]> git.uio.no Git - u/mrichter/AliRoot.git/blame - STAT/TStatToolkit.cxx
git://git.uio.no / u / mrichter / AliRoot.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
Updated version of the PID calibration
[u/mrichter/AliRoot.git] / STAT / TStatToolkit.cxx
CommitLineData
21f3a443 1/**************************************************************************
2 * Copyright(c) 1998-1999, ALICE Experiment at CERN, All rights reserved. *
3 * *
4 * Author: The ALICE Off-line Project. *
5 * Contributors are mentioned in the code where appropriate. *
6 * *
7 * Permission to use, copy, modify and distribute this software and its *
8 * documentation strictly for non-commercial purposes is hereby granted *
9 * without fee, provided that the above copyright notice appears in all *
10 * copies and that both the copyright notice and this permission notice *
11 * appear in the supporting documentation. The authors make no claims *
12 * about the suitability of this software for any purpose. It is *
13 * provided "as is" without express or implied warranty. *
14 **************************************************************************/
15
16
17///////////////////////////////////////////////////////////////////////////
18// Class TStatToolkit
19//
20// Subset of matheamtical functions not included in the TMath
21//
22
23///////////////////////////////////////////////////////////////////////////
24#include "TMath.h"
25#include "Riostream.h"
26#include "TH1F.h"
3240a856 27#include "TH2F.h"
21f3a443 28#include "TH3.h"
29#include "TF1.h"
30#include "TTree.h"
31#include "TChain.h"
32#include "TObjString.h"
33#include "TLinearFitter.h"
3d7cc0b4 34#include "TGraph2D.h"
35#include "TGraph.h"
b3453fe7 36#include "TGraphErrors.h"
377a7d60 37#include "TMultiGraph.h"
38#include "TCanvas.h"
39#include "TLatex.h"
40#include "TCut.h"
21f3a443 41
42//
43// includes neccessary for test functions
44//
45#include "TSystem.h"
46#include "TRandom.h"
47#include "TStopwatch.h"
48#include "TTreeStream.h"
49
50#include "TStatToolkit.h"
51
52
53ClassImp(TStatToolkit) // Class implementation to enable ROOT I/O
54
55TStatToolkit::TStatToolkit() : TObject()
56{
57 //
58 // Default constructor
59 //
60}
61///////////////////////////////////////////////////////////////////////////
62TStatToolkit::~TStatToolkit()
63{
64 //
65 // Destructor
66 //
67}
68
69
70//_____________________________________________________________________________
71void TStatToolkit::EvaluateUni(Int_t nvectors, Double_t *data, Double_t &mean
72 , Double_t &sigma, Int_t hh)
73{
74 //
75 // Robust estimator in 1D case MI version - (faster than ROOT version)
76 //
77 // For the univariate case
78 // estimates of location and scatter are returned in mean and sigma parameters
79 // the algorithm works on the same principle as in multivariate case -
80 // it finds a subset of size hh with smallest sigma, and then returns mean and
81 // sigma of this subset
82 //
83
84 if (hh==0)
85 hh=(nvectors+2)/2;
86 Double_t faclts[]={2.6477,2.5092,2.3826,2.2662,2.1587,2.0589,1.9660,1.879,1.7973,1.7203,1.6473};
87 Int_t *index=new Int_t[nvectors];
88 TMath::Sort(nvectors, data, index, kFALSE);
89
90 Int_t nquant = TMath::Min(Int_t(Double_t(((hh*1./nvectors)-0.5)*40))+1, 11);
91 Double_t factor = faclts[TMath::Max(0,nquant-1)];
92
93 Double_t sumx =0;
94 Double_t sumx2 =0;
95 Int_t bestindex = -1;
96 Double_t bestmean = 0;
97 Double_t bestsigma = (data[index[nvectors-1]]-data[index[0]]+1.); // maximal possible sigma
98 bestsigma *=bestsigma;
99
100 for (Int_t i=0; i<hh; i++){
101 sumx += data[index[i]];
102 sumx2 += data[index[i]]*data[index[i]];
103 }
104
105 Double_t norm = 1./Double_t(hh);
bd7b4d18 106 Double_t norm2 = (hh-1)>0 ? 1./Double_t(hh-1):1;
21f3a443 107 for (Int_t i=hh; i<nvectors; i++){
108 Double_t cmean = sumx*norm;
109 Double_t csigma = (sumx2 - hh*cmean*cmean)*norm2;
110 if (csigma<bestsigma){
111 bestmean = cmean;
112 bestsigma = csigma;
113 bestindex = i-hh;
114 }
115
116 sumx += data[index[i]]-data[index[i-hh]];
117 sumx2 += data[index[i]]*data[index[i]]-data[index[i-hh]]*data[index[i-hh]];
118 }
119
120 Double_t bstd=factor*TMath::Sqrt(TMath::Abs(bestsigma));
121 mean = bestmean;
122 sigma = bstd;
123 delete [] index;
124
125}
126
127
128
129void TStatToolkit::EvaluateUniExternal(Int_t nvectors, Double_t *data, Double_t &mean, Double_t &sigma, Int_t hh, Float_t externalfactor)
130{
131 // Modified version of ROOT robust EvaluateUni
132 // robust estimator in 1D case MI version
133 // added external factor to include precision of external measurement
134 //
135
136 if (hh==0)
137 hh=(nvectors+2)/2;
138 Double_t faclts[]={2.6477,2.5092,2.3826,2.2662,2.1587,2.0589,1.9660,1.879,1.7973,1.7203,1.6473};
139 Int_t *index=new Int_t[nvectors];
140 TMath::Sort(nvectors, data, index, kFALSE);
141 //
142 Int_t nquant = TMath::Min(Int_t(Double_t(((hh*1./nvectors)-0.5)*40))+1, 11);
143 Double_t factor = faclts[0];
144 if (nquant>0){
145 // fix proper normalization - Anja
146 factor = faclts[nquant-1];
147 }
148
149 //
150 //
151 Double_t sumx =0;
152 Double_t sumx2 =0;
153 Int_t bestindex = -1;
154 Double_t bestmean = 0;
155 Double_t bestsigma = -1;
156 for (Int_t i=0; i<hh; i++){
157 sumx += data[index[i]];
158 sumx2 += data[index[i]]*data[index[i]];
159 }
160 //
161 Double_t kfactor = 2.*externalfactor - externalfactor*externalfactor;
162 Double_t norm = 1./Double_t(hh);
163 for (Int_t i=hh; i<nvectors; i++){
164 Double_t cmean = sumx*norm;
165 Double_t csigma = (sumx2*norm - cmean*cmean*kfactor);
166 if (csigma<bestsigma || bestsigma<0){
167 bestmean = cmean;
168 bestsigma = csigma;
169 bestindex = i-hh;
170 }
171 //
172 //
173 sumx += data[index[i]]-data[index[i-hh]];
174 sumx2 += data[index[i]]*data[index[i]]-data[index[i-hh]]*data[index[i-hh]];
175 }
176
177 Double_t bstd=factor*TMath::Sqrt(TMath::Abs(bestsigma));
178 mean = bestmean;
179 sigma = bstd;
180 delete [] index;
181}
182
183
184//_____________________________________________________________________________
185Int_t TStatToolkit::Freq(Int_t n, const Int_t *inlist
186 , Int_t *outlist, Bool_t down)
187{
188 //
189 // Sort eleements according occurancy
190 // The size of output array has is 2*n
191 //
192
193 Int_t * sindexS = new Int_t[n]; // temp array for sorting
194 Int_t * sindexF = new Int_t[2*n];
b8072cce 195 for (Int_t i=0;i<n;i++) sindexS[i]=0;
196 for (Int_t i=0;i<2*n;i++) sindexF[i]=0;
21f3a443 197 //
198 TMath::Sort(n,inlist, sindexS, down);
199 Int_t last = inlist[sindexS[0]];
200 Int_t val = last;
201 sindexF[0] = 1;
202 sindexF[0+n] = last;
203 Int_t countPos = 0;
204 //
205 // find frequency
206 for(Int_t i=1;i<n; i++){
