy(), getData(), and init().

{
  // first remove "
  name.ReplaceAll("\"", "");
  name.ReplaceAll("  ", " ");
  name.ReplaceAll(">=", ".GE.");
  name.ReplaceAll(">", ".GT.");
  name.ReplaceAll("<=", ".LE.");
  name.ReplaceAll("<", ".LT.");
  name.ReplaceAll("==", ".EQ.");
  name.ReplaceAll("!=", ".NEQ.");
  name.ReplaceAll("=", ".EQ.");
  name.ReplaceAll("&&", ".AND.");
  name.ReplaceAll("||", ".OR.");
  for (Int_t i=0; i<name.Length(); i++)
    if (!isalnum(name[i])) name[i]='_';
  
  return name;
}

virtual RooArgSet* rarDatasets::getFullFObs

(

)

[inline, virtual]

Return full observables.

Returns:

The full (fundamental) observables defined

Definition at line 44 of file rarDatasets.hh.

References _fullFObs.

Referenced by rarBasePdf::init().

{return _fullFObs;}

virtual RooArgSet* rarDatasets::getPrimaryObs

(

)

[inline, virtual]

Return primary observables in dataset files.

Returns:

The primary observables defined

Reimplemented from rarConfig.

Definition at line 36 of file rarDatasets.hh.

References _dsd, and rarDatasetDef::getPrimaryObs().

Referenced by rarBasePdf::getPrimaryObs().

{return _dsd->getPrimaryObs();}

TString rarDatasets::getWeightVarName

(

TString

datasetName

)

[protected]

return the name of the weighted variable to be used in this dataset

It sets weight var according to configs

Definition at line 156 of file rarDatasets.cc.

References _fullFObs, and rarConfig::readConfStr().

Referenced by init().

{
  TString swvStr = readConfStr("setWeightVar", "no");
  if (swvStr.BeginsWith("no")) return(""); // no weight variable
  
  // do we have a default weight variable for all datasets?
  rarStrParser swvStrParser = swvStr;
  TString swvName = swvStrParser[0]; // either dataset name or default weight variable
  // check variable exists
  if (_fullFObs->find(swvName)) {
    cout << "Will use " << swvName << " as weight for dataset " << datasetName << endl;
  } else {
    cout << "Can not find " << swvName << " as a variable to use as a weight." << endl;
    swvName="";
  }
  return(swvName);
}

void rarDatasets::init

(

)

[protected, virtual]

Initial function called by ctor.

init is called by the ctor. It first instantiates a rarDatasetDef class for dataset definition. It then reads in the info of datasets, and prints out the datasets configured int the config section. It creates those datasets by calling createDataSet and adds them to the dataset list, _dataSets. It also checks if the datasets need to be tabulated for output.

Reimplemented from rarConfig.

Definition at line 83 of file rarDatasets.cc.

References _actionSec, rarConfig::_configFile, rarConfig::_configSec, rarConfig::_createFundamental, _dataSets, _dsd, _fullFObs, rarConfig::_fullNameSchema, rarConfig::_fullObs, rarConfig::_rarVars, rarConfig::createDataSet(), getDSName(), rarConfig::getFullObs(), getWeightVarName(), rarStrParser::nArgs(), rarConfig::readConfStr(), tabulateDatasets(), and ubStr().

Referenced by rarDatasets().

{
  cout<<"init of rarDatasets:"<<endl;
  // first set #_createFundamental and #_fullNameSchema
  _createFundamental=kTRUE;
  _fullNameSchema="self";

  // first create dsd object
  TString dataDefSec=readConfStr("dsdSec", "Dataset Definition");
  _dsd=new rarDatasetDef(_configFile, dataDefSec);
  _fullObs=_dsd->getFullObs();
  // create full fundamental obs
  _fullFObs=new RooArgSet("theFullFObs");
  RooArgList fullObs(*_fullObs);
  for (Int_t i=0; i<fullObs.getSize(); i++) {
    RooAbsArg *theVar=fullObs.at(i);
    _fullFObs->add(*((RooAbsArg*)_rarVars.FindObject(theVar->GetName())));
  }
  
