Inheritance diagram for DeepTauId:

Public Member Functions
	DeepTauId (const edm::ParameterSet &cfg, const deep_tau::DeepTauCache *cache)

Public Member Functions inherited from deep_tau::DeepTauBase
	DeepTauBase (const edm::ParameterSet &cfg, const OutputCollection &outputs, const DeepTauCache *cache)

void	produce (edm::Event &event, const edm::EventSetup &es) override

	~DeepTauBase () override

Public Member Functions inherited from edm::stream::EDProducer< edm::GlobalCache< DeepTauCache > >
	EDProducer ()=default

bool	hasAbilityToProduceInLumis () const final

bool	hasAbilityToProduceInRuns () const final

Static Public Member Functions
static void	fillDescriptions (edm::ConfigurationDescriptions &descriptions)

static const OutputCollection &	GetOutputs ()

static void	globalEndJob (const deep_tau::DeepTauCache *cache_)

static std::unique_ptr< deep_tau::DeepTauCache >	initializeGlobalCache (const edm::ParameterSet &cfg)

Static Public Member Functions inherited from deep_tau::DeepTauBase
static void	globalEndJob (const DeepTauCache *cache)

static std::unique_ptr< DeepTauCache >	initializeGlobalCache (const edm::ParameterSet &cfg)

Static Public Attributes
static float	default_value = -999.

Private Member Functions
void	checkInputs (const tensorflow::Tensor &inputs, const char *block_name, int n_inputs, int n_eta=1, int n_phi=1) const

void	createConvFeatures (const TauType &tau, const reco::Vertex &pv, double rho, const pat::ElectronCollection &electrons, const pat::MuonCollection &muons, const pat::PackedCandidateCollection &pfCands, const CellGrid &grid, bool is_inner)

void	createEgammaBlockInputs (const TauType &tau, const reco::Vertex &pv, double rho, const pat::ElectronCollection &electrons, const pat::PackedCandidateCollection &pfCands, const Cell &cell_map, bool is_inner)

void	createHadronsBlockInputs (const TauType &tau, const reco::Vertex &pv, double rho, const pat::PackedCandidateCollection &pfCands, const Cell &cell_map, bool is_inner)

template<typename dnn >
tensorflow::Tensor	createInputsV1 (const TauType &tau, const ElectronCollection &electrons, const MuonCollection &muons) const

void	createMuonBlockInputs (const TauType &tau, const reco::Vertex &pv, double rho, const pat::MuonCollection &muons, const pat::PackedCandidateCollection &pfCands, const Cell &cell_map, bool is_inner)

void	createTauBlockInputs (const TauType &tau, const reco::Vertex &pv, double rho)

template<typename Collection >
void	fillGrids (const TauType &tau, const Collection &objects, CellGrid &inner_grid, CellGrid &outer_grid)

void	getPartialPredictions (tensorflow::Tensor &convTensor, bool is_inner, int eta_index, int phi_index)

tensorflow::Tensor	getPredictions (edm::Event &event, const edm::EventSetup &es, edm::Handle< TauCollection > taus) override

void	getPredictionsV1 (const TauType &tau, const pat::ElectronCollection &electrons, const pat::MuonCollection &muons, std::vector< tensorflow::Tensor > &pred_vector)

void	getPredictionsV2 (const TauType &tau, const pat::ElectronCollection &electrons, const pat::MuonCollection &muons, const pat::PackedCandidateCollection &pfCands, const reco::Vertex &pv, double rho, std::vector< tensorflow::Tensor > &pred_vector)

void	setCellConvFeatures (tensorflow::Tensor &convTensor, const tensorflow::Tensor &features, int eta_index, int phi_index)

Static Private Member Functions
static void	calculateElectronClusterVars (const pat::Electron *ele, float &elecEe, float &elecEgamma)

static bool	calculateElectronClusterVarsV2 (const pat::Electron &ele, float &cc_ele_energy, float &cc_gamma_energy, int &cc_n_gamma)

static bool	calculateGottfriedJacksonAngleDifference (const pat::Tau &tau, double &gj_diff)

static float	calculateGottfriedJacksonAngleDifference (const pat::Tau &tau)

static const pat::Electron *	findMatchedElectron (const pat::Tau &tau, const pat::ElectronCollection &electrons, double deltaR)

static double	getInnerSignalConeRadius (double pt)

template<typename T >
static float	getValue (T value)

template<typename T >
static float	getValueLinear (T value, float min_value, float max_value, bool positive)

template<typename T >
static float	getValueNorm (T value, float mean, float sigma, float n_sigmas_max=5)

static bool	isInEcalCrack (double eta)

template<typename CandidateCollection >
static void	processIsolationPFComponents (const pat::Tau &tau, const CandidateCollection &candidates, LorentzVectorXYZ &p4, float &pt, float &d_eta, float &d_phi, float &m, float &n)

template<typename CandidateCollection >
static void	processSignalPFComponents (const pat::Tau &tau, const CandidateCollection &candidates, LorentzVectorXYZ &p4_inner, LorentzVectorXYZ &p4_outer, float &pt_inner, float &dEta_inner, float &dPhi_inner, float &m_inner, float &pt_outer, float &dEta_outer, float &dPhi_outer, float &m_outer, float &n_inner, float &n_outer)