207 val = inlist[sindexS[i]];
208 if (last == val) sindexF[countPos]++;
209 else{
210 countPos++;
211 sindexF[countPos+n] = val;
212 sindexF[countPos]++;
213 last =val;
214 }
215 }
216 if (last==val) countPos++;
217 // sort according frequency
218 TMath::Sort(countPos, sindexF, sindexS, kTRUE);
219 for (Int_t i=0;i<countPos;i++){
220 outlist[2*i ] = sindexF[sindexS[i]+n];
221 outlist[2*i+1] = sindexF[sindexS[i]];
222 }
223 delete [] sindexS;
224 delete [] sindexF;
225
226 return countPos;
227
228}
229
230//___TStatToolkit__________________________________________________________________________
3d7cc0b4 231void TStatToolkit::TruncatedMean(const TH1 * his, TVectorD *param, Float_t down, Float_t up, Bool_t verbose){
21f3a443 232 //
233 //
234 //
235 Int_t nbins = his->GetNbinsX();
236 Float_t nentries = his->GetEntries();
237 Float_t sum =0;
238 Float_t mean = 0;
239 Float_t sigma2 = 0;
240 Float_t ncumul=0;
241 for (Int_t ibin=1;ibin<nbins; ibin++){
242 ncumul+= his->GetBinContent(ibin);
243 Float_t fraction = Float_t(ncumul)/Float_t(nentries);
244 if (fraction>down && fraction<up){
245 sum+=his->GetBinContent(ibin);
246 mean+=his->GetBinCenter(ibin)*his->GetBinContent(ibin);
247 sigma2+=his->GetBinCenter(ibin)*his->GetBinCenter(ibin)*his->GetBinContent(ibin);
248 }
249 }
250 mean/=sum;
251 sigma2= TMath::Sqrt(TMath::Abs(sigma2/sum-mean*mean));
252 if (param){
253 (*param)[0] = his->GetMaximum();
254 (*param)[1] = mean;
255 (*param)[2] = sigma2;
256
257 }
258 if (verbose) printf("Mean\t%f\t Sigma2\t%f\n", mean,sigma2);
259}
260
261void TStatToolkit::LTM(TH1F * his, TVectorD *param , Float_t fraction, Bool_t verbose){
262 //
263 // LTM
264 //
265 Int_t nbins = his->GetNbinsX();
266 Int_t nentries = (Int_t)his->GetEntries();
267 Double_t *data = new Double_t[nentries];
268 Int_t npoints=0;
269 for (Int_t ibin=1;ibin<nbins; ibin++){
270 Float_t entriesI = his->GetBinContent(ibin);
271 Float_t xcenter= his->GetBinCenter(ibin);
272 for (Int_t ic=0; ic<entriesI; ic++){
273 if (npoints<nentries){
274 data[npoints]= xcenter;
275 npoints++;
276 }
277 }
278 }
279 Double_t mean, sigma;
280 Int_t npoints2=TMath::Min(Int_t(fraction*Float_t(npoints)),npoints-1);
281 npoints2=TMath::Max(Int_t(0.5*Float_t(npoints)),npoints2);
282 TStatToolkit::EvaluateUni(npoints, data, mean,sigma,npoints2);
283 delete [] data;
284 if (verbose) printf("Mean\t%f\t Sigma2\t%f\n", mean,sigma);if (param){
285 (*param)[0] = his->GetMaximum();
286 (*param)[1] = mean;
287 (*param)[2] = sigma;
288 }
289}
290
94a43b22 291Double_t TStatToolkit::FitGaus(TH1* his, TVectorD *param, TMatrixD */*matrix*/, Float_t xmin, Float_t xmax, Bool_t verbose){
21f3a443 292 //
293 // Fit histogram with gaussian function
294 //
295 // Prameters:
296 // return value- chi2 - if negative ( not enough points)
297 // his - input histogram
298 // param - vector with parameters
299 // xmin, xmax - range to fit - if xmin=xmax=0 - the full histogram range used
300 // Fitting:
301 // 1. Step - make logarithm
302 // 2. Linear fit (parabola) - more robust - always converge
303 // 3. In case of small statistic bins are averaged
304 //
305 static TLinearFitter fitter(3,"pol2");
306 TVectorD par(3);
307 TVectorD sigma(3);
308 TMatrixD mat(3,3);
309 if (his->GetMaximum()<4) return -1;
310 if (his->GetEntries()<12) return -1;
311 if (his->GetRMS()<mat.GetTol()) return -1;
312 Float_t maxEstimate = his->GetEntries()*his->GetBinWidth(1)/TMath::Sqrt((TMath::TwoPi()*his->GetRMS()));
313 Int_t dsmooth = TMath::Nint(6./TMath::Sqrt(maxEstimate));
314
315 if (maxEstimate<1) return -1;
316 Int_t nbins = his->GetNbinsX();
317 Int_t npoints=0;
318 //
319
320
321 if (xmin>=xmax){
322 xmin = his->GetXaxis()->GetXmin();
323 xmax = his->GetXaxis()->GetXmax();
324 }
325 for (Int_t iter=0; iter<2; iter++){
326 fitter.ClearPoints();
327 npoints=0;
328 for (Int_t ibin=1;ibin<nbins+1; ibin++){
329 Int_t countB=1;
330 Float_t entriesI = his->GetBinContent(ibin);
331 for (Int_t delta = -dsmooth; delta<=dsmooth; delta++){
332 if (ibin+delta>1 &&ibin+delta<nbins-1){
333 entriesI += his->GetBinContent(ibin+delta);
334 countB++;
335 }
336 }
337 entriesI/=countB;
338 Double_t xcenter= his->GetBinCenter(ibin);
339 if (xcenter<xmin || xcenter>xmax) continue;
340 Double_t error=1./TMath::Sqrt(countB);
341 Float_t cont=2;
342 if (iter>0){
343 if (par[0]+par[1]*xcenter+par[2]*xcenter*xcenter>20) return 0;
344 cont = TMath::Exp(par[0]+par[1]*xcenter+par[2]*xcenter*xcenter);
345 if (cont>1.) error = 1./TMath::Sqrt(cont*Float_t(countB));
346 }
347 if (entriesI>1&&cont>1){
348 fitter.AddPoint(&xcenter,TMath::Log(Float_t(entriesI)),error);
349 npoints++;
350 }
351 }
352 if (npoints>3){
353 fitter.Eval();
354 fitter.GetParameters(par);
355 }else{
356 break;
357 }
358 }
359 if (npoints<=3){
360 return -1;
361 }
362 fitter.GetParameters(par);
363 fitter.GetCovarianceMatrix(mat);
364 if (TMath::Abs(par[1])<mat.GetTol()) return -1;
365 if (TMath::Abs(par[2])<mat.GetTol()) return -1;
366 Double_t chi2 = fitter.GetChisquare()/Float_t(npoints);
367 //fitter.GetParameters();
368 if (!param) param = new TVectorD(3);
cb1d20de 369 // if (!matrix) matrix = new TMatrixD(3,3); // Covariance matrix to be implemented
21f3a443 370 (*param)[1] = par[1]/(-2.*par[2]);
371 (*param)[2] = 1./TMath::Sqrt(TMath::Abs(-2.*par[2]));
372 (*param)[0] = TMath::Exp(par[0]+ par[1]* (*param)[1] + par[2]*(*param)[1]*(*param)[1]);
373 if (verbose){
374 par.Print();
375 mat.Print();
376 param->Print();
377 printf("Chi2=%f\n",chi2);
378 TF1 * f1= new TF1("f1","[0]*exp(-(x-[1])^2/(2*[2]*[2]))",his->GetXaxis()->GetXmin(),his->GetXaxis()->GetXmax());
379 f1->SetParameter(0, (*param)[0]);
380 f1->SetParameter(1, (*param)[1]);
381 f1->SetParameter(2, (*param)[2]);
382 f1->Draw("same");
383 }
384 return chi2;
385}
386
cb1d20de 387Double_t TStatToolkit::FitGaus(Float_t *arr, Int_t nBins, Float_t xMin, Float_t xMax, TVectorD *param, TMatrixD */*matrix*/, Bool_t verbose){
21f3a443 388 //
389 // Fit histogram with gaussian function
390 //
391 // Prameters:
392 // nbins: size of the array and number of histogram bins
393 // xMin, xMax: histogram range
394 // param: paramters of the fit (0-Constant, 1-Mean, 2-Sigma)
395 // matrix: covariance matrix -- not implemented yet, pass dummy matrix!!!
396 //
397 // Return values:
398 // >0: the chi2 returned by TLinearFitter
399 // -3: only three points have been used for the calculation - no fitter was used
400 // -2: only two points have been used for the calculation - center of gravity was uesed for calculation
401 // -1: only one point has been used for the calculation - center of gravity was uesed for calculation
402 // -4: invalid result!!
