| 1 |
buchmann |
1.1 |
#include <iostream>
|
| 2 |
|
|
#include <sstream>
|
| 3 |
|
|
#include <iomanip>
|
| 4 |
|
|
#include <TFile.h>
|
| 5 |
|
|
#include <TTree.h>
|
| 6 |
|
|
#include <TH1.h>
|
| 7 |
|
|
#include <TF1.h>
|
| 8 |
|
|
#include <TMath.h>
|
| 9 |
|
|
#include <TCanvas.h>
|
| 10 |
|
|
#include <vector>
|
| 11 |
|
|
#include <TROOT.h>
|
| 12 |
|
|
#include <TLine.h>
|
| 13 |
|
|
#include <TLegend.h>
|
| 14 |
|
|
#include <TLatex.h>
|
| 15 |
|
|
#include <TRandom.h>
|
| 16 |
|
|
#ifndef GeneralToolBoxLoaded
|
| 17 |
|
|
#include "GeneralToolBox.C"
|
| 18 |
|
|
#endif
|
| 19 |
|
|
#ifndef Verbosity
|
| 20 |
|
|
#define Verbosity 0
|
| 21 |
|
|
#endif
|
| 22 |
|
|
|
| 23 |
|
|
using namespace std;
|
| 24 |
|
|
|
| 25 |
|
|
Double_t LogParabola(Double_t *x,Double_t *par)
|
| 26 |
|
|
{
|
| 27 |
|
|
return par[0]*TMath::Exp(-par[1]*(x[0]-par[2])*(x[0]-par[2])); // we're adding a "logarithmic parabola" :-)
|
| 28 |
|
|
//note: the abs() around the first parameter ensures that, when fitting, no negative values are chosen.
|
| 29 |
|
|
}
|
| 30 |
|
|
|
| 31 |
buchmann |
1.2 |
Double_t LogParabolaP(Double_t *x,Double_t *par)
|
| 32 |
|
|
{
|
| 33 |
|
|
float fitval = par[0]*TMath::Exp(-par[1]*(x[0]-par[2])*(x[0]-par[2])); // we're adding a "logarithmic parabola" :-)
|
| 34 |
|
|
fitval+= statErrorP(fitval);
|
| 35 |
|
|
return fitval;
|
| 36 |
|
|
}
|
| 37 |
|
|
|
| 38 |
|
|
Double_t LogParabolaN(Double_t *x,Double_t *par)
|
| 39 |
|
|
{
|
| 40 |
|
|
float fitval = par[0]*TMath::Exp(-par[1]*(x[0]-par[2])*(x[0]-par[2])); // we're adding a "logarithmic parabola" :-)
|
| 41 |
|
|
fitval-= statErrorN(fitval);
|
| 42 |
|
|
return fitval;
|
| 43 |
|
|
}
|
| 44 |
|
|
|
| 45 |
|
|
|
| 46 |
|
|
|
| 47 |
buchmann |
1.1 |
bool doreject=false;
|
| 48 |
|
|
float low_reject=-10;
|
| 49 |
|
|
float hi_reject=10;
|
| 50 |
|
|
|
| 51 |
|
|
bool dofixed=true;
|
| 52 |
|
|
|
| 53 |
|
|
|
| 54 |
|
|
bool addparabola=true;
|
| 55 |
|
|
float parabola_height=0;
|
| 56 |
|
|
float parabola_inclination=0;
|
| 57 |
|
|
float parabola_pointzero=0;
|
| 58 |
|
|
|
| 59 |
|
|
float find_KM_peak(TH1F *all, TH1F *ttbar, float minfit, float maxfit, bool is_data, float &error,float &Sigma);
|
| 60 |
|
|
float find_Gauss_peak(TH1F *all, TH1F *ttbar, float minfit, float maxfit, bool is_data, float &error,float &Sigma,int numsig);
|
| 61 |
|
|
|
| 62 |
|
|
|
| 63 |
|
|
Double_t KrystalMallLogPar(double *x, double *par)
|
| 64 |
|
|
{
|
| 65 |
|
|
//parameters:
|
| 66 |
|
|
//N: the way we scale the function
|
| 67 |
|
|
//alpha (where the function changes)
|
| 68 |
|
|
//n: exponent of the power expression in the left area
|
| 69 |
|
|
//xbar: peak of the gaussian part (RHS)
|
| 70 |
|
|
//sigma: width of the gaussian part (RHS)
|
| 71 |
|
|
float N=par[0];
|
| 72 |
|
|
float alpha=par[1];
|
| 73 |
|
|
float n=par[2];
|
| 74 |
|
|
float xbar=par[3];
|
| 75 |
|
|
float sigma=par[4];
|
| 76 |
|
|
float altX=x[0];
|
| 77 |
|
|
float result=-999;
|
| 78 |
|
|
if(doreject&&x[0]>low_reject&&x[0]<hi_reject)
|
| 79 |
|
|
{
|
| 80 |
|
|
TF1::RejectPoint();
|
| 81 |
|
|
return 0;
|
| 82 |
|
|
}
|
| 83 |
|
|
if(((altX-xbar)/sigma>-alpha)&&((altX-xbar)/sigma<alpha)){
|
| 84 |
|
|
result=N*TMath::Exp(-(altX-xbar)*(altX-xbar)/(2*sigma*sigma));
|
| 85 |
|
|
}
|
| 86 |
|
|
else
|
| 87 |
|
|
{
|
| 88 |
|
|
//if we are outside the central (Gaussian) area things become more difficult ...