Private Attributes
std::array< std::shared_ptr< tensorflow::Tensor >, 2 >	convTensor_

const int	debug_level

const bool	disable_dxy_pca_

std::array< std::shared_ptr< tensorflow::Tensor >, 2 >	eGammaTensor_

edm::EDGetTokenT< ElectronCollection >	electrons_token_

std::array< std::shared_ptr< tensorflow::Tensor >, 2 >	hadronsTensor_

std::string	input_layer_

edm::EDGetTokenT< MuonCollection >	muons_token_

std::array< std::shared_ptr< tensorflow::Tensor >, 2 >	muonTensor_

std::string	output_layer_

edm::EDGetTokenT< double >	rho_token_

std::shared_ptr< tensorflow::Tensor >	tauBlockTensor_

const unsigned	version

std::array< std::shared_ptr< tensorflow::Tensor >, 2 >	zeroOutputTensor_

Static Private Attributes
static float	pi = M_PI

Additional Inherited Members
Public Types inherited from deep_tau::DeepTauBase
using	Cutter = TauWPThreshold

using	CutterPtr = std::unique_ptr< Cutter >

using	ElectronCollection = pat::ElectronCollection

using	LorentzVectorXYZ = ROOT::Math::LorentzVector< ROOT::Math::PxPyPzE4D< double >>

using	MuonCollection = pat::MuonCollection

using	OutputCollection = std::map< std::string, Output >

using	TauCollection = std::vector< TauType >

using	TauDiscriminator = pat::PATTauDiscriminator

using	TauRef = edm::Ref< TauCollection >

using	TauRefProd = edm::RefProd< TauCollection >

using	TauType = pat::Tau

using	WPMap = std::map< std::string, CutterPtr >

Public Types inherited from edm::stream::EDProducer< edm::GlobalCache< DeepTauCache > >
typedef CacheContexts< T... >	CacheTypes

typedef CacheTypes::GlobalCache	GlobalCache

typedef AbilityChecker< T... >	HasAbility

typedef CacheTypes::LuminosityBlockCache	LuminosityBlockCache

typedef LuminosityBlockContextT< LuminosityBlockCache, RunCache, GlobalCache >	LuminosityBlockContext

typedef CacheTypes::LuminosityBlockSummaryCache	LuminosityBlockSummaryCache

typedef CacheTypes::RunCache	RunCache

typedef RunContextT< RunCache, GlobalCache >	RunContext

typedef CacheTypes::RunSummaryCache	RunSummaryCache

Protected Attributes inherited from deep_tau::DeepTauBase
const DeepTauCache *	cache_

OutputCollection	outputs_

edm::EDGetTokenT< pat::PackedCandidateCollection >	pfcandToken_

edm::EDGetTokenT< TauCollection >	tausToken_

edm::EDGetTokenT< reco::VertexCollection >	vtxToken_

std::map< std::string, WPMap >	workingPoints_

Detailed Description

Definition at line 551 of file DeepTauId.cc.

Constructor & Destructor Documentation

DeepTauId::DeepTauId	(	const edm::ParameterSet &	cfg,
		const deep_tau::DeepTauCache *	cache
	)

inlineexplicit

Definition at line 599 of file DeepTauId.cc.

References Exception, gen::n, dataset::name, and jets_cff::version.

                                                                                       :
         DeepTauBase(cfg, GetOutputs(), cache),
         electrons_token_(consumes<ElectronCollection>(cfg.getParameter<edm::InputTag>("electrons"))),
         muons_token_(consumes<MuonCollection>(cfg.getParameter<edm::InputTag>("muons"))),
         rho_token_(consumes<double>(cfg.getParameter<edm::InputTag>("rho"))),
         version(cfg.getParameter<unsigned>("version")),
         debug_level(cfg.getParameter<int>("debug_level")),
     disable_dxy_pca_(cfg.getParameter<bool>("disable_dxy_pca"))
     {
         if(version == 1) {
             input_layer_ = cache_->getGraph().node(0).name();
             output_layer_ = cache_->getGraph().node(cache_->getGraph().node_size() - 1).name();
             const auto& shape = cache_->getGraph().node(0).attr().at("shape").shape();
             if(shape.dim(1).size() != dnn_inputs_2017v1::NumberOfInputs)
                 throw cms::Exception("DeepTauId") << "number of inputs does not match the expected inputs for the given version";
         } else if(version == 2) {
             tauBlockTensor_ =  std::make_shared<tensorflow::Tensor>(tensorflow::DT_FLOAT, tensorflow::TensorShape{ 1,
                 dnn_inputs_2017_v2::TauBlockInputs::NumberOfInputs});
             for(size_t n = 0; n < 2; ++n) {
                 const bool is_inner = n == 0;
                 const auto n_cells = is_inner ? dnn_inputs_2017_v2::number_of_inner_cell
                                               : dnn_inputs_2017_v2::number_of_outer_cell;
                 eGammaTensor_[is_inner] = std::make_shared<tensorflow::Tensor>(tensorflow::DT_FLOAT,
                     tensorflow::TensorShape{1, 1, 1, dnn_inputs_2017_v2::EgammaBlockInputs::NumberOfInputs});
                 muonTensor_[is_inner] = std::make_shared<tensorflow::Tensor>(tensorflow::DT_FLOAT,
                     tensorflow::TensorShape{1, 1, 1, dnn_inputs_2017_v2::MuonBlockInputs::NumberOfInputs});
                 hadronsTensor_[is_inner] = std::make_shared<tensorflow::Tensor>(tensorflow::DT_FLOAT,
                     tensorflow::TensorShape{1, 1, 1, dnn_inputs_2017_v2::HadronBlockInputs::NumberOfInputs});
                 convTensor_[is_inner] = std::make_shared<tensorflow::Tensor>(tensorflow::DT_FLOAT,
                     tensorflow::TensorShape{1, n_cells, n_cells, dnn_inputs_2017_v2::number_of_conv_features});
                 zeroOutputTensor_[is_inner] = std::make_shared<tensorflow::Tensor>(tensorflow::DT_FLOAT,
                     tensorflow::TensorShape{1, 1, 1, dnn_inputs_2017_v2::number_of_conv_features});
 
 
                 eGammaTensor_[is_inner]->flat<float>().setZero();
                 muonTensor_[is_inner]->flat<float>().setZero();
                 hadronsTensor_[is_inner]->flat<float>().setZero();
                 getPartialPredictions(*zeroOutputTensor_[is_inner], is_inner, 0, 0);
 
             }
         } else {
             throw cms::Exception("DeepTauId") << "version " << version << " is not supported.";
         }
 
     }

Member Function Documentation

static void DeepTauId::calculateElectronClusterVars	(	const pat::Electron *	ele,
		float &	elecEe,
		float &	elecEgamma
	)

inlinestaticprivate

Definition at line 1578 of file DeepTauId.cc.

References pat::Electron::superCluster().