403 //
404 // Fitting:
405 // 1. Step - make logarithm
406 // 2. Linear fit (parabola) - more robust - always converge
407 //
408 static TLinearFitter fitter(3,"pol2");
409 static TMatrixD mat(3,3);
410 static Double_t kTol = mat.GetTol();
411 fitter.StoreData(kFALSE);
412 fitter.ClearPoints();
413 TVectorD par(3);
414 TVectorD sigma(3);
3d7cc0b4 415 TMatrixD matA(3,3);
21f3a443 416 TMatrixD b(3,1);
417 Float_t rms = TMath::RMS(nBins,arr);
418 Float_t max = TMath::MaxElement(nBins,arr);
419 Float_t binWidth = (xMax-xMin)/(Float_t)nBins;
420
421 Float_t meanCOG = 0;
422 Float_t rms2COG = 0;
423 Float_t sumCOG = 0;
424
425 Float_t entries = 0;
426 Int_t nfilled=0;
427
428 for (Int_t i=0; i<nBins; i++){
429 entries+=arr[i];
430 if (arr[i]>0) nfilled++;
431 }
432
433 if (max<4) return -4;
434 if (entries<12) return -4;
435 if (rms<kTol) return -4;
436
437 Int_t npoints=0;
438 //
439
440 //
441 for (Int_t ibin=0;ibin<nBins; ibin++){
442 Float_t entriesI = arr[ibin];
443 if (entriesI>1){
444 Double_t xcenter = xMin+(ibin+0.5)*binWidth;
445
446 Float_t error = 1./TMath::Sqrt(entriesI);
447 Float_t val = TMath::Log(Float_t(entriesI));
448 fitter.AddPoint(&xcenter,val,error);
449 if (npoints<3){
3d7cc0b4 450 matA(npoints,0)=1;
451 matA(npoints,1)=xcenter;
452 matA(npoints,2)=xcenter*xcenter;
21f3a443 453 b(npoints,0)=val;
454 meanCOG+=xcenter*entriesI;
455 rms2COG +=xcenter*entriesI*xcenter;
456 sumCOG +=entriesI;
457 }
458 npoints++;
459 }
460 }
461
462
463 Double_t chi2 = 0;
464 if (npoints>=3){
465 if ( npoints == 3 ){
466 //analytic calculation of the parameters for three points
3d7cc0b4 467 matA.Invert();
21f3a443 468 TMatrixD res(1,3);
3d7cc0b4 469 res.Mult(matA,b);
21f3a443 470 par[0]=res(0,0);
471 par[1]=res(0,1);
472 par[2]=res(0,2);
473 chi2 = -3.;
474 } else {
475 // use fitter for more than three points
476 fitter.Eval();
477 fitter.GetParameters(par);
478 fitter.GetCovarianceMatrix(mat);
479 chi2 = fitter.GetChisquare()/Float_t(npoints);
480 }
481 if (TMath::Abs(par[1])<kTol) return -4;
482 if (TMath::Abs(par[2])<kTol) return -4;
483
484 if (!param) param = new TVectorD(3);
cb1d20de 485 //if (!matrix) matrix = new TMatrixD(3,3); // !!!!might be a memory leek. use dummy matrix pointer to call this function! // Covariance matrix to be implemented
21f3a443 486
487 (*param)[1] = par[1]/(-2.*par[2]);
488 (*param)[2] = 1./TMath::Sqrt(TMath::Abs(-2.*par[2]));
489 Double_t lnparam0 = par[0]+ par[1]* (*param)[1] + par[2]*(*param)[1]*(*param)[1];
490 if ( lnparam0>307 ) return -4;
491 (*param)[0] = TMath::Exp(lnparam0);
492 if (verbose){
493 par.Print();
494 mat.Print();
495 param->Print();
496 printf("Chi2=%f\n",chi2);
497 TF1 * f1= new TF1("f1","[0]*exp(-(x-[1])^2/(2*[2]*[2]))",xMin,xMax);
498 f1->SetParameter(0, (*param)[0]);
499 f1->SetParameter(1, (*param)[1]);
500 f1->SetParameter(2, (*param)[2]);
501 f1->Draw("same");
502 }
503 return chi2;
504 }
505
506 if (npoints == 2){
507 //use center of gravity for 2 points
508 meanCOG/=sumCOG;
509 rms2COG /=sumCOG;
510 (*param)[0] = max;
511 (*param)[1] = meanCOG;
512 (*param)[2] = TMath::Sqrt(TMath::Abs(meanCOG*meanCOG-rms2COG));
513 chi2=-2.;
514 }
515 if ( npoints == 1 ){
516 meanCOG/=sumCOG;
517 (*param)[0] = max;
518 (*param)[1] = meanCOG;
519 (*param)[2] = binWidth/TMath::Sqrt(12);
520 chi2=-1.;
521 }
522 return chi2;
523
524}
525
526
3d7cc0b4 527Float_t TStatToolkit::GetCOG(const Short_t *arr, Int_t nBins, Float_t xMin, Float_t xMax, Float_t *rms, Float_t *sum)
21f3a443 528{
529 //
530 // calculate center of gravity rms and sum for array 'arr' with nBins an a x range xMin to xMax
531 // return COG; in case of failure return xMin
532 //
533 Float_t meanCOG = 0;
534 Float_t rms2COG = 0;
535 Float_t sumCOG = 0;
536 Int_t npoints = 0;
537
538 Float_t binWidth = (xMax-xMin)/(Float_t)nBins;
539
540 for (Int_t ibin=0; ibin<nBins; ibin++){
541 Float_t entriesI = (Float_t)arr[ibin];
542 Double_t xcenter = xMin+(ibin+0.5)*binWidth;
543 if ( entriesI>0 ){
544 meanCOG += xcenter*entriesI;
545 rms2COG += xcenter*entriesI*xcenter;
546 sumCOG += entriesI;
547 npoints++;
548 }
549 }
550 if ( sumCOG == 0 ) return xMin;
551 meanCOG/=sumCOG;
552
553 if ( rms ){
554 rms2COG /=sumCOG;
555 (*rms) = TMath::Sqrt(TMath::Abs(meanCOG*meanCOG-rms2COG));
556 if ( npoints == 1 ) (*rms) = binWidth/TMath::Sqrt(12);
557 }
558
559 if ( sum )
560 (*sum) = sumCOG;
561
562 return meanCOG;
563}
564
565
566
567///////////////////////////////////////////////////////////////
568////////////// TEST functions /////////////////////////
569///////////////////////////////////////////////////////////////
570
571
572
573
574
575void TStatToolkit::TestGausFit(Int_t nhistos){
576 //
577 // Test performance of the parabolic - gaussian fit - compare it with
578 // ROOT gauss fit
579 // nhistos - number of histograms to be used for test
580 //
581 TTreeSRedirector *pcstream = new TTreeSRedirector("fitdebug.root");
582
583 Float_t *xTrue = new Float_t[nhistos];
584 Float_t *sTrue = new Float_t[nhistos];
585 TVectorD **par1 = new TVectorD*[nhistos];
586 TVectorD **par2 = new TVectorD*[nhistos];
587 TMatrixD dummy(3,3);
588
589
590 TH1F **h1f = new TH1F*[nhistos];
591 TF1 *myg = new TF1("myg","gaus");
592 TF1 *fit = new TF1("fit","gaus");
593 gRandom->SetSeed(0);
594
595 //init
596 for (Int_t i=0;i<nhistos; i++){
597 par1[i] = new TVectorD(3);
598 par2[i] = new TVectorD(3);
599 h1f[i] = new TH1F(Form("h1f%d",i),Form("h1f%d",i),20,-10,10);
600 xTrue[i]= gRandom->Rndm();
601 gSystem->Sleep(2);
602 sTrue[i]= .75+gRandom->Rndm()*.5;
603 myg->SetParameters(1,xTrue[i],sTrue[i]);
604 h1f[i]->FillRandom("myg");
605 }
606
607 TStopwatch s;
608 s.Start();
609 //standard gaus fit
610 for (Int_t i=0; i<nhistos; i++){
611 h1f[i]->Fit(fit,"0q");
612 (*par1[i])(0) = fit->GetParameter(0);
613 (*par1[i])(1) = fit->GetParameter(1);
614 (*par1[i])(2) = fit->GetParameter(2);
615 }
616 s.Stop();
617 printf("Gaussian fit\t");
618 s.Print();
619
620 s.Start();
621 //TStatToolkit gaus fit
622 for (Int_t i=0; i<nhistos; i++){
623 TStatToolkit::FitGaus(h1f[i]->GetArray()+1,h1f[i]->GetNbinsX(),h1f[i]->GetXaxis()->GetXmin(),h1f[i]->GetXaxis()->GetXmax(),par2[i],&dummy);
624 }
625
626 s.Stop();
627 printf("Parabolic fit\t");
628 s.Print();
629 //write stream
630 for (Int_t i=0;i<nhistos; i++){
631 Float_t xt = xTrue[i];
632 Float_t st = sTrue[i];
633 (*pcstream)<<"data"
634 <<"xTrue="<<xt
635 <<"sTrue="<<st
636 <<"pg.="<<(par1[i])
637 <<"pa.="<<(par2[i])
638 <<"\n";
639 }
640 //delete pointers
641 for (Int_t i=0;i<nhistos; i++){
642 delete par1[i];
643 delete par2[i];
644 delete h1f[i];
645 }
646 delete pcstream;
647 delete []h1f;
648 delete []xTrue;
649 delete []sTrue;
650 //
651 delete []par1;
652 delete []par2;
653
654}
655
656
657
658TGraph2D * TStatToolkit::MakeStat2D(TH3 * his, Int_t delta0, Int_t delta1, Int_t type){
659 //
660 //
661 //
662 // delta - number of bins to integrate
663 // type - 0 - mean value
664
665 TAxis * xaxis = his->GetXaxis();
666 TAxis * yaxis = his->GetYaxis();
667 // TAxis * zaxis = his->GetZaxis();
668 Int_t nbinx = xaxis->GetNbins();
669 Int_t nbiny = yaxis->GetNbins();
670 char name[1000];
671 Int_t icount=0;
672 TGraph2D *graph = new TGraph2D(nbinx*nbiny);
673 TF1 f1("f1","gaus");
674 for (Int_t ix=0; ix<nbinx;ix++)
675 for (Int_t iy=0; iy<nbiny;iy++){
676 Float_t xcenter = xaxis->GetBinCenter(ix);
677 Float_t ycenter = yaxis->GetBinCenter(iy);
cb1d20de 678 snprintf(name,1000,"%s_%d_%d",his->GetName(), ix,iy);
21f3a443 679 TH1 *projection = his->ProjectionZ(name,ix-delta0,ix+delta0,iy-delta1,iy+delta1);
680 Float_t stat= 0;
681 if (type==0) stat = projection->GetMean();
682 if (type==1) stat = projection->GetRMS();
683 if (type==2 || type==3){
684 TVectorD vec(3);
685 TStatToolkit::LTM((TH1F*)projection,&vec,0.7);
686 if (type==2) stat= vec[1];
687 if (type==3) stat= vec[0];