|
| 89 |
|
|
float A=TMath::Power(n/TMath::Abs(alpha),n)*TMath::Exp(-alpha*alpha/2);
|
| 90 |
|
|
float B=n/TMath::Abs(alpha) - TMath::Abs(alpha);
|
| 91 |
|
|
if((altX-xbar)/sigma<=-alpha) result=N*A*TMath::Power((B-((altX-xbar)/sigma)),-n);
|
| 92 |
|
|
if((altX-xbar)/sigma>=alpha) result=N*A*TMath::Power((B+((altX-xbar)/sigma)),-n);
|
| 93 |
|
|
}
|
| 94 |
|
|
|
| 95 |
|
|
if(addparabola) {
|
| 96 |
|
|
if(dofixed) {
|
| 97 |
|
|
result+=parabola_height*TMath::Exp(-parabola_inclination*(x[0]-parabola_pointzero)*(x[0]-parabola_pointzero)); // we're adding a "logarithmic parabola" :-)
|
| 98 |
|
|
}
|
| 99 |
|
|
else {
|
| 100 |
|
|
result+=par[5]*TMath::Exp(-par[6]*(x[0]-par[7])*(x[0]-par[7])); // we're adding a "logarithmic parabola" :-)
|
| 101 |
|
|
if(par[5]<0) return -999; // there can be no negative ttbar contribution, so just return a value which is going to be a horrible fit.
|
| 102 |
|
|
if(par[6]<0) return -999; // the parabola needs to close (i.e. tend to negative values for large |jzb|, not to large positive values)
|
| 103 |
|
|
}
|
| 104 |
|
|
}
|
| 105 |
|
|
return result;
|
| 106 |
|
|
}
|
| 107 |
|
|
|
| 108 |
|
|
|
| 109 |
|
|
void do_ttbar_fit(TH1F *ttbar,TF1 *logpar, TF1 *KM)
|
| 110 |
|
|
{
|
| 111 |
|
|
logpar->SetParameters(10,2,3);
|
| 112 |
|
|
ttbar->Fit(logpar,"NQ");
|
| 113 |
|
|
ttbar->Fit(logpar,"NQ");
|
| 114 |
|
|
ttbar->Fit(logpar,"NQ");
|
| 115 |
|
|
ttbar->Fit(logpar,"NQ");
|
| 116 |
|
|
ttbar->SetStats(0);
|
| 117 |
|
|
parabola_height=logpar->GetParameter(0);
|
| 118 |
|
|
parabola_inclination=logpar->GetParameter(1);
|
| 119 |
|
|
parabola_pointzero=logpar->GetParameter(2);
|
| 120 |
|
|
}
|
| 121 |
|
|
|
| 122 |
|
|
void draw_complete_fit(TH1F *all, TH1F *ttbar, float minfit, float maxfit, bool is_data, TF1 *KM)
|
| 123 |
|
|
{
|
| 124 |
|
|
TCanvas *fitsummary;
|
| 125 |
|
|
if(is_data) {
|
| 126 |
|
|
fitsummary= new TCanvas("fitsummary","Fit Summary",1000,500);
|
| 127 |
|
|
fitsummary->Divide(2,1);
|
| 128 |
|
|
}
|
| 129 |
|
|
else {
|
| 130 |
|
|
fitsummary= new TCanvas("fitsummary","Fit Summary",1200,400);
|
| 131 |
|
|
fitsummary->Divide(3,1);
|
| 132 |
|
|
}
|
| 133 |
|
|
TF1 *logpar = new TF1("logpar",LogParabola,minfit,maxfit,3);
|
| 134 |
|
|
|
| 135 |
|
|
logpar->SetParameters(KM->GetParameter(5),KM->GetParameter(6),KM->GetParameter(7));
|
| 136 |
|
|
logpar->SetLineColor(kOrange);
|
| 137 |
|
|
logpar->SetLineStyle(2);
|
| 138 |
|
|
if(!is_data)
|
| 139 |
|
|
{
|
| 140 |
|
|
ttbar->GetXaxis()->SetTitle("JZB (GeV/c)");
|
| 141 |
|
|
ttbar->GetYaxis()->SetTitle("events");
|
| 142 |
|
|
ttbar->GetXaxis()->CenterTitle();
|
| 143 |
|
|
ttbar->GetYaxis()->CenterTitle();
|
| 144 |
|
|
ttbar->SetLineColor(kRed);
|
| 145 |
|
|
fitsummary->cd(1);
|
| 146 |
|
|
ttbar->Draw();
|
| 147 |
|
|
fitsummary->cd(1);
|
| 148 |
|
|
logpar->Draw("same");
|
| 149 |
|
|
TLegend *leg = new TLegend(0.3,0.25,0.65,0.4);
|
| 150 |
|
|
leg->AddEntry(ttbar,"t#bar{t} (mc)","l");
|
| 151 |
|
|
leg->AddEntry(logpar,"Fit with Log. Parabola","l");
|
| 152 |
|
|
leg->SetLineColor(kWhite);
|
| 153 |
|
|
leg->SetFillColor(kWhite);
|
| 154 |
|
|
leg->Draw();
|
| 155 |
|
|
TText *title1=write_title("t#bar{t} Distribution and Fit");
|
| 156 |
|
|
title1->Draw();
|
| 157 |
|
|
}
|
| 158 |
|
|
fitsummary->cd(2-int(is_data));
|
| 159 |
|
|
fitsummary->cd(2-int(is_data))->SetLogy(1);
|
