TH1 *
YieldMean_HighExtrapolationHisto(TH1 *h, TF1 *f, Double_t max, Double_t binwidth)
{
/* find highest edge in histo */
Int_t binhi;
Double_t hi;
for (Int_t ibin = h->GetNbinsX(); ibin > 0; ibin--) {
if (h->GetBinContent(ibin) != 0.) {
binhi = ibin + 1;
hi = h->GetBinLowEdge(ibin + 1);
break;
}
}
if(max<hi) {
Printf("Warning! You should probably set a higher max value (Max = %f, hi = %f)", max, hi);
return 0x0;
}
Int_t nbins = (max - hi) / binwidth;
if(nbins<1)
return 0x0;
TH1 *hhi = new TH1F("hhi", "", nbins, hi, max);
/* integrate function in histogram bins */
Double_t cont, err, width;
for (Int_t ibin = 0; ibin < hhi->GetNbinsX(); ibin++) {
width = hhi->GetBinWidth(ibin + 1);
cont = f->Integral(hhi->GetBinLowEdge(ibin + 1), hhi->GetBinLowEdge(ibin + 2), (Double_t *)0, 1.e-6);
err = f->IntegralError(hhi->GetBinLowEdge(ibin + 1), hhi->GetBinLowEdge(ibin + 2), (Double_t *)0, (Double_t *)0, 1.e-6);
hhi->SetBinContent(ibin + 1, cont / width);
hhi->SetBinError(ibin + 1, err / width);
}
return hhi;
}
TH1 *
YieldMean_LowExtrapolationHisto(TH1 *h, TF1 *f, Double_t min, Double_t binwidth)
{
/* find lowest edge in histo */
Int_t binlo;
Double_t lo;
for (Int_t ibin = 1; ibin < h->GetNbinsX() + 1; ibin++) {
if (h->GetBinContent(ibin) != 0.) {
binlo = ibin;
lo = h->GetBinLowEdge(ibin);
break;
}
}
Int_t nbins = (lo - min) / binwidth;
if(nbins<1)
return 0x0;
TH1 *hlo = new TH1F("hlo", "", nbins, min, lo);
/* integrate function in histogram bins */
Double_t cont, err, width;
for (Int_t ibin = 0; ibin < hlo->GetNbinsX(); ibin++) {
width = hlo->GetBinWidth(ibin + 1);
cont = f->Integral(hlo->GetBinLowEdge(ibin + 1), hlo->GetBinLowEdge(ibin + 2), (Double_t *)0, 1.e-6);
err = f->IntegralError(hlo->GetBinLowEdge(ibin + 1), hlo->GetBinLowEdge(ibin + 2), (Double_t *)0, (Double_t *)0, 1.e-6);
hlo->SetBinContent(ibin + 1, cont / width);
hlo->SetBinError(ibin + 1, err / width);
}
return hlo;
}
Double_t fitfulllang( char* hs ) {
TH1 *h = (TH1*)gDirectory->Get(hs);
if( h == NULL ){
cout << hs << " does not exist\n";
return 0;
}
double aa = h->GetEntries();//normalization
// find peak:
int ipk = h->GetMaximumBin();
double xpk = h->GetBinCenter(ipk);
double sm = xpk / 9; // sigma
double ns = sm; // noise
// fit range:
int ib0 = ipk/2;
int ib9 = h->GetNbinsX() - 1;
double x0 = h->GetBinLowEdge(ib0);
double x9 = h->GetBinLowEdge(ib9) + h->GetBinWidth(ib9);
// create a TF1 with the range from x0 to x9 and 4 parameters
TF1 *fitFcn = new TF1( "fitFcn", fitLandauGauss, x0, x9, 4 );
fitFcn->SetParName( 0, "peak" );
fitFcn->SetParName( 1, "sigma" );
fitFcn->SetParName( 2, "area" );
fitFcn->SetParName( 3, "smear" );
fitFcn->SetNpx(500);
fitFcn->SetLineWidth(4);
fitFcn->SetLineColor(kMagenta);
// set start values:
fitFcn->SetParameter( 0, xpk ); // peak position, defined above
fitFcn->SetParameter( 1, sm ); // width
fitFcn->SetParameter( 2, aa ); // area
fitFcn->SetParameter( 3, ns ); // noise
h->Fit("fitFcn", "NQR", "ep" );// R = range from fitFcn
return fitFcn->GetParameter(0);
}
TH1 *
UnfoldMe_MB2(const Char_t *data, const Char_t *mc, const Char_t *anatag, Int_t bin, Bool_t useMBcorr , Bool_t usecorrfit , Bool_t ismc , Float_t smooth , Int_t iter , Int_t regul , Float_t weight , Bool_t bayesian , Int_t nloop )
{
// MF comments:
// usedMBcorr: changes the matrix used for unfonding, from effMatrix to bin matrix (I think this is just to use mult dependent v s mb correction_)
// usecorrfit: if I understand correctly, fits the response matrix and uses fit to extrapolate it
TFile *fdt =0;
if (ismc)
fdt = TFile::Open(data);
else
fdt = TFile::Open(data);
TFile *fmc = TFile::Open(mc);
TList *ldt = (TList *)fdt->Get(Form("%s", anatag));
TList *lmc = (TList *)fmc->Get(Form("%s", anatag));
TH2 *hmatdt = (TH2 *)ldt->FindObject(Form(responseMatrix, bin));
TH2 *hmatmc = 0;
if (useMBcorr){
hmatmc = (TH2 *)lmc->FindObject("effMatrix");
std::cout << "USING MB" << std::endl;
}
else {
hmatmc = (TH2 *)lmc->FindObject(Form(responseMatrix, bin));
}
TH1 *hdata = hmatdt->ProjectionY("hdata");
// TH1 *hdata = hmatdt->ProjectionY("htrue"); // For truth Only Calculations
hdata->Sumw2();
hdata->SetBinContent(1, 0.);
hdata->SetBinError(1, 0.);
// hdata->Scale(1. / hdata->Integral());
hdata->SetMarkerStyle(25);
TH1 *htrue = hmatdt->ProjectionX("htrue");
htrue->Sumw2();
// htrue->Scale(1. / htrue->Integral());
htrue->SetMarkerStyle(7);
htrue->SetMarkerColor(2);
htrue->SetBinContent(1, 0.);
htrue->SetBinError(1, 0.);
TH2 *hcorr = (TH2 *)hmatmc->Clone("hcorr");
TH1 *hinit = (TH1 *)hdata->Clone("hinit");
TH1 *hresu = (TH1 *)hdata->Clone("hresu");
TH1 *hbias = (TH1 *)hdata->Clone("hbias");
hresu->SetMarkerStyle(20);
hresu->SetMarkerColor(4);
hresu->Reset();
TH1 *hnum = hcorr->ProjectionY("hnum");
TH1 *hden = hcorr->ProjectionY("hden");
TH1 *heff = hcorr->ProjectionY("heff");
hnum->Reset();
hnum->Sumw2();
hden->Reset();
hden->Sumw2();
heff->Reset();
for (Int_t i = 0; i < heff->GetNbinsX(); i++) {
Float_t int1 = hcorr->Integral(i + 1, i + 1, 0, -1);
if (int1 <= 0.) continue;
Float_t int2 = hcorr->Integral(i + 1, i + 1, 2, -1);
hnum->SetBinContent(i + 1, int2);
hnum->SetBinError(i + 1, TMath::Sqrt(int2));
hden->SetBinContent(i + 1, int1);
hden->SetBinError(i + 1, TMath::Sqrt(int1));
}
TCanvas *cEfficiency = new TCanvas("cEfficiency", "cEfficiency");
cEfficiency->SetLogx();
cEfficiency->SetLogy();
heff->Divide(hnum, hden, 1., 1., "B");
heff->Draw();
#if 0
for (Int_t ii = 0; ii < heff->GetNbinsX(); ii++) {
heff->SetBinContent(ii + 1, 1.);
heff->SetBinError(ii + 1, 0.);
}
#endif
for (Int_t i = 0; i < hcorr->GetNbinsX(); i++) {
hcorr->SetBinContent(i + 1, 1, 0.);
hcorr->SetBinError(i + 1, 1, 0.);
}
for (Int_t i = 0; i < hcorr->GetNbinsY(); i++) {
hcorr->SetBinContent(1, i + 1, 0.);
hcorr->SetBinError(1, i + 1, 0.);
}
TH2 *hcorrfit = ReturnCorrFromFit(hcorr);
// Docs from AliUnfolding
//Int_t AliUnfolding::Unfold(TH2* correlation, TH1* efficiency, TH1* measured, TH1* initialConditions, TH1* result, Bool_t check)
// unfolds with unfolding method fgMethodType
//
// parameters:
// correlation: response matrix as measured vs. generated
// efficiency: (optional) efficiency that is applied on the unfolded spectrum, i.e. it has to be in unfolded variables. If 0 no efficiency is applied.