  // get number of datasets, first check action section
  TString datasetsStr=readConfStr("Datasets", "notSet", _actionSec);
  // if not found, go to dsi section
  if ("notSet"==datasetsStr) datasetsStr=readConfStr("Datasets", "");
  rarStrParser datasetsStrParser=datasetsStr;
  Int_t nDataset=datasetsStrParser.nArgs();
  if (nDataset<=0) {
    cout<<"no datasets defined in RooStringVar \"Datasets\" in section \""
        <<_configSec<<"\""<<endl;
    exit(-1);
  }
  // read in datasets info
  RooArgSet datasetList("Dataset List");
  for (Int_t i=0; i<nDataset; i++) {
    RooStringVar *dataset=new RooStringVar
      (datasetsStrParser[i], datasetsStrParser[i], "notSet", 8192);
    datasetList.addOwned(*dataset);
  }
  datasetList.readFromFile(_configFile, 0, _configSec);
  //datasetList.Print("v");
  // list configed datasets
  cout<<endl
      <<"The datasets defined in config file:"<<endl;
  for (Int_t i=0; i<nDataset; i++) {
    TString datasetStr=(RooStringVar&)datasetList[datasetsStrParser[i]];
    cout<<Form(" dataset%02d ",i)<<datasetsStrParser[i]<<" "<<datasetStr<<endl;
  }
  cout<<endl;
  // now read in the datasets
  for (Int_t i=0; i<nDataset; i++) {
    Bool_t isUB=kFALSE;
    TString datasetStr=(RooStringVar&)datasetList[datasetsStrParser[i]];
    // get name of weight variable
    TString wgtVarName = getWeightVarName(getDSName(datasetsStrParser[i]));

    RooDataSet *data=createDataSet(datasetsStrParser[i]+" "+datasetStr, isUB, wgtVarName);
    data->SetName(getDSName(datasetsStrParser[i]));
    _dataSets.Add(data);
    if (isUB) ubStr(getDSName(datasetsStrParser[i]), "Unblinded");
    //data->Print("v");
    //data->get()->Print("v");
  }

  cout<<endl<<"Datasets read in:"<<endl;
  _dataSets.Print();
  cout<<endl;
  
  // do we need to tabulate it?
  tabulateDatasets();
}

Bool_t rarDatasets::isBlind

(

TString

dsName

)

[virtual]

Check if the named dataset is blind or not.

Parameters:

dsName

The name of the dataset

Returns:

Boolean for blind (true) or unblind (false)

Definition at line 318 of file rarDatasets.cc.

References rarStrParser::Have(), rarStrParser::nArgs(), rarConfig::readConfStr(), rarStrParser::Remove(), and ubStr().

Referenced by rarMLFitter::chkBlind(), and getData().

{
  // magic override
  if (getenv("RARFITUNBLIND")) return kFALSE;

  // rest of the routine
  Bool_t retVal(kTRUE);
  TString ubStrVal=ubStr(dsName);
  if ("notSet"==ubStrVal) {
    cout<<" W A R N I N G !"<<endl
        <<" Somehow the ub status for "<<dsName<<" is "<<ubStrVal<<endl
        <<" So "<<dsName<<" remains blind. Please double check"<<endl;
    return retVal;
  }
  if ("Unblinded"==ubStrVal) return kFALSE;
  rarStrParser ubIDParser=ubStrVal;
  rarStrParser dsiUBStrParser=readConfStr("ub_"+dsName, "");
  while(dsiUBStrParser.nArgs()>0) {
    TString dsiUBStr=dsiUBStrParser[0];
    dsiUBStrParser.Remove();
    if (ubIDParser.Have(dsiUBStr)) return kFALSE;
  }
  
  return retVal;
}

void rarDatasets::tabulateDatasets

(

const char *

dsName = 0

)

[protected, virtual]

Tabulate datasets wrt cats.

Parameters:

dsName

The dataset name if provided

It tabulates datasets wrt categories defined according to configs

Definition at line 178 of file rarDatasets.cc.

References _dataSets, rarConfig::getCats(), and rarConfig::readConfStr().

Referenced by getData(), and init().