| 160 |
|
|
all->GetYaxis()->SetTitle("events");
|
| 161 |
|
|
all->GetYaxis()->CenterTitle();
|
| 162 |
|
|
all->Draw();
|
| 163 |
|
|
ttbar->SetLineColor(kRed);
|
| 164 |
|
|
if(!is_data) ttbar->Draw("same");
|
| 165 |
|
|
KM->SetLineWidth(1);
|
| 166 |
|
|
KM->Draw("same");
|
| 167 |
|
|
logpar->SetLineWidth(1);
|
| 168 |
|
|
logpar->Draw("same");
|
| 169 |
|
|
if(!is_data)ttbar->Draw("same");
|
| 170 |
|
|
TLegend *leg2 = new TLegend(0.65,0.65,0.89,0.89);
|
| 171 |
|
|
if(is_data) leg2->AddEntry(all,"Data","l");
|
| 172 |
|
|
else leg2->AddEntry(all,"Stacked MC","l");
|
| 173 |
|
|
leg2->AddEntry(KM,"Fitted KM Function","l");
|
| 174 |
|
|
if(!is_data) leg2->AddEntry(ttbar,"t#bar{t} MC","l");
|
| 175 |
|
|
leg2->AddEntry(logpar,"t#bar{t} (Fit)","l");
|
| 176 |
|
|
leg2->SetFillColor(kWhite);
|
| 177 |
|
|
leg2->SetLineColor(kWhite);
|
| 178 |
|
|
leg2->Draw();
|
| 179 |
|
|
TText *title2=write_title("Distribution and Fits (log.)");
|
| 180 |
|
|
title2->Draw();
|
| 181 |
|
|
fitsummary->cd(3-is_data);
|
| 182 |
|
|
all->Draw();
|
| 183 |
|
|
KM->Draw("same");
|
| 184 |
|
|
float peaklocation=KM->GetParameter(3);
|
| 185 |
|
|
TLine *muline = new TLine(peaklocation,0,peaklocation,all->GetMaximum());
|
| 186 |
|
|
muline->SetLineColor(kBlue);
|
| 187 |
|
|
muline->SetLineStyle(2);
|
| 188 |
|
|
muline->Draw();
|
| 189 |
|
|
TLegend *leg = new TLegend(0.6,0.6,0.89,0.89);
|
| 190 |
|
|
if(is_data) leg2->AddEntry(all,"Data","l");
|
| 191 |
|
|
else leg->AddEntry(all,"Stacked MC","l");
|
| 192 |
|
|
leg->AddEntry(KM,"Fitted KM Function","l");
|
| 193 |
|
|
stringstream mulinelabel;
|
| 194 |
|
|
mulinelabel<<"Peak position at #mu="<<peaklocation;
|
| 195 |
|
|
leg->AddEntry(muline,mulinelabel.str().c_str(),"l");
|
| 196 |
|
|
leg->SetLineColor(kWhite);
|
| 197 |
|
|
leg->SetFillColor(kWhite);
|
| 198 |
|
|
leg->Draw();
|
| 199 |
|
|
mulinelabel<<"+/-"<<TMath::Abs(KM->GetParError(3));
|
| 200 |
|
|
TText *title3=write_title("Distribution and Fits");
|
| 201 |
|
|
title3->Draw();
|
| 202 |
|
|
TText *titlel=write_title_low(mulinelabel.str().c_str());
|
| 203 |
|
|
titlel->Draw();
|
| 204 |
|
|
|
| 205 |
|
|
stringstream printtop;
|
| 206 |
|
|
printtop << "#mu="<<std::setprecision(3)<<KM->GetParameter(3)<<"+/-"<<std::setprecision(3)<<KM->GetParError(3);
|
| 207 |
|
|
TLatex *toptext = new TLatex(0,all->GetMaximum()*1.3,printtop.str().c_str());
|
| 208 |
|
|
toptext->SetTextAlign(22);
|
| 209 |
|
|
// toptext->Draw();
|
| 210 |
|
|
|
| 211 |
|
|
doreject=false;
|
| 212 |
|
|
TF1 *wholefitfunc=(TF1*)KM->Clone();
|
| 213 |
|
|
doreject=true;
|
| 214 |
|
|
wholefitfunc->SetLineColor(kRed);
|
| 215 |
|
|
wholefitfunc->SetLineStyle(2);
|
| 216 |
|
|
wholefitfunc->Draw("same");
|
| 217 |
|
|
|
| 218 |
|
|
fitsummary->cd(2-is_data);
|
| 219 |
|
|
wholefitfunc->Draw("same");
|
| 220 |
|
|
|
| 221 |
|
|
if(is_data) CompleteSave(fitsummary, "fit/Fit_Summary_Data");
|
| 222 |
|
|
else CompleteSave(fitsummary,"fit/Fit_Summary_MC");
|
| 223 |
|
|
|
| 224 |
|
|
}
|
| 225 |
|
|
|
| 226 |
|
|
float Kostas_algorithm(TH1F *hist, float &error, float &sigma, TF1* fitFunc, float lowlimit, float highlimit,bool is_data)
|
| 227 |
|
|
{
|
| 228 |
|
|
float mean = hist->GetBinCenter( hist->GetMaximumBin());
|
| 229 |
|
|
float rms = hist->GetRMS();
|
| 230 |
|
|
|
| 231 |
|
|
mean = hist->GetBinCenter( hist->GetMaximumBin());