// measured: the measured spectrum
// initialConditions: (optional) initial conditions for the unfolding. if 0 the measured spectrum is used as initial conditions.
// result: target for the unfolded result
// check: depends on the unfolding method, see comments in specific functions
for (Int_t iloop = 0; iloop < nloop; iloop++) {
if (bayesian) {
AliUnfolding::SetUnfoldingMethod(AliUnfolding::kBayesian);
AliUnfolding::SetBayesianParameters(smooth, iter);
} else {
AliUnfolding::SetUnfoldingMethod(AliUnfolding::kChi2Minimization);
AliUnfolding::SetChi2Regularization(AliUnfolding::RegularizationType(regul), weight);
}
AliUnfolding::SetSkip0BinInChi2(kTRUE);
AliUnfolding::SetSkipBinsBegin(1);
AliUnfolding::SetNbins(150, 150);
AliUnfolding::Unfold(usecorrfit ? hcorrfit : hcorr, heff, hdata, hinit, hresu);
hinit = (TH1 *)hresu->Clone(Form("hinit_%d", iloop));
}
printf("hdata->Integral(2, -1) = %f\n", hdata->Integral(2, -1));
printf("hresu->Integral(2, -1) = %f\n", hresu->Integral(2, -1));
TCanvas *cUnfolded = new TCanvas ("cUnfolded", "cUnfolded", 400, 800);
cUnfolded->Divide(1, 2);
cUnfolded->cd(1)->SetLogx();
cUnfolded->cd(1)->SetLogy();
hdata->Draw();
hresu->Draw("same");
htrue->Draw("same");
cUnfolded->cd(2)->SetLogx();
cUnfolded->cd(2)->DrawFrame(1., 0, 300., 10);
TH1 *hrat = (TH1 *)hresu->Clone("hrat");
hrat->Divide(htrue);
hrat->Draw("same");
TH1 *htrig = (TH1 *)hresu->Clone("htrig");
htrig->Multiply(heff);
Float_t dndeta_resu = 0.;
Float_t integr_resu = 0.;
Float_t dndeta_trig = 0.;
Float_t integr_trig = 0.;
for (Int_t i = 1; i < hresu->GetNbinsX(); i++) {
dndeta_resu += hresu->GetBinContent(i + 1) * hresu->GetBinLowEdge(i + 1);
integr_resu += hresu->GetBinContent(i + 1);
dndeta_trig += htrig->GetBinContent(i + 1) * htrig->GetBinLowEdge(i + 1);
integr_trig += htrig->GetBinContent(i + 1);
}
cUnfolded->SaveAs("unfold_efficiency.pdf");
integr_eff = integr_trig / integr_resu;
integr_eff_err = TMath::Sqrt(integr_eff * (1. - integr_eff) / integr_resu);
dndeta_eff = dndeta_trig / dndeta_resu;
dndeta_eff_err = TMath::Sqrt(dndeta_eff * (1. - dndeta_eff) / dndeta_resu);
printf("INEL > 0 efficiency: %.3f +- %.3f\n", integr_eff, integr_eff_err);
printf("dN/dEta correction: %.3f +- %.3f\n", dndeta_eff, dndeta_eff_err);
return hresu;
}
TH1 *
UnfoldMe(Char_t *data, Char_t *mc, Char_t *anatag, Int_t bin, Bool_t useMBcorr = kTRUE, Bool_t usecorrfit = kFALSE, Bool_t ismc = kFALSE, Float_t smooth = 0.001, Int_t iter = 50, Int_t regul = AliUnfolding::kPowerLaw, Float_t weight = 100., Bool_t bayesian = kTRUE, Int_t nloop = 1)
{
if (ismc)
TFile *fdt = TFile::Open(data);
else
TFile *fdt = TFile::Open(data);
TFile *fmc = TFile::Open(mc);
TList *ldt = (TList *)fdt->Get(Form("clist_%s", anatag));
TList *lmc = (TList *)fmc->Get(Form("clist_%s", anatag));
TH2 *hmatdt = (TH2 *)ldt->FindObject(Form("b%d_corrMatrix", bin));
if (useMBcorr)
TH2 *hmatmc = (TH2 *)lmc->FindObject("effMatrix");
else
TH2 *hmatmc = (TH2 *)lmc->FindObject(Form("b%d_corrMatrix", bin));
TH1 *hdata = hmatdt->ProjectionY("hdata");
hdata->Sumw2();
hdata->SetBinContent(1, 0.);
hdata->SetBinError(1, 0.);
// hdata->Scale(1. / hdata->Integral());
hdata->SetMarkerStyle(25);
TH1 *htrue = hmatdt->ProjectionX("htrue");
htrue->Sumw2();
// htrue->Scale(1. / htrue->Integral());
htrue->SetMarkerStyle(7);
htrue->SetMarkerColor(2);
htrue->SetBinContent(1, 0.);
htrue->SetBinError(1, 0.);
TH2 *hcorr = (TH2 *)hmatmc->Clone("hcorr");
TH1 *hinit = (TH1 *)hdata->Clone("hinit");
TH1 *hresu = (TH1 *)hdata->Clone("hresu");
TH1 *hbias = (TH1 *)hdata->Clone("hbias");
hresu->SetMarkerStyle(20);
hresu->SetMarkerColor(4);
hresu->Reset();
TH1 *hnum = hcorr->ProjectionY("hnum");
TH1 *hden = hcorr->ProjectionY("hden");
TH1 *heff = hcorr->ProjectionY("heff");
hnum->Reset();
hnum->Sumw2();
hden->Reset();
hden->Sumw2();
heff->Reset();
for (Int_t i = 0; i < heff->GetNbinsX(); i++) {
Float_t int1 = hcorr->Integral(i + 1, i + 1, 0, -1);
if (int1 <= 0.) continue;
Float_t int2 = hcorr->Integral(i + 1, i + 1, 2, -1);
hnum->SetBinContent(i + 1, int2);
hnum->SetBinError(i + 1, TMath::Sqrt(int2));
hden->SetBinContent(i + 1, int1);
hden->SetBinError(i + 1, TMath::Sqrt(int1));
}
new TCanvas("cEfficiency");
heff->Divide(hnum, hden, 1., 1., "B");
heff->Draw();
#if 0
for (Int_t ii = 0; ii < heff->GetNbinsX(); ii++) {
heff->SetBinContent(ii + 1, 1.);
heff->SetBinError(ii + 1, 0.);
}
#endif
for (Int_t i = 0; i < hcorr->GetNbinsX(); i++) {
hcorr->SetBinContent(i + 1, 1, 0.);
hcorr->SetBinError(i + 1, 1, 0.);
}
for (Int_t i = 0; i < hcorr->GetNbinsY(); i++) {
hcorr->SetBinContent(1, i + 1, 0.);
hcorr->SetBinError(1, i + 1, 0.);
}
TH2 *hcorrfit = ReturnCorrFromFit(hcorr);
for (Int_t iloop = 0; iloop < nloop; iloop++) {
if (bayesian) {
AliUnfolding::SetUnfoldingMethod(AliUnfolding::kBayesian);
AliUnfolding::SetBayesianParameters(smooth, iter);
} else {
