Plotting/Modules/EdgeLimit.C

#include <iostream>


#include <TVirtualIndex.h>

#include <RooRealVar.h>
#include <RooArgSet.h>
#include <RooDataSet.h>
#include <RooMCStudy.h>
#include <RooCategory.h>

#include <RooPlot.h>
#include <RooGaussian.h>
#include <RooConstVar.h>
#include <RooSimultaneous.h>
#include <RooAddPdf.h>
#include <RooFitResult.h>
#include <RooVoigtian.h>
#include <RooMsgService.h>

#include <RooStats/ModelConfig.h>
#include "RooStats/ProfileLikelihoodCalculator.h"
#include "RooStats/LikelihoodInterval.h"
#include "RooStats/HypoTestResult.h"
#include "RooStats/SimpleLikelihoodRatioTestStat.h"
#include "RooStats/ProfileLikelihoodTestStat.h"
#include "RooStats/HybridCalculatorOriginal.h"
#include "RooStats/HypoTestInverterOriginal.h"

//#include "ParametrizedEdge.C"
#include "EdgeModules/RooSUSYTPdf.cxx"
#include "EdgeModules/RooSUSYBkgPdf.cxx"

#include "md5/md5.h"
#include "md5/md5.C"

using namespace std;
using namespace PlottingSetup;


ShapeDroplet LimitsFromEdge(float low_fullCLs, float high_CLs, TTree *events, string addcut, string name, string mcjzb, string datajzb, vector<float> jzbbins, float jzbpeakerrormc, float jzbpeakerrordata) {
  write_error(__FUNCTION__,"Not implemented edge limits yet (returning crap)");
  ShapeDroplet beta;beta.observed=-12345;beta.SignalIntegral=1;return beta;
}


void PrepareEdgeShapes(string mcjzb, string datajzb, vector<float> jzbbins, float jzbpeakerrordata) {
  write_error(__FUNCTION__,"Not implemented edge shape storage yet");
}
  

///------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------


namespace EdgeFitter {
  
  void DoEdgeFit(string mcjzb, string datajzb, float DataPeakError, float MCPeakError, float jzb_cut, int icut, int is_data, TCut cut, TTree*);
  void DoEdgeFit(string mcjzb, string datajzb, float DataPeakError, float MCPeakError, vector<float> jzb_cut, int is_data, TCut cut, TTree*);
  void getMedianLimit(RooWorkspace *ws,vector<RooDataSet> theToys,float &median,float &sigmaDown, float &sigmaUp, float &twoSigmaDown, float &twoSigmaUp);
  void InitializeVariables(float _mllmin, float _mllmax, float _jzbmax, TCut _cut);
  void PrepareDatasets(int);
  void DrawDatasetContent(int);
  void DoFit(int is_data, float jzb_cut);
  string RandomStorageFile();
  Yield Get_Z_estimate(float,int);
  Yield Get_T_estimate(float,int);
  float calcExclusion(RooWorkspace *ws, RooDataSet data, bool calcExclusion);
  vector<RooDataSet> generateToys(RooWorkspace *ws, int nToys);
  void prepareLimits(RooWorkspace *ws, bool ComputeBands);
  TGraph* prepareLM(float mass, float nEv);
  
  float jzbmax;
  float mllmin;
  float mllmax;
  TCut cut;
  
  RooDataSet* AllData;
  RooDataSet* SFSample;
  RooDataSet* OFSample; 
  
  bool MarcoDebug=true;
  
  float FixedMEdge=-1;
  float FixedMEdgeChi2_H1=-1;
  float FixedMEdgeChi2_H0=-1;
  
  bool RejectPointIfNoConvergence=false;
  
  string Mode="UndefinedMode";
  
  bool AllowTriangle=true;
}

TGraph* EdgeFitter::prepareLM(float mass, float nEv) {
  float massLM[1];
  massLM[0]=mass;
  float accEffLM[1];
  accEffLM[0]=nEv/PlottingSetup::luminosity;
  TGraph *lm = new TGraph(1, massLM, accEffLM);
  lm->GetXaxis()->SetNoExponent(true);
  lm->GetXaxis()->SetTitle("m_{cut} [GeV]");
  lm->GetYaxis()->SetTitle("#sigma #times A [pb] 95% CL UL");
  lm->GetXaxis()->SetLimits(0.,300.);
  lm->GetYaxis()->SetRangeUser(0.,0.08);
  lm->SetMarkerStyle(34);
  lm->SetMarkerColor(kRed);
  return lm;
}

vector<RooDataSet> EdgeFitter::generateToys(RooWorkspace *ws, int nToys) {
  ws->ls();
  ws->var("nSig")->setVal(0.);
  ws->var("nSig")->setConstant(true);
  RooFitResult* fit = ws->pdf("combModel")->fitTo(*ws->data("data_obs"),RooFit::Save());
  vector<RooDataSet> theToys;
  
  RooMCStudy mcEE(*ws->pdf("combModel"),RooArgSet(*ws->var("inv")),RooFit::Slice(*ws->cat("cat"),"EE"));
  mcEE.generate(nToys,44,true);
  RooMCStudy mcMM(*ws->pdf("combModel"),RooArgSet(*ws->var("inv")),RooFit::Slice(*ws->cat("cat"),"MM"));
  mcMM.generate(nToys,44,true);
  RooMCStudy mcOSOF(*ws->pdf("combModel"),RooArgSet(*ws->var("inv")),RooFit::Slice(*ws->cat("cat"),"OSOF"));
  mcOSOF.generate(nToys,44,true);
  