|
| 232 |
|
|
|
| 233 |
|
|
fitFunc->SetParameter(1,mean);
|
| 234 |
|
|
|
| 235 |
|
|
hist->Fit(fitFunc,"QLL0","",mean-10,mean+10);
|
| 236 |
|
|
|
| 237 |
|
|
mean=fitFunc->GetParameter(1);
|
| 238 |
|
|
rms=fitFunc->GetParameter(2);
|
| 239 |
|
|
error=fitFunc->GetParError(1);
|
| 240 |
|
|
|
| 241 |
|
|
bool printOut = false; // print the peak estimate in the i-th iteration
|
| 242 |
|
|
|
| 243 |
|
|
// --- perform iterations
|
| 244 |
|
|
int numIterations=5;
|
| 245 |
|
|
|
| 246 |
|
|
if(printOut) std::cout << " ( ";
|
| 247 |
|
|
for(int i=1;i<numIterations+1;i++) //--- modify the number of iterations until peak is stable
|
| 248 |
|
|
{
|
| 249 |
|
|
hist->Fit(fitFunc,"QLLN","same",mean - lowlimit*rms, mean + highlimit*rms); // fit -2 +1 sigma from previous iteration
|
| 250 |
|
|
mean=fitFunc->GetParameter(1);
|
| 251 |
|
|
fitFunc->SetRange(mean - lowlimit*rms, mean + highlimit*rms);
|
| 252 |
|
|
if(printOut) std::cout << mean << ",";
|
| 253 |
|
|
}
|
| 254 |
|
|
if(printOut) std::cout << " ) ";
|
| 255 |
|
|
if(printOut) std::cout << endl;
|
| 256 |
|
|
mean=fitFunc->GetParameter(1);
|
| 257 |
|
|
sigma=fitFunc->GetParameter(2);
|
| 258 |
|
|
error=1.0*fitFunc->GetParError(1);
|
| 259 |
|
|
|
| 260 |
|
|
// below this point we're merely doing cosmetics :-)
|
| 261 |
|
|
TCanvas *fitcanvas = new TCanvas("fitcanvas","Fitting Canvas");
|
| 262 |
|
|
fitcanvas->cd();
|
| 263 |
|
|
hist->SetMinimum(0);
|
| 264 |
|
|
if(is_data) hist->Draw("e1");
|
| 265 |
|
|
else hist->Draw("histo");
|
| 266 |
|
|
fitFunc->SetLineColor(kBlue);
|
| 267 |
|
|
fitFunc->SetLineWidth(1);
|
| 268 |
|
|
fitFunc->Draw("same");
|
| 269 |
|
|
hist->SetStats(0);
|
| 270 |
|
|
TLegend *leg;
|
| 271 |
|
|
if(is_data) {
|
| 272 |
|
|
leg= make_legend("Fit (Data)");
|
| 273 |
|
|
leg->AddEntry(hist,"Data","p");
|
| 274 |
|
|
}
|
| 275 |
|
|
else {
|
| 276 |
|
|
leg= make_legend("Fit (MC)");
|
| 277 |
|
|
leg->AddEntry(hist,"MC","l");
|
| 278 |
|
|
}
|
| 279 |
|
|
|
| 280 |
|
|
leg->AddEntry(fitFunc,"Fit","l");
|
| 281 |
|
|
leg->Draw();
|
| 282 |
|
|
|
| 283 |
|
|
TText *ftitle=write_text(0.20,0.86,"Fit results:");
|
| 284 |
|
|
ftitle->SetTextSize(0.03);
|
| 285 |
|
|
ftitle->SetTextAlign(11);
|
| 286 |
|
|
stringstream fitresult;
|
| 287 |
|
|
fitresult << "#mu=" << std::setprecision(4) << mean << "+/-" << std::setprecision(4) << error;
|
| 288 |
|
|
// TText *title1=write_text(0.20,0.96,fitresult.str().c_str());
|
| 289 |
|
|
TText *title1=write_text(0.20,0.82,fitresult.str().c_str());
|
| 290 |
|
|
title1->SetTextSize(0.03);
|
| 291 |
|
|
title1->SetTextAlign(11);
|
| 292 |
|
|
stringstream sigmainfo;
|
| 293 |
|
|
sigmainfo << "#sigma=" << std::setprecision(4) << fitFunc->GetParameter(2) << "+/-" << std::setprecision(4) << fitFunc->GetParError(2);
|
| 294 |
|
|
// TText *sigmatext=write_text(0.80,0.96,sigmainfo.str().c_str());
|
| 295 |
|
|
TText *sigmatext=write_text(0.20,0.78,sigmainfo.str().c_str());
|
| 296 |
|
|
sigmatext->SetTextSize(0.03);
|
| 297 |
|
|
sigmatext->SetTextAlign(11);
|
| 298 |
|
|
|
| 299 |
|
|
TText* toptitle;
|
| 300 |
|
|
if(is_data) toptitle = write_title("Fit Result (data)");
|
| 301 |
|
|
else toptitle = write_title("Fit Result (MC)");
|
| 302 |
|
|
toptitle->Draw();
|
| 303 |
|
|
ftitle->Draw();
|
| 304 |
|
|
title1->Draw();
|
| 305 |
|
|
sigmatext->Draw();
|
| 306 |
|
|