     {
         if(ele) {
             elecEe = elecEgamma = 0;
             auto superCluster = ele->superCluster();
             if(superCluster.isNonnull() && superCluster.isAvailable() && superCluster->clusters().isNonnull()
                     && superCluster->clusters().isAvailable()) {
                 for(auto iter = superCluster->clustersBegin(); iter != superCluster->clustersEnd(); ++iter) {
                     const double energy = (*iter)->energy();
                     if(iter == superCluster->clustersBegin()) elecEe += energy;
                     else elecEgamma += energy;
                 }
             }
         } else {
             elecEe = elecEgamma = default_value;
         }
     }

static bool DeepTauId::calculateElectronClusterVarsV2	(	const pat::Electron &	ele,
		float &	cc_ele_energy,
		float &	cc_gamma_energy,
		int &	cc_n_gamma
	)

inlinestaticprivate

Definition at line 682 of file DeepTauId.cc.

References pat::Electron::superCluster().

     {
         cc_ele_energy = cc_gamma_energy = 0;
         cc_n_gamma = 0;
         const auto& superCluster = ele.superCluster();
         if(superCluster.isNonnull() && superCluster.isAvailable() && superCluster->clusters().isNonnull()
             && superCluster->clusters().isAvailable()) {
             for(auto iter = superCluster->clustersBegin(); iter != superCluster->clustersEnd(); ++iter) {
                 const float energy = static_cast<float>((*iter)->energy());
                 if(iter == superCluster->clustersBegin())
                     cc_ele_energy += energy;
                 else {
                     cc_gamma_energy += energy;
                     ++cc_n_gamma;
                 }
             }
             return true;
         } else
             return false;
     }

static bool DeepTauId::calculateGottfriedJacksonAngleDifference	(	const pat::Tau &	tau,
		double &	gj_diff
	)

inlinestaticprivate

Definition at line 1659 of file DeepTauId.cc.

References funct::abs(), constexpr, pat::Tau::flightLength(), pat::Tau::hasSecondaryVertex(), reco::LeafCandidate::p(), reco::LeafCandidate::p4(), and funct::pow().

     {
         if(tau.hasSecondaryVertex()) {
             static constexpr double mTau = 1.77682;
             const double mAOne = tau.p4().M();
             const double pAOneMag = tau.p();
             const double argumentThetaGJmax = (std::pow(mTau,2) - std::pow(mAOne,2) ) / ( 2 * mTau * pAOneMag );
             const double argumentThetaGJmeasured = tau.p4().Vect().Dot(tau.flightLength())
                     / ( pAOneMag * tau.flightLength().R() );
             if ( std::abs(argumentThetaGJmax) <= 1. && std::abs(argumentThetaGJmeasured) <= 1. ) {
                 double thetaGJmax = std::asin( argumentThetaGJmax );
                 double thetaGJmeasured = std::acos( argumentThetaGJmeasured );
                 gj_diff = thetaGJmeasured - thetaGJmax;
                 return true;
             }
         }
         return false;
     }

static float DeepTauId::calculateGottfriedJacksonAngleDifference ( const pat::Tau & tau )

inlinestaticprivate

Definition at line 1678 of file DeepTauId.cc.

     {
         double gj_diff;
         if(calculateGottfriedJacksonAngleDifference(tau, gj_diff))
             return static_cast<float>(gj_diff);
         return default_value;
     }

void DeepTauId::checkInputs	(	const tensorflow::Tensor &	inputs,
		const char *	block_name,
		int	n_inputs,
		int	n_eta = `1`,
		int	n_phi = `1`
	)		const

inlineprivate

Definition at line 704 of file DeepTauId.cc.

References gather_cfg::cout, PVValHelper::eta, Exception, alignBH_cfg::fixed, objects.autophobj::float, input, edm::isNotFinite(), and gen::k.

     {
         if(debug_level >= 1){
             for(int eta = 0; eta < n_eta; ++eta){
                 for(int phi = 0; phi < n_phi; phi++){
                     for(int k = 0; k < n_inputs; ++k) {
                         const float input = n_eta == 1 && n_phi == 1
                                           ? inputs.matrix<float>()(0, k) : inputs.tensor<float,4>()(0, eta, phi, k);
                         if(edm::isNotFinite(input))
                             throw cms::Exception("DeepTauId") << "in the " << block_name << ", input is not finite, i.e. infinite or NaN, for eta_index = "
                                                               << n_eta << ", phi_index = " << n_phi << ", input_index = " << k;
                         if(debug_level >= 2)
                             std::cout << block_name << "," << eta << ","<< phi << "," << k << "," << std::setprecision(5) << std::fixed << input << '\n';
                     }
                 }
             }
         }
     }

void DeepTauId::createConvFeatures	(	const TauType &	tau,
		const reco::Vertex &	pv,
		double	rho,
		const pat::ElectronCollection &	electrons,
		const pat::MuonCollection &	muons,
		const pat::PackedCandidateCollection &	pfCands,
		const CellGrid &	grid,
		bool	is_inner
	)

inlineprivate

Definition at line 849 of file DeepTauId.cc.

References PVValHelper::eta.

     {
         tensorflow::Tensor& convTensor = *convTensor_.at(is_inner);
         for(int eta = -grid.maxEtaIndex(); eta <= grid.maxEtaIndex(); ++eta) {
             for(int phi = -grid.maxPhiIndex(); phi <= grid.maxPhiIndex(); ++phi) {
                 const CellIndex cell_index{eta, phi};
                 const int eta_index = grid.getEtaTensorIndex(cell_index);
                 const int phi_index = grid.getPhiTensorIndex(cell_index);
 
                 const auto cell_iter = grid.find(cell_index);
                 if(cell_iter != grid.end()) {
                     const Cell& cell = cell_iter->second;
                     createEgammaBlockInputs(tau, pv, rho, electrons, pfCands, cell, is_inner);
                     createMuonBlockInputs(tau, pv, rho, muons, pfCands, cell, is_inner);
                     createHadronsBlockInputs(tau, pv, rho, pfCands, cell, is_inner);
                     getPartialPredictions(convTensor, is_inner, eta_index, phi_index);
                 } else {
                     setCellConvFeatures(convTensor, *zeroOutputTensor_[is_inner], eta_index, phi_index);
                 }
             }
         }
     }

void DeepTauId::createEgammaBlockInputs	(	const TauType &	tau,
		const reco::Vertex &	pv,
		double	rho,
		const pat::ElectronCollection &	electrons,
		const pat::PackedCandidateCollection &	pfCands,
		const Cell &	cell_map,
		bool	is_inner
	)

inlineprivate

Definition at line 975 of file DeepTauId.cc.