  RooRealVar mll("m_{ll}","m_{ll}",mllmin,mllmax,"GeV/c^{2}");
  RooRealVar id1("id1","id1",0,1,"GeV/c^{2}");
  RooRealVar id2("id2","id2",0,1,"GeV/c^{2}");
  RooRealVar jzb("jzb","jzb",-jzbmax,jzbmax,"GeV/c");
  RooRealVar weight("weight","weight",0,1000,"");
  RooArgSet observables(mll,jzb,id1,id2,weight);

  for(int i=0;i<nToys;i++) {
    RooDataSet* toyEE    = (RooDataSet*)mcEE.genData(i);
    RooDataSet* toyMM    = (RooDataSet*)mcMM.genData(i);
    RooDataSet* toyOSOF  = (RooDataSet*)mcOSOF.genData(i);
    stringstream toyname;
    toyname << "theToy_" << i;
    write_warning(__FUNCTION__,"Problem while adding toys");
    RooDataSet toyData = RooDataSet(toyname.str().c_str(),toyname.str().c_str(),observables,RooFit::Index(const_cast<RooCategory&>(*ws->cat("cat"))),RooFit::Import("OSOF",*toyOSOF),RooFit::Import("EE",*toyEE),RooFit::Import("MM",*toyMM));
    theToys.push_back(toyData);
  }
  ws->var("nSig")->setVal(17.0);
  ws->var("nSig")->setConstant(false);
  return theToys;
}

void EdgeFitter::getMedianLimit(RooWorkspace *ws,vector<RooDataSet> theToys,float &median,float &sigmaDown, float &sigmaUp, float &twoSigmaDown, float &twoSigmaUp) {
  TH1F *gauLimit = new TH1F("gausLimit","gausLimit",60,0.,80./PlottingSetup::luminosity);
  vector<float> theLimits;
  for(int itoy=0;itoy<(int)theToys.size();itoy++) {
    float theLimit = calcExclusion(ws,theToys[itoy],false);
    if(theLimit > 0 ) gauLimit->Fill(theLimit);
  }
  const Int_t nQ = 4;
  Double_t yQ[nQ] = {0.,0.,0.,0.};
  Double_t xQ[nQ] = {0.022750132,0.1586552539,0.84134474609999998,0.977249868};
  gauLimit->GetQuantiles(nQ,yQ,xQ);
  median = gauLimit->GetMean();
//  median = median1(gauLimit);
  twoSigmaDown = abs(yQ[0]-median);
  sigmaDown = abs(yQ[1]-median);
  sigmaUp = abs(yQ[2]-median);
  twoSigmaUp = abs(yQ[3]-median);
  dout << median * PlottingSetup::luminosity << " " << sigmaUp * PlottingSetup::luminosity << endl;
}

void EdgeFitter::prepareLimits(RooWorkspace *ws, bool ComputeBands) {
  if(ComputeBands) {
    vector<RooDataSet> theToys = EdgeFitter::generateToys(ws,50);
    vector<float> medVals;
    vector<float> medLimits;
    vector<float> sigmaLimitsDown;
    vector<float> sigmaLimitsUp;
    vector<float> twoSigmaLimitsDown;
    vector<float> twoSigmaLimitsUp;
    for(int i=20;i<=320;i+=40) {
      ws->var("nSig")->setVal(10.0);
      medVals.push_back((float)i);
      ws->var("m0")->setVal((float)i);
      ws->var("m0")->setConstant(true);
      float Smedian,SsigmaDown,SsigmaUp,StwoSigmaDown,StwoSigmaUp;
      EdgeFitter::getMedianLimit(ws,theToys,Smedian,SsigmaDown,SsigmaUp,StwoSigmaDown,StwoSigmaUp);
      medLimits.push_back(Smedian);
      sigmaLimitsDown.push_back(SsigmaDown);
      sigmaLimitsUp.push_back(SsigmaUp);
      twoSigmaLimitsDown.push_back(StwoSigmaDown);
      twoSigmaLimitsUp.push_back(StwoSigmaUp);
    }
    write_warning(__FUNCTION__,"Still need to store limits");
  } else {
    vector<float> theVals;
    vector<float> theLimits;
    for(int i=20;i<300;i+=5) {
      ws->var("nSig")->setVal(0.0);
      theVals.push_back((float)i);
      ws->var("m0")->setVal((float)i);
      ws->var("m0")->setConstant(true);
//      theLimits.push_back(calcExclusion(ws,(RooDataSet)*ws->data("data_obs"),true));
      write_error(__FUNCTION__,"Error while casting roo data set");
    }
    
    for(int i=0;i<(int)theLimits.size();i++) {
      if((theLimits[i]<2.0/PlottingSetup::luminosity)||(theLimits[i]>40.0/PlottingSetup::luminosity)) {
        dout << i << " : " << theVals[i] << endl;
        theLimits[i] = (theLimits[i+2]+theLimits[i-2])/2.0;
        write_warning(__FUNCTION__,"Need to store limits");
      }
    write_warning(__FUNCTION__,"Need to store limits");
    }
}
}
  

float EdgeFitter::calcExclusion(RooWorkspace *ws, RooDataSet data, bool LoadDataObs) {
  int numberOfToys=50;
  RooRealVar mu("mu","nSig",0,10000,"");
  RooArgSet poi = RooArgSet(mu);
  RooArgSet *nullParams = (RooArgSet*)poi.snapshot();
  nullParams->setRealValue("nSig",0);
  RooStats::ModelConfig *model = new RooStats::ModelConfig();
  model->SetWorkspace(*ws);
  model->SetPdf("combModel");
  model->SetParametersOfInterest(poi);
//  if(LoadDataObs) data = (RooDataSet)*ws->data("data_obs");

  RooStats::ProfileLikelihoodCalculator plc(data, *model);
  plc.SetNullParameters(*nullParams);
  plc.SetTestSize(0.05);
  
  RooStats::LikelihoodInterval* interval = plc.GetInterval();
  RooStats::HypoTestResult *htr = plc.GetHypoTest();
  double theLimit = interval->UpperLimit( mu );
//  double significance = htr->Significance();
  
  ws->defineSet("obs","nB");
  ws->defineSet("poi","nSig");
  
  RooStats::ModelConfig b_model = RooStats::ModelConfig("B_model", ws);
  b_model.SetPdf(*ws->pdf("combModel"));
  b_model.SetObservables(*ws->set("obs"));
  b_model.SetParametersOfInterest(*ws->set("poi"));
  ws->var("nSig")->setVal(0.0);  //# important!
  b_model.SetSnapshot(*ws->set("poi"));
  
  RooStats::ModelConfig sb_model = RooStats::ModelConfig("S+B_model", ws);
  sb_model.SetPdf(*ws->pdf("combModel"));
  sb_model.SetObservables(*ws->set("obs"));
  sb_model.SetParametersOfInterest(*ws->set("poi"));
  ws->var("nSig")->setVal(64.0); //# important!
  sb_model.SetSnapshot(*ws->set("poi"));
  