if(!is_data) CompleteSave(fitcanvas,"fit/Fit_Summary_MC");
|
| 307 |
|
|
if(is_data) CompleteSave(fitcanvas,"fit/Fit_Summary_Data");
|
| 308 |
|
|
|
| 309 |
|
|
|
| 310 |
|
|
|
| 311 |
|
|
// cout << "[" << fitFunc->GetParameter(1) << " , " << fitFunc->GetParError(1) << "]" << endl;
|
| 312 |
|
|
return mean;
|
| 313 |
|
|
}
|
| 314 |
|
|
|
| 315 |
|
|
|
| 316 |
|
|
|
| 317 |
|
|
float find_peak(TH1F *all, TH1F *ttbar, float minfit, float maxfit, bool is_data, float &error,float &Sigma, int method)
|
| 318 |
|
|
{
|
| 319 |
|
|
float peak_position;
|
| 320 |
|
|
if(method==0||method>1) {
|
| 321 |
|
|
//looking at a gaus request
|
| 322 |
|
|
int numsig=1;
|
| 323 |
|
|
if(method>1) numsig=method;
|
| 324 |
|
|
peak_position=find_Gauss_peak(all,ttbar,minfit,maxfit,is_data,error,Sigma,numsig);
|
| 325 |
|
|
}
|
| 326 |
|
|
if(method==1) {
|
| 327 |
|
|
//looking at a KM request
|
| 328 |
|
|
peak_position=find_KM_peak(all,ttbar,minfit,maxfit,is_data,error,Sigma);
|
| 329 |
|
|
}
|
| 330 |
|
|
if(method==-99) { // KOSTAS!!
|
| 331 |
|
|
TF1 *f1 = new TF1("f1","gaus",-40,40);
|
| 332 |
|
|
peak_position=Kostas_algorithm(all,error,Sigma,f1,2.5,2.5,is_data);
|
| 333 |
|
|
}
|
| 334 |
|
|
return peak_position;
|
| 335 |
|
|
}
|
| 336 |
|
|
|
| 337 |
|
|
|
| 338 |
|
|
float get_Gaussian_peak(TH1F *hist, float &error, float &sigma, TF1* fitFunc, float lowlimit, float highlimit,bool is_data,int numsig)
|
| 339 |
|
|
{
|
| 340 |
|
|
TCanvas *fitcanvas = new TCanvas("fitcanvas","fitcanvas");
|
| 341 |
|
|
float mean = hist->GetBinCenter( hist->GetMaximumBin());
|
| 342 |
|
|
float rms = hist->GetRMS();
|
| 343 |
|
|
|
| 344 |
|
|
mean = hist->GetBinCenter( hist->GetMaximumBin());
|
| 345 |
|
|
|
| 346 |
|
|
fitFunc->SetParameter(1,mean);
|
| 347 |
|
|
|
| 348 |
|
|
hist->Fit(fitFunc,"QLL0","",mean-10,mean+10);
|
| 349 |
|
|
|
| 350 |
|
|
mean=fitFunc->GetParameter(1);
|
| 351 |
|
|
rms=fitFunc->GetParameter(2);
|
| 352 |
|
|
error=fitFunc->GetParError(1);
|
| 353 |
|
|
|
| 354 |
|
|
bool printOut = false; // print the peak estimate in the i-th iteration
|
| 355 |
|
|
|
| 356 |
|
|
// --- perform iterations
|
| 357 |
|
|
int numIterations=5;
|
| 358 |
|
|
|
| 359 |
|
|
if(printOut) std::cout << " ( ";
|
| 360 |
|
|
for(int i=1;i<numIterations+1;i++) //--- modify the number of iterations until peak is stable
|
| 361 |
|
|
{
|
| 362 |
|
|
hist->Fit(fitFunc,"QLLN","same",mean - numsig*rms, mean + numsig*rms); // fit -2 +1 sigma from previous iteration
|
| 363 |
|
|
mean=fitFunc->GetParameter(1);
|
| 364 |
|
|
fitFunc->SetRange(mean - numsig*rms, mean + numsig*rms);
|
| 365 |
|
|
if(printOut) std::cout << mean << ",";
|
| 366 |
|
|
}
|
| 367 |
|
|
if(printOut) std::cout << " ) ";
|
| 368 |
|
|
if(printOut) std::cout << endl;
|
| 369 |
|
|
mean=fitFunc->GetParameter(1);
|
| 370 |
|
|
sigma=fitFunc->GetParameter(2);
|
| 371 |
|
|
error=1.0*fitFunc->GetParError(1);
|
| 372 |
|
|
fitcanvas->cd();
|
| 373 |
|
|
hist->SetMinimum(0);
|
| 374 |
|
|
if(is_data) hist->Draw("e1");
|
| 375 |
|
|
else hist->Draw("histo");
|
| 376 |
|
|
fitFunc->SetLineColor(kBlue);
|
| 377 |
|
|
fitFunc->SetLineWidth(1);
|
| 378 |
|
|
fitFunc->Draw("same");
|
| 379 |
|
|
hist->SetStats(0);
|
| 380 |
|
|
TLegend *leg;
|
| 381 |
|
|
if(is_data) {
|
| 382 |
|
|
leg= make_legend("Fit (Data)");
|
| 383 |
|
|
leg->AddEntry(hist,"Data","p");
|
| 384 |
|
|
}
|