References funct::abs(), particleFlow_cfi::dPhi, nanoDQM_cff::Electron, f, pat::Tau::flightLength(), objects.autophobj::float, reco::JetExtendedAssociation::getValue(), PatBasicFWLiteJetAnalyzer_Selector_cfg::inputs, reco::LeafCandidate::polarP4(), reco::Vertex::position(), lowPtElectrons_cff::rho, runTauDisplay::tau_eta, and runTauDisplay::tau_pt.

     {
         namespace dnn = dnn_inputs_2017_v2::EgammaBlockInputs;
 
         tensorflow::Tensor& inputs = *eGammaTensor_.at(is_inner);
         inputs.flat<float>().setZero();
 
         const auto& get = [&](int var_index) -> float& {
             return inputs.tensor<float,4>()(0, 0, 0, var_index);
         };
 
         const bool valid_index_pf_ele = cell_map.count(CellObjectType::PfCand_electron);
         const bool valid_index_pf_gamma = cell_map.count(CellObjectType::PfCand_gamma);
         const bool valid_index_ele = cell_map.count(CellObjectType::Electron);
 
         if(!cell_map.empty()){
             get(dnn::rho) = getValueNorm(rho, 21.49f, 9.713f);
             get(dnn::tau_pt) =  getValueLinear(tau.polarP4().pt(), 20.f, 1000.f, true);
             get(dnn::tau_eta) = getValueLinear(tau.polarP4().eta(), -2.3f, 2.3f, false);
             get(dnn::tau_inside_ecal_crack) = getValue(isInEcalCrack(tau.polarP4().eta()));
         }
         if(valid_index_pf_ele){
             size_t index_pf_ele = cell_map.at(CellObjectType::PfCand_electron);
 
             get(dnn::pfCand_ele_valid) = valid_index_pf_ele;
             get(dnn::pfCand_ele_rel_pt) = getValueNorm(pfCands.at(index_pf_ele).polarP4().pt() / tau.polarP4().pt(),
                 is_inner ? 0.9792f : 0.304f, is_inner ? 0.5383f : 1.845f);
             get(dnn::pfCand_ele_deta) = getValueLinear(pfCands.at(index_pf_ele).polarP4().eta() - tau.polarP4().eta(),
                 is_inner ? -0.1f : -0.5f, is_inner ? 0.1f : 0.5f, false);
             get(dnn::pfCand_ele_dphi) = getValueLinear(dPhi(tau.polarP4(), pfCands.at(index_pf_ele).polarP4()),
                 is_inner ? -0.1f : -0.5f, is_inner ? 0.1f : 0.5f, false);
             get(dnn::pfCand_ele_pvAssociationQuality) = getValueLinear<int>(pfCands.at(index_pf_ele).pvAssociationQuality(), 0, 7, true);
             get(dnn::pfCand_ele_puppiWeight) = getValue(pfCands.at(index_pf_ele).puppiWeight());
             get(dnn::pfCand_ele_charge) = getValue(pfCands.at(index_pf_ele).charge());
             get(dnn::pfCand_ele_lostInnerHits) = getValue<int>(pfCands.at(index_pf_ele).lostInnerHits());
             get(dnn::pfCand_ele_numberOfPixelHits) = getValueLinear(pfCands.at(index_pf_ele).numberOfPixelHits(), 0, 10, true);
             get(dnn::pfCand_ele_vertex_dx) = getValueNorm(pfCands.at(index_pf_ele).vertex().x() - pv.position().x(), 0.f, 0.1221f);
             get(dnn::pfCand_ele_vertex_dy) = getValueNorm(pfCands.at(index_pf_ele).vertex().y() - pv.position().y(), 0.f, 0.1226f);
             get(dnn::pfCand_ele_vertex_dz) = getValueNorm(pfCands.at(index_pf_ele).vertex().z() - pv.position().z(), 0.001f, 1.024f);
             get(dnn::pfCand_ele_vertex_dx_tauFL) = getValueNorm(pfCands.at(index_pf_ele).vertex().x() -
                 pv.position().x() - tau.flightLength().x(), 0.f, 0.3411f);
             get(dnn::pfCand_ele_vertex_dy_tauFL) = getValueNorm(pfCands.at(index_pf_ele).vertex().y() -
                 pv.position().y() - tau.flightLength().y(), 0.0003f, 0.3385f);
             get(dnn::pfCand_ele_vertex_dz_tauFL) = getValueNorm(pfCands.at(index_pf_ele).vertex().z() -
                 pv.position().z() - tau.flightLength().z(), 0.f, 1.307f);
 
