|
|
|
00001 #include "RecoEgamma/EgammaTools/interface/ConversionLikelihoodCalculator.h" 00002 00003 00004 ConversionLikelihoodCalculator::ConversionLikelihoodCalculator() 00005 { 00006 reader_ = new TMVA::Reader("!Color:Silent"); 00007 00008 // std::cout << "Init Reader()" << std::endl; 00009 00010 reader_->AddVariable("log(e_over_p)", &log_e_over_p_); 00011 reader_->AddVariable("log(abs(cot_theta))", &log_abs_cot_theta_); 00012 reader_->AddVariable("log(abs(delta_phi))", &log_abs_delta_phi_); 00013 reader_->AddVariable("log(chi2_max_pt)", &log_chi2_max_pt_); 00014 reader_->AddVariable("log(chi2_min_pt)", &log_chi2_min_pt_); 00015 00016 } 00017 00018 void ConversionLikelihoodCalculator::setWeightsFile(const char * weightsFile) 00019 { 00020 // std::cout << "Before BookMVA " << weightsFile << std::endl; 00021 reader_->BookMVA("Likelihood", weightsFile); 00022 // std::cout << "After BookMVA" << std::endl; 00023 } 00024 00025 double ConversionLikelihoodCalculator::calculateLikelihood(reco::ConversionRef conversion) 00026 { 00027 if (conversion->nTracks() != 2) return -1.; 00028 00029 log_e_over_p_ = log(conversion->EoverP()); 00030 00031 log_abs_cot_theta_ = log(fabs(conversion->pairCotThetaSeparation())); 00032 00033 double delta_phi = conversion->tracks()[0]->innerMomentum().phi()-conversion->tracks()[1]->innerMomentum().phi(); 00034 double pi = 3.14159265; 00035 // phi normalization 00036 while (delta_phi > pi) delta_phi -= 2*pi; 00037 while (delta_phi < -pi) delta_phi += 2*pi; 00038 log_abs_delta_phi_ = log(fabs(delta_phi)); 00039 00040 double chi2_1 = conversion->tracks()[0]->normalizedChi2(); 00041 double pt_1 = conversion->tracks()[0]->pt(); 00042 00043 double chi2_2 = conversion->tracks()[1]->normalizedChi2(); 00044 double pt_2 = conversion->tracks()[1]->pt(); 00045 00046 double chi2_max_pt=chi2_1; 00047 double chi2_min_pt=chi2_2; 00048 00049 if (pt_2 > pt_1) { 00050 chi2_max_pt=chi2_2; 00051 chi2_min_pt=chi2_1; 00052 } 00053 00054 log_chi2_max_pt_ = log(chi2_max_pt); 00055 log_chi2_min_pt_ = log(chi2_min_pt); 00056 00057 // std::cout << "log_e_over_p_ " << log_e_over_p_ << std::endl; 00058 // std::cout << "log_abs_cot_theta_ " << log_abs_cot_theta_ << std::endl; 00059 // std::cout << "log_abs_delta_phi_ " << log_abs_delta_phi_ << std::endl; 00060 // std::cout << "log_chi2_max_pt_ " << log_chi2_max_pt_ << std::endl; 00061 // std::cout << "log_chi2_min_pt_ " << log_chi2_min_pt_ << std::endl; 00062 std::vector<Float_t> inputVec; 00063 inputVec.push_back(log_e_over_p_ ); 00064 inputVec.push_back(log_abs_cot_theta_ ); 00065 inputVec.push_back(log_abs_delta_phi_ ); 00066 inputVec.push_back(log_chi2_max_pt_ ); 00067 inputVec.push_back(log_chi2_min_pt_ ); 00068 float like = reader_->EvaluateMVA(inputVec,"Likelihood"); 00069 // std::cout << "reader_->EvaluateMVA(\"Likelihood\") " << reader_->EvaluateMVA(inputVec,"Likelihood") << std::endl; 00070 00071 return like; 00072 } 00073 00074 double ConversionLikelihoodCalculator::calculateLikelihood(reco::Conversion& conversion) 00075 { 00076 00077 if (conversion.nTracks() != 2) return -1.; 00078 00079 log_e_over_p_ = log(conversion.EoverP()); 00080 00081 log_abs_cot_theta_ = log(fabs(conversion.pairCotThetaSeparation())); 00082 00083 00084 double delta_phi = conversion.tracksPin()[0].phi()-conversion.tracksPin()[1].phi(); 00085 double pi = 3.14159265; 00086 // phi normalization 00087 while (delta_phi > pi) delta_phi -= 2*pi; 00088 while (delta_phi < -pi) delta_phi += 2*pi; 00089 log_abs_delta_phi_ = log(fabs(delta_phi)); 00090 00091 double chi2_1 = conversion.tracks()[0]->normalizedChi2(); 00092 double pt_1 = conversion.tracks()[0]->pt(); 00093 00094 double chi2_2 = conversion.tracks()[1]->normalizedChi2(); 00095 double pt_2 = conversion.tracks()[1]->pt(); 00096 00097 double chi2_max_pt=chi2_1; 00098 double chi2_min_pt=chi2_2; 00099 00100 if (pt_2 > pt_1) { 00101 chi2_max_pt=chi2_2; 00102 chi2_min_pt=chi2_1; 00103 } 00104 00105 log_chi2_max_pt_ = log(chi2_max_pt); 00106 log_chi2_min_pt_ = log(chi2_min_pt); 00107 00108 // std::cout << "log_e_over_p_ " << log_e_over_p_ << std::endl; 00109 // std::cout << "log_abs_cot_theta_ " << log_abs_cot_theta_ << std::endl; 00110 // std::cout << "log_abs_delta_phi_ " << log_abs_delta_phi_ << std::endl; 00111 // std::cout << "log_chi2_max_pt_ " << log_chi2_max_pt_ << std::endl; 00112 // std::cout << "log_chi2_min_pt_ " << log_chi2_min_pt_ << std::endl; 00113 std::vector<Float_t> inputVec; 00114 inputVec.push_back(log_e_over_p_ ); 00115 inputVec.push_back(log_abs_cot_theta_ ); 00116 inputVec.push_back(log_abs_delta_phi_ ); 00117 inputVec.push_back(log_chi2_max_pt_ ); 00118 inputVec.push_back(log_chi2_min_pt_ ); 00119 float like = reader_->EvaluateMVA(inputVec,"Likelihood"); 00120 // std::cout << "reader_->EvaluateMVA(\"Likelihood\") " << reader_->EvaluateMVA(inputVec,"Likelihood") << std::endl; 00121 00122 return like; 00123 } 00124 00125
1.7.3