CDCDedxRunGainAlgorithm.cc

/**************************************************************************
 * basf2 (Belle II Analysis Software Framework)                           *
 * Author: The Belle II Collaboration                                     *
 *                                                                        *
 * See git log for contributors and copyright holders.                    *
 * This file is licensed under LGPL-3.0, see LICENSE.md.                  *
 **************************************************************************/
 
#include <reconstruction/calibration/CDCdEdx/CDCDedxRunGainAlgorithm.h>
 
#include <TF1.h>
#include <TCanvas.h>
#include <TLine.h>
#include <iostream>
#include <fstream>
#include <TH1I.h>
 
using namespace Belle2;
 
//-----------------------------------------------------------------
//                 Implementation
//-----------------------------------------------------------------
CDCDedxRunGainAlgorithm::CDCDedxRunGainAlgorithm() :
  CalibrationAlgorithm("CDCDedxElectronCollector"),
  isMakePlots(true),
  isMergePayload(true),
  fsrun(""),
  fSigLim(2.5),
  fdEdxBins(1000),
  fdEdxMin(0.0),
  fdEdxMax(5.0),
  fAdjust(1.00)
{
  // Set module properties
  setDescription("A calibration algorithm for CDC dE/dx run gains");
}
 
//-----------------------------------------------------------------
//                 Run the calibration
//-----------------------------------------------------------------
CalibrationAlgorithm::EResult CDCDedxRunGainAlgorithm::calibrate()
{
  B2INFO("Preparing dE/dx calibration for CDC dE/dx run gain");
 
  const auto expRun = getRunList()[0];
  updateDBObjPtrs(1, expRun.second, expRun.first);
  double ExistingRG = m_DBRunGain->getRunGain();
 
  // Get data objects
  auto ttree = getObjectPtr<TTree>("tree");
  if (!ttree)return c_NotEnoughData;
 
  double dedx;
  int run = 0, runtemp = 0;
  TString status = "";
  ttree->SetBranchAddress("dedx", &dedx);
  ttree->SetBranchAddress("run", &run);
 
  TH1D* hDedx = new TH1D("hDedx", "hDedx;dEdx rel (no hadsat);entries", fdEdxBins, fdEdxMin, fdEdxMax);
  TH1D* hDedxAbs = new TH1D("hDedxAbs", "hDedxAbs;dEdx abs(no hadsat);entries", fdEdxBins, fdEdxMin, fdEdxMax);
 
  for (int i = 0; i < ttree->GetEntries(); ++i) {
    ttree->GetEvent(i);
    if (dedx <= 0) continue;
    hDedx->Fill(dedx);
    hDedxAbs->Fill(dedx * ExistingRG);
    runtemp = run;
  }
 
  //min 500 good events otherwise merge to next run
  if (hDedx->Integral() < 1000) {
    B2INFO("Not enough data for run (" << runtemp << ") so far only " << hDedx->Integral() << " therefore adding next run");
    fsrun += Form("%d_", runtemp);
    delete hDedx;
    delete hDedxAbs;
    return c_NotEnoughData;
  }
 
  B2INFO("Total track for this run: " << runtemp << " = " << hDedx->Integral());
  fsrun += Form("%d", runtemp);
  hDedx->SetName(Form("%s_%s", hDedx->GetName(), fsrun.Data()));
  hDedxAbs->SetName(Form("%s_%s", hDedxAbs->GetName(), fsrun.Data()));
 
  //Fit absolute run for the reference only
  double RGabs_Mean = 1.0;
  FitGaussianWRange(hDedxAbs, status);
  if (status != "FitOK") {
    hDedxAbs->SetTitle(Form("dE/dx abs, Run %s (%s)", fsrun.Data(), status.Data()));
  } else {
    RGabs_Mean = hDedxAbs->GetFunction("gaus")->GetParameter(1);
    double RGabs_MeanErr = hDedxAbs->GetFunction("gaus")->GetParError(1);
    double RGabs_Sigma = hDedxAbs->GetFunction("gaus")->GetParameter(2);
    hDedxAbs->SetTitle(Form("dE/dx abs, Run %s (%s), #mu_{fit}: %0.04f#pm%0.04f,, #sigma_{fit}: %0.04f", fsrun.Data(), status.Data(),
                            RGabs_Mean, RGabs_MeanErr, RGabs_Sigma));
  }
 
  //Fit relative run for actual calibration
  double RGrel_Mean = 1.0, RGrel_MeanErr = -99.0, RGrel_Sigma = -99.0;
  FitGaussianWRange(hDedx, status);
  if (status != "FitOK") {
    RGrel_Mean = hDedx->GetMean(); //mean for hist for fit failure
    hDedx->SetTitle(Form("dE/dx abs, Run %s (%s)", fsrun.Data(), status.Data()));
  } else {
    RGrel_Mean = hDedx->GetFunction("gaus")->GetParameter(1);
    RGrel_MeanErr = hDedx->GetFunction("gaus")->GetParError(1);
    RGrel_Sigma = hDedx->GetFunction("gaus")->GetParameter(2);
    hDedx->SetTitle(Form("dE/dx rel, Run %s (%s), #mu_{fit}: %0.04f#pm%0.04f,, #sigma_{fit}: %0.04f", fsrun.Data(), status.Data(),
                         RGrel_Mean, RGrel_MeanErr, RGrel_Sigma));
  }
 
  if (RGrel_Mean <= 0.0)RGrel_Mean = 1.0; //safe button
 
  //for validation purpose
  if (isMakePlots) {
 
    std::ofstream fileRGout;
    fileRGout.open(Form("fRG_Constants_%s.txt", fsrun.Data()));
    fileRGout << expRun.second << " " << ExistingRG << " " << RGrel_Mean << " " << RGrel_MeanErr << " " << RGrel_Sigma << "\n";
    fileRGout.close();
 