AliUnfolding::SetUnfoldingMethod(AliUnfolding::kChi2Minimization);
AliUnfolding::SetChi2Regularization(regul, weight);
}
AliUnfolding::SetSkip0BinInChi2(kTRUE);
AliUnfolding::SetSkipBinsBegin(1);
AliUnfolding::SetNbins(150, 150);
AliUnfolding::Unfold(usecorrfit ? hcorrfit : hcorr, heff, hdata, hinit, hresu);
hinit = (TH1 *)hresu->Clone(Form("hinit_%d", iloop));
}
printf("hdata->Integral(2, -1) = %f\n", hdata->Integral(2, -1));
printf("hresu->Integral(2, -1) = %f\n", hresu->Integral(2, -1));
TCanvas *cUnfolded = new TCanvas ("cUnfolded", "", 400, 800);
cUnfolded->Divide(1, 2);
cUnfolded->cd(1)->SetLogx();
cUnfolded->cd(1)->SetLogy();
hdata->Draw();
hresu->Draw("same");
htrue->Draw("same");
cUnfolded->cd(2)->SetLogx();
cUnfolded->cd(2)->DrawFrame(1., 0.75, 300., 1.25);
TH1 *hrat = (TH1 *)hresu->Clone("hrat");
hrat->Divide(htrue);
hrat->Draw("same");
TH1 *htrig = (TH1 *)hresu->Clone("htrig");
htrig->Multiply(heff);
Float_t dndeta_resu = 0.;
Float_t integr_resu = 0.;
Float_t dndeta_trig = 0.;
Float_t integr_trig = 0.;
for (Int_t i = 1; i < hresu->GetNbinsX(); i++) {
dndeta_resu += hresu->GetBinContent(i + 1) * hresu->GetBinLowEdge(i + 1);
integr_resu += hresu->GetBinContent(i + 1);
dndeta_trig += htrig->GetBinContent(i + 1) * htrig->GetBinLowEdge(i + 1);
integr_trig += htrig->GetBinContent(i + 1);
}
// dndeta_resu /= integr_resu;
// dndeta_trig /= integr_trig;
integr_eff = integr_trig / integr_resu;
integr_eff_err = TMath::Sqrt(integr_eff * (1. - integr_eff) / integr_resu);
dndeta_eff = dndeta_trig / dndeta_resu;
dndeta_eff_err = TMath::Sqrt(dndeta_eff * (1. - dndeta_eff) / dndeta_resu);
printf("INEL > 0 efficiency: %.3f +- %.3f\n", integr_eff, integr_eff_err);
printf("dN/dEta correction: %.3f +- %.3f\n", dndeta_eff, dndeta_eff_err);
return hresu;
}
void Fit(TString SIGNAL,TString HISTO,double scaleSGN)
{
gROOT->ForceStyle();
TFile *fDat = TFile::Open("Histo_flatTree_data_tmva"+SIGNAL+".root");
TFile *fBkg = TFile::Open("Histo_flatTree_qcd_weights_tmva"+SIGNAL+".root");
TFile *fSgn = TFile::Open("Histo_flatTree_"+SIGNAL+"_weights_tmva"+SIGNAL+".root");
TH1 *hDat = (TH1*)fDat->Get(HISTO);
TH1 *hBkgRaw = (TH1*)fBkg->Get(HISTO);
TH1 *hSgn = (TH1*)fSgn->Get(HISTO);
TH1 *hDat_JESlo = (TH1*)fDat->Get(HISTO+"_JESlo");
TH1 *hBkgRaw_JESlo = (TH1*)fBkg->Get(HISTO+"_JESlo");
TH1 *hSgn_JESlo = (TH1*)fSgn->Get(HISTO+"_JESlo");
TH1 *hDat_JESup = (TH1*)fDat->Get(HISTO+"_JESup");
TH1 *hBkgRaw_JESup = (TH1*)fBkg->Get(HISTO+"_JESup");
TH1 *hSgn_JESup = (TH1*)fSgn->Get(HISTO+"_JESup");
TH1F *hBkg = (TH1F*)hBkgRaw->Clone("Bkg");
TH1F *hBkg_JESlo = (TH1F*)hBkgRaw_JESlo->Clone("Bkg_JESlo");
TH1F *hBkg_JESup = (TH1F*)hBkgRaw_JESup->Clone("Bkg_JESup");
hBkg->Smooth(2);
hBkg_JESlo->Smooth(2);
hBkg_JESup->Smooth(2);
hSgn->Smooth(2);
hSgn_JESlo->Smooth(2);
hSgn_JESup->Smooth(2);
double lumi = 4967;
hBkg->Scale(lumi);
hBkg_JESlo->Scale(lumi);
hBkg_JESup->Scale(lumi);
double k_factor = hDat->Integral()/hBkg->Integral();
double k_factor_JESlo = hDat->Integral()/hBkg_JESlo->Integral();
double k_factor_JESup = hDat->Integral()/hBkg_JESup->Integral();
hBkg->Scale(k_factor);
cout<<"Signal entries = "<<hSgn->GetEntries()<<endl;
hSgn->Scale(lumi/scaleSGN);
hBkg_JESlo->Scale(k_factor_JESlo);
hSgn_JESlo->Scale(lumi/scaleSGN);
hBkg_JESup->Scale(k_factor_JESup);
hSgn_JESup->Scale(lumi/scaleSGN);
hSgn_JESlo->Scale(hSgn->Integral()/hSgn_JESlo->Integral());
hSgn_JESup->Scale(hSgn->Integral()/hSgn_JESup->Integral());
TH1 *hBkg_STATlo = (TH1*)hBkg->Clone(HISTO+"_STATlo");
TH1 *hSgn_STATlo = (TH1*)hSgn->Clone(HISTO+"_STATlo");
TH1 *hBkg_STATup = (TH1*)hBkg->Clone(HISTO+"_STATup");
TH1 *hSgn_STATup = (TH1*)hSgn->Clone(HISTO+"_STATup");
float y1,e1;
for(int i=0;i<hBkg->GetNbinsX();i++) {
y1 = hBkg->GetBinContent(i+1);
e1 = hBkg->GetBinError(i+1);
hBkg_STATlo->SetBinContent(i+1,y1-e1);
hBkg_STATup->SetBinContent(i+1,y1+e1);
y1 = hSgn->GetBinContent(i+1);
e1 = hSgn->GetBinError(i+1);
hSgn_STATlo->SetBinContent(i+1,y1-e1);
hSgn_STATup->SetBinContent(i+1,y1+e1);
}
hBkg_STATlo->Scale(hBkg->Integral()/hBkg_STATlo->Integral());
hBkg_STATup->Scale(hBkg->Integral()/hBkg_STATup->Integral());
hSgn_STATlo->Scale(hSgn->Integral()/hSgn_STATlo->Integral());
hSgn_STATup->Scale(hSgn->Integral()/hSgn_STATup->Integral());
double xMIN = hBkg->GetBinLowEdge(1);
double xMAX = hBkg->GetBinLowEdge(hBkg->GetNbinsX()+1);
double xMIN2 = hDat->GetBinLowEdge(hDat->FindFirstBinAbove(0.5));
double xMAX2 = hDat->GetBinLowEdge(hDat->FindLastBinAbove(0.5)+1);
RooRealVar x("x","x",xMIN2,xMAX2);
RooDataHist data("data","dataset with x",x,hDat);
RooDataHist bkg("qcd","bkg with x",x,hBkg);
RooDataHist sgn("signal","sgn with x",x,hSgn);
RooHistPdf bkgPDF("bkgPDF","bkgPDF",x,bkg,0);
RooHistPdf sgnPDF("sgnPDF","sgnPDF",x,sgn,0);
RooRealVar f("f","f",0,0.,1.);
RooAddPdf model("model","model",RooArgList(sgnPDF,bkgPDF),RooArgList(f));
RooFitResult* r = model.fitTo(data,Save());