688 }
689 if (type==4|| type==5){
690 projection->Fit(&f1);
691 if (type==4) stat= f1.GetParameter(1);
692 if (type==5) stat= f1.GetParameter(2);
693 }
694 //printf("%d\t%f\t%f\t%f\n", icount,xcenter, ycenter, stat);
695 graph->SetPoint(icount,xcenter, ycenter, stat);
696 icount++;
697 }
698 return graph;
699}
700
701TGraph * TStatToolkit::MakeStat1D(TH3 * his, Int_t delta1, Int_t type){
702 //
703 //
704 //
705 // delta - number of bins to integrate
706 // type - 0 - mean value
707
708 TAxis * xaxis = his->GetXaxis();
709 TAxis * yaxis = his->GetYaxis();
710 // TAxis * zaxis = his->GetZaxis();
711 Int_t nbinx = xaxis->GetNbins();
712 Int_t nbiny = yaxis->GetNbins();
713 char name[1000];
714 Int_t icount=0;
715 TGraph *graph = new TGraph(nbinx);
716 TF1 f1("f1","gaus");
717 for (Int_t ix=0; ix<nbinx;ix++){
718 Float_t xcenter = xaxis->GetBinCenter(ix);
719 // Float_t ycenter = yaxis->GetBinCenter(iy);
cb1d20de 720 snprintf(name,1000,"%s_%d",his->GetName(), ix);
21f3a443 721 TH1 *projection = his->ProjectionZ(name,ix-delta1,ix+delta1,0,nbiny);
722 Float_t stat= 0;
723 if (type==0) stat = projection->GetMean();
724 if (type==1) stat = projection->GetRMS();
725 if (type==2 || type==3){
726 TVectorD vec(3);
727 TStatToolkit::LTM((TH1F*)projection,&vec,0.7);
728 if (type==2) stat= vec[1];
729 if (type==3) stat= vec[0];
730 }
731 if (type==4|| type==5){
732 projection->Fit(&f1);
733 if (type==4) stat= f1.GetParameter(1);
734 if (type==5) stat= f1.GetParameter(2);
735 }
736 //printf("%d\t%f\t%f\t%f\n", icount,xcenter, ycenter, stat);
737 graph->SetPoint(icount,xcenter, stat);
738 icount++;
739 }
740 return graph;
741}
742
743
744
745
746
88b1c775 747TString* TStatToolkit::FitPlane(TTree *tree, const char* drawCommand, const char* formula, const char* cuts, Double_t & chi2, Int_t &npoints, TVectorD &fitParam, TMatrixD &covMatrix, Float_t frac, Int_t start, Int_t stop,Bool_t fix0){
21f3a443 748 //
749 // fit an arbitrary function, specified by formula into the data, specified by drawCommand and cuts
750 // returns chi2, fitParam and covMatrix
751 // returns TString with fitted formula
752 //
dd46129c 753
21f3a443 754 TString formulaStr(formula);
755 TString drawStr(drawCommand);
756 TString cutStr(cuts);
dd46129c 757 TString ferr("1");
758
759 TString strVal(drawCommand);
760 if (strVal.Contains(":")){
761 TObjArray* valTokens = strVal.Tokenize(":");
762 drawStr = valTokens->At(0)->GetName();
763 ferr = valTokens->At(1)->GetName();
09d5920f 764 delete valTokens;
dd46129c 765 }
766
21f3a443 767
768 formulaStr.ReplaceAll("++", "~");
769 TObjArray* formulaTokens = formulaStr.Tokenize("~");
770 Int_t dim = formulaTokens->GetEntriesFast();
771
772 fitParam.ResizeTo(dim);
773 covMatrix.ResizeTo(dim,dim);
774
775 TLinearFitter* fitter = new TLinearFitter(dim+1, Form("hyp%d",dim));
776 fitter->StoreData(kTRUE);
777 fitter->ClearPoints();
778
779 Int_t entries = tree->Draw(drawStr.Data(), cutStr.Data(), "goff", stop-start, start);
09d5920f 780 if (entries == -1) {
781 delete formulaTokens;
782 return new TString("An ERROR has occured during fitting!");
783 }
bd7b4d18 784 Double_t **values = new Double_t*[dim+1] ;
785 for (Int_t i=0; i<dim+1; i++) values[i]=NULL;
dd46129c 786 //
787 entries = tree->Draw(ferr.Data(), cutStr.Data(), "goff", stop-start, start);
b8072cce 788 if (entries == -1) {
09d5920f 789 delete formulaTokens;
b8072cce 790 delete []values;
791 return new TString("An ERROR has occured during fitting!");
792 }
dd46129c 793 Double_t *errors = new Double_t[entries];
794 memcpy(errors, tree->GetV1(), entries*sizeof(Double_t));
21f3a443 795
796 for (Int_t i = 0; i < dim + 1; i++){
797 Int_t centries = 0;
798 if (i < dim) centries = tree->Draw(((TObjString*)formulaTokens->At(i))->GetName(), cutStr.Data(), "goff", stop-start,start);
799 else centries = tree->Draw(drawStr.Data(), cutStr.Data(), "goff", stop-start,start);
800
b8072cce 801 if (entries != centries) {
802 delete []errors;
803 delete []values;
804 return new TString("An ERROR has occured during fitting!");
805 }
21f3a443 806 values[i] = new Double_t[entries];
807 memcpy(values[i], tree->GetV1(), entries*sizeof(Double_t));
808 }
809
810 // add points to the fitter
811 for (Int_t i = 0; i < entries; i++){
812 Double_t x[1000];
813 for (Int_t j=0; j<dim;j++) x[j]=values[j][i];
dd46129c 814 fitter->AddPoint(x, values[dim][i], errors[i]);
21f3a443 815 }
816
817 fitter->Eval();
2c629c56 818 if (frac>0.5 && frac<1){
819 fitter->EvalRobust(frac);
88b1c775 820 }else{
821 if (fix0) {
822 fitter->FixParameter(0,0);
823 fitter->Eval();
824 }
2c629c56 825 }
21f3a443 826 fitter->GetParameters(fitParam);
827 fitter->GetCovarianceMatrix(covMatrix);
828 chi2 = fitter->GetChisquare();
b8072cce 829 npoints = entries;
21f3a443 830 TString *preturnFormula = new TString(Form("( %f+",fitParam[0])), &returnFormula = *preturnFormula;
831
832 for (Int_t iparam = 0; iparam < dim; iparam++) {
833 returnFormula.Append(Form("%s*(%f)",((TObjString*)formulaTokens->At(iparam))->GetName(),fitParam[iparam+1]));
834 if (iparam < dim-1) returnFormula.Append("+");
835 }
836 returnFormula.Append(" )");
4d61c301 837
838
b8072cce 839 for (Int_t j=0; j<dim+1;j++) delete [] values[j];
4d61c301 840
841
cb1d20de 842 delete formulaTokens;
843 delete fitter;
844 delete[] values;
b8072cce 845 delete[] errors;
cb1d20de 846 return preturnFormula;
847}
848
849TString* TStatToolkit::FitPlaneConstrain(TTree *tree, const char* drawCommand, const char* formula, const char* cuts, Double_t & chi2, Int_t &npoints, TVectorD &fitParam, TMatrixD &covMatrix, Float_t frac, Int_t start, Int_t stop,Double_t constrain){
850 //
851 // fit an arbitrary function, specified by formula into the data, specified by drawCommand and cuts
852 // returns chi2, fitParam and covMatrix
853 // returns TString with fitted formula
854 //
855
856 TString formulaStr(formula);
857 TString drawStr(drawCommand);
858 TString cutStr(cuts);
859 TString ferr("1");
860
861 TString strVal(drawCommand);
862 if (strVal.Contains(":")){
863 TObjArray* valTokens = strVal.Tokenize(":");
864 drawStr = valTokens->At(0)->GetName();
865 ferr = valTokens->At(1)->GetName();
09d5920f 866 delete valTokens;
cb1d20de 867 }
868
869
870 formulaStr.ReplaceAll("++", "~");
871 TObjArray* formulaTokens = formulaStr.Tokenize("~");
872 Int_t dim = formulaTokens->GetEntriesFast();
873
874 fitParam.ResizeTo(dim);
875 covMatrix.ResizeTo(dim,dim);
876
877 TLinearFitter* fitter = new TLinearFitter(dim+1, Form("hyp%d",dim));
878 fitter->StoreData(kTRUE);
879 fitter->ClearPoints();
880
881 Int_t entries = tree->Draw(drawStr.Data(), cutStr.Data(), "goff", stop-start, start);
09d5920f 882 if (entries == -1) {
883 delete formulaTokens;
884 return new TString("An ERROR has occured during fitting!");
885 }
cb1d20de 886 Double_t **values = new Double_t*[dim+1] ;
bd7b4d18 887 for (Int_t i=0; i<dim+1; i++) values[i]=NULL;
cb1d20de 888 //
889 entries = tree->Draw(ferr.Data(), cutStr.Data(), "goff", stop-start, start);
b8072cce 890 if (entries == -1) {
09d5920f 891 delete formulaTokens;
b8072cce 892 delete [] values;
893 return new TString("An ERROR has occured during fitting!");
894 }
cb1d20de 895 Double_t *errors = new Double_t[entries];
896 memcpy(errors, tree->GetV1(), entries*sizeof(Double_t));
897
898 for (Int_t i = 0; i < dim + 1; i++){
899 Int_t centries = 0;
900 if (i < dim) centries = tree->Draw(((TObjString*)formulaTokens->At(i))->GetName(), cutStr.Data(), "goff", stop-start,start);
901 else centries = tree->Draw(drawStr.Data(), cutStr.Data(), "goff", stop-start,start);
902
b8072cce 903 if (entries != centries) {
904 delete []errors;
905 delete []values;
09d5920f 906 delete formulaTokens;
b8072cce 907 return new TString("An ERROR has occured during fitting!");
908 }
cb1d20de 909 values[i] = new Double_t[entries];