             const bool hasTrackDetails = pfCands.at(index_pf_ele).hasTrackDetails();
             if(hasTrackDetails){
                 get(dnn::pfCand_ele_hasTrackDetails) = hasTrackDetails;
                 get(dnn::pfCand_ele_dxy) = getValueNorm(pfCands.at(index_pf_ele).dxy(), 0.f, 0.171f);
                 get(dnn::pfCand_ele_dxy_sig) = getValueNorm(std::abs(pfCands.at(index_pf_ele).dxy()) /
                     pfCands.at(index_pf_ele).dxyError(), 1.634f, 6.45f);
                 get(dnn::pfCand_ele_dz) =  getValueNorm(pfCands.at(index_pf_ele).dz(), 0.001f, 1.02f);
                 get(dnn::pfCand_ele_dz_sig) =  getValueNorm(std::abs(pfCands.at(index_pf_ele).dz()) /
                     pfCands.at(index_pf_ele).dzError(), 24.56f, 210.4f);
                 get(dnn::pfCand_ele_track_chi2_ndof) = getValueNorm(pfCands.at(index_pf_ele).pseudoTrack().chi2() /
                     pfCands.at(index_pf_ele).pseudoTrack().ndof(), 2.272f, 8.439f);
                 get(dnn::pfCand_ele_track_ndof) = getValueNorm(pfCands.at(index_pf_ele).pseudoTrack().ndof(), 15.18f, 3.203f);
             }
         }
         if(valid_index_pf_gamma){
             size_t index_pf_gamma = cell_map.at(CellObjectType::PfCand_gamma);
             get(dnn::pfCand_gamma_valid) = valid_index_pf_gamma;
             get(dnn::pfCand_gamma_rel_pt) = getValueNorm(pfCands.at(index_pf_gamma).polarP4().pt() / tau.polarP4().pt(),
                 is_inner ? 0.6048f : 0.02576f, is_inner ? 1.669f : 0.3833f);
             get(dnn::pfCand_gamma_deta) = getValueLinear(pfCands.at(index_pf_gamma).polarP4().eta() - tau.polarP4().eta(),
                 is_inner ? -0.1f : -0.5f, is_inner ? 0.1f : 0.5f, false);
             get(dnn::pfCand_gamma_dphi) = getValueLinear(dPhi(tau.polarP4(), pfCands.at(index_pf_gamma).polarP4()),
                 is_inner ? -0.1f : -0.5f, is_inner ? 0.1f : 0.5f, false);
             get(dnn::pfCand_gamma_pvAssociationQuality) =
                 getValueLinear<int>(pfCands.at(index_pf_gamma).pvAssociationQuality(), 0, 7, true);
             get(dnn::pfCand_gamma_fromPV) = getValueLinear<int>(pfCands.at(index_pf_gamma).fromPV(), 0, 3, true);
             get(dnn::pfCand_gamma_puppiWeight) = getValue(pfCands.at(index_pf_gamma).puppiWeight());
             get(dnn::pfCand_gamma_puppiWeightNoLep) = getValue(pfCands.at(index_pf_gamma).puppiWeightNoLep());
             get(dnn::pfCand_gamma_lostInnerHits) = getValue<int>(pfCands.at(index_pf_gamma).lostInnerHits());
             get(dnn::pfCand_gamma_numberOfPixelHits) = getValueLinear(pfCands.at(index_pf_gamma).numberOfPixelHits(), 0, 7, true);
             get(dnn::pfCand_gamma_vertex_dx) = getValueNorm(pfCands.at(index_pf_gamma).vertex().x() - pv.position().x(), 0.f, 0.0067f);
             get(dnn::pfCand_gamma_vertex_dy) = getValueNorm(pfCands.at(index_pf_gamma).vertex().y() - pv.position().y(), 0.f, 0.0069f);
             get(dnn::pfCand_gamma_vertex_dz) = getValueNorm(pfCands.at(index_pf_gamma).vertex().z() - pv.position().z(), 0.f, 0.0578f);
             get(dnn::pfCand_gamma_vertex_dx_tauFL) = getValueNorm(pfCands.at(index_pf_gamma).vertex().x() -
                 pv.position().x() - tau.flightLength().x(), 0.001f, 0.9565f);
             get(dnn::pfCand_gamma_vertex_dy_tauFL) = getValueNorm(pfCands.at(index_pf_gamma).vertex().y() -
                 pv.position().y() - tau.flightLength().y(), 0.0008f, 0.9592f);
             get(dnn::pfCand_gamma_vertex_dz_tauFL) = getValueNorm(pfCands.at(index_pf_gamma).vertex().z() -
                 pv.position().z() - tau.flightLength().z(), 0.0038f, 2.154f);
 
             const bool hasTrackDetails = pfCands.at(index_pf_gamma).hasTrackDetails();
             if(hasTrackDetails){
                 get(dnn::pfCand_gamma_hasTrackDetails) = hasTrackDetails;
                 get(dnn::pfCand_gamma_dxy) = getValueNorm(pfCands.at(index_pf_gamma).dxy(), 0.0004f, 0.882f);
                 get(dnn::pfCand_gamma_dxy_sig) = getValueNorm(std::abs(pfCands.at(index_pf_gamma).dxy()) /
                     pfCands.at(index_pf_gamma).dxyError(), 4.271f, 63.78f);
                 get(dnn::pfCand_gamma_dz) =  getValueNorm(pfCands.at(index_pf_gamma).dz(), 0.0071f, 5.285f);
                 get(dnn::pfCand_gamma_dz_sig) =  getValueNorm(std::abs(pfCands.at(index_pf_gamma).dz()) /
                     pfCands.at(index_pf_gamma).dzError(), 162.1f, 622.4f);
                 get(dnn::pfCand_gamma_track_chi2_ndof) = pfCands.at(index_pf_gamma).pseudoTrack().ndof() > 0 ?
                     getValueNorm(pfCands.at(index_pf_gamma).pseudoTrack().chi2() /
                                  pfCands.at(index_pf_gamma).pseudoTrack().ndof(), 4.268f, 15.47f) : 0;
                 get(dnn::pfCand_gamma_track_ndof) = pfCands.at(index_pf_gamma).pseudoTrack().ndof() > 0 ?
                     getValueNorm(pfCands.at(index_pf_gamma).pseudoTrack().ndof(), 12.25f, 4.774f) : 0;
             }
         }
         if(valid_index_ele){
             size_t index_ele = cell_map.at(CellObjectType::Electron);
 
             get(dnn::ele_valid) = valid_index_ele;
             get(dnn::ele_rel_pt) = getValueNorm(electrons.at(index_ele).polarP4().pt() / tau.polarP4().pt(),
                 is_inner ? 1.067f : 0.5111f, is_inner ? 1.521f : 2.765f);
             get(dnn::ele_deta) = getValueLinear(electrons.at(index_ele).polarP4().eta() - tau.polarP4().eta(),
                 is_inner ? -0.1f : -0.5f, is_inner ? 0.1f : 0.5f, false);
             get(dnn::ele_dphi) = getValueLinear(dPhi(tau.polarP4(), electrons.at(index_ele).polarP4()),
                 is_inner ? -0.1f : -0.5f, is_inner ? 0.1f : 0.5f, false);
 
