void makeDataset( TString fname, TString tname, TString outfname ) {

  RooWorkspace *w = new RooWorkspace("w","w");

  w->factory( "Dst_M[1950.,2070.]" );
  w->factory( "D0_M[1810.,1920.]" );
  w->factory( "D0_LTIME_ps[0.00,5.]" );

  RooArgSet *observables = new RooArgSet();
  observables->add( *w->var("Dst_M") );
  observables->add( *w->var("D0_M") );
  observables->add( *w->var("D0_LTIME_ps") );
  w->defineSet("observables", *observables);

  w->var("Dst_M")->setBins(240);
  w->var("D0_M")->setBins(220);
  w->var("D0_LTIME_ps")->setBins(200);

  double Dst_M = -999.;
  double D0_M = -999.;
  double D0_LTIME_ps = -999.;

  TFile *tf = TFile::Open(fname);
  TTree *tree = (TTree*)tf->Get(tname);

  tree->SetBranchAddress( "Dst_M", &Dst_M );
  tree->SetBranchAddress( "D0_M" , &D0_M );
  tree->SetBranchAddress( "D0_LTIME_ps", &D0_LTIME_ps );

  RooDataSet *data = new RooDataSet("Data","Data",*observables);
  RooDataHist *dataH = new RooDataHist("DataHist","Data",*observables);

  for ( int ev=0; ev<tree->GetEntries(); ev++) {

    tree->GetEntry(ev);

    if ( ev%10000 == 0 ) cout << ev << " / " << tree->GetEntries() << endl;

    if ( Dst_M < w->var("Dst_M")->getMin() || Dst_M > w->var("Dst_M")->getMax() ) continue;
    if ( D0_M < w->var("D0_M")->getMin() || D0_M > w->var("D0_M")->getMax() ) continue;
    if ( D0_LTIME_ps < w->var("D0_LTIME_ps")->getMin() || D0_LTIME_ps > w->var("D0_LTIME_ps")->getMax() ) continue;

    w->var("Dst_M")->setVal(Dst_M);
    w->var("D0_M")->setVal(D0_M);
    w->var("D0_LTIME_ps")->setVal(D0_LTIME_ps);

    data->add( *observables );
    dataH->add( *observables );

  }
  tf->Close();

  w->import(*data);
  w->import(*dataH);

  w->writeToFile(outfname);

}
int main(int argc, char *argv[]){

	OptionParser(argc,argv);
	

	RooMsgService::instance().setGlobalKillBelow(RooFit::ERROR);
	RooMsgService::instance().setSilentMode(true);


  
	system(Form("mkdir -p %s",outdir_.c_str()));

	vector<string> procs;
	split(infilenames_,infilenamesStr_,boost::is_any_of(","));
  
   TPython::Exec("import os,imp,re");
   for (unsigned int i =0 ; i<infilenames_.size() ; i++){
     TFile *infile =  TFile::Open(infilenames_[i].c_str());
	   string outname  =(string) TPython::Eval(Form("'%s'.split(\"/\")[-1].replace('root','_reduced.root')",infilenames_[i].c_str())); 
     TFile *outfile = TFile::Open(outname.c_str(),"RECREATE") ;
    TDirectory* saveDir = outfile->mkdir("tagsDumper");
    saveDir->cd();