    TCanvas* ctmp = new TCanvas("tmp", "tmp", 1000, 500);
    ctmp->SetBatch(kTRUE);
    ctmp->Divide(2, 1);
 
    ctmp->cd(1);
    hDedx->SetFillColorAlpha(kYellow, 0.30);
    hDedx->DrawCopy("");
 
    TLine* tl = new TLine();
    tl->SetLineColor(kRed);
    tl->SetX1(RGrel_Mean); tl->SetX2(RGrel_Mean);
    tl->SetY1(0); tl->SetY2(hDedx->GetMaximum());
    tl->DrawClone("same");
 
    ctmp->cd(2);
    hDedxAbs->SetFillColorAlpha(kBlue, 0.10);
    hDedxAbs->DrawCopy("");
 
    TLine* tlAbs = new TLine();
    tlAbs->SetLineColor(kRed);
    tlAbs->SetX1(RGabs_Mean); tlAbs->SetX2(RGabs_Mean);
    tlAbs->SetY1(0); tlAbs->SetY2(hDedxAbs->GetMaximum());
    tlAbs->DrawClone("same");
 
    ctmp->Print(Form("cdcdedx_rungain_fits_%s_%s.pdf", fsrun.Data(), status.Data()));
 
    TCanvas* cstats = new TCanvas("cstats", "cstats", 1000, 500);
    cstats->SetBatch(kTRUE);
    cstats->Divide(2, 1);
    cstats->cd(1);
    auto hestats = getObjectPtr<TH1I>("hestats");
    if (hestats)hestats->DrawCopy("");
    cstats->cd(2);
    auto htstats = getObjectPtr<TH1I>("htstats");
    if (htstats)htstats->DrawCopy("");
    cstats->Print(Form("cdcdedx_rungain_stats_%s.pdf", fsrun.Data()));
 
    delete ctmp;
    delete tl;
    delete tlAbs;
    delete cstats;
  }
 
  B2INFO("Saving new rung for (Exp, Run) : (" << expRun.first << "," << expRun.second << ")");
  generateNewPayloads(RGrel_Mean, ExistingRG);
 
  fsrun.Resize(0);
  delete hDedx;
  delete hDedxAbs;
  return c_OK;
}
 
 
void CDCDedxRunGainAlgorithm::generateNewPayloads(double RGrel_Mean, double ExistingRG)
{
  if (isMergePayload) {
    B2INFO("--> RunGain: Previous = " << ExistingRG << ", Relative = " << RGrel_Mean << ", Adjustment = " << fAdjust <<
           ", Merged (saved) = " << RGrel_Mean*
           ExistingRG * fAdjust);
    RGrel_Mean *= ExistingRG;
    RGrel_Mean *= fAdjust; //default is 1.0
  } else {
    B2INFO("--> RunGain: Previous = " << ExistingRG << ", Relative (saved) = " << RGrel_Mean);
  }
 
  CDCDedxRunGain* gain = new CDCDedxRunGain(RGrel_Mean);
  saveCalibration(gain, "CDCDedxRunGain");
}
 
 
void CDCDedxRunGainAlgorithm::FitGaussianWRange(TH1D*& temphist, TString& status)
{
 
  double histmean = temphist->GetMean();
  double histrms = temphist->GetRMS();
  temphist->GetXaxis()->SetRangeUser(histmean - 5.0 * histrms, histmean + 5.0 * histrms);
 
  int fs = temphist->Fit("gaus", "Q0");
  if (fs != 0) {
    B2INFO(Form("\tFit (round 1) for hist (%s) failed (status = %d)", temphist->GetName(), fs));
    status = "FitFailed";
    return;
  } else {
    double mean = temphist->GetFunction("gaus")->GetParameter(1);
    double width = temphist->GetFunction("gaus")->GetParameter(2);
    temphist->GetXaxis()->SetRangeUser(mean - 5.0 * width, mean + 5.0 * width);
    fs = temphist->Fit("gaus", "QR", "", mean - fSigLim * width, mean + fSigLim * width);
    if (fs != 0) {
      B2INFO(Form("\tFit (round 2) for hist (%s) failed (status = %d)", temphist->GetName(), fs));
      status = "FitFailed";
      return;
    } else {
      temphist->GetXaxis()->SetRangeUser(mean - 5.0 * width, mean + 5.0 * width);
      B2INFO(Form("\tFit for hist (%s) successful (status = %d)", temphist->GetName(), fs));
      status = "FitOK";
    }
  }
}