ConversionLikelihoodCalculator.cc

Go to the documentation of this file.

#include "RecoEgamma/EgammaTools/interface/ConversionLikelihoodCalculator.h"

#include "DataFormats/EgammaCandidates/interface/Conversion.h"

ConversionLikelihoodCalculator::ConversionLikelihoodCalculator() {
  reader_ = std::make_unique<TMVA::Reader>("!Color:Silent");

  //   std::cout << "Init Reader()" << std::endl;

  reader_->AddVariable("log(e_over_p)", &log_e_over_p_);
  reader_->AddVariable("log(abs(cot_theta))", &log_abs_cot_theta_);
  reader_->AddVariable("log(abs(delta_phi))", &log_abs_delta_phi_);
  reader_->AddVariable("log(chi2_max_pt)", &log_chi2_max_pt_);
  reader_->AddVariable("log(chi2_min_pt)", &log_chi2_min_pt_);
}

void ConversionLikelihoodCalculator::setWeightsFile(const char* weightsFile) {
  //   std::cout << "Before BookMVA " << weightsFile << std::endl;
  reader_->BookMVA("Likelihood", weightsFile);
  //   std::cout << "After  BookMVA" << std::endl;
}

double ConversionLikelihoodCalculator::calculateLikelihood(reco::ConversionRef conversion) {
  if (conversion->nTracks() != 2)
    return -1.;

  log_e_over_p_ = log(conversion->EoverP());

  log_abs_cot_theta_ = log(fabs(conversion->pairCotThetaSeparation()));

  double delta_phi = conversion->tracks()[0]->innerMomentum().phi() - conversion->tracks()[1]->innerMomentum().phi();
  double pi = 3.14159265;
  // phi normalization
  while (delta_phi > pi)
    delta_phi -= 2 * pi;
  while (delta_phi < -pi)
    delta_phi += 2 * pi;
  log_abs_delta_phi_ = log(fabs(delta_phi));

  double chi2_1 = conversion->tracks()[0]->normalizedChi2();
  double pt_1 = conversion->tracks()[0]->pt();

  double chi2_2 = conversion->tracks()[1]->normalizedChi2();
  double pt_2 = conversion->tracks()[1]->pt();

  double chi2_max_pt = chi2_1;
  double chi2_min_pt = chi2_2;

  if (pt_2 > pt_1) {
    chi2_max_pt = chi2_2;
    chi2_min_pt = chi2_1;
  }

  log_chi2_max_pt_ = log(chi2_max_pt);
  log_chi2_min_pt_ = log(chi2_min_pt);

  //   std::cout << "log_e_over_p_ " << log_e_over_p_ << std::endl;
  //   std::cout << "log_abs_cot_theta_ " << log_abs_cot_theta_ << std::endl;
  //   std::cout << "log_abs_delta_phi_ " << log_abs_delta_phi_ << std::endl;
  //   std::cout << "log_chi2_max_pt_ " << log_chi2_max_pt_ << std::endl;
  //   std::cout << "log_chi2_min_pt_ " << log_chi2_min_pt_ << std::endl;
  std::vector<Float_t> inputVec;
  inputVec.push_back(log_e_over_p_);
  inputVec.push_back(log_abs_cot_theta_);
  inputVec.push_back(log_abs_delta_phi_);
  inputVec.push_back(log_chi2_max_pt_);
  inputVec.push_back(log_chi2_min_pt_);
  float like = reader_->EvaluateMVA(inputVec, "Likelihood");
  //   std::cout << "reader_->EvaluateMVA(\"Likelihood\") " << reader_->EvaluateMVA(inputVec,"Likelihood") << std::endl;

  return like;
}

double ConversionLikelihoodCalculator::calculateLikelihood(reco::Conversion& conversion) {
  if (conversion.nTracks() != 2)
    return -1.;

  log_e_over_p_ = log(conversion.EoverP());

  log_abs_cot_theta_ = log(fabs(conversion.pairCotThetaSeparation()));

  double delta_phi = conversion.tracksPin()[0].phi() - conversion.tracksPin()[1].phi();
  double pi = 3.14159265;
  // phi normalization
  while (delta_phi > pi)
    delta_phi -= 2 * pi;
  while (delta_phi < -pi)
    delta_phi += 2 * pi;
  log_abs_delta_phi_ = log(fabs(delta_phi));

  double chi2_1 = conversion.tracks()[0]->normalizedChi2();
  double pt_1 = conversion.tracks()[0]->pt();

  double chi2_2 = conversion.tracks()[1]->normalizedChi2();
  double pt_2 = conversion.tracks()[1]->pt();

  double chi2_max_pt = chi2_1;
  double chi2_min_pt = chi2_2;

  if (pt_2 > pt_1) {
    chi2_max_pt = chi2_2;
    chi2_min_pt = chi2_1;
  }

  log_chi2_max_pt_ = log(chi2_max_pt);
  log_chi2_min_pt_ = log(chi2_min_pt);

  //   std::cout << "log_e_over_p_ " << log_e_over_p_ << std::endl;
  //   std::cout << "log_abs_cot_theta_ " << log_abs_cot_theta_ << std::endl;
  //   std::cout << "log_abs_delta_phi_ " << log_abs_delta_phi_ << std::endl;
  //   std::cout << "log_chi2_max_pt_ " << log_chi2_max_pt_ << std::endl;
  //   std::cout << "log_chi2_min_pt_ " << log_chi2_min_pt_ << std::endl;
  std::vector<Float_t> inputVec;
  inputVec.push_back(log_e_over_p_);
  inputVec.push_back(log_abs_cot_theta_);
  inputVec.push_back(log_abs_delta_phi_);
  inputVec.push_back(log_chi2_max_pt_);
  inputVec.push_back(log_chi2_min_pt_);
  float like = reader_->EvaluateMVA(inputVec, "Likelihood");
  //   std::cout << "reader_->EvaluateMVA(\"Likelihood\") " << reader_->EvaluateMVA(inputVec,"Likelihood") << std::endl;

  return like;
}