  RooStats::SimpleLikelihoodRatioTestStat slrts = RooStats::SimpleLikelihoodRatioTestStat(*b_model.GetPdf(),*sb_model.GetPdf());
  slrts.SetNullParameters(*b_model.GetSnapshot());
  slrts.SetAltParameters(*sb_model.GetSnapshot());
  RooStats::ProfileLikelihoodTestStat profll = RooStats::ProfileLikelihoodTestStat(*b_model.GetPdf());
  
  RooStats::HybridCalculatorOriginal hc = RooStats::HybridCalculatorOriginal(data, sb_model, b_model,0,0);
  hc.SetTestStatistic(2);
  hc.SetNumberOfToys(numberOfToys);
  
  RooStats::HypoTestInverterOriginal hcInv =  RooStats::HypoTestInverterOriginal(hc,*ws->var("nSig"));
  hcInv.SetTestSize(0.05);
  hcInv.UseCLs(true);
  hcInv.RunFixedScan(5,theLimit-0.5,theLimit+0.5);;
  RooStats::HypoTestInverterResult* hcInt = hcInv.GetInterval();
  float ulError = hcInt->UpperLimitEstimatedError();
  theLimit = hcInt->UpperLimit();
  dout << "Found upper limit : " << theLimit << "+/-" << ulError << endl;
  
  return theLimit/PlottingSetup::luminosity;
  
}

string GetSkimName(int isample) {
  return removefunnystring(allsamples.collection[isample].filename);
}

TTree* SkimTree(int isample) {
  string TreeName = GetSkimName(isample);
  cout << "About to skim " << TreeName << " for sample id " << isample << endl;
  TTree* newTree = new TTree(TreeName.c_str(),TreeName.c_str());
  
  float mll,edgeWeight;
  int id1,id2;
  
  newTree->Branch("edgeWeight",&edgeWeight,"edgeWeight/F");
  newTree->Branch("mll",&mll,"mll/F");
  newTree->Branch("id1",&id1,"id1/I");
  newTree->Branch("id2",&id2,"id2/I");
  newTree->Branch("isample",&isample,"isample/I");
  
  float xsweight=1.0;
  if(allsamples.collection[isample].is_data==false) xsweight=luminosity*allsamples.collection[isample].weight;
  if(EdgeFitter::MarcoDebug) {
    dout << "   Going to reduce it with cut " << EdgeFitter::cut << endl;
    dout << "   Original tree contains " << allsamples.collection[isample].events->GetEntries() << endl;
  }
  
  float tmll;
  int tid1,tid2;
  allsamples.collection[isample].events->SetBranchAddress("mll",&tmll);
  allsamples.collection[isample].events->SetBranchAddress("id1",&tid1);
  allsamples.collection[isample].events->SetBranchAddress("id2",&tid2);
  
  TTreeFormula *select = new TTreeFormula("select", EdgeFitter::cut, allsamples.collection[isample].events);
  TTreeFormula *Weight = new TTreeFormula("Weight", cutWeight, allsamples.collection[isample].events);
  
  float wgt=1.0;
  for (Int_t entry = 0 ; entry < allsamples.collection[isample].events->GetEntries() ; entry++) {
   allsamples.collection[isample].events->LoadTree(entry);
   if (select->EvalInstance()) {
     allsamples.collection[isample].events->GetEntry(entry);
     mll=tmll;
     id1=tid1;
     id2=tid2;
     wgt=Weight->EvalInstance();
     edgeWeight=wgt*xsweight;
     newTree->Fill();
   }
  }
  
  if(EdgeFitter::MarcoDebug) dout << "   Reduced tree contains " << newTree->GetEntries() << " entries " << endl;
  return newTree;
}

void EdgeFitter::InitializeVariables(float _mllmin, float _mllmax, float _jzbmax, TCut _cut) {
  mllmin=_mllmin;
  mllmax=_mllmax;
  jzbmax=_jzbmax;
  cut=_cut;
}

TTree* MergeTrees(vector<TTree*> trees) {
  assert(trees.size()>0);
  TTree * newtree = (TTree*)trees[0]->CloneTree(0);
  newtree->SetTitle("FullTree");
  newtree->SetName("FullTree");
  trees[0]->GetListOfClones()->Remove(newtree);
  trees[0]->ResetBranchAddresses();
  newtree->ResetBranchAddresses();
  
  for(int itree=0;itree<trees.size();itree++) {
    newtree->CopyAddresses(trees[itree]);
    Long64_t nentries = trees[itree]->GetEntries();
    for (Long64_t iEntry=0;iEntry<nentries;iEntry++) {
      trees[itree]->GetEntry(iEntry);
      newtree->Fill();
    }
    trees[itree]->ResetBranchAddresses(); // Disconnect from new tree
    if (newtree->GetTreeIndex()) {
      newtree->GetTreeIndex()->Append(trees[itree]->GetTreeIndex(),kTRUE);
    }
    if (newtree && newtree->GetTreeIndex()) {
      newtree->GetTreeIndex()->Append(0,kFALSE); // Force the sorting
    }
  }
  return newtree;
}
    
      
void EdgeFitter::PrepareDatasets(int is_data) {
  write_warning(__FUNCTION__,"Need to make this function ready for scans as well (use signal from scan samples)");
  
  ensure_directory_exists(PlottingSetup::cbafbasedir+"/EdgeCache/");
  
  stringstream FileName;
  FileName << PlottingSetup::cbafbasedir << "/EdgeCache/" << md5((const char*)cut) << ".root";
  