             float cc_ele_energy, cc_gamma_energy;
             int cc_n_gamma;
             const bool cc_valid = calculateElectronClusterVarsV2(electrons.at(index_ele), cc_ele_energy, cc_gamma_energy, cc_n_gamma);
             if(cc_valid){
                 get(dnn::ele_cc_valid) = cc_valid;
                 get(dnn::ele_cc_ele_rel_energy) = getValueNorm(cc_ele_energy / electrons.at(index_ele).polarP4().pt(), 1.729f, 1.644f);
                 get(dnn::ele_cc_gamma_rel_energy) = getValueNorm(cc_gamma_energy / cc_ele_energy, 0.1439f, 0.3284f);
                 get(dnn::ele_cc_n_gamma) = getValueNorm(cc_n_gamma, 1.794f, 2.079f);
             }
             get(dnn::ele_rel_trackMomentumAtVtx) = getValueNorm(electrons.at(index_ele).trackMomentumAtVtx().R() /
                 electrons.at(index_ele).polarP4().pt(), 1.531f, 1.424f);
             get(dnn::ele_rel_trackMomentumAtCalo) =  getValueNorm(electrons.at(index_ele).trackMomentumAtCalo().R() /
                 electrons.at(index_ele).polarP4().pt(), 1.531f, 1.424f);
             get(dnn::ele_rel_trackMomentumOut) = getValueNorm(electrons.at(index_ele).trackMomentumOut().R() /
                 electrons.at(index_ele).polarP4().pt(), 0.7735f, 0.935f);
             get(dnn::ele_rel_trackMomentumAtEleClus) = getValueNorm(electrons.at(index_ele).trackMomentumAtEleClus().R() /
                 electrons.at(index_ele).polarP4().pt(), 0.7735f, 0.935f);
             get(dnn::ele_rel_trackMomentumAtVtxWithConstraint) =
                 getValueNorm(electrons.at(index_ele).trackMomentumAtVtxWithConstraint().R() /
                 electrons.at(index_ele).polarP4().pt(), 1.625f, 1.581f);
             get(dnn::ele_rel_ecalEnergy) = getValueNorm(electrons.at(index_ele).ecalEnergy() /
                 electrons.at(index_ele).polarP4().pt(), 1.993f, 1.308f);
             get(dnn::ele_ecalEnergy_sig) =  getValueNorm(electrons.at(index_ele).ecalEnergy() /
                 electrons.at(index_ele).ecalEnergyError(), 70.25f, 58.16f);
             get(dnn::ele_eSuperClusterOverP) =  getValueNorm(electrons.at(index_ele).eSuperClusterOverP(), 2.432f, 15.13f);
             get(dnn::ele_eSeedClusterOverP) = getValueNorm(electrons.at(index_ele).eSeedClusterOverP(), 2.034f, 13.96f);
             get(dnn::ele_eSeedClusterOverPout) = getValueNorm(electrons.at(index_ele).eSeedClusterOverPout(), 6.64f, 36.8f);
             get(dnn::ele_eEleClusterOverPout) = getValueNorm(electrons.at(index_ele).eEleClusterOverPout(), 4.183f, 20.63f);
             get(dnn::ele_deltaEtaSuperClusterTrackAtVtx) =
                 getValueNorm(electrons.at(index_ele).deltaEtaSuperClusterTrackAtVtx(),0.f, 0.0363f);
             get(dnn::ele_deltaEtaSeedClusterTrackAtCalo) =
                 getValueNorm(electrons.at(index_ele).deltaEtaSeedClusterTrackAtCalo(), -0.0001f, 0.0512f);
             get(dnn::ele_deltaEtaEleClusterTrackAtCalo) =
                 getValueNorm(electrons.at(index_ele).deltaEtaEleClusterTrackAtCalo(), -0.0001f, 0.0541f);
             get(dnn::ele_deltaPhiEleClusterTrackAtCalo) =
                 getValueNorm(electrons.at(index_ele).deltaPhiEleClusterTrackAtCalo(), 0.0002f, 0.0553f);
             get(dnn::ele_deltaPhiSuperClusterTrackAtVtx) =
                 getValueNorm(electrons.at(index_ele).deltaPhiSuperClusterTrackAtVtx(), 0.0001f, 0.0523f);
             get(dnn::ele_deltaPhiSeedClusterTrackAtCalo) =
                 getValueNorm(electrons.at(index_ele).deltaPhiSeedClusterTrackAtCalo(), 0.0004f, 0.0777f);
             get(dnn::ele_mvaInput_earlyBrem) = getValue(electrons.at(index_ele).mvaInput().earlyBrem);
             get(dnn::ele_mvaInput_lateBrem) = getValue(electrons.at(index_ele).mvaInput().lateBrem);
             get(dnn::ele_mvaInput_sigmaEtaEta) = getValueNorm(electrons.at(index_ele).mvaInput().sigmaEtaEta,0.0008f, 0.0052f);
             get(dnn::ele_mvaInput_hadEnergy) = getValueNorm(electrons.at(index_ele).mvaInput().hadEnergy, 14.04f, 69.48f);
             get(dnn::ele_mvaInput_deltaEta) = getValueNorm(electrons.at(index_ele).mvaInput().deltaEta, 0.0099f, 0.0851f);
 