{
  TString tdsStr=readConfStr("tabulateDatasets", "no");
  if (tdsStr.BeginsWith("no")) return;
  
  cout << "Tabulate Datsets by Category\n" << endl;
  RooArgList catList(*getCats());
  for (Int_t dIdx=0; dIdx<_dataSets.GetSize(); dIdx++) {
    RooDataSet *data=(RooDataSet*) _dataSets.At(dIdx);
    if (dsName&&TString(dsName)!=data->GetName()) continue;
    for (Int_t i=0; i<catList.getSize(); i++) {
      RooAbsCategory *theCat=(RooAbsCategory *)&(catList[i]);
      Roo1DTable *theTable(0);
      if (theCat&&(theTable=data->table(*theCat))) {
        theTable->Print("v");
        // get frac table
        TIterator* catTypeIter = theCat->typeIterator();
        RooCatType *theType(0);
        while(theType=(RooCatType*)catTypeIter->Next()) {
          TString typeName=theType->GetName();
          cout<<"    "<<typeName<<"\t"<<theTable->getFrac(typeName)<<endl;
        }
        cout<<endl;
        delete catTypeIter;
      }
    }
  }
}

TString rarDatasets::ubStr	(	TString	dsName,
		const char *	ubStrVal = 0
	)			[virtual]

Return/set the unblind string for the dataset.

Parameters:

	dsName	The name of the dataset
	ubStrVal	The ubStr value (to set)

Returns:

The unblind string for the dataset

Definition at line 271 of file rarDatasets.cc.

References _UBs, and getData().

Referenced by rarMLFitter::chkBlind(), rarMLFitter::doToyStudy(), getData(), init(), and isBlind().

{
  TString ubStrName="ub_"+dsName;
  // have we calculated it?
  RooStringVar *theStr=(RooStringVar*)_UBs.find(ubStrName);
  if (theStr&&("notSet"!=TString(theStr->getVal()))) {
    if (ubStrVal) theStr->setVal(ubStrVal);
    return theStr->getVal();
  }
  if (!theStr) {
    theStr=new RooStringVar(ubStrName, ubStrName, "notSet", 8192);
    _UBs.add(*theStr);
  }
  if (ubStrVal) {
    theStr->setVal(ubStrVal);
    return theStr->getVal();
  }
  // compute chksum
  TMD5 chksum;
  stringstream o;
  // read in all the data entries
  RooDataSet *theData=getData(dsName);
  if (!theData) {
    cout<<" W A R N I N G !"<<endl
        <<" Can not find dataset named "<<dsName<<" for ub calculation!"<<endl;
    return theStr->getVal();
  }
  Int_t nEvt=theData->numEntries();
  Int_t nStep=nEvt/10000;
  nStep++;
  Int_t i=0;
  while (i<nEvt) {
    RooArgList evt(*theData->get(i));
    evt.writeToStream((ostream&)o, kTRUE);
    i+=nStep;
  }
  string chkStr=o.str();
  chksum.Update((UChar_t*)chkStr.c_str(), chkStr.length());
  chksum.Final();
  theStr->setVal(chksum.AsString());
  
  return theStr->getVal();
}

---

Member Data Documentation

TString rarDatasets::_actionSec [protected]

Action config section name.

Definition at line 63 of file rarDatasets.hh.

Referenced by init().

TList rarDatasets::_dataSets [protected]

Defined datasets.

Definition at line 65 of file rarDatasets.hh.

Referenced by getData(), getDatasetList(), init(), and tabulateDatasets().

rarDatasetDef* rarDatasets::_dsd [protected]

Dataset definition object.

Definition at line 64 of file rarDatasets.hh.

Referenced by getAddOnCols(), getPrimaryObs(), and init().

RooArgSet* rarDatasets::_fullFObs [protected]

Full set of fundamental observables.

Definition at line 66 of file rarDatasets.hh.

Referenced by getFullFObs(), getWeightVarName(), and init().

RooArgSet rarDatasets::_UBs [protected]

Unblind strings for datasets.

Definition at line 67 of file rarDatasets.hh.

Referenced by ubStr().

---

The documentation for this class was generated from the following files:

• rarDatasets.hh
• rarDatasets.cc

---