  TFile *fEdgeCache = new TFile(FileName.str().c_str(),"UPDATE");
  if(fEdgeCache->GetNkeys()==0) {
    ofstream CacheOverview;
    CacheOverview.open((PlottingSetup::cbafbasedir+"/EdgeCache/CacheOverview").c_str(),ios::app);
    CacheOverview << md5((const char*)cut) << ";" << (const char*) cut << endl;
    CacheOverview.close();
  }
  
  vector<TTree*> SkimmedTrees;
  TTree *SkimmedTree[(int)allsamples.collection.size()];
  for(int isample=0;isample<(int)allsamples.collection.size();isample++) {
    if(!allsamples.collection[isample].is_active) continue;
    if(is_data==1&&allsamples.collection[isample].is_data==false) continue;//kick all samples that aren't data if we're looking for data.
    if(is_data==1&&allsamples.collection[isample].is_data==false) continue;//kick all samples that aren't data if we're looking for data.
    if(is_data!=1&&allsamples.collection[isample].is_data==true) continue;//kick all data samples when looking for MC
    if(is_data!=2&&allsamples.collection[isample].is_signal==true) continue;//remove signal sample if we don't want it.
    if(EdgeFitter::MarcoDebug) dout << "Considering : " << allsamples.collection[isample].samplename << endl;
    
    string SkimName = GetSkimName(isample);
    SkimmedTree[isample] = (TTree*)fEdgeCache->Get(SkimName.c_str());
    if(!SkimmedTree[isample]) {
      dout << "Need to generate tree for " << allsamples.collection[isample].filename << endl;
      SkimmedTree[isample] = SkimTree(isample);
      fEdgeCache->cd();
      SkimmedTree[isample]->Write();
    } else {
      dout << "Loaded tree for " << allsamples.collection[isample].filename << " from cache file " << endl;
    }
      
    SkimmedTrees.push_back(SkimmedTree[isample]);
  }
  
  TTree *completetree  = MergeTrees(SkimmedTrees);
  
  if(EdgeFitter::MarcoDebug) dout << "Complete tree now contains " << completetree->GetEntries() << " entries " << endl;
  
  RooRealVar mll("mll","m_{ll}",mllmin,mllmax,"GeV/c^{2}");
  RooRealVar id1("id1","id1",0,1,"GeV/c^{2}");
  RooRealVar id2("id2","id2",0,1,"GeV/c^{2}");
  RooRealVar edgeWeight("edgeWeight","edgeWeight",0,1000,"");
  RooArgSet observables(mll,id1,id2,edgeWeight);
  
  string title="CMS Data";
  if(is_data!=1) title="CMS SIMULATION";
  RooDataSet LAllData("LAllData",title.c_str(),completetree,observables,"","edgeWeight");
  fEdgeCache->Close();
  dout << "Edge cache closed." << endl;
    
  EdgeFitter::SFSample = (RooDataSet*)LAllData.reduce("id1==id2");
  EdgeFitter::OFSample = (RooDataSet*)LAllData.reduce("id1!=id2");
  EdgeFitter::AllData  = (RooDataSet*)LAllData.reduce("id1!=id2||id1==id2");
  
  SFSample->SetName("SFSample");
  OFSample->SetName("OFSample");
  AllData->SetName("AllData");
  
  if(EdgeFitter::MarcoDebug) {
    dout << "Number of (weighted) events in full sample = " << AllData->sumEntries() << " (raw events : " << AllData->numEntries() << ")" <<  endl;
    dout << "Number of (weighted) events in eemm sample = " << SFSample->sumEntries() << " (raw events : " << SFSample->numEntries() << ")" <<  endl;
    dout << "Number of (weighted) events in em sample = " << OFSample->sumEntries() << " (raw events : " << OFSample->numEntries() << ")" <<  endl;
  }
  
}

void EdgeFitter::DrawDatasetContent(int is_data) {
  TCanvas* cSFdata = new TCanvas("cSFdata","cSFdata") ;
  
  RooRealVar mll("mll","m_{ll}",mllmin,mllmax,"GeV/c^{2}");
  RooPlot* frame1 = mll.frame(RooFit::Bins(int((mllmax-mllmin)/5.0)),RooFit::Title("SF sample")) ;
  frame1->GetXaxis()->CenterTitle(1);
  frame1->GetYaxis()->CenterTitle(1);
  SFSample->plotOn(frame1,RooFit::Name("SFdata")) ;
  
  RooPlot* frame2 = mll.frame(RooFit::Bins(int((mllmax-mllmin)/5.0)),RooFit::Title("OF sample")) ;
  frame2->GetXaxis()->CenterTitle(1);
  frame2->GetYaxis()->CenterTitle(1);
  OFSample->plotOn(frame2,RooFit::Name("OFdata")) ;
  
//  THStack sf_ref = allsamples.DrawStack("sf_ref","mll",int((mllmax-mllmin)/5.0),mllmin,mllmax, "m_{ll} [GeV]", "events", cut&&TCut("id1==id2"),is_data, luminosity);
//  THStack of_ref = allsamples.DrawStack("of_ref","mll",int((mllmax-mllmin)/5.0),mllmin,mllmax, "m_{ll} [GeV]", "events", cut&&TCut("id1!=id2"),is_data, luminosity);
  
  TH1F *sf_ref = allsamples.Draw("sf_ref","mll",int((mllmax-mllmin)/5.0),mllmin,mllmax, "m_{ll} [GeV]", "events", cut&&TCut("id1==id2"),is_data, luminosity);
  TH1F *of_ref = allsamples.Draw("of_ref","mll",int((mllmax-mllmin)/5.0),mllmin,mllmax, "m_{ll} [GeV]", "events", cut&&TCut("id1!=id2"),is_data, luminosity);
  
  TH1F *sf_fit = (TH1F*)SFSample->createHistogram("sf_fit",mll,RooFit::Binning(int((mllmax-mllmin)/5)),RooFit::SumW2Error(true));
  TH1F *of_fit = (TH1F*)OFSample->createHistogram("of_fit",mll,RooFit::Binning(int((mllmax-mllmin)/5)),RooFit::SumW2Error(true));
  
  TLegend *leg = allsamples.allbglegend();
  leg->AddEntry(sf_fit,"RooDataSet content","p");
  leg->SetX1(0.58);

  cSFdata->cd() ;
  gPad->SetLeftMargin(0.15);
  frame1->GetYaxis()->SetTitleOffset(1.4);
  frame1->Draw();
  sf_ref->Draw("histo,same");
  sf_fit->Draw("e1,same");
  leg->Draw("same");
  if(is_data==data) DrawPrelim();
  else DrawPrelim(PlottingSetup::luminosity,true);
  CompleteSave(cSFdata,"Edge/SF_NoFit");
  