r->Print("v");
double N = hDat->Integral();
double B = hBkg->Integral();
double S = hSgn->Integral();
double m = f.getVal();
double e = f.getError();
cout<<"k-factor = "<<k_factor<<endl;
cout<<N<<" "<<B<<" "<<S<<endl;
cout<<"Total cross section = "<<N/lumi<<" pb"<<endl;
cout<<"Model cross section = "<<S/lumi<<" pb"<<endl;
cout<<"Fitted signal strength = "<<m*N/S<<endl;
cout<<"Fitted signal error = "<<e*N/S<<endl;
double p = 0.95;
double xup = (N/S)*(m+sqrt(2.)*e*TMath::ErfInverse((1-p)*TMath::Erf(m/e)+p));
cout<<"Bayesian Upper limit = "<<xup<<endl;
RooPlot* frame1 = x.frame();
data.plotOn(frame1);
model.plotOn(frame1,Components("sgnPDF*"),LineStyle(1),LineWidth(2),LineColor(kGreen+1));
model.plotOn(frame1,Components("bkgPDF*"),LineStyle(1),LineWidth(2),LineColor(kRed));
model.plotOn(frame1,LineStyle(1),LineWidth(2),LineColor(kBlue));
//cout<<frame1->chiSquare()<<endl;
RooHist* hresid = frame1->residHist();
RooHist* hpull = frame1->pullHist();
RooPlot* frame2 = x.frame(Title("Residual Distribution"));
frame2->addPlotable(hresid,"P") ;
// Create a new frame to draw the pull distribution and add the distribution to the frame
RooPlot* frame3 = x.frame(Title("Pull Distribution"));
frame3->addPlotable(hpull,"P");
TCanvas* cFit = new TCanvas("fitANN_"+SIGNAL,"fitANN_"+SIGNAL,900,600);
gPad->SetLogy();
frame1->SetMaximum(1e+4);
frame1->SetMinimum(0.5);
frame1->GetXaxis()->SetTitle("ANN Output");
frame1->GetYaxis()->SetTitle("Events");
frame1->Draw();
cout<<frame1->nameOf(3)<<endl;
TLegend *leg = new TLegend(0.7,0.65,0.9,0.9);
leg->SetHeader(SIGNAL);
leg->AddEntry(frame1->findObject("h_data"),"data","P");
leg->AddEntry(frame1->findObject("model_Norm[x]"),"QCD+Signal","L");
leg->AddEntry(frame1->findObject("model_Norm[x]_Comp[bkgPDF*]"),"QCD","L");
leg->AddEntry(frame1->findObject("model_Norm[x]_Comp[sgnPDF*]"),"Signal","L");
leg->SetFillColor(0);
leg->SetBorderSize(0);
leg->SetTextFont(42);
leg->SetTextSize(0.04);
leg->Draw();
TCanvas* cPull = new TCanvas("pullANN_"+SIGNAL,"pullANN_"+SIGNAL,900,400);
frame3->GetXaxis()->SetTitle("ANN Output");
frame3->GetYaxis()->SetTitle("Pull");
frame3->Draw();
cout<<"Creating datacard"<<endl;
ofstream datacard;
datacard.open("datacard_"+SIGNAL+"_"+HISTO+".txt");
datacard.setf(ios::right);
datacard<<"imax 1"<<"\n";
datacard<<"jmax 1"<<"\n";
datacard<<"kmax *"<<"\n";
datacard<<"----------------"<<"\n";
datacard<<"shapes * * "<<SIGNAL+"_"+HISTO+"_input.root $PROCESS $PROCESS_$SYSTEMATIC"<<"\n";
datacard<<"----------------"<<"\n";
datacard<<"bin 1"<<"\n";
datacard<<"observation "<<N<<"\n";
datacard<<"----------------"<<"\n";
datacard<<"bin 1 1"<<"\n";
datacard<<"process signal background "<<"\n";
datacard<<"process 0 1"<<"\n";
datacard<<"rate "<<S<<" "<<B<<"\n";
datacard<<"----------------"<<"\n";
datacard<<"lumi lnN 1.022 1.022"<<"\n";
datacard<<"jes shape 1 1"<<"\n";
datacard<<"mcstat shape 1 1"<<"\n";
datacard<<"jer shape 0 0"<<"\n";
datacard.close();
TFile *out = new TFile(SIGNAL+"_"+HISTO+"_input.root","RECREATE");
out->cd();
hDat->Write("data_obs");
hBkg->Write("background");
hSgn->Write("signal");
hDat_JESlo->Write("data_obs_jesDown");
hBkg_JESlo->Write("background_jesDown");
hBkg_STATlo->Write("background_mcstatDown");
hSgn_STATlo->Write("signal_mcstatDown");
hSgn_JESlo->Write("signal_jesDown");
hDat_JESup->Write("data_obs_jesUp");
hBkg_JESup->Write("background_jesUp");
hBkg_STATup->Write("background_mcstatUp");
hSgn_JESup->Write("signal_jesUp");
hSgn_STATup->Write("signal_mcstatUp");
//----- JER placeholder ----------------
hBkg_JESlo->Write("background_jerDown");
hSgn_JESlo->Write("signal_jerDown");
hBkg_JESup->Write("background_jerUp");
hSgn_JESup->Write("signal_jerUp");
}
void correlation_plots2(bool rwt = true, TString catName = "e3je2t"){
int numhists = 0;
// jet and tag cut for each category
TString jettagcut = "";
if (catName == "ge4je2t"){
jettagcut = "numJets>3 && numTaggedJets==2";
numhists = 7;
}
else if(catName == "e3je2t"){
jettagcut = "numJets==3 && numTaggedJets==2";
numhists = 5;
}
else std::cout << "Error, wrong category name" << std::endl;
//IN GLOBAL SCOPE
// const int numhists = 5; // 5;
const int numprofiles = pow(numhists,2);
TProfile** profiles_Data = new TProfile * [numprofiles];
TProfile** profiles_MC = new TProfile * [numprofiles];
//IN MAIN FUNCTION
char** histnames = new char * [numhists];
char** histXaxis = new char * [numhists];
int* histbins = new int[numhists];
double* histmin = new double[numhists];
double* histmax = new double[numhists];
if(catName == "e3je2t"){
histnames[0] = (char*) "sum_pt";
histXaxis[0] = (char*) "p_{T}(l, jets)";
histbins[0] = 13;
histmin[0] = 150;
histmax[0] = 930;
histnames[1] = (char*) "min_dr_jets";
histXaxis[1] = (char*) "minimum #DeltaR(j, j)";
histbins[1] = 12;
histmin[1] = 0.5;
histmax[1] = 3.5;
histnames[2] = (char*) "avg_btag_disc_non_btags";