    RooWorkspace *inWS = (RooWorkspace*) infile->Get("tagsDumper/cms_hgg_13TeV");
    RooRealVar *intLumi = (RooRealVar*)inWS->var("IntLumi");
    RooWorkspace *outWS = new RooWorkspace("cms_hgg_13TeV");
    outWS->import(*intLumi);
    std::list<RooAbsData*> data =  (inWS->allData()) ;
    std::cout <<" [INFO] Reading WS dataset contents: "<< std::endl;
        for (std::list<RooAbsData*>::const_iterator iterator = data.begin(), end = data.end(); iterator != end; ++iterator )  {
              RooDataSet *dataset = dynamic_cast<RooDataSet *>( *iterator );
              if (dataset) {
              RooDataSet *datasetReduced = (RooDataSet*) dataset->emptyClone(dataset->GetName(),dataset->GetName());
                TRandom3 r;
                r.Rndm();
                double x[dataset->numEntries()];
                r.RndmArray(dataset->numEntries(),x);
                int desiredEntries = floor(0.5+ dataset->numEntries()*fraction_);
                int modFraction = floor(0.5+ 1/fraction_);
                int finalEventCount=0;
                for (int j =0; j < dataset->numEntries() ; j++){
                    if( j%modFraction==0){
                      finalEventCount++;
                    }
                 }
                float average_weight= dataset->sumEntries()/finalEventCount;
                for (int j =0; j < dataset->numEntries() ; j++){
                    if( j%modFraction==0){
                    dataset->get(j);
                    datasetReduced->add(*(dataset->get(j)),average_weight);
                    }
                }
              float entriesIN =dataset->sumEntries();
              float entriesOUT =datasetReduced->sumEntries();
           if(verbose_){
              std::cout << "Original dataset " << *dataset <<std::endl;
              std::cout << "Reduced       dataset " << *datasetReduced <<std::endl;
              std::cout << "********************************************" <<std::endl;
              std::cout << "fraction (obs) : " << entriesOUT/entriesIN << std::endl;
              std::cout << "fraction (exp) : " << fraction_ << std::endl;
              std::cout << "********************************************" <<std::endl;
              std::cout << "" <<std::endl;
              std::cout << "" <<std::endl;
              std::cout << "" <<std::endl;
              std::cout << "********************************************" <<std::endl;
              }
               outWS->import(*datasetReduced);
                }
                
             RooDataHist *datahist = dynamic_cast<RooDataHist *>( *iterator );

              if (datahist) {
              RooDataHist *datahistOUT = (RooDataHist*) datahist->emptyClone(datahist->GetName(),datahist->GetName());
                TRandom3 r;
                r.Rndm();
                for (int j =0; j < datahist->numEntries() ; j++){
                    
                    datahistOUT->add(*(datahist->get(j)),datahist->weight());
                }
              float w =datahistOUT->sumEntries();
              float z =datahist->sumEntries();
           if(verbose_){
              std::cout << "Original datahist " << *datahist <<std::endl;
              std::cout << "Reduced  datahist " << *datahistOUT<<std::endl;
              std::cout << "********************************************" <<std::endl;
              std::cout << "WH fraction (obs) : " << w/(z) <<std::endl;
              std::cout << "WH fraction (exp) : " << fraction_ << std::endl;
              std::cout << "********************************************" <<std::endl;
              std::cout << "" <<std::endl;
              std::cout << "" <<std::endl;
              std::cout << "" <<std::endl;
              std::cout << "********************************************" <<std::endl;
              }
               outWS->import(*datahistOUT);
                }
                  }
   saveDir->cd();
   outWS->Write();
   outfile->Close();
   infile->Close();
   }
}
int main(int argc, char *argv[]){

	OptionParser(argc,argv);
	

	RooMsgService::instance().setGlobalKillBelow(RooFit::ERROR);
	RooMsgService::instance().setSilentMode(true);


  
	system(Form("mkdir -p %s",outdir_.c_str()));

	vector<string> procs;
	split(infilenames_,infilenamesStr_,boost::is_any_of(","));
  
   TPython::Exec("import os,imp,re");
    const char * env = gSystem->Getenv("CMSSW_BASE") ; 
      std::string globeRt = env;
      TPython::Exec(Form("buildSMHiggsSignalXSBR = imp.load_source('*', '%s/src/flashggFinalFit/Signal/python/buildSMHiggsSignalXSBR.py')",globeRt.c_str()));
      TPython::Eval(Form("buildSMHiggsSignalXSBR.Init%dTeV()", 13));
   for (unsigned int i =0 ; i<infilenames_.size() ; i++){
	    int mH  =(int) TPython::Eval(Form("int(re.search('_M(.+?)_','%s').group(1))",infilenames_[i].c_str())); 
	   double WH_XS  =  (double)TPython::Eval(Form("buildSMHiggsSignalXSBR.getXS(%d,'%s')",mH,"WH"));
	   double ZH_XS  =  (double)TPython::Eval(Form("buildSMHiggsSignalXSBR.getXS(%d,'%s')",mH,"ZH"));
     float tot_XS = WH_XS + ZH_XS;
     float wFrac=  WH_XS /tot_XS ;
     float zFrac=  ZH_XS /tot_XS ;
      std::cout << "mass "<< mH << " wh fraction "<< WH_XS /tot_XS << ", zh fraction "<< ZH_XS /tot_XS <<std::endl; 
     TFile *infile =  TFile::Open(infilenames_[i].c_str());
	   string outname  =(string) TPython::Eval(Form("'%s'.split(\"/\")[-1].replace(\"VH\",\"WH_VH\")",infilenames_[i].c_str())); 
     TFile *outfile = TFile::Open(outname.c_str(),"RECREATE") ;
    TDirectory* saveDir = outfile->mkdir("tagsDumper");
    saveDir->cd();