             const auto& gsfTrack = electrons.at(index_ele).gsfTrack();
             if(gsfTrack.isNonnull()){
                 get(dnn::ele_gsfTrack_normalizedChi2) = getValueNorm(gsfTrack->normalizedChi2(), 3.049f, 10.39f);
                 get(dnn::ele_gsfTrack_numberOfValidHits) = getValueNorm(gsfTrack->numberOfValidHits(), 16.52f, 2.806f);
                 get(dnn::ele_rel_gsfTrack_pt) = getValueNorm(gsfTrack->pt() / electrons.at(index_ele).polarP4().pt(), 1.355f, 16.81f);
                 get(dnn::ele_gsfTrack_pt_sig) = getValueNorm(gsfTrack->pt() / gsfTrack->ptError(), 5.046f, 3.119f);
             }
             const auto& closestCtfTrack = electrons.at(index_ele).closestCtfTrackRef();
             const bool has_closestCtfTrack = closestCtfTrack.isNonnull();
             if(has_closestCtfTrack){
                 get(dnn::ele_has_closestCtfTrack) = has_closestCtfTrack;
                 get(dnn::ele_closestCtfTrack_normalizedChi2) = getValueNorm(closestCtfTrack->normalizedChi2(), 2.411f, 6.98f);
                 get(dnn::ele_closestCtfTrack_numberOfValidHits) = getValueNorm(closestCtfTrack->numberOfValidHits(), 15.16f, 5.26f);
             }
         }
         checkInputs(inputs, is_inner ? "egamma_inner_block" : "egamma_outer_block", dnn::NumberOfInputs);
     }

void DeepTauId::createHadronsBlockInputs	(	const TauType &	tau,
		const reco::Vertex &	pv,
		double	rho,
		const pat::PackedCandidateCollection &	pfCands,
		const Cell &	cell_map,
		bool	is_inner
	)

inlineprivate

Definition at line 1283 of file DeepTauId.cc.

References funct::abs(), particleFlow_cfi::dPhi, f, pat::Tau::flightLength(), objects.autophobj::float, edm::Ptr< T >::get(), reco::JetExtendedAssociation::getValue(), PatBasicFWLiteJetAnalyzer_Selector_cfg::inputs, pat::Tau::leadChargedHadrCand(), reco::LeafCandidate::polarP4(), reco::Vertex::position(), lowPtElectrons_cff::rho, runTauDisplay::tau_eta, and runTauDisplay::tau_pt.

     {
         namespace dnn = dnn_inputs_2017_v2::HadronBlockInputs;
 
         tensorflow::Tensor& inputs = *hadronsTensor_.at(is_inner);
         inputs.flat<float>().setZero();
 
 
         const auto& get = [&](int var_index) -> float& {
             return inputs.tensor<float,4>()(0, 0, 0, var_index);
         };
 
         const bool valid_chH = cell_map.count(CellObjectType::PfCand_chargedHadron);
         const bool valid_nH = cell_map.count(CellObjectType::PfCand_neutralHadron);
 
         if(!cell_map.empty()){
             get(dnn::rho) = getValueNorm(rho, 21.49f, 9.713f);
             get(dnn::tau_pt) =  getValueLinear(tau.polarP4().pt(), 20.f, 1000.f, true);
             get(dnn::tau_eta) = getValueLinear(tau.polarP4().eta(), -2.3f, 2.3f, false);
             get(dnn::tau_inside_ecal_crack) = getValue(isInEcalCrack(tau.polarP4().eta()));
         }
         if(valid_chH){
             size_t index_chH = cell_map.at(CellObjectType::PfCand_chargedHadron);
 
             get(dnn::pfCand_chHad_valid) = valid_chH;
             get(dnn::pfCand_chHad_rel_pt) = getValueNorm(pfCands.at(index_chH).polarP4().pt() / tau.polarP4().pt(),
                 is_inner ? 0.2564f : 0.0194f, is_inner ? 0.8607f : 0.1865f);
             get(dnn::pfCand_chHad_deta) = getValueLinear(pfCands.at(index_chH).polarP4().eta() - tau.polarP4().eta(),
                 is_inner ? -0.1f : -0.5f, is_inner ? 0.1f : 0.5f, false);
             get(dnn::pfCand_chHad_dphi) = getValueLinear(dPhi(tau.polarP4(),pfCands.at(index_chH).polarP4()),
                 is_inner ? -0.1f : -0.5f, is_inner ? 0.1f : 0.5f, false);
             get(dnn::pfCand_chHad_leadChargedHadrCand) = getValue(&pfCands.at(index_chH) ==
                 dynamic_cast<const pat::PackedCandidate*>(tau.leadChargedHadrCand().get()));
             get(dnn::pfCand_chHad_pvAssociationQuality) =
                 getValueLinear<int>(pfCands.at(index_chH).pvAssociationQuality(), 0, 7, true);
             get(dnn::pfCand_chHad_fromPV) = getValueLinear<int>(pfCands.at(index_chH).fromPV(), 0, 3, true);
             get(dnn::pfCand_chHad_puppiWeight) = getValue(pfCands.at(index_chH).puppiWeight());
             get(dnn::pfCand_chHad_puppiWeightNoLep) = getValue(pfCands.at(index_chH).puppiWeightNoLep());
             get(dnn::pfCand_chHad_charge) =  getValue(pfCands.at(index_chH).charge());
             get(dnn::pfCand_chHad_lostInnerHits) = getValue<int>(pfCands.at(index_chH).lostInnerHits());
             get(dnn::pfCand_chHad_numberOfPixelHits) = getValueLinear(pfCands.at(index_chH).numberOfPixelHits(), 0, 12, true);
             get(dnn::pfCand_chHad_vertex_dx) = getValueNorm(pfCands.at(index_chH).vertex().x() - pv.position().x(), 0.0005f, 1.735f);
             get(dnn::pfCand_chHad_vertex_dy) = getValueNorm(pfCands.at(index_chH).vertex().y() - pv.position().y(), -0.0008f, 1.752f);
             get(dnn::pfCand_chHad_vertex_dz) = getValueNorm(pfCands.at(index_chH).vertex().z() - pv.position().z(), -0.0201f, 8.333f);
             get(dnn::pfCand_chHad_vertex_dx_tauFL) = getValueNorm(pfCands.at(index_chH).vertex().x() - pv.position().x()
                 - tau.flightLength().x(), -0.0014f, 1.93f);
             get(dnn::pfCand_chHad_vertex_dy_tauFL) = getValueNorm(pfCands.at(index_chH).vertex().y() - pv.position().y()
                 - tau.flightLength().y(), 0.0022f, 1.948f);
             get(dnn::pfCand_chHad_vertex_dz_tauFL) = getValueNorm(pfCands.at(index_chH).vertex().z() - pv.position().z()
                 - tau.flightLength().z(), -0.0138f, 8.622f);
 