| 385 |
|
|
else {
|
| 386 |
|
|
leg= make_legend("Fit (MC)");
|
| 387 |
|
|
leg->AddEntry(hist,"MC","l");
|
| 388 |
|
|
}
|
| 389 |
|
|
|
| 390 |
|
|
leg->AddEntry(fitFunc,"Fit","l");
|
| 391 |
|
|
leg->Draw();
|
| 392 |
|
|
|
| 393 |
|
|
TText *ftitle=write_text(0.20,0.86,"Fit results:");
|
| 394 |
|
|
ftitle->SetTextSize(0.03);
|
| 395 |
|
|
ftitle->SetTextAlign(11);
|
| 396 |
|
|
stringstream fitresult;
|
| 397 |
|
|
fitresult << "#mu=" << std::setprecision(4) << mean << "+/-" << std::setprecision(4) << error;
|
| 398 |
|
|
// TText *title1=write_text(0.20,0.96,fitresult.str().c_str());
|
| 399 |
|
|
TText *title1=write_text(0.20,0.82,fitresult.str().c_str());
|
| 400 |
|
|
title1->SetTextSize(0.03);
|
| 401 |
|
|
title1->SetTextAlign(11);
|
| 402 |
|
|
stringstream sigmainfo;
|
| 403 |
|
|
sigmainfo << "#sigma=" << std::setprecision(4) << fitFunc->GetParameter(2) << "+/-" << std::setprecision(4) << fitFunc->GetParError(2);
|
| 404 |
|
|
// TText *sigmatext=write_text(0.80,0.96,sigmainfo.str().c_str());
|
| 405 |
|
|
TText *sigmatext=write_text(0.20,0.78,sigmainfo.str().c_str());
|
| 406 |
|
|
sigmatext->SetTextSize(0.03);
|
| 407 |
|
|
sigmatext->SetTextAlign(11);
|
| 408 |
|
|
|
| 409 |
|
|
TText* toptitle;
|
| 410 |
|
|
if(is_data) toptitle = write_title("Fit Result (data)");
|
| 411 |
|
|
else toptitle = write_title("Fit Result (MC)");
|
| 412 |
|
|
toptitle->Draw();
|
| 413 |
|
|
ftitle->Draw();
|
| 414 |
|
|
title1->Draw();
|
| 415 |
|
|
sigmatext->Draw();
|
| 416 |
|
|
if(!is_data) CompleteSave(fitcanvas,"fit/Fit_Summary_MC");
|
| 417 |
|
|
if(is_data) CompleteSave(fitcanvas,"fit/Fit_Summary_Data");
|
| 418 |
|
|
|
| 419 |
|
|
|
| 420 |
|
|
// cout << "[" << fitFunc->GetParameter(1) << " , " << fitFunc->GetParError(1) << "]" << endl;
|
| 421 |
|
|
return mean;
|
| 422 |
|
|
}
|
| 423 |
|
|
|
| 424 |
|
|
|
| 425 |
|
|
float find_Gauss_peak(TH1F *all, TH1F *ttbar, float minfit, float maxfit, bool is_data, float &error,float &Sigma,int numsig)
|
| 426 |
|
|
{
|
| 427 |
|
|
TF1 *fitfunc = new TF1("fitfunc","gaus",minfit,maxfit);
|
| 428 |
|
|
float peakpos = get_Gaussian_peak(all,error,Sigma,fitfunc, minfit, maxfit,is_data,numsig);
|
| 429 |
|
|
return peakpos;
|
| 430 |
|
|
}
|
| 431 |
|
|
|
| 432 |
|
|
float find_KM_peak(TH1F *all, TH1F *ttbar, float minfit, float maxfit, bool is_data, float &error,float &Sigma)
|
| 433 |
|
|
{
|
| 434 |
|
|
all->SetLineColor(kBlue);
|
| 435 |
|
|
all->SetStats(0);
|
| 436 |
|
|
all->SetTitle("");
|
| 437 |
|
|
all->GetXaxis()->SetTitle("JZB (GeV/c)");
|
| 438 |
|
|
all->GetYaxis()->SetTitle("events");
|
| 439 |
|
|
all->GetXaxis()->CenterTitle();
|
| 440 |
|
|
all->GetYaxis()->CenterTitle();
|
| 441 |
|
|
TF1 *fitfunc = new TF1("fitfunc",KrystalMallLogPar,0.8*minfit,0.8*maxfit,8);
|
| 442 |
|
|
if(!is_data)
|
| 443 |
|
|
{
|
| 444 |
|
|
TF1 *logpar = new TF1("logpar",LogParabola,minfit,maxfit,3);
|
| 445 |
|
|
do_ttbar_fit(ttbar,logpar,fitfunc);
|
| 446 |
|
|
fitfunc->SetParameters(1000,2,2.5,-1.6,4,logpar->GetParameter(0),logpar->GetParameter(1),logpar->GetParameter(2));
|
| 447 |
|
|
parabola_height=logpar->GetParameter(0);
|
| 448 |
|
|
parabola_inclination=logpar->GetParameter(1);
|
| 449 |
|
|
parabola_pointzero=logpar->GetParameter(2);
|
| 450 |
|
|
dofixed=true;//ttbar is known so we can fix the parameters and don't need to use them for fitting!