    RooWorkspace *inWS = (RooWorkspace*) infile->Get("tagsDumper/cms_hgg_13TeV");
    RooRealVar *intLumi = (RooRealVar*)inWS->var("IntLumi");
    RooWorkspace *outWS = new RooWorkspace("cms_hgg_13TeV");
    outWS->import(*intLumi);
    std::list<RooAbsData*> data =  (inWS->allData()) ;
    std::cout <<" [INFO] Reading WS dataset contents: "<< std::endl;
        for (std::list<RooAbsData*>::const_iterator iterator = data.begin(), end = data.end(); iterator != end; ++iterator )  {
              RooDataSet *dataset = dynamic_cast<RooDataSet *>( *iterator );
              if (dataset) {
	            string zhname  =(string) TPython::Eval(Form("'%s'.replace(\"wzh\",\"zh\")",dataset->GetName())); 
	            string whname  =(string) TPython::Eval(Form("'%s'.replace(\"wzh\",\"wh\")",dataset->GetName())); 
              RooDataSet *datasetZH = (RooDataSet*) dataset->emptyClone(zhname.c_str(),zhname.c_str());
              RooDataSet *datasetWH = (RooDataSet*) dataset->emptyClone(whname.c_str(),whname.c_str());
                TRandom3 r;
                r.Rndm();
                double x[dataset->numEntries()];
                r.RndmArray(dataset->numEntries(),x);
                for (int j =0; j < dataset->numEntries() ; j++){
                    
                    if( x[j] < wFrac){
                    dataset->get(j);
                    datasetWH->add(*(dataset->get(j)),dataset->weight());
                    } else{
                    dataset->get(j);
                    datasetZH->add(*(dataset->get(j)),dataset->weight());
                    }
                }
              float w =datasetWH->sumEntries();
              float z =datasetZH->sumEntries();
           if(verbose_){
              std::cout << "Original dataset " << *dataset <<std::endl;
              std::cout << "WH       dataset " << *datasetWH <<std::endl;
              std::cout << "ZH       dataset " << *datasetZH <<std::endl;
              std::cout << "********************************************" <<std::endl;
              std::cout << "WH fraction (obs) : WH " << w/(w+z) <<",   ZH "<< z/(w+z) << std::endl;
              std::cout << "WH fraction (exp) : WH " << wFrac <<",   ZH "<< zFrac << std::endl;
              std::cout << "********************************************" <<std::endl;
              std::cout << "" <<std::endl;
              std::cout << "" <<std::endl;
              std::cout << "" <<std::endl;
              std::cout << "********************************************" <<std::endl;
              }
               outWS->import(*datasetWH);
               outWS->import(*datasetZH);
                }
             RooDataHist *datahist = dynamic_cast<RooDataHist *>( *iterator );

              if (datahist) {
	            string zhname  =(string) TPython::Eval(Form("'%s'.replace(\"wzh\",\"zh\")",datahist->GetName())); 
	            string whname  =(string) TPython::Eval(Form("'%s'.replace(\"wzh\",\"wh\")",datahist->GetName())); 
              RooDataHist *datahistZH = (RooDataHist*) datahist->emptyClone(zhname.c_str(),zhname.c_str());
              RooDataHist *datahistWH = (RooDataHist*) datahist->emptyClone(whname.c_str(),whname.c_str());
                TRandom3 r;
                r.Rndm();
                double x[datahist->numEntries()];
                r.RndmArray(datahist->numEntries(),x);
                for (int j =0; j < datahist->numEntries() ; j++){
                    