histXaxis[2] = (char*) "#mu^{CSV}(non b-tags)";
histbins[2] = 10;
histmin[2] = 0.0;
histmax[2] = 0.68;
histnames[3] = (char*) "avg_btag_disc_btags";
histXaxis[3] = (char*) "#mu^{CSV}";
histbins[3] = 10;
histmin[3] = 0.7;
histmax[3] = 1;
histnames[4] = (char*) "BDTG_e3je2t";
histXaxis[4] = (char*) "BDT output";
histbins[4] = 10;
histmin[4] = -1.;
histmax[4] = 1.;
}
else if(catName == "ge4je2t"){
histnames[0] = (char*) "sum_pt";
histXaxis[0] = (char*) "p_{T}(l, jets)";
histbins[0] = 15;
histmin[0] = 200;
histmax[0] = 1400;
histnames[1] = (char*) "min_dr_jets";
histXaxis[1] = (char*) "minimum #DeltaR(j, j)";
histbins[1] = 11;
histmin[1] = 0.5;
histmax[1] = 2.7;
histnames[2] = (char*) "avg_btag_disc_non_btags";
histXaxis[2] = (char*) "#mu^{CSV}(non b-tags)";
histbins[2] = 11;
histmin[2] = 0.0;
histmax[2] = 0.605;
histnames[3] = (char*) "higgsLike_dijet_mass";
histXaxis[3] = (char*) "higgsLike dijet mass";
histbins[3] = 17;
histmin[3] = 34;
histmax[3] = 255;
histnames[4] = (char*) "higgsLike_dijet_mass2";
histXaxis[4] = (char*) "higgsLike dijet mass2";
histbins[4] = 17;
histmin[4] = 30;
histmax[4] = 370;
histnames[5] = (char*) "numJets";
histXaxis[5] = (char*) "N_{jets}";
histbins[5] = 5;
histmin[5] = 4;
histmax[5] = 9;
histnames[6] = (char*) "BDTG_ge4je2t";
histXaxis[6] = (char*) "BDT output";
histbins[6] = 12;
histmin[6] = -0.95;
histmax[6] = 0.85;
}
else std::cout << "Error2, wrong category name" << std::endl;
////
TString cuts = "(oppositeLepCharge == 1) && (dR_leplep > 0.2) && (mass_leplep > 12) && isCleanEvent && PassZmask==1 && ";
cuts += jettagcut;
//// sample info
TString htitle[21] = {"_singlet_s","_singlet_tW","_singlet_t","_singletbar_s","_singletbar_tW","_singletbar_t","_ttbarW","_ttbarZ","_ttbar_cc","_ttbar_bb","_ttbar_b","_ttbar","_wjets","_zjets","_zjets_lowmass","_ww","_wz","_zz","_MuEG","_DoubleElectron","_DoubleMu",};
// Float_t lumi = 19450;
////////
//SKIPPING HISTOGRAM INITIALIZATION FOR OTHER 49 VARIABLES (USE ABOVE THREE EXAMPLES FOR YOUR USE)
////////
TDirectory *currentDir = gDirectory;
for (int i = 0; i < numhists; i++){
std::cout << "-->Figuring out binning for " << histnames[i] << std::endl;
//Set the binning on the x-axis by looking at the data and combining bins (if necessary) to make sure there are no poorly populated bins
//Define poorly populated arbitrarily as N < 100
currentDir->cd(); //This is because the "Draw" command only finds
TString histName = Form("tempData_%s",histnames[i]);
TH1 * tempHist = new TH1D(histName,"",histbins[i],histmin[i],histmax[i]);
//Only look at data
for (int isam=18; isam<21; isam++){
TString sample = htitle[isam];
TString fileName = "/afs/crc.nd.edu/user/w/wluo1/LHCP_2013/CMSSW_5_3_8_patch1/src/BEAN/DrawPlots/bin/treeFiles/dilSummaryTrees" +sample + "_2012_53x_July5th_all.root";
std::cout << " -->Including " << fileName << std::endl;
TFile * tmpfile = new TFile(fileName);
TTree * tmpTree = (TTree*)tmpfile->Get("summaryTree");
currentDir->cd(); //This is because the "Draw" command only finds
TString selection = "(" + cuts + "&&";
if (isam == 18) selection+= "MuonEle && isMuEGTriggerPass";
else if (isam == 19) selection += "TwoEle && isDoubleElectronTriggerPass";
else if (isam == 20) selection += "TwoMuon && isDoubleMuTriggerPass";
selection += ")";
TString var = string(histnames[i])+">>+"+histName ;
tmpTree->Draw(var, selection, "goff");
tmpfile->Close();
delete tmpfile;
}
std::cout << "tempHist name = " << tempHist->GetName() << std::endl;
std::cout << "tempHist: Entries = " << tempHist->GetEntries() << ", Integral = " << tempHist->Integral() << std::endl;
double threshold = (tempHist->Integral())/20;
//Now, look through the bins of the histograms, combining when necessary
int nBinsUsed = 0;
double *xBins = new double[histbins[i]+1];
xBins[0] = histmin[i]; //First bin always starts at minimum
double binSum = 0;
for (int iBin = 1; iBin <= histbins[i]; ++iBin) {
binSum += tempHist->GetBinContent(iBin);
std::cout << "binSum = " << binSum << ", iBin = " << iBin << std::endl;
if (binSum >= threshold) {
std::cout << "Setting bin edge!" << std::endl;
++nBinsUsed;
xBins[nBinsUsed] = tempHist->GetBinLowEdge(iBin+1);
binSum = 0.;
}
}
//Now check: if binSum > 0, that means the last bin was too small to be a bin
//by itself. If so, we need to make sure it's combined with the bin on it's left by setting the bin edge to the histogram max
if (binSum > 0) {
xBins[nBinsUsed] = tempHist->GetBinLowEdge(histbins[i]+1);
}
std::cout << " -->Proposed binning: " << histbins[i] << ", " << histmin[i] << ", " << histmax[i] << std::endl;
std::cout << " -->Using " << nBinsUsed << " bins: ";
for (int iBin = 0; iBin < nBinsUsed; ++iBin) std::cout << xBins[iBin] << ", ";
std::cout << xBins[nBinsUsed] << std::endl;
for (int j = 0; j < numhists; j++)
{
currentDir->cd(); //This is because the "Draw" command only finds