  TCanvas* cOFdata = new TCanvas("cOFdata","cOFdata") ;
  cOFdata->cd() ;
  gPad->SetLeftMargin(0.15);
  frame2->SetMaximum(frame1->GetMaximum());
  frame2->GetYaxis()->SetTitleOffset(1.4);
  frame2->Draw();
  of_ref->Draw("histo,same");
  of_fit->Draw("e1,same");
  leg->Draw("same");
  if(is_data==data) DrawPrelim();
  else DrawPrelim(PlottingSetup::luminosity,true);
  CompleteSave(cOFdata,"Edge/OF_NoFit");

  delete sf_fit;
  delete of_fit;
  
  delete sf_ref;
  delete of_ref;
  
}
  
string WriteWithError(float central, float error, int digits) {
  float ref=central;
  if(ref<0) ref=-central;
  int HighestSigDigit = 0;
  if(ref>1) HighestSigDigit = int(log(ref)/log(10))+1;
  else HighestSigDigit = int(log(ref)/(log(10)));
  
  float divider=pow(10.0,(double(HighestSigDigit-digits)));
  
  stringstream result;
  result << divider*int(central/divider+0.5) << " #pm " << divider*int(error/divider+0.5);
  return result.str();
}


string EdgeFitter::RandomStorageFile() {
  TRandom3 *r = new TRandom3(0);
  int rho = (int)r->Uniform(1,10000000); 
  stringstream RandomFile;
  RandomFile << PlottingSetup::cbafbasedir << "/exchange/TempEdgeFile_" << rho << ".root";
  delete r;
  return RandomFile.str();
}

Yield EdgeFitter::Get_Z_estimate(float jzb_cut, int icut) {
  if(MarcoDebug) write_error(__FUNCTION__,"Returning random Z yield"); 
  Yield a(123,45,67); return a;
  return PlottingSetup::allresults.predictions[icut].Zbkg;
}

Yield EdgeFitter::Get_T_estimate(float jzb_cut, int icut) {
  if(MarcoDebug) write_error(__FUNCTION__,"Returning random TTbar yield"); 
  Yield a(1234,56,78); return a;
  return PlottingSetup::allresults.predictions[icut].Flavorsym;
}

void EdgeFitter::DoFit(int is_data, float jzb_cut) {
  
  stringstream prefix;
  if(is_data==data) prefix << "data_";
  if(is_data==mc) prefix << "mc_";
  if(is_data==mcwithsignal) prefix << "mcwithS_";
  
  prefix << EdgeFitter::Mode << "_" << jzb_cut;
  
  if(EdgeFitter::AllowTriangle) prefix << "__H1";
  else prefix << "__H0";
  
  RooRealVar mll("mll","m_{ll}",mllmin,mllmax,"GeV/c^{2}");
  RooRealVar edgeWeight("edgeWeight","edgeWeight",0,1000,"");
  RooCategory sample("sample","sample") ;
  sample.defineType("SF");
  //sample.defineType("mm");
  sample.defineType("OF");
  
  //RooDataSet combData("combData","combined data",mll,Index(sample),Import("SF",*SFSample),Import("mm",*mmSample),Import("OF",*OFSample));
  RooDataSet combData("combData","combined data",RooArgSet(mll,edgeWeight),RooFit::Index(sample),RooFit::Import("SF",*SFSample),RooFit::Import("OF",*OFSample),RooFit::WeightVar(edgeWeight));
  
  
  //First we make a fit to opposite flavor
  RooRealVar fttbarOF("fttbarOF", "fttbarOF", 100, 0, 1.5*combData.sumEntries());
  RooRealVar par1ttbarOF("par1ttbarOF", "par1ttbarOF", 1.6, 0.01, 4.0);
  RooRealVar par2ttbarOF("par2ttbarOF", "par2ttbarOF", 1.0);
  RooRealVar par3ttbarOF("par3ttbarOF", "par3ttbarOF", 0.028, 0.001, 1.0);
  RooRealVar par4ttbarOF("par4ttbarOF", "par4ttbarOF", 2.0);
  RooSUSYBkgPdf ttbarOF("ttbarOF","ttbarOF", mll , par1ttbarOF, par2ttbarOF, par3ttbarOF, par4ttbarOF);
  RooAddPdf model_OF("model_OF","model_OF", ttbarOF, fttbarOF);
  RooSimultaneous simPdfOF("simPdfOF","simultaneous pdf", sample) ;
  simPdfOF.addPdf(model_OF,"OF");
  RooFitResult *resultOF = simPdfOF.fitTo(combData, RooFit::Save(),RooFit::Extended(),RooFit::Minos(true));
  dout << "============================= < OF results > =============================" << endl;
  resultOF->Print();
  dout << "============================= < /OF results > =============================" << endl;
  
  
  RooPlot* frameO = mll.frame(RooFit::Bins(int((mllmax-mllmin)/5.0)),RooFit::Title("OF sample"));
  frameO->GetXaxis()->CenterTitle(1);
  frameO->GetYaxis()->CenterTitle(1);
  combData.plotOn(frameO,RooFit::Name("OFdata"),RooFit::Cut("sample==sample::OF")) ;
  simPdfOF.plotOn(frameO,RooFit::Slice(sample,"OF"),RooFit::Name("FullFit"),RooFit::ProjWData(sample,combData), RooFit::LineColor(kBlack)) ;
  simPdfOF.plotOn(frameO,RooFit::Slice(sample,"OF"),RooFit::Name("TTbarOFonly"),RooFit::Components("ttbarOF"),RooFit::ProjWData(sample,combData),RooFit::LineStyle(kDashed)) ;
  