                    datahistWH->add(*(datahist->get(j)),datahist->weight()*wFrac);
                    datahistZH->add(*(datahist->get(j)),datahist->weight()*zFrac);
                }
              float w =datahistWH->sumEntries();
              float z =datahistZH->sumEntries();
           if(verbose_){
              std::cout << "Original datahist " << *datahist <<std::endl;
              std::cout << "WH       datahist " << *datahistWH <<std::endl;
              std::cout << "ZH       datahist " << *datahistZH <<std::endl;
              std::cout << "********************************************" <<std::endl;
              std::cout << "WH fraction (obs) : WH " << w/(w+z) <<",   ZH "<< z/(w+z) << std::endl;
              std::cout << "WH fraction (exp) : WH " << wFrac <<",   ZH "<< zFrac << std::endl;
              std::cout << "********************************************" <<std::endl;
              std::cout << "" <<std::endl;
              std::cout << "" <<std::endl;
              std::cout << "" <<std::endl;
              std::cout << "********************************************" <<std::endl;
              }
               outWS->import(*datahistWH);
               outWS->import(*datahistZH);
                }
                  }
   saveDir->cd();
   outWS->Write();
   outfile->Close();
   infile->Close();
   }
}
Beispiel #4
0
RooDataHist* withSmartBinning(TH1F* source, RooRealVar* var, float min, float max, float minWeight, float sizeSig, int type) {
  
  // DEFINE BINNING
  // type:
  // 0 = background
  // 1 = HRes gg signal
  // 2 = POWHEG gg signal
  // 2 = VBF signal
  float lowlimit;
  std::vector<float> sizes;
  std::vector<float> highlimits;  
  if (type == 0) {
    lowlimit = min;
    sizes.push_back(1.);
    highlimits.push_back(max);
  } else if (type == 1 || type == 2) {
    lowlimit = min;
    sizes.push_back(sizeSig);        
    // sizes.push_back(8.);        
    // sizes.push_back(4.);
    // sizes.push_back(8.);
    // highlimits.push_back(150.); 
    // highlimits.push_back(200.);
    // highlimits.push_back(320.);
    highlimits.push_back(max);
  } else {
    lowlimit = min;
    sizes.push_back(sizeSig);
    highlimits.push_back(max);
  }

  // ROOREALVAR BINNING
  var->setMin(lowlimit);
  var->setMax(highlimits.back());
  RooBinning rb(lowlimit,highlimits.back());
  float leftlimit = lowlimit;
  for (unsigned int ii = 0; ii < highlimits.size(); ii++) {
    float rightlimit = highlimits.at(ii);
    int howManyBins = (rightlimit-leftlimit)/sizes.at(ii);
    rb.addUniform(howManyBins,leftlimit,rightlimit);
    leftlimit = rightlimit;
  }
  var->setBinning(rb);

  // FILL ROODATAHIST
  RooDataHist* result = new RooDataHist("result","A dataset",RooArgList(*var));
  float nWeight = source->GetEntries();
  if (type == 2) nWeight = 0.00215;
  float binWidth = source->GetBinWidth(1);

  float thisWeight = 0.;
  float thisWeightsum = 0.;
  int j = 0;
  int whichInterval = 0;
  for (Int_t i=1; i<=source->GetNbinsX(); i++) {

    float thispt = source->GetXaxis()->GetBinCenter(i);
    if (thispt > lowlimit && thispt < highlimits.back()) {
      bool lastbin = false;
      if (source->GetXaxis()->GetBinCenter(i+1) >= highlimits.back()) lastbin = true;
      if (thispt > highlimits.at(whichInterval)) whichInterval++;
      var->setVal(thispt);
      thisWeight += source->GetBinContent(i)*nWeight;
      thisWeightsum += source->GetBinError(i)*nWeight*source->GetBinError(i)*nWeight;
      j++;
      cout << "Entries in TH1 bin " << i << " = " << source->GetBinContent(i)*nWeight << " +/- " << source->GetBinError(i)*nWeight << endl;   
      if (binWidth*j >= sizes.at(whichInterval) || lastbin) {
        // cout << "Entro" << endl; 
	if (thisWeight <= 0) thisWeight = minWeight; 
        if (fabs(thisWeight/sizes.at(whichInterval) - 0.0829038) < 0.00001) thisWeight /= 10.;
	if (sqrt(thisWeightsum) > thisWeight) thisWeightsum = thisWeight*thisWeight*0.9;
	cout << " " << var->getVal() << " " << thisWeight << " " << thisWeightsum << " " << sizes.at(whichInterval) << endl; 
	result->add(RooArgSet(*var),
		    thisWeight/sizes.at(whichInterval),
		    thisWeightsum/(sizes.at(whichInterval)*sizes.at(whichInterval)));
	thisWeight = 0.;
	thisWeightsum = 0.;
	j = 0;
      }
    }
  } 

  return result;
}