910 memcpy(values[i], tree->GetV1(), entries*sizeof(Double_t));
911 }
912
913 // add points to the fitter
914 for (Int_t i = 0; i < entries; i++){
915 Double_t x[1000];
916 for (Int_t j=0; j<dim;j++) x[j]=values[j][i];
917 fitter->AddPoint(x, values[dim][i], errors[i]);
918 }
919 if (constrain>0){
920 for (Int_t i = 0; i < dim; i++){
921 Double_t x[1000];
922 for (Int_t j=0; j<dim;j++) if (i!=j) x[j]=0;
923 x[i]=1.;
924 fitter->AddPoint(x, 0, constrain);
925 }
926 }
927
928
929 fitter->Eval();
930 if (frac>0.5 && frac<1){
931 fitter->EvalRobust(frac);
932 }
933 fitter->GetParameters(fitParam);
934 fitter->GetCovarianceMatrix(covMatrix);
935 chi2 = fitter->GetChisquare();
936 npoints = entries;
cb1d20de 937
938 TString *preturnFormula = new TString(Form("( %f+",fitParam[0])), &returnFormula = *preturnFormula;
939
940 for (Int_t iparam = 0; iparam < dim; iparam++) {
941 returnFormula.Append(Form("%s*(%f)",((TObjString*)formulaTokens->At(iparam))->GetName(),fitParam[iparam+1]));
942 if (iparam < dim-1) returnFormula.Append("+");
943 }
944 returnFormula.Append(" )");
945
b8072cce 946 for (Int_t j=0; j<dim+1;j++) delete [] values[j];
cb1d20de 947
948
949
950 delete formulaTokens;
951 delete fitter;
952 delete[] values;
b8072cce 953 delete[] errors;
cb1d20de 954 return preturnFormula;
955}
956
957
958
959TString* TStatToolkit::FitPlaneFixed(TTree *tree, const char* drawCommand, const char* formula, const char* cuts, Double_t & chi2, Int_t &npoints, TVectorD &fitParam, TMatrixD &covMatrix, Float_t frac, Int_t start, Int_t stop){
960 //
961 // fit an arbitrary function, specified by formula into the data, specified by drawCommand and cuts
962 // returns chi2, fitParam and covMatrix
963 // returns TString with fitted formula
964 //
965
966 TString formulaStr(formula);
967 TString drawStr(drawCommand);
968 TString cutStr(cuts);
969 TString ferr("1");
970
971 TString strVal(drawCommand);
972 if (strVal.Contains(":")){
973 TObjArray* valTokens = strVal.Tokenize(":");
974 drawStr = valTokens->At(0)->GetName();
09d5920f 975 ferr = valTokens->At(1)->GetName();
976 delete valTokens;
cb1d20de 977 }
978
979
980 formulaStr.ReplaceAll("++", "~");
981 TObjArray* formulaTokens = formulaStr.Tokenize("~");
982 Int_t dim = formulaTokens->GetEntriesFast();
983
984 fitParam.ResizeTo(dim);
985 covMatrix.ResizeTo(dim,dim);
986 TString fitString="x0";
987 for (Int_t i=1; i<dim; i++) fitString+=Form("++x%d",i);
988 TLinearFitter* fitter = new TLinearFitter(dim, fitString.Data());
989 fitter->StoreData(kTRUE);
990 fitter->ClearPoints();
991
992 Int_t entries = tree->Draw(drawStr.Data(), cutStr.Data(), "goff", stop-start, start);
09d5920f 993 if (entries == -1) {
994 delete formulaTokens;
995 return new TString("An ERROR has occured during fitting!");
996 }
cb1d20de 997 Double_t **values = new Double_t*[dim+1] ;
bd7b4d18 998 for (Int_t i=0; i<dim+1; i++) values[i]=NULL;
cb1d20de 999 //
1000 entries = tree->Draw(ferr.Data(), cutStr.Data(), "goff", stop-start, start);
b8072cce 1001 if (entries == -1) {
1002 delete []values;
09d5920f 1003 delete formulaTokens;
b8072cce 1004 return new TString("An ERROR has occured during fitting!");
1005 }
cb1d20de 1006 Double_t *errors = new Double_t[entries];
1007 memcpy(errors, tree->GetV1(), entries*sizeof(Double_t));
1008
1009 for (Int_t i = 0; i < dim + 1; i++){
1010 Int_t centries = 0;
1011 if (i < dim) centries = tree->Draw(((TObjString*)formulaTokens->At(i))->GetName(), cutStr.Data(), "goff", stop-start,start);
1012 else centries = tree->Draw(drawStr.Data(), cutStr.Data(), "goff", stop-start,start);
1013
b8072cce 1014 if (entries != centries) {
1015 delete []errors;
1016 delete []values;
09d5920f 1017 delete formulaTokens;
b8072cce 1018 return new TString("An ERROR has occured during fitting!");
1019 }
cb1d20de 1020 values[i] = new Double_t[entries];
1021 memcpy(values[i], tree->GetV1(), entries*sizeof(Double_t));
1022 }
1023
1024 // add points to the fitter
1025 for (Int_t i = 0; i < entries; i++){
1026 Double_t x[1000];
1027 for (Int_t j=0; j<dim;j++) x[j]=values[j][i];
1028 fitter->AddPoint(x, values[dim][i], errors[i]);
1029 }
1030
1031 fitter->Eval();
1032 if (frac>0.5 && frac<1){
1033 fitter->EvalRobust(frac);
1034 }
1035 fitter->GetParameters(fitParam);
1036 fitter->GetCovarianceMatrix(covMatrix);
1037 chi2 = fitter->GetChisquare();
1038 npoints = entries;
cb1d20de 1039
1040 TString *preturnFormula = new TString("("), &returnFormula = *preturnFormula;
1041
1042 for (Int_t iparam = 0; iparam < dim; iparam++) {
1043 returnFormula.Append(Form("%s*(%f)",((TObjString*)formulaTokens->At(iparam))->GetName(),fitParam[iparam]));
1044 if (iparam < dim-1) returnFormula.Append("+");
1045 }
1046 returnFormula.Append(" )");
1047
1048
b8072cce 1049 for (Int_t j=0; j<dim+1;j++) delete [] values[j];
cb1d20de 1050
21f3a443 1051 delete formulaTokens;
1052 delete fitter;
1053 delete[] values;
b8072cce 1054 delete[] errors;
21f3a443 1055 return preturnFormula;
1056}
7c9cf6e4 1057
1058
1059
1060
1061
3d7cc0b4 1062Int_t TStatToolkit::GetFitIndex(const TString fString, const TString subString){
7c9cf6e4 1063 //
1064 // fitString - ++ separated list of fits
1065 // substring - ++ separated list of the requiered substrings
1066 //
1067 // return the last occurance of substring in fit string
1068 //
1069 TObjArray *arrFit = fString.Tokenize("++");
1070 TObjArray *arrSub = subString.Tokenize("++");
1071 Int_t index=-1;
1072 for (Int_t i=0; i<arrFit->GetEntries(); i++){
1073 Bool_t isOK=kTRUE;
1074 TString str =arrFit->At(i)->GetName();
1075 for (Int_t isub=0; isub<arrSub->GetEntries(); isub++){
1076 if (str.Contains(arrSub->At(isub)->GetName())==0) isOK=kFALSE;
1077 }
1078 if (isOK) index=i;
1079 }
09d5920f 1080 delete arrFit;
1081 delete arrSub;
7c9cf6e4 1082 return index;
1083}
1084
1085
3d7cc0b4 1086TString TStatToolkit::FilterFit(const TString &input, const TString filter, TVectorD &param, TMatrixD & covar){
7c9cf6e4 1087 //
1088 // Filter fit expression make sub-fit
1089 //
1090 TObjArray *array0= input.Tokenize("++");
1091 TObjArray *array1= filter.Tokenize("++");
1092 //TString *presult=new TString("(0");
1093 TString result="(0.0";
1094 for (Int_t i=0; i<array0->GetEntries(); i++){
1095 Bool_t isOK=kTRUE;
1096 TString str(array0->At(i)->GetName());
1097 for (Int_t j=0; j<array1->GetEntries(); j++){
1098 if (str.Contains(array1->At(j)->GetName())==0) isOK=kFALSE;
1099 }
1100 if (isOK) {
1101 result+="+"+str;
1102 result+=Form("*(%f)",param[i+1]);
1103 printf("%f\t%f\t%s\n",param[i+1], TMath::Sqrt(covar(i+1,i+1)),str.Data());
1104 }
1105 }
1106 result+="-0.)";
09d5920f 1107 delete array0;
1108 delete array1;
7c9cf6e4 1109 return result;
1110}
1111
1112void TStatToolkit::Update1D(Double_t delta, Double_t sigma, Int_t s1, TMatrixD &vecXk, TMatrixD &covXk){
1113 //
1114 // Update parameters and covariance - with one measurement
1115 // Input:
1116 // vecXk - input vector - Updated in function
1117 // covXk - covariance matrix - Updated in function
1118 // delta, sigma, s1 - new measurement, rms of new measurement and the index of measurement
1119 const Int_t knMeas=1;
1120 Int_t knElem=vecXk.GetNrows();
1121
1122 TMatrixD mat1(knElem,knElem); // update covariance matrix
1123 TMatrixD matHk(1,knElem); // vector to mesurement
1124 TMatrixD vecYk(knMeas,1); // Innovation or measurement residual
1125 TMatrixD matHkT(knElem,knMeas); // helper matrix Hk transpose
1126 TMatrixD matSk(knMeas,knMeas); // Innovation (or residual) covariance
1127 TMatrixD matKk(knElem,knMeas); // Optimal Kalman gain
1128 TMatrixD covXk2(knElem,knElem); // helper matrix
1129 TMatrixD covXk3(knElem,knElem); // helper matrix
1130 TMatrixD vecZk(1,1);
1131 TMatrixD measR(1,1);
1132 vecZk(0,0)=delta;
1133 measR(0,0)=sigma*sigma;
1134 //
1135 // reset matHk
1136 for (Int_t iel=0;iel<knElem;iel++)