|
| 451 |
|
|
}
|
| 452 |
|
|
else
|
| 453 |
|
|
{
|
| 454 |
|
|
fitfunc->SetParameters(1000,2,2.5,-1.6,4,5.45039,0.000324593,12.3528);
|
| 455 |
|
|
dofixed=false;
|
| 456 |
|
|
}
|
| 457 |
|
|
|
| 458 |
|
|
vector<float> chi2values;
|
| 459 |
|
|
addparabola=true;
|
| 460 |
|
|
for (int ifit=0;ifit<100;ifit++)
|
| 461 |
|
|
{
|
| 462 |
|
|
all->Fit(fitfunc,"NQ");
|
| 463 |
|
|
chi2values.push_back(fitfunc->GetChisquare());
|
| 464 |
|
|
if(ifit>5&&chi2values[ifit-2]==chi2values[ifit]) break;
|
| 465 |
|
|
}
|
| 466 |
|
|
/*
|
| 467 |
|
|
The parameters represent the following quantities:
|
| 468 |
|
|
float N=par[0];
|
| 469 |
|
|
float alpha=par[1];
|
| 470 |
|
|
float n=par[2];
|
| 471 |
|
|
float xbar=par[3];
|
| 472 |
|
|
float sigma=par[4];
|
| 473 |
|
|
*/
|
| 474 |
|
|
//we are clearing an area of two sigma to the left and to the right of the center of the function for the "real fit".
|
| 475 |
|
|
low_reject=-2*fitfunc->GetParameter(4)+fitfunc->GetParameter(3);
|
| 476 |
|
|
hi_reject=fitfunc->GetParameter(3)+2*fitfunc->GetParameter(4);
|
| 477 |
|
|
if(low_reject>-15) low_reject=-10;
|
| 478 |
|
|
if(hi_reject<15) hi_reject=10;
|
| 479 |
|
|
doreject=true;//activating the rejection :-)
|
| 480 |
|
|
|
| 481 |
|
|
for (int ifit=0;ifit<100;ifit++)
|
| 482 |
|
|
{
|
| 483 |
|
|
all->Fit(fitfunc,"NQ");
|
| 484 |
|
|
chi2values.push_back(fitfunc->GetChisquare());
|
| 485 |
|
|
if(ifit>5&&chi2values[ifit-2]==chi2values[ifit]) break;
|
| 486 |
|
|
}
|
| 487 |
|
|
|
| 488 |
|
|
draw_complete_fit(all,ttbar,minfit,maxfit,is_data,fitfunc);
|
| 489 |
|
|
doreject=true;
|
| 490 |
|
|
error=fitfunc->GetParError(3);
|
| 491 |
|
|
Sigma=fitfunc->GetParameter(4);//sigma
|
| 492 |
|
|
|
| 493 |
|
|
return fitfunc->GetParameter(3);
|
| 494 |
|
|
}
|
| 495 |
|
|
|
| 496 |
|
|
Double_t InvCrystalBall(Double_t *x,Double_t *par)
|
| 497 |
|
|
{
|
| 498 |
|
|
Double_t arg1=0,arg2=0,A=0,B=0;
|
| 499 |
|
|
Double_t f1=0;
|
| 500 |
|
|
Double_t f2=0;
|
| 501 |
|
|
Double_t lim=0;
|
| 502 |
|
|
Double_t fitval=0;
|
| 503 |
|
|
Double_t N=0;
|
| 504 |
|
|
Double_t n=par[4];
|
| 505 |
|
|
|
| 506 |
|
|
Double_t invX = -x[0];
|
| 507 |
|
|
|
| 508 |
|
|
if (par[2] != 0)
|
| 509 |
|
|
arg1 = (invX-par[1])/par[2];
|
| 510 |
|
|
|
| 511 |
|
|
arg2 = ( -pow( TMath::Abs(par[3]) , 2 ) ) / 2 ;
|
| 512 |
|
|
|
| 513 |
|
|
if (par[3] != 0)
|
| 514 |
|
|
A = pow( ( n / TMath::Abs( par[3] ) ) , n) * TMath::Exp(arg2);
|
| 515 |
|
|
|
| 516 |
|
|
if (par[3] != 0)
|
| 517 |
|
|
B = n / TMath::Abs(par[3]) - TMath::Abs(par[3]);
|
| 518 |
|
|
|
| 519 |
|
|
f1 = TMath::Exp(-0.5*arg1*arg1);
|
| 520 |
|
|
if (par[2] != 0)
|
| 521 |
|
|
f2 = A * pow( ( B - (invX - par[1])/par[2] ) , -n );
|
| 522 |
|
|
|
| 523 |
|
|
if (par[2] != 0)
|
| 524 |
|
|
lim = ( par[1] - invX ) / par[2] ;
|
| 525 |
|
|
|
| 526 |
|
|
N = par[0];
|
| 527 |
|
|
|
| 528 |
|
|
|
| 529 |
|
|
|