profiles_Data[numhists*i+j] = new TProfile(Form("profiles_Data_%s_%s",histnames[i],histnames[j]),"",nBinsUsed,xBins);
profiles_Data[numhists*i+j]->Sumw2();
profiles_Data[numhists*i+j]->SetLineWidth(2);
profiles_MC[numhists*i+j] = new TProfile(Form("profiles_MC_%s_%s",histnames[i],histnames[j]),"",nBinsUsed,xBins);
profiles_MC[numhists*i+j]->Sumw2();
profiles_MC[numhists*i+j]->SetLineWidth(2);
}
delete[] xBins;
}
// loop over samples
for (int isam=0; isam<21; isam++){
/// get the tree file for each sample
TString sample = htitle[isam];
TString fileName = "/afs/crc.nd.edu/user/w/wluo1/LHCP_2013/CMSSW_5_3_8_patch1/src/BEAN/DrawPlots/bin/treeFiles/dilSummaryTrees" +sample + "_2012_53x_July5th_all.root";
std::cout << "-->start processing sample " << fileName << std::endl;
TFile * tmpfile = new TFile(fileName);
TTree * tmpTree = (TTree*)tmpfile->Get("summaryTree");
TString weight = "(19450*Xsec/nGen)*weight*topPtWgt*lepTotalSF*triggerSF*csvWgtlf*csvWgthf*"; ///// before rwt
if (rwt) weight = "(19450*Xsec/nGen)*weight*topPtWgt*lepTotalSF*triggerSF*csvWgtlf*csvWgthf*"; ///// after rwt
TString lepCut = "((MuonEle && isMuEGTriggerPass) || (TwoEle && isDoubleElectronTriggerPass) || (TwoMuon && isDoubleMuTriggerPass))";
if (isam==18) lepCut = "(MuonEle && isMuEGTriggerPass)";
if (isam==19) lepCut = "(TwoEle && isDoubleElectronTriggerPass)";
if (isam==20) lepCut = "(TwoMuon && isDoubleMuTriggerPass)";
/// selection cut
TString selection = weight;
// selection += Form("%f*",lumi); //Adjust for sample normalization;
selection += ("(" + cuts + "&&" + lepCut + ")");
if (isam>17) selection = ("(" + cuts + "&&" + lepCut + ")"); // no weights for data
// std::cout << " -->selection is: " << selection << std::endl;
///// loop through variables
for (int i = 0; i < numhists; i++){
for (int j = 0; j < numhists; j++){
TString proName = profiles_MC[numhists*i+j]->GetName();
if (isam >= 18) proName = profiles_Data[numhists*i+j]->GetName();
TString var = string(histnames[j]) + ":" + string(histnames[i])+">>+"+proName ;
currentDir->cd(); //This is because the "Draw" command only finds
tmpTree->Draw(var, selection, "goff");
}
}
tmpfile->Close();
delete tmpfile;
} // end sample loop
//Format the plots the way Robin does
for (int i = 0; i < numhists; i++) {
for (int j = 0; j < numhists; j++) {
profiles_MC[numhists*i+j]->SetFillColor(2);
profiles_MC[numhists*i+j]->SetLineColor(0);
profiles_MC[numhists*i+j]->SetMarkerSize(0);
profiles_MC[numhists*i+j]->SetFillStyle(3013);
profiles_MC[numhists*i+j]->GetXaxis()->SetTitle(histXaxis[i]);
profiles_MC[numhists*i+j]->GetYaxis()->SetTitle(histXaxis[j]);
profiles_MC[numhists*i+j]->GetYaxis()->SetTitleOffset(1.3);
double minval = 9999999.0;
double maxval = -9999999.0;
for(int ibin = 0; ibin < profiles_Data[numhists*i+j]->GetNbinsX(); ibin++){
if(profiles_Data[numhists*i+j]->GetBinContent(ibin+1) != 0.0){
if(profiles_Data[numhists*i+j]->GetBinContent(ibin+1)-profiles_Data[numhists*i+j]->GetBinError(ibin+1) < minval)
minval = profiles_Data[numhists*i+j]->GetBinContent(ibin+1)-profiles_Data[numhists*i+j]->GetBinError(ibin+1);
if(profiles_Data[numhists*i+j]->GetBinContent(ibin+1)+profiles_Data[numhists*i+j]->GetBinError(ibin+1) > maxval)
maxval = profiles_Data[numhists*i+j]->GetBinContent(ibin+1)+profiles_Data[numhists*i+j]->GetBinError(ibin+1);
}
if(profiles_MC[numhists*i+j]->GetBinContent(ibin+1) != 0.0){
if(profiles_MC[numhists*i+j]->GetBinContent(ibin+1)-profiles_MC[numhists*i+j]->GetBinError(ibin+1) < minval)
minval = profiles_MC[numhists*i+j]->GetBinContent(ibin+1)-profiles_MC[numhists*i+j]->GetBinError(ibin+1);
if(profiles_MC[numhists*i+j]->GetBinContent(ibin+1)+profiles_MC[numhists*i+j]->GetBinError(ibin+1) > maxval)
maxval = profiles_MC[numhists*i+j]->GetBinContent(ibin+1)+profiles_MC[numhists*i+j]->GetBinError(ibin+1);
}
}
if(minval < 0)
{
profiles_MC[numhists*i+j]->SetMaximum(maxval + 0.4*fabs(maxval-1.1*minval));
profiles_MC[numhists*i+j]->SetMinimum(1.1*minval);
}
else
{
profiles_MC[numhists*i+j]->SetMaximum(maxval + 0.4*fabs(maxval-0.9*minval));
profiles_MC[numhists*i+j]->SetMinimum(0.9*minval);
}
} //end loop j
} //end loop i
//Save plots in output file
TString outFileName = "corrPlots";
if (rwt) outFileName += "_rwt_";
outFileName += catName;
outFileName += ".root";
TFile *outputFile = TFile::Open(outFileName,"RECREATE");
outputFile->cd();
for (int i = 0; i < numhists*numhists; ++i) {
profiles_MC[i]->Write();
profiles_Data[i]->Write();
}
outputFile->Close();
}
float PurityFit::fit_specific( const TH1* h_, const TH1* sig_, const TH1* bkg_,
string name, // unique name of the result
string outname , // output file name, where to save results
map<string,float> *pars // to gather additional params
)
{
// 1) perform preliminary checks
if ( h_ == NULL ) { cout<<"[PurityFit]::[fit_specific]::[ERROR] no target histogram"<<endl; return -1;}
if ( sig_ == NULL ) { cout<<"[PurityFit]::[fit_specific]::[ERROR] no sig histogram"<<endl; return -1;}