  TCanvas* pof = new TCanvas("pof","pof") ;
  pof->cd() ;
  gPad->SetLeftMargin(0.15);
  frameO->GetYaxis()->SetTitleOffset(1.4);
  frameO->Draw();
  if(is_data==data) DrawPrelim();
  else DrawPrelim(PlottingSetup::luminosity,true);
  if(EdgeFitter::FixedMEdge>=0) CompleteSave(pof,"Edge/OF__OFFitonly_"+prefix.str()+"__MEdgeFix_"+any2string(EdgeFitter::FixedMEdge),false,false);
  else CompleteSave(pof,"Edge/OF__OFFitonly_"+prefix.str(),false,false);
  delete pof;
    
  if(resultOF->covQual()!=3) {
    write_error(__FUNCTION__,"OF fit did not converge!!! Cannot continue!");
    dout << "covQual is " << resultOF->covQual() << endl;
    if(EdgeFitter::AllowTriangle) EdgeFitter::FixedMEdgeChi2_H1=-1;
    else EdgeFitter::FixedMEdgeChi2_H0=-1;
    if(EdgeFitter::RejectPointIfNoConvergence) return;
  } else {
    write_info(__FUNCTION__,"OF fit converged");
  }
  
  float StartingMedge=70;
  if(EdgeFitter::FixedMEdge>0) StartingMedge=EdgeFitter::FixedMEdge;

 
  RooDataSet *ZDataSet = (RooDataSet*)EdgeFitter::AllData->reduce("id1==id2 && abs(mll-91.2)<20");

  float maxZ = ZDataSet->sumEntries();
  dout << "maxZ was set to " << maxZ << endl;
  delete ZDataSet;
  
  
  // Now same flavor  
  RooRealVar fzSF("fzSF", "fzSF", 39, 39, maxZ);
  RooRealVar meanzSF("meanzSF", "meanzSF", 91.1876, 89, 95);
  //RooRealVar sigmazSF("sigmazSF", "sigmazSF", 0.5);
  RooRealVar sigmazSF("sigmazSF", "sigmazSF", 5, 0.5, 5);
  RooRealVar widthzSF("widthzSF", "widthzSF", 2.94);
  widthzSF.setConstant(1);
  
  /*RooRealVar fttbarSFx("fttbarSFx","fttbarSFx",0.2,1.8);
  RooRealVar mRsfof("mRsfof","mRsfof",1.02);
  RooRealVar wRsfof("wRsfof","wRsfof",0.07);
  
  RooGaussian RSFOF("RSFOF","RSFOF",fttbarSFx,mRsfof,wRsfof);
  RooProdPdf fttbarSF("fttbarSF","fttbarOF*RSFOF",RooArgList(RSFOF,fttbarOF));*/
  RooRealVar fttbarSF("fttbarSF", "fttbarSF", fttbarOF.getVal(), 0.2*fttbarOF.getVal(), 1.5*fttbarOF.getVal());
  
  RooRealVar par1ttbarSF("par1ttbarSF", "par1ttbarSF", 1.02*par1ttbarOF.getVal(), (1.02-0.07)*par1ttbarOF.getVal(), (1.02+0.07)*par1ttbarOF.getVal());
 
  RooRealVar fsignalSF("fsignalSF", "fsignalSF", 0, 0, 300);
  RooRealVar par1signalSF("par1signalSF", "par1signalSF", 45, 20, 100);
  RooRealVar par2signalSF("par2signalSF", "par2signalSF", 2, 1, 10);
  RooRealVar par3signalSF("par3signalSF", "par3signalSF", StartingMedge, 0, 300);

  RooVoigtian zSF("zSF", "zSF", mll, meanzSF, widthzSF, sigmazSF);

  if(EdgeFitter::FixedMEdge>0) par3signalSF.setConstant();
  
  RooSUSYBkgPdf ttbarSF("ttbarSF","ttbarSF", mll , par1ttbarSF, par2ttbarOF, par3ttbarOF, par4ttbarOF);
  RooSUSYTPdf signalSF("signalSF","signalSF", mll , par1signalSF, sigmazSF, par3signalSF);

  RooAddPdf model_SF("model_SF","model_SF", RooArgList(zSF, ttbarSF, signalSF), RooArgList(fzSF, fttbarSF, fsignalSF));
  
  
  if(!EdgeFitter::AllowTriangle) {
    fsignalSF.setVal(0.0); // kill off the signal if we don't want the triangle
    fsignalSF.setConstant(1);
    par1signalSF.setConstant(1);
    par2signalSF.setConstant(1);
    par3signalSF.setConstant(1);
  }
  
  RooSimultaneous simPdf("simPdf","simultaneous pdf",sample) ;
  simPdf.addPdf(model_SF,"SF") ;
  simPdf.addPdf(model_OF,"OF") ;

  
  RooFitResult *result = simPdf.fitTo(combData, RooFit::Save(), RooFit::Extended(),RooFit::Minos(true));
  
  if(result->covQual()!=3) {
    write_error(__FUNCTION__,"Full fit did not converge!!! Cannot continue!");
    dout << "covQual is " << result->covQual() << endl;
    if(EdgeFitter::AllowTriangle) EdgeFitter::FixedMEdgeChi2_H1=-1;
    else EdgeFitter::FixedMEdgeChi2_H0=-1;
    if(EdgeFitter::RejectPointIfNoConvergence) return;
  } else {
    write_info(__FUNCTION__,"Full fit converged");
  }

  dout << "============================= < Full results > =============================" << endl;
  result->Print();
  dout << "============================= < /Full results > =============================" << endl;

  
  RooPlot* frame1 = mll.frame(RooFit::Bins(int((mllmax-mllmin)/5.0)),RooFit::Title("EE sample")) ;
  frame1->GetXaxis()->CenterTitle(1);
  frame1->GetYaxis()->CenterTitle(1);
  combData.plotOn(frame1,RooFit::Name("SFdata"),RooFit::Cut("sample==sample::SF")) ;
  simPdf.plotOn(frame1,RooFit::Slice(sample,"SF"),RooFit::Name("FullFit"),RooFit::ProjWData(sample,combData), RooFit::LineColor(kBlack)) ;
  simPdf.plotOn(frame1,RooFit::Slice(sample,"SF"),RooFit::Name("TTbarSFonly"),RooFit::Components("ttbarSF"),RooFit::ProjWData(sample,combData),RooFit::LineStyle(kDashed)) ;
  simPdf.plotOn(frame1,RooFit::Slice(sample,"SF"),RooFit::Name("DYSFonly"),RooFit::Components("zSF"), RooFit::ProjWData(sample, combData), RooFit::LineStyle(kDashed), RooFit::LineColor(kRed));
  if(EdgeFitter::AllowTriangle) simPdf.plotOn(frame1,RooFit::Slice(sample,"SF"),RooFit::Name("SignalSFonly"),RooFit::Components("signalSF"), RooFit::ProjWData(sample, combData), RooFit::LineStyle(kDashed), RooFit::LineColor(kGreen));  
  