| 530 |
|
|
if(lim < par[3])
|
| 531 |
|
|
fitval = N * f1;
|
| 532 |
|
|
if(lim >= par[3])
|
| 533 |
|
|
fitval = N * f2;
|
| 534 |
|
|
|
| 535 |
|
|
return fitval;
|
| 536 |
|
|
}
|
| 537 |
|
|
|
| 538 |
|
|
Double_t InvCrystalBallP(Double_t *x,Double_t *par)
|
| 539 |
|
|
{
|
| 540 |
|
|
Double_t arg1=0,arg2=0,A=0,B=0;
|
| 541 |
|
|
Double_t f1=0;
|
| 542 |
|
|
Double_t f2=0;
|
| 543 |
|
|
Double_t lim=0;
|
| 544 |
|
|
Double_t fitval=0;
|
| 545 |
|
|
Double_t N=0;
|
| 546 |
|
|
Double_t n=par[4];
|
| 547 |
|
|
|
| 548 |
|
|
Double_t invX = -x[0];
|
| 549 |
|
|
|
| 550 |
|
|
if (par[2] != 0)
|
| 551 |
|
|
arg1 = (invX-par[1])/par[2];
|
| 552 |
|
|
|
| 553 |
|
|
arg2 = ( -pow( TMath::Abs(par[3]) , 2 ) ) / 2 ;
|
| 554 |
|
|
|
| 555 |
|
|
if (par[3] != 0)
|
| 556 |
|
|
A = pow( ( n / TMath::Abs( par[3] ) ) , n) * TMath::Exp(arg2);
|
| 557 |
|
|
|
| 558 |
|
|
if (par[3] != 0)
|
| 559 |
|
|
B = n / TMath::Abs(par[3]) - TMath::Abs(par[3]);
|
| 560 |
|
|
|
| 561 |
|
|
f1 = TMath::Exp(-0.5*arg1*arg1);
|
| 562 |
|
|
if (par[2] != 0)
|
| 563 |
|
|
f2 = A * pow( ( B - (invX - par[1])/par[2] ) , -n );
|
| 564 |
|
|
|
| 565 |
|
|
if (par[2] != 0)
|
| 566 |
|
|
lim = ( par[1] - invX ) / par[2] ;
|
| 567 |
|
|
|
| 568 |
|
|
N = par[0];
|
| 569 |
|
|
|
| 570 |
|
|
|
| 571 |
|
|
|
| 572 |
|
|
if(lim < par[3])
|
| 573 |
|
|
fitval = N * f1;
|
| 574 |
|
|
if(lim >= par[3])
|
| 575 |
|
|
fitval = N * f2;
|
| 576 |
|
|
|
| 577 |
|
|
fitval+= statErrorP(fitval);
|
| 578 |
|
|
return fitval;
|
| 579 |
|
|
}
|
| 580 |
|
|
|
| 581 |
|
|
Double_t InvCrystalBallN(Double_t *x,Double_t *par)
|
| 582 |
|
|
{
|
| 583 |
|
|
Double_t arg1=0,arg2=0,A=0,B=0;
|
| 584 |
|
|
Double_t f1=0;
|
| 585 |
|
|
Double_t f2=0;
|
| 586 |
|
|
Double_t lim=0;
|
| 587 |
|
|
Double_t fitval=0;
|
| 588 |
|
|
Double_t N=0;
|
| 589 |
|
|
Double_t n=par[4];
|
| 590 |
|
|
|
| 591 |
|
|
Double_t invX = -x[0];
|
| 592 |
|
|
|
| 593 |
|
|
if (par[2] != 0)
|
| 594 |
|
|
arg1 = (invX-par[1])/par[2];
|
| 595 |
|
|
|
| 596 |
|
|
arg2 = ( -pow( TMath::Abs(par[3]) , 2 ) ) / 2 ;
|
| 597 |
|
|
|
| 598 |
|
|
if (par[3] != 0)
|
| 599 |
|
|
A = pow( ( n / TMath::Abs( par[3] ) ) , n) * TMath::Exp(arg2);
|
| 600 |
|
|
|
| 601 |
|
|
if (par[3] != 0)
|
| 602 |
|
|
B = n / TMath::Abs(par[3]) - TMath::Abs(par[3]);
|
| 603 |
|
|
|
| 604 |
|
|
f1 = TMath::Exp(-0.5*arg1*arg1);
|
| 605 |
|
|
if (par[2] != 0)
|
| 606 |
|
|
f2 = A * pow( ( B - (invX - par[1])/par[2] ) , -n );
|
| 607 |
|
|
|
| 608 |
|
|
if (par[2] != 0)
|
| 609 |
|
|
lim = ( par[1] - invX ) / par[2] ;
|
| 610 |
|
|
|
| 611 |
|
|
N = par[0];
|
| 612 |
|
|
|
| 613 |
|
|
|
| 614 |
|
|
|
| 615 |
|
|
if(lim < par[3])
|
| 616 |
|
|
fitval = N * f1;
|
| 617 |
|
|
if(lim >= par[3])
|
| 618 |
|
|
fitval = N * f2;
|
| 619 |
|
|
|
| 620 |
|
|
fitval-= statErrorN(fitval);
|
| 621 |
|
|
return fitval;
|
| 622 |
|
|
}
|
| 623 |
|
|
|
| 624 |
|
|
|