if ( bkg_ == NULL ) { cout<<"[PurityFit]::[fit_specific]::[ERROR] no bkg histogram"<<endl; return -1;}
if (sig_ -> Integral() == 0 ) { cout<<"[PurityFit]::[fit_specific]::[ERROR] sig integrall is NULL"<<endl; return -2;}
if (bkg_ -> Integral() == 0 ) { cout<<"[PurityFit]::[fit_specific]::[ERROR] bkg integrall is NULL"<<endl; return -2;}
if (verbose_ >0) cout <<"[PurityFit]::[fit_specific]::[INFO] fitting "<<h_->GetName() << " " << sig_->GetName()<<" "<<bkg_->GetName()<<endl;
// 1.5) Clone
TH1 * sig = (TH1*)sig_ -> Clone(Form("%s_fitspecific_clone",sig_->GetName()));
TH1 * bkg = (TH1*)bkg_ -> Clone(Form("%s_fitspecific_clone",bkg_->GetName()));
TH1 * h = (TH1*)h_ -> Clone(Form("%s_fitspecific_clone",h_->GetName()));
// 1.6) check no negative entries otherwise 0
for(int i=1;i<=sig->GetNbinsX();++i)
if(sig->GetBinContent(i) <0) sig->SetBinContent(i,0);
for(int i=1;i<=bkg->GetNbinsX();++i)
if(bkg->GetBinContent(i) <0) bkg->SetBinContent(i,0);
// 2) scale templates: template normalization is meaningless
sig->Sumw2();
bkg->Sumw2();
if (bkg->Integral() <=0)
{
for(int i=1;i<=bkg->GetNbinsX();++i)
bkg->SetBinContent(i,1.0);
}
// 2.5) fit background with exponential
int nbins = h->GetNbinsX();
float xmin = h->GetBinLowEdge(1);
float xmax = h->GetBinLowEdge(nbins+1);
double prevChi2=-1;
vector<TCanvas*> cbkgs;
vector<TF1> expos;
double cached_pars[50]; //cache
for(int i=0;i< sizeof(cached_pars)/sizeof(double); ++i) cached_pars[i]=0;
int poln=0;
for( poln=0;poln<5; ++poln)
{
string formula = "TMath::Exp(-[0]*x) * ( ";
for(int i=0;i<=poln;++i)
{
if (i>0 ) formula += " + ";
formula += Form("[%d]",i+1);
if (i>0) formula += Form("*TMath::Power(x,%d)",i);
}
formula += ")";
cout <<"Considering Formula"<<formula<<endl;
TF1 expo(Form("expo%d",poln),formula.c_str(),xmin,xmax);
expo.SetParameter(0,0.01);
expo.SetParameter(1,0.05);
expo.SetParLimits(0,TMath::Min(bkg->Integral()*.1,0.001), bkg->Integral()*10);
expo.SetParLimits(1,1e-9, 1.);
if (poln>0)for(int k=0;k < poln+1;++k) expo.SetParameter(k, cached_pars[k]);
bkg->Fit( &expo ,"QN") ;
bkg->Fit( &expo ,"QNM") ;
expos.push_back(expo);
//cache parameters
for(int k=0;k < poln+1;++k) cached_pars[k]=expo.GetParameter(k);
double chi2=expo.GetChisquare();
double prob = 0 ;
if (poln > 0)
{
int n = 0;
for(int i=1;i<=bkg->GetNbinsX();++i) if ( bkg->GetBinContent(i)> 0 ) ++n;
int dof = n - (poln+2) + 1;
double f = (prevChi2 - chi2) / ( chi2 /dof) ;
prob= 1.- TMath::FDistI (f , 1, dof ) ;
}
prevChi2 = chi2;
cout<<"----------- BKG PARAMETERS ARE -------"<<endl;
cout << "Prob = "<<prob<<endl;
cout << "chi2 = "<<chi2<<endl;
cout<<" 0 : "<< expo.GetParameter(0) <<endl;
for(int i=0; i<=poln;++i)
cout<<" "<< i+1 <<" : "<< expo.GetParameter(i+1) <<endl;
cout<<"--------------------------------------"<<endl;
TCanvas *cbkg=new TCanvas( (string(name)+"_bkgfit"+ Form("pol%d",poln)).c_str(),"Canvas"); cbkgs.push_back(cbkg);
bkg->Clone("bkgForDraw")->Draw("P E"); // FIXME, memory leak
expo.DrawClone("L SAME");
if (prob >0.05 and poln> 0) break; // ---------------------- EXIT BEFORE UPDATING
}
poln -= 1; // the last has a negligible improvement
for(int i=1;i<=bkg->GetNbinsX() ;++i)
bkg->SetBinContent(i,expos[poln].Integral( bkg->GetBinLowEdge(i),bkg->GetBinLowEdge(i+1) ) ) ;
if (sig->Integral() >0)
sig -> Scale( 1./sig->Integral() );
if (bkg->Integral() >0)
bkg -> Scale( 1./bkg->Integral() );
// 3) estimate paramaters
// float fracEstimator=0.5;
// {
// // 3.1) compute frac as around the max of sig
// float sigMax= sig->GetMaximum();
// int sigBinMax= sig->GetMaximumBin();
// float sigInt= sig->Integral();
// float bkgUnderMax=bkg->GetBinContent(sigBinMax);
// float targetMax = h->GetMaximum();
// float targetInt = h->Integral();
// float frac1=(targetMax/targetInt)/( (sigMax-bkgUnderMax)/sigInt);
// // 3.2) compute frac as tail
// float bkgInt= bkg->Integral();
// int bkgN=bkg->GetNbinsX();
// int nTailSum= bkgN/10;
// float bkgTail=0;for(int i=0;i<nTailSum;i++) bkgTail+= bkg->GetBinContent(bkgN-i);
// int targetN=h->GetNbinsX();
// float targetTail=0;for(int i=0;i<nTailSum;i++) targetTail+= h->GetBinContent(targetN-i);
// float frac2=1.- (targetTail/targetInt)/(bkgTail/bkgInt);
// // 3.3) set estimator
// fracEstimator = frac1;
// if(fracEstimator<0.05 or fracEstimator>0.95) fracEstimator=frac2;
// if(fracEstimator<0.05 or fracEstimator>0.95) fracEstimator=0.8;
// }
// 4) create roofit variables
RooRealVar f("f","fraction",0.95,0.3,1.0);
//if ( TMath::IsNaN( fracEstimator ) )
// f.setVal(0.05);
//else
// f.setVal(fracEstimator);
f.setRange(0.3,1.0);
f.setConstant(false);
RooRealVar x("x","EtMiss",xmin,xmax);
// 5) create roo data hist
RooDataHist HistSig("sig","hist sig",x,sig);
RooDataHist HistBkg("bkg","hist bkg",x,bkg);
RooDataHist HistToFit("target","hist target",x,h);
// 6) create roo hist pdf