1137 for (Int_t ip=0;ip<knMeas;ip++) matHk(ip,iel)=0;
1138 //mat1
1139 for (Int_t iel=0;iel<knElem;iel++) {
1140 for (Int_t jel=0;jel<knElem;jel++) mat1(iel,jel)=0;
1141 mat1(iel,iel)=1;
1142 }
1143 //
1144 matHk(0, s1)=1;
1145 vecYk = vecZk-matHk*vecXk; // Innovation or measurement residual
1146 matHkT=matHk.T(); matHk.T();
1147 matSk = (matHk*(covXk*matHkT))+measR; // Innovation (or residual) covariance
1148 matSk.Invert();
1149 matKk = (covXk*matHkT)*matSk; // Optimal Kalman gain
1150 vecXk += matKk*vecYk; // updated vector
1151 covXk2= (mat1-(matKk*matHk));
1152 covXk3 = covXk2*covXk;
1153 covXk = covXk3;
1154 Int_t nrows=covXk3.GetNrows();
1155
1156 for (Int_t irow=0; irow<nrows; irow++)
1157 for (Int_t icol=0; icol<nrows; icol++){
1158 // rounding problems - make matrix again symteric
1159 covXk(irow,icol)=(covXk3(irow,icol)+covXk3(icol,irow))*0.5;
1160 }
1161}
1162
1163
1164
3d7cc0b4 1165void TStatToolkit::Constrain1D(const TString &input, const TString filter, TVectorD &param, TMatrixD & covar, Double_t mean, Double_t sigma){
7c9cf6e4 1166 //
1167 // constrain linear fit
1168 // input - string description of fit function
1169 // filter - string filter to select sub fits
1170 // param,covar - parameters and covariance matrix of the fit
1171 // mean,sigma - new measurement uning which the fit is updated
1172 //
ae45c94d 1173
7c9cf6e4 1174 TObjArray *array0= input.Tokenize("++");
1175 TObjArray *array1= filter.Tokenize("++");
1176 TMatrixD paramM(param.GetNrows(),1);
1177 for (Int_t i=0; i<=array0->GetEntries(); i++){paramM(i,0)=param(i);}
1178
ae45c94d 1179 if (filter.Length()==0){
1180 TStatToolkit::Update1D(mean, sigma, 0, paramM, covar);//
1181 }else{
1182 for (Int_t i=0; i<array0->GetEntries(); i++){
1183 Bool_t isOK=kTRUE;
1184 TString str(array0->At(i)->GetName());
1185 for (Int_t j=0; j<array1->GetEntries(); j++){
1186 if (str.Contains(array1->At(j)->GetName())==0) isOK=kFALSE;
1187 }
1188 if (isOK) {
1189 TStatToolkit::Update1D(mean, sigma, i+1, paramM, covar);//
1190 }
7c9cf6e4 1191 }
1192 }
1193 for (Int_t i=0; i<=array0->GetEntries(); i++){
1194 param(i)=paramM(i,0);
1195 }
09d5920f 1196 delete array0;
1197 delete array1;
7c9cf6e4 1198}
1199
ae45c94d 1200TString TStatToolkit::MakeFitString(const TString &input, const TVectorD &param, const TMatrixD & covar, Bool_t verbose){
7c9cf6e4 1201 //
1202 //
1203 //
1204 TObjArray *array0= input.Tokenize("++");
ae45c94d 1205 TString result=Form("(%f",param[0]);
1206 printf("%f\t%f\t\n", param[0], TMath::Sqrt(covar(0,0)));
7c9cf6e4 1207 for (Int_t i=0; i<array0->GetEntries(); i++){
1208 TString str(array0->At(i)->GetName());
1209 result+="+"+str;
1210 result+=Form("*(%f)",param[i+1]);
ae45c94d 1211 if (verbose) printf("%f\t%f\t%s\n", param[i+1], TMath::Sqrt(covar(i+1,i+1)),str.Data());
7c9cf6e4 1212 }
1213 result+="-0.)";
09d5920f 1214 delete array0;
7c9cf6e4 1215 return result;
1216}
df0a2a0a 1217
1a8f4649 1218TGraphErrors * TStatToolkit::MakeGraphErrors(TTree * tree, const char * expr, const char * cut, Int_t mstyle, Int_t mcolor, Float_t msize, Float_t offset){
1219 //
1220 // Query a graph errors
1221 // return TGraphErrors specified by expr and cut
1222 // Example usage TStatToolkit::MakeGraphError(tree,"Y:X:ErrY","X>0", 25,2,0.4)
1223 // tree - tree with variable
1224 // expr - examp
1225 const Int_t entries = tree->Draw(expr,cut,"goff");
1226 if (entries<=0) {
1227 TStatToolkit t;
1228 t.Error("TStatToolkit::MakeGraphError",Form("Empty or Not valid expression (%s) or cut *%s)", expr,cut));
1229 return 0;
1230 }
1231 if ( tree->GetV2()==0){
1232 TStatToolkit t;
1233 t.Error("TStatToolkit::MakeGraphError",Form("Not valid expression (%s) ", expr));
1234 return 0;
1235 }
1236 TGraphErrors * graph=0;
1237 if ( tree->GetV3()!=0){
1238 graph = new TGraphErrors (entries, tree->GetV2(),tree->GetV1(),0,tree->GetV3());
1239 }else{
1240 graph = new TGraphErrors (entries, tree->GetV2(),tree->GetV1(),0,0);
1241 }
1242 graph->SetMarkerStyle(mstyle);
1243 graph->SetMarkerColor(mcolor);
1244 graph->SetLineColor(mcolor);
1245 if (msize>0) graph->SetMarkerSize(msize);
1246 for(Int_t i=0;i<graph->GetN();i++) graph->GetX()[i]+=offset;
1247 return graph;
1248
1249}
1250
df0a2a0a 1251
377a7d60 1252TGraph * TStatToolkit::MakeGraphSparse(TTree * tree, const char * expr, const char * cut, Int_t mstyle, Int_t mcolor, Float_t msize, Float_t offset){
df0a2a0a 1253 //
1254 // Make a sparse draw of the variables
d02b8756 1255 // Format of expr : Var:Run or Var:Run:ErrorY or Var:Run:ErrorY:ErrorX
1256 // offset : points can slightly be shifted in x for better visibility with more graphs
1257 //
1258 // Written by Weilin.Yu
1259 // updated & merged with QA-code by Patrick Reichelt
1260 //
1261 const Int_t entries = tree->Draw(expr,cut,"goff");
8fb3bea1 1262 if (entries<=0) {
1263 TStatToolkit t;
d02b8756 1264 t.Error("TStatToolkit::MakeGraphSparse",Form("Empty or Not valid expression (%s) or cut (%s)", expr, cut));
8fb3bea1 1265 return 0;
1266 }
df0a2a0a 1267 // TGraph * graph = (TGraph*)gPad->GetPrimitive("Graph"); // 2D
d02b8756 1268
1269 Double_t *graphY, *graphX;
1270 graphY = tree->GetV1();
1271 graphX = tree->GetV2();
1272
1273 // sort according to run number
eda18694 1274 Int_t *index = new Int_t[entries*4];
d02b8756 1275 TMath::Sort(entries,graphX,index,kFALSE);
df0a2a0a 1276
d02b8756 1277 // define arrays for the new graph
1278 Double_t *unsortedX = new Double_t[entries];
1279 Int_t *runNumber = new Int_t[entries];
df0a2a0a 1280 Double_t count = 0.5;
d02b8756 1281
1282 // evaluate arrays for the new graph according to the run-number
baa0041d 1283 Int_t icount=0;
d02b8756 1284 //first entry
1285 unsortedX[index[0]] = count;
1286 runNumber[0] = graphX[index[0]];
1287 // loop the rest of entries
1288 for(Int_t i=1;i<entries;i++)
1289 {
1290 if(graphX[index[i]]==graphX[index[i-1]])
1291 unsortedX[index[i]] = count;
1292 else if(graphX[index[i]]!=graphX[index[i-1]]){
df0a2a0a 1293 count++;
baa0041d 1294 icount++;
d02b8756 1295 unsortedX[index[i]] = count;
1296 runNumber[icount]=graphX[index[i]];
df0a2a0a 1297 }
1298 }
d02b8756 1299
1300 // count the number of xbins (run-wise) for the new graph
df0a2a0a 1301 const Int_t newNbins = int(count+0.5);
1302 Double_t *newBins = new Double_t[newNbins+1];
1303 for(Int_t i=0; i<=count+1;i++){
1304 newBins[i] = i;
1305 }
d02b8756 1306
1307 // define and fill the new graph
b3453fe7 1308 TGraph *graphNew = 0;
d02b8756 1309 if (tree->GetV3()) {
1310 if (tree->GetV4()) {
1311 graphNew = new TGraphErrors(entries,unsortedX,graphY,tree->GetV4(),tree->GetV3());
1312 }
1313 else { graphNew = new TGraphErrors(entries,unsortedX,graphY,0,tree->GetV3()); }
1314 }
1315 else { graphNew = new TGraphErrors(entries,unsortedX,graphY,0,0); }
1316 // with "Set(...)", the x-axis is being sorted
df0a2a0a 1317 graphNew->GetXaxis()->Set(newNbins,newBins);
d02b8756 1318
1319 // set the bins for the x-axis, apply shifting of points
df0a2a0a 1320 Char_t xName[50];
df0a2a0a 1321 for(Int_t i=0;i<count;i++){
d02b8756 1322 snprintf(xName,50,"%d",runNumber[i]);
df0a2a0a 1323 graphNew->GetXaxis()->SetBinLabel(i+1,xName);
d02b8756 1324 graphNew->GetX()[i]+=offset;
df0a2a0a 1325 }
d02b8756 1326
df0a2a0a 1327 graphNew->GetHistogram()->SetTitle("");
d02b8756 1328 graphNew->SetMarkerStyle(mstyle);
b3453fe7 1329 graphNew->SetMarkerColor(mcolor);
70989f8d 1330 if (msize>0) graphNew->SetMarkerSize(msize);
d02b8756 1331 delete [] unsortedX;
1332 delete [] runNumber;
df0a2a0a 1333 delete [] index;
1334 delete [] newBins;
d02b8756 1335 //
df0a2a0a 1336 return graphNew;
1337}
1338
377a7d60 1339
1340
1341//
d02b8756 1342// functions used for the trending
377a7d60 1343//
1344
1345Int_t TStatToolkit::MakeStatAlias(TTree * tree, const char * expr, const char * cut, const char * alias)
1346{
1347 //
1348 // Add alias using statistical values of a given variable.