  if(EdgeFitter::AllowTriangle) EdgeFitter::FixedMEdgeChi2_H1=frame1->chiSquare("FullFit", "SFdata", 3);
  else EdgeFitter::FixedMEdgeChi2_H0=frame1->chiSquare("FullFit", "SFdata", 3);

  dout << "Result   : " << endl;
  if(EdgeFitter::AllowTriangle) dout << "f signal : " << fsignalSF.getVal() << " +/- " << fsignalSF.getError() << endl;
//  dout << "f ttbar  : " << fttbarSF.getVal() << " +/- " << fttbarSF.getError() << endl;
  dout << "f ttbar  : " << fttbarSF.getVal() << " +/- NO ERROR CUZ ITS A PDF "<< endl;
  dout << "f tt OF  : " << fttbarOF.getVal() << " +/- " << fttbarOF.getError() << endl;
  dout << "f z SF   : " << fzSF.getVal() << " +/- " << fzSF.getError() << endl;
  if(EdgeFitter::AllowTriangle) dout << "#Chi^{2}/NDF     : " << EdgeFitter::FixedMEdgeChi2_H1 << endl;
  else dout << "#Chi^{2}/NDF     : " << EdgeFitter::FixedMEdgeChi2_H0 << endl;
  
  // The same plot for the cointrol sample slice
  RooPlot* frame3 = mll.frame(RooFit::Bins(int((mllmax-mllmin)/5.0)),RooFit::Title("OF sample")) ;
  frame3->GetXaxis()->CenterTitle(1);
  frame3->GetYaxis()->CenterTitle(1);
  frame3->SetMaximum(frame1->GetMaximum());
  combData.plotOn(frame3,RooFit::Cut("sample==sample::OF")) ;
  simPdf.plotOn(frame3,RooFit::Slice(sample,"OF"),RooFit::ProjWData(sample,combData), RooFit::LineColor(kBlack)) ;
  simPdf.plotOn(frame3,RooFit::Slice(sample,"OF"),RooFit::Components("ttbarOF"),RooFit::ProjWData(sample,combData),RooFit::LineStyle(kDashed)) ;
  
  
  TCanvas* c = new TCanvas("rf501_simultaneouspdf","rf403_simultaneouspdf") ;
  c->cd() ;
  gPad->SetLeftMargin(0.15);
  frame1->GetYaxis()->SetTitleOffset(1.4);
  frame1->Draw();
  if(is_data==data) DrawPrelim();
  else DrawPrelim(PlottingSetup::luminosity,true);
  stringstream infotext;
  infotext << "#splitline{Fit results (" << EdgeFitter::Mode << ">" << jzb_cut << "): }{#splitline{";
  infotext << "N(Data) = " << EdgeFitter::SFSample->sumEntries() << "}{#splitline{";
  infotext << "N(Z+Jets) = " << WriteWithError(fzSF.getVal(),fzSF.getError(),3) << "}{#splitline{";
  //infotext << "N(t#bar{t}) = " << WriteWithError(fttbarSF.getVal(),fttbarSF.getError(),3) << "}{#splitline{";
  infotext << "N(t#bar{t}) = " << WriteWithError(fttbarSF.getVal(),0,3) << "}{#splitline{";
  write_warning(any2string(__LINE__),"Don't have the error yet, need to complete this");
  if(EdgeFitter::AllowTriangle) {
    infotext << "N(signal) = " << WriteWithError(fsignalSF.getVal(),fsignalSF.getError(),3) << "}{";
    infotext << "m_{edge} = " << WriteWithError(par3signalSF.getVal(),par3signalSF.getError(),3) << "}}}}}";  
  } else infotext << "}{}}}}}";  

  TLatex *infobox = new TLatex(0.57,0.75,infotext.str().c_str());
  infobox->SetNDC();
  infobox->SetTextSize(0.03);
  infobox->Draw();
  if(EdgeFitter::FixedMEdge>=0) CompleteSave(c,"Edge/"+prefix.str()+"_SF__MEdgeFix_"+any2string(EdgeFitter::FixedMEdge),false,false);
  else CompleteSave(c,"Edge/"+prefix.str()+"_SF",false,false);
  delete c;
  
  TCanvas* e = new TCanvas("rf501_simultaneouspdfem","rf403_simultaneouspdfem") ;
  e->cd();
  gPad->SetLeftMargin(0.15);
  frame3->GetYaxis()->SetTitleOffset(1.4);
  frame3->Draw();
  if(is_data==data) DrawPrelim();
  else DrawPrelim(PlottingSetup::luminosity,true);
  if(EdgeFitter::FixedMEdge>=0) CompleteSave(e,"Edge/"+prefix.str()+"_OF__MEdgeFix_"+any2string(EdgeFitter::FixedMEdge),false,false);
  else CompleteSave(e,"Edge/"+prefix.str()+"_OF",false,false);
  delete e;
  
  
/*  TCanvas* f = new TCanvas("rf501_simultaneouspdfem","rf403_simultaneouspdfem") ;
  f->cd();
  gPad->SetLeftMargin(0.15);
  frame4->GetYaxis()->SetTitleOffset(1.4);
  frame4->Draw();
  if(is_data==data) DrawPrelim();
  else DrawPrelim(PlottingSetup::luminosity,true);
  CompleteSave(f,"Edge/"+prefix.str()+"_SF");
  delete f;*/
 

/*
  float maxZ=200;
  RooWorkspace* wspace = new RooWorkspace();
  stringstream mllvar;
  mllvar << "mll[" << (mllmax-mllmin)/2 << "," << mllmin << "," << mllmax << "]";
  wspace->factory(mllvar.str().c_str());
  wspace->var("mll")->setBins(30);
  wspace->factory("nSig[1.,0.,100.]");
  wspace->factory(("nZ[0.04.,0.,"+any2string(maxZ)+"]").c_str());
  wspace->factory("rME[1.12,1.05,1.19]");
  wspace->factory("effUncert[1.]");
  EdgeFitter::prepareLimits(wspace, true);
*/

write_warning(__FUNCTION__," A lot missing here to calculate limits");