RooHistPdf PdfSig("pdfsig","pdfsig",x,HistSig,0);
RooHistPdf PdfBkg("pdfbkg","pdfbkg",x,HistBkg,10); //last number is interpolation
// 7) create model
RooAddPdf PdfModel("model","model",RooArgList(PdfSig,PdfBkg),f);
// 8) fit
RooFitResult *r;
RooPlot *frame=x.frame();
RooMsgService::instance().setSilentMode(true);
PdfModel.fitTo(HistToFit,
SumW2Error(kTRUE),
Save(),
PrintEvalErrors(-1),
PrintLevel(-1),
//Minos(kTRUE),
Warnings(0)
);
r = PdfModel.fitTo(HistToFit,
SumW2Error(kTRUE),
Save(),
PrintEvalErrors(-1),
PrintLevel(-1),
Minos(kTRUE),
Warnings(0)
);
cout <<" -------------- FLOATING -------------"<<endl;
r-> floatParsInit() . Print("V");
cout <<" -------------- FINAL ----------------"<<endl;
r-> floatParsFinal() . Print("V");
cout <<" -------------------------------------"<<endl;
// 8.5) save additional results
if (pars != NULL ) {
(*pars)["fracErrorHigh"] = f.getAsymErrorHi();
(*pars)["fracErrorLow" ] = f.getAsymErrorLo();
}
// 9) plot
HistToFit.plotOn(frame,DataError(RooAbsData::SumW2));
PdfModel.plotOn(frame, LineColor(kBlack));
PdfModel.plotOn(frame, Components(PdfBkg),LineColor(kRed));
PdfModel.plotOn(frame, Components(PdfSig),LineColor(kBlue),LineStyle(kDashed));
TCanvas *c=new TCanvas((string(name)+"_canvas").c_str(),"Canvas");
c->cd();
c->Draw();
frame->Draw();
TLatex* txt = new TLatex();//Form("Fraction=%.1f\%",f.getVal()*100) ;
txt->SetTextSize(0.03) ;
txt->SetTextColor(kBlack) ;
txt->SetNDC();
txt->SetTextAlign(22);
//txt->AddText(Form("Fraction=%.1f\%",f.getVal()*100) );
txt->DrawLatex(.3,.85,Form("Fraction=%.1f%%",float(f.getVal()*100)));
// get pt from name
{
size_t n = name.find("pt");
float ptmin = -1;
float ptmax = -1;
if (n != string::npos)
{
string pts = "";
//name[n+0] == 'p'
//name[n+1] =='t'
n +=2;
//all string are null terminated in C/C++
while ( name[n] != '_' and name[n] != '\0' ) { pts+=name[n]; ++n;}
ptmin = atof( pts.c_str() ) ;
pts="";
++n;
while ( name[n] != '_' and name[n] != '\0' ) { pts+=name[n]; ++n;}
ptmax = atof( pts.c_str() ) ;
txt->DrawLatex(.3,.82,Form("%.1f<P_{T}<%.1f",ptmin,ptmax));
}
} // end get pt from name
if ( outname != "")
{
TFile *fOut=TFile::Open(outname.c_str(),"UPDATE");
fOut->cd();
c->Write();
for( auto c : cbkgs ) c->Write();
r->Write(Form("%s_roofit",name.c_str() ) );
fOut->Close();
}
// delete the clone
for( auto c : cbkgs ) c->Delete();
cbkgs.clear();
sig -> Delete();
bkg -> Delete();
h -> Delete();
return f.getVal();
}
Bool_t fitsHere(TLegend *l,Double_t x1, Double_t y1, Double_t x2, Double_t y2)
{
Bool_t fits = true;
TList *list = l->GetListOfPrimitives();
for (Int_t k = 0; list->At(k) != 0 && fits; k++)
{
TObject *obj = ((TLegendEntry*)(list->At(k)))->GetObject();
if (obj == 0) continue;
TClass *cl = obj->IsA();
//Histogram, drawn as a histogram
if (cl->InheritsFrom("TH1") && !cl->InheritsFrom("TH2") && !cl->InheritsFrom("TH3")
&& cl != TProfile::Class() && ((TH1*)obj)->GetMarkerColor() == kWhite)
{
Int_t where = 0;
TH1 *h = (TH1*)obj;
for (Int_t i = 1; i <= h->GetNbinsX() && fits; i++)
{
if (h->GetBinLowEdge(i) + h->GetBinWidth(i) < x1) continue; //to the left of the legend
if (h->GetBinLowEdge(i) > x2) continue; //to the right of the legend
if (h->GetBinContent(i) > y1 && h->GetBinContent(i) < y2) fits = false; //inside the legend
if (h->GetBinContent(i) < y1)
{
if (where == 0) where = -1; //below the legend
if (where == 1) fits = false; //a previous bin was above it so there's a vertical line through it
}
if (h->GetBinContent(i) > y2)
{
if (where == 0) where = 1; //above the legend
if (where == -1) fits = false; //a previous bin was below it so there's a vertical line through it
}
}
continue;
}
//Histogram, drawn with Draw("P")
else if (cl->InheritsFrom("TH1") && !cl->InheritsFrom("TH2") && !cl->InheritsFrom("TH3")
&& cl != TProfile::Class())
//Probably TProfile would be the same but I haven't tested it
{
TH1 *h = (TH1*)obj;
for (Int_t i = 1; i <= h->GetNbinsX() && fits; i++)
{
if (h->GetBinLowEdge(i) + h->GetBinWidth(i)/2 < x1) continue;
if (h->GetBinLowEdge(i) > x2) continue;
if (h->GetBinContent(i) > y1 && h->GetBinContent(i) < y2) fits = false;
if (h->GetBinContent(i) + h->GetBinError(i) > y2 && h->GetBinContent(i) - h->GetBinError(i) < y2) fits = false;
if (h->GetBinContent(i) + h->GetBinError(i) > y1 && h->GetBinContent(i) - h->GetBinError(i) < y1) fits = false;
}
}
else if (cl->InheritsFrom("TF1") && !cl->InheritsFrom("TF2"))
{
//TF1 *f = (TF1*)obj;
//Double_t max = f->GetMaximum(x1,x2);
//Double_t min = f->GetMinimum(x1,x2);
//if (min < y2 && max > y1) fits = false;
}
// else if (cl->InheritsFrom(...... add more objects here
else
{
cout << "Don't know how to place the legend around objects of type " << obj->ClassName() << "." << endl
<< "Add this class into placeLegend.C if you want it to work properly." << endl
<< "The legend will still be placed around any other objects." << endl;
}
}
return fits;
}