1349 // (by MI, Patrick Reichelt)
1350 //
1351 // tree - input tree
1352 // expr - variable expression
1353 // cut - selection criteria
1354 // Output - return number of entries used to define variable
1355 // In addition mean, rms, median, and robust mean and rms (choosing fraction of data with smallest RMS)
1356 //
d02b8756 1357 /* Example usage:
1358 1.) create the robust estimators for variable expr="QA.TPC.CPass1.meanTPCncl" and create a corresponding
1359 aliases with the prefix alias[0]="ncl", calculated using fraction alias[1]="0.90"
1360
1361 TStatToolkit::MakeStatAlias(tree,"QA.TPC.CPass1.meanTPCncl","QA.TPC.CPass1.status>0","ncl:0.9");
1362 root [4] tree->GetListOfAliases().Print()
1363 OBJ: TNamed ncl_Median (130.964333+0)
1364 OBJ: TNamed ncl_Mean (122.120387+0)
1365 OBJ: TNamed ncl_RMS (33.509623+0)
1366 OBJ: TNamed ncl_Mean90 (131.503862+0)
1367 OBJ: TNamed ncl_RMS90 (3.738260+0)
1368 */
377a7d60 1369 //
d02b8756 1370 Int_t entries = tree->Draw(expr,cut,"goff");
377a7d60 1371 if (entries<=1){
1372 printf("Expression or cut not valid:\t%s\t%s\n", expr, cut);
1373 return 0;
1374 }
1375 //
1376 TObjArray* oaAlias = TString(alias).Tokenize(":");
1377 if (oaAlias->GetEntries()<2) return 0;
1378 Float_t entryFraction = atof( oaAlias->At(1)->GetName() );
1379 //
1380 Double_t median = TMath::Median(entries,tree->GetV1());
d02b8756 1381 Double_t mean = TMath::Mean(entries,tree->GetV1());
377a7d60 1382 Double_t rms = TMath::RMS(entries,tree->GetV1());
1383 Double_t meanEF=0, rmsEF=0;
1384 TStatToolkit::EvaluateUni(entries, tree->GetV1(), meanEF, rmsEF, entries*entryFraction);
1385 //
d02b8756 1386 tree->SetAlias(Form("%s_Median",oaAlias->At(0)->GetName()), Form("(%f+0)",median));
377a7d60 1387 tree->SetAlias(Form("%s_Mean",oaAlias->At(0)->GetName()), Form("(%f+0)",mean));
1388 tree->SetAlias(Form("%s_RMS",oaAlias->At(0)->GetName()), Form("(%f+0)",rms));
377a7d60 1389 tree->SetAlias(Form("%s_Mean%d",oaAlias->At(0)->GetName(),Int_t(entryFraction*100)), Form("(%f+0)",meanEF));
1390 tree->SetAlias(Form("%s_RMS%d",oaAlias->At(0)->GetName(),Int_t(entryFraction*100)), Form("(%f+0)",rmsEF));
1391 delete oaAlias;
1392 return entries;
1393}
1394
1395Int_t TStatToolkit::SetStatusAlias(TTree * tree, const char * expr, const char * cut, const char * alias)
1396{
1397 //
1398 // Add alias to trending tree using statistical values of a given variable.
1399 // (by MI, Patrick Reichelt)
1400 //
1401 // format of expr : varname (e.g. meanTPCncl)
1402 // format of cut : char like in TCut
1403 // format of alias: alias:query:entryFraction(EF) (fraction of entries used for uniformity evaluation)
1404 // e.g.: varname_Out:(abs(varname-meanEF)>6.*rmsEF):0.8
d02b8756 1405 // available internal variables are: 'varname, Median, Mean, MeanEF, RMS, RMSEF'
1406 // in the alias, 'varname' will be replaced by its content, and 'EF' by the percentage (e.g. MeanEF -> Mean80)
377a7d60 1407 //
1408 /* Example usage:
d02b8756 1409 1.) Define robust mean (possible, but easier done with TStatToolkit::MakeStatAlias(...))
1410 TStatToolkit::SetStatusAlias(tree, "meanTPCnclF", "meanTPCnclF>0", "meanTPCnclF_MeanEF:MeanEF:0.80") ;
377a7d60 1411 root [10] tree->GetListOfAliases()->Print()
1412 Collection name='TList', class='TList', size=1
d02b8756 1413 OBJ: TNamed meanTPCnclF_Mean80 0.899308
377a7d60 1414 2.) create alias outlyers - 6 sigma cut
d02b8756 1415 TStatToolkit::SetStatusAlias(tree, "meanTPCnclF", "meanTPCnclF>0", "meanTPCnclF_Out:(abs(meanTPCnclF-MeanEF)>6.*RMSEF):0.8")
1416 meanTPCnclF_Out ==> (abs(meanTPCnclF-0.899308)>6.*0.016590)
377a7d60 1417 3.) the same functionality as in 2.)
d02b8756 1418 TStatToolkit::SetStatusAlias(tree, "meanTPCnclF", "meanTPCnclF>0", "varname_Out2:(abs(varname-MeanEF)>6.*RMSEF):0.8")
1419 meanTPCnclF_Out2 ==> (abs(meanTPCnclF-0.899308)>6.*0.016590)
377a7d60 1420 */
1421 //
d02b8756 1422 Int_t entries = tree->Draw(expr,cut,"goff");
1423 if (entries<=1){
1424 printf("Expression or cut not valid:\t%s\t%s\n", expr, cut);
1425 return 0;
1426 }
1427 //
377a7d60 1428 TObjArray* oaVar = TString(expr).Tokenize(":");
1429 char varname[50];
377a7d60 1430 snprintf(varname,50,"%s", oaVar->At(0)->GetName());
d02b8756 1431 //
377a7d60 1432 TObjArray* oaAlias = TString(alias).Tokenize(":");
d02b8756 1433 if (oaAlias->GetEntries()<3) return 0;
377a7d60 1434 Float_t entryFraction = atof( oaAlias->At(2)->GetName() );
1435 //
377a7d60 1436 Double_t median = TMath::Median(entries,tree->GetV1());
d02b8756 1437 Double_t mean = TMath::Mean(entries,tree->GetV1());
377a7d60 1438 Double_t rms = TMath::RMS(entries,tree->GetV1());
1439 Double_t meanEF=0, rmsEF=0;
1440 TStatToolkit::EvaluateUni(entries, tree->GetV1(), meanEF, rmsEF, entries*entryFraction);
377a7d60 1441 //
1442 TString sAlias( oaAlias->At(0)->GetName() );
1443 sAlias.ReplaceAll("varname",varname);
d02b8756 1444 sAlias.ReplaceAll("MeanEF", Form("Mean%1.0f",entryFraction*100) );
1445 sAlias.ReplaceAll("RMSEF", Form("RMS%1.0f",entryFraction*100) );
377a7d60 1446 TString sQuery( oaAlias->At(1)->GetName() );
1447 sQuery.ReplaceAll("varname",varname);
1448 sQuery.ReplaceAll("MeanEF", Form("%f",meanEF) );
d02b8756 1449 sQuery.ReplaceAll("RMSEF", Form("%f",rmsEF) ); //make sure to replace 'RMSEF' before 'RMS'...
377a7d60 1450 sQuery.ReplaceAll("Median", Form("%f",median) );
d02b8756 1451 sQuery.ReplaceAll("Mean", Form("%f",mean) );
377a7d60 1452 sQuery.ReplaceAll("RMS", Form("%f",rms) );
d02b8756 1453 printf("define alias:\t%s = %s\n", sAlias.Data(), sQuery.Data());
377a7d60 1454 //
1455 char query[200];
1456 char aname[200];
1457 snprintf(query,200,"%s", sQuery.Data());
1458 snprintf(aname,200,"%s", sAlias.Data());
1459 tree->SetAlias(aname, query);
d02b8756 1460 delete oaVar;
1461 delete oaAlias;
377a7d60 1462 return entries;
1463}
1464
1465TMultiGraph* TStatToolkit::MakeStatusMultGr(TTree * tree, const char * expr, const char * cut, const char * alias, Int_t igr)
1466{
1467 //
1468 // Compute a trending multigraph that shows for which runs a variable has outliers.