}

void EdgeFitter::DoEdgeFit(string mcjzb, string datajzb, float DataPeakError, float MCPeakError, float jzb_cut, int icut, int is_data, TCut cut, TTree *signalevents=0) {
  
  TCut _cut(cut&&PlottingSetup::basiccut&&PlottingSetup::passtrig);
  
  TFile *f = new TFile("workingfile.root","RECREATE");

  EdgeFitter::InitializeVariables(PlottingSetup::iMllLow,PlottingSetup::iMllHigh,PlottingSetup::jzbHigh,_cut);
  
  EdgeFitter::PrepareDatasets(is_data);
  
  EdgeFitter::DrawDatasetContent(is_data);
  
  RooFit::MsgLevel msglevel = RooMsgService::instance().globalKillBelow();
  RooMsgService::instance().setGlobalKillBelow(RooFit::FATAL);
  
  
  EdgeFitter::AllowTriangle=false;
  EdgeFitter::DoFit(is_data, jzb_cut);
  
  write_info(__FUNCTION__,"TAKING SHORTCUT");return;
  
  EdgeFitter::AllowTriangle=true;
  
  bool ScanMassRange=false;
  float ScanSteps=5.0;//GeV
  
  
  if(ScanMassRange) {
    TFile *fscan = new TFile("fscan.root","UPDATE");
    TGraph *gr = new TGraph();
    TGraph *Rgr = new TGraph();
    stringstream GrName;
    GrName << "ScanGraphFor_" << EdgeFitter::Mode << "_" << jzb_cut;
    stringstream RGrName;
    RGrName << "ScanRatioGraphFor_" << EdgeFitter::Mode << "_" << jzb_cut;
    gr->SetName(GrName.str().c_str());
    Rgr->SetName(RGrName.str().c_str());

    int i=0;
    for(float tempMedge=10;tempMedge<=300;tempMedge+=ScanSteps) {
      write_info(__FUNCTION__,"Now testing Medge="+any2string(tempMedge)+" for "+EdgeFitter::Mode+">"+any2string(jzb_cut));
      EdgeFitter::FixedMEdge=tempMedge;
      EdgeFitter::DoFit(is_data, jzb_cut);
      if(EdgeFitter::FixedMEdgeChi2_H1<0) continue;
      gr->SetPoint(i,tempMedge,EdgeFitter::FixedMEdgeChi2_H1);
      Rgr->SetPoint(i,tempMedge,EdgeFitter::FixedMEdgeChi2_H1/EdgeFitter::FixedMEdgeChi2_H0);
      i++;
    }
    
    TCanvas *ScanCan = new TCanvas("ScanCan","ScanCan",500,500);
    gr->GetXaxis()->SetTitle("m_{edge}");
    gr->GetXaxis()->CenterTitle();
    gr->GetYaxis()->SetTitle("#Chi^{2} / NDF");
    gr->GetYaxis()->CenterTitle();
    gr->GetYaxis()->SetTitleOffset(0.95);
    gr->GetXaxis()->SetTitleOffset(0.9);
    gr->SetLineColor(kBlue);
    gr->SetTitle("");
    gr->Draw("AL");
    stringstream ScanCanSave;
    ScanCanSave << "Edge/MEdgeScan_"+EdgeFitter::Mode+"_" << jzb_cut;
    if(is_data) DrawPrelim();
    else DrawMCPrelim();
    CompleteSave(ScanCan,ScanCanSave.str());

    Rgr->GetXaxis()->SetTitle("m_{edge}");
    Rgr->GetXaxis()->CenterTitle();
    Rgr->GetYaxis()->SetTitle("#Chi^{2} / NDF");
    Rgr->GetYaxis()->CenterTitle();
    Rgr->GetYaxis()->SetTitleOffset(0.95);
    Rgr->GetXaxis()->SetTitleOffset(0.9);
    Rgr->SetLineColor(kBlue);
    Rgr->SetTitle("");
    Rgr->Draw("AL");
    ScanCanSave.str("");
    ScanCanSave << "Edge/MEdgeScan_Ratio_"+EdgeFitter::Mode+"_" << jzb_cut;
    if(is_data) DrawPrelim();
    else DrawMCPrelim();
    CompleteSave(ScanCan,ScanCanSave.str());
    fscan->cd();
    gr->Write();
    delete ScanCan;
    fscan->Close();
  } else {
    EdgeFitter::DoFit(is_data, jzb_cut);
    dout << "Chi^2 (H0) = " << EdgeFitter::FixedMEdgeChi2_H0 << endl;
    dout << "Chi^2 (H1) = " << EdgeFitter::FixedMEdgeChi2_H1 << endl;
  }
  
  
  RooMsgService::instance().setGlobalKillBelow(msglevel);
 
  f->Close();

}

void DoEdgeFit(string mcjzb, string datajzb, float DataPeakError, float MCPeakError, vector<float> jzb_cut, int is_data, TCut cut, TTree *signalevents=0) {
  
  EdgeFitter::Mode="JZB";
  if(mcjzb=="met[4]") EdgeFitter::Mode="MET";
  
  for(int icut=0;icut<(int)jzb_cut.size();icut++) {
    stringstream addcut;
    if(is_data==1) addcut << "(" << datajzb << ">" << jzb_cut[icut] << ")";
    if(is_data!=1) addcut << "(" << mcjzb << ">" << jzb_cut[icut] << ")";
    TCut jcut(addcut.str().c_str());
    
    
    EdgeFitter::DoEdgeFit(mcjzb, datajzb, DataPeakError, MCPeakError, jzb_cut[icut], icut, is_data, jcut&&cut, signalevents);
    
  }
}
Revision:	1.23
Committed:	Thu Jun 20 13:16:43 2013 UTC (11 years, 10 months ago) by buchmann
Content type:	text/plain
Branch:	MAIN
Changes since 1.22:	+38 -22 lines
Log Message:	Added edgeCache: storing passed trees for edge fit in a special cache according to the cuts used; this helps speeding up the process tremendously