             const bool hasTrackDetails = pfCands.at(index_chH).hasTrackDetails();
             if(hasTrackDetails){
                 get(dnn::pfCand_chHad_hasTrackDetails) = hasTrackDetails;
                 get(dnn::pfCand_chHad_dxy) = getValueNorm(pfCands.at(index_chH).dxy(), -0.012f, 2.386f);
                 get(dnn::pfCand_chHad_dxy_sig) = getValueNorm(std::abs(pfCands.at(index_chH).dxy()) /
                     pfCands.at(index_chH).dxyError(), 6.417f, 36.28f);
                 get(dnn::pfCand_chHad_dz) =  getValueNorm(pfCands.at(index_chH).dz(), -0.0246f, 7.618f);
                 get(dnn::pfCand_chHad_dz_sig) =  getValueNorm(std::abs(pfCands.at(index_chH).dz()) /
                     pfCands.at(index_chH).dzError(), 301.3f, 491.1f);
                 get(dnn::pfCand_chHad_track_chi2_ndof) = pfCands.at(index_chH).pseudoTrack().ndof() > 0 ?
                     getValueNorm(pfCands.at(index_chH).pseudoTrack().chi2() /
                     pfCands.at(index_chH).pseudoTrack().ndof(), 0.7876f, 3.694f) : 0;
                 get(dnn::pfCand_chHad_track_ndof) = pfCands.at(index_chH).pseudoTrack().ndof() > 0 ?
                     getValueNorm(pfCands.at(index_chH).pseudoTrack().ndof(), 13.92f, 6.581f) : 0;
             }
         float hcal_fraction = 0.;
         if(pfCands.at(index_chH).pdgId() == 1 || pfCands.at(index_chH).pdgId() == 130) {
           hcal_fraction = pfCands.at(index_chH).hcalFraction();
         } else if(pfCands.at(index_chH).isIsolatedChargedHadron()) {
           hcal_fraction = pfCands.at(index_chH).rawHcalFraction();
         }
             get(dnn::pfCand_chHad_hcalFraction) = getValue(hcal_fraction);
             get(dnn::pfCand_chHad_rawCaloFraction) = getValueLinear(pfCands.at(index_chH).rawCaloFraction(), 0.f, 2.6f, true);
         }
         if(valid_nH){
             size_t index_nH = cell_map.at(CellObjectType::PfCand_neutralHadron);
 
             get(dnn::pfCand_nHad_valid) = valid_nH;
             get(dnn::pfCand_nHad_rel_pt) = getValueNorm(pfCands.at(index_nH).polarP4().pt() / tau.polarP4().pt(),
                 is_inner ? 0.3163f : 0.0502f, is_inner ? 0.2769f : 0.4266f);
             get(dnn::pfCand_nHad_deta) = getValueLinear(pfCands.at(index_nH).polarP4().eta() - tau.polarP4().eta(),
                 is_inner ? -0.1f : -0.5f, is_inner ? 0.1f : 0.5f, false);
             get(dnn::pfCand_nHad_dphi) = getValueLinear(dPhi(tau.polarP4(),pfCands.at(index_nH).polarP4()),
                 is_inner ? -0.1f : -0.5f, is_inner ? 0.1f : 0.5f, false);
             get(dnn::pfCand_nHad_puppiWeight) = getValue(pfCands.at(index_nH).puppiWeight());
             get(dnn::pfCand_nHad_puppiWeightNoLep) = getValue(pfCands.at(index_nH).puppiWeightNoLep());
         float hcal_fraction = 0.;
         if(pfCands.at(index_nH).pdgId() == 1 || pfCands.at(index_nH).pdgId() == 130) {
           hcal_fraction = pfCands.at(index_nH).hcalFraction();
         } else if(pfCands.at(index_nH).isIsolatedChargedHadron()) {
           hcal_fraction = pfCands.at(index_nH).rawHcalFraction();
         }
             get(dnn::pfCand_nHad_hcalFraction) = getValue(hcal_fraction);
         }
         checkInputs(inputs, is_inner ? "hadron_inner_block" : "hadron_outer_block", dnn::NumberOfInputs);
     }

template<typename dnn >

tensorflow::Tensor DeepTauId::createInputsV1	(	const TauType &	tau,
		const ElectronCollection &	electrons,
		const MuonCollection &	muons
	)		const

inlineprivate

Definition at line 1384 of file DeepTauId.cc.

References funct::abs(), tauProducer_cfi::chargedIsoPtSum, constexpr, boostedTaus_cff::decayMode, pat::Tau::decayMode(), particleFlow_cfi::dEta, particleFlow_cfi::dPhi, PVValHelper::dxy, pat::Tau::dxy(), pat::Tau::dxy_Sig(), PVValHelper::dz, pat::Tau::ecalEnergyLeadChargedHadrCand(), pat::Tau::emFraction_MVA(), reco::tau::eratio(), PVValHelper::eta, pat::Tau::etaAtEcalEntranceLeadChargedCand(), Exception, pat::Tau::flightLength(), pat::Tau::flightLengthSig(), edm::Ptr< T >::get(), pat::Tau::hasSecondaryVertex(), pat::Tau::hcalEnergyLeadChargedHadrCand(), PatBasicFWLiteJetAnalyzer_Selector_cfg::inputs, pat::Tau::ip3d(), electrons_cff::ip3d, pat::Tau::ip3d_Sig(), edm::Ptr< T >::isNonnull(), pat::Tau::leadChargedHadrCand(), pat::Tau::leadingTrackNormChi2(), pat::Tau::leadPFChargedHadrCand(), ResonanceBuilder::mass, reco::tau::n_photons_total(), tauProducer_cfi::neutralIsoPtSum, tauProducer_cfi::neutralIsoPtSumWeight, reco::LeafCandidate::p4(), tauProducer_cfi::photonPtSumOutsideSignalCone, EnergyCorrector::pt, reco::tau::pt_weighted_deta_strip(), reco::tau::pt_weighted_dphi_strip(), reco::tau::pt_weighted_dr_iso(), reco::tau::pt_weighted_dr_signal(), pat::Tau::ptLeadChargedCand(), tauProducer_