1469 // (by MI, Patrick Reichelt)
1470 //
1471 // format of expr : varname:xaxis (e.g. meanTPCncl:run)
1472 // format of cut : char like in TCut
1473 // format of alias: (1):(varname_Out==0):(varname_Out)[:(varname_Warning):...]
1474 // in the alias, 'varname' will be replaced by its content (e.g. varname_Out -> meanTPCncl_Out)
d02b8756 1475 // note: the aliases 'varname_Out' etc have to be defined by function TStatToolkit::SetStatusAlias(...)
377a7d60 1476 // counter igr is used to shift the multigraph in y when filling a TObjArray.
1477 //
1478 TObjArray* oaVar = TString(expr).Tokenize(":");
d02b8756 1479 if (oaVar->GetEntries()<2) return 0;
377a7d60 1480 char varname[50];
1481 char var_x[50];
1482 snprintf(varname,50,"%s", oaVar->At(0)->GetName());
1483 snprintf(var_x ,50,"%s", oaVar->At(1)->GetName());
d02b8756 1484 //
377a7d60 1485 TString sAlias(alias);
1486 sAlias.ReplaceAll("varname",varname);
1487 TObjArray* oaAlias = TString(sAlias.Data()).Tokenize(":");
d02b8756 1488 if (oaAlias->GetEntries()<3) return 0;
377a7d60 1489 //
1490 char query[200];
1491 TMultiGraph* multGr = new TMultiGraph();
1492 Int_t marArr[6] = {24+igr%2, 20+igr%2, 20+igr%2, 20+igr%2, 22, 23};
1493 Int_t colArr[6] = {kBlack, kBlack, kRed, kOrange, kMagenta, kViolet};
1494 Double_t sizArr[6] = {1.2, 1.1, 1.0, 1.0, 1, 1};
1495 const Int_t ngr = oaAlias->GetEntriesFast();
1496 for (Int_t i=0; i<ngr; i++){
1497 if (i==2) continue; // the Fatal(Out) graph will be added in the end to be plotted on top!
1498 snprintf(query,200, "%f*(%s-0.5):%s", 1.+igr, oaAlias->At(i)->GetName(), var_x);
d02b8756 1499 multGr->Add( (TGraphErrors*) TStatToolkit::MakeGraphSparse(tree,query,cut,marArr[i],colArr[i],sizArr[i]) );
377a7d60 1500 }
1501 snprintf(query,200, "%f*(%s-0.5):%s", 1.+igr, oaAlias->At(2)->GetName(), var_x);
d02b8756 1502 multGr->Add( (TGraphErrors*) TStatToolkit::MakeGraphSparse(tree,query,cut,marArr[2],colArr[2],sizArr[2]) );
377a7d60 1503 //
1504 multGr->SetName(varname);
1505 multGr->SetTitle(varname); // used for y-axis labels. // details to be included!
d02b8756 1506 delete oaVar;
1507 delete oaAlias;
377a7d60 1508 return multGr;
1509}
1510
1511
1512void TStatToolkit::AddStatusPad(TCanvas* c1, Float_t padratio, Float_t bottommargin)
1513{
1514 //
1515 // add pad to bottom of canvas for Status graphs (by Patrick Reichelt)
1516 // call function "DrawStatusGraphs(...)" afterwards
1517 //
1518 TCanvas* c1_clone = (TCanvas*) c1->Clone("c1_clone");
1519 c1->Clear();
1520 // produce new pads
1521 c1->cd();
1522 TPad* pad1 = new TPad("pad1", "pad1", 0., padratio, 1., 1.);
1523 pad1->Draw();
1524 pad1->SetNumber(1); // so it can be called via "c1->cd(1);"
1525 c1->cd();
1526 TPad* pad2 = new TPad("pad2", "pad2", 0., 0., 1., padratio);
1527 pad2->Draw();
1528 pad2->SetNumber(2);
1529 // draw original canvas into first pad
1530 c1->cd(1);
1531 c1_clone->DrawClonePad();
1532 pad1->SetBottomMargin(0.001);
1533 pad1->SetRightMargin(0.01);
1534 // set up second pad
1535 c1->cd(2);
1536 pad2->SetGrid(3);
1537 pad2->SetTopMargin(0);
1538 pad2->SetBottomMargin(bottommargin); // for the long x-axis labels (runnumbers)
1539 pad2->SetRightMargin(0.01);
1540}
1541
1542
1543void TStatToolkit::DrawStatusGraphs(TObjArray* oaMultGr)
1544{
1545 //
1546 // draw Status graphs into active pad of canvas (by MI, Patrick Reichelt)
1547 // ...into bottom pad, if called after "AddStatusPad(...)"
1548 //
1549 const Int_t nvars = oaMultGr->GetEntriesFast();
1550 TGraph* grAxis = (TGraph*) ((TMultiGraph*) oaMultGr->At(0))->GetListOfGraphs()->At(0);
1551 grAxis->SetMaximum(0.5*nvars+0.5);
1552 grAxis->SetMinimum(0);
1553 grAxis->GetYaxis()->SetLabelSize(0);
1554 Int_t entries = grAxis->GetN();
1555 printf("entries (via GetN()) = %d\n",entries);
1556 grAxis->GetXaxis()->SetLabelSize(5.7*TMath::Min(TMath::Max(5./entries,0.01),0.03));
1557 grAxis->GetXaxis()->LabelsOption("v");
1558 grAxis->Draw("ap");
1559 //
1560 // draw multigraphs & names of status variables on the y axis
1561 for (Int_t i=0; i<nvars; i++){
1562 ((TMultiGraph*) oaMultGr->At(i))->Draw("p");
1563 TLatex* ylabel = new TLatex(-0.1, 0.5*i+0.5, ((TMultiGraph*) oaMultGr->At(i))->GetTitle());
1564 ylabel->SetTextAlign(32); //hor:right & vert:centered
1565 ylabel->SetTextSize(0.025/gPad->GetHNDC());
1566 ylabel->Draw();
1567 }
1568}
376089b6 1569
1570
1571void TStatToolkit::DrawHistogram(TTree * tree, const char* drawCommand, const char* cuts, const char* histoname, const char* histotitle, Int_t nsigma, Float_t fraction )
1572{
1573 //
1574 // Draw histogram from TTree with robust range
1575 // Only for 1D so far!
1576 //
1577 // Parameters:
1578 // - histoname: name of histogram
1579 // - histotitle: title of histgram
1580 // - fraction: fraction of data to define the robust mean
1581 // - nsigma: nsigma value for range
1582 //
1583
1584 TString drawStr(drawCommand);
1585 TString cutStr(cuts);
1586 Int_t dim = 1;
1587
376089b6 1588 if(!tree) {
1589 cerr<<" Tree pointer is NULL!"<<endl;
1590 return;
1591 }
1592
3240a856 1593 // get entries
376089b6 1594 Int_t entries = tree->Draw(drawStr.Data(), cutStr.Data(), "goff");
1595 if (entries == -1) {
1596 cerr<<"TTree draw returns -1"<<endl;
1597 return;
1598 }
1599
3240a856 1600 // get dimension
1601 if(tree->GetV1()) dim = 1;
1602 if(tree->GetV2()) dim = 2;
1603 if(tree->GetV3()) dim = 3;
1604 if(dim > 2){
1605 cerr<<"TTree has more than 2 dimensions (not yet supported)"<<endl;
1606 return;
1607 }
1608
1609 // draw robust
1610 Double_t meanX, rmsX=0;
1611 Double_t meanY, rmsY=0;
1612 TStatToolkit::EvaluateUni(entries, tree->GetV1(),meanX,rmsX, fraction*entries);
1613 if(dim==2){
1614 TStatToolkit::EvaluateUni(entries, tree->GetV1(),meanY,rmsY, fraction*entries);
1615 TStatToolkit::EvaluateUni(entries, tree->GetV2(),meanX,rmsX, fraction*entries);
1616 }
1617 TH1* hOut;
1618 if(dim==1){
1619 hOut = new TH1F(histoname, histotitle, 200, meanX-nsigma*rmsX, meanX+nsigma*rmsX);
1620 for (Int_t i=0; i<entries; i++) hOut->Fill(tree->GetV1()[i]);
1621 hOut->GetXaxis()->SetTitle(tree->GetHistogram()->GetXaxis()->GetTitle());
1622 hOut->Draw();
1623 }
1624 else if(dim==2){
1625 hOut = new TH2F(histoname, histotitle, 200, meanX-nsigma*rmsX, meanX+nsigma*rmsX,200, meanY-nsigma*rmsY, meanY+nsigma*rmsY);
1626 for (Int_t i=0; i<entries; i++) hOut->Fill(tree->GetV2()[i],tree->GetV1()[i]);
1627 hOut->GetXaxis()->SetTitle(tree->GetHistogram()->GetXaxis()->GetTitle());
1628 hOut->GetYaxis()->SetTitle(tree->GetHistogram()->GetYaxis()->GetTitle());
1629 hOut->Draw("colz");
1630 }
376089b6 1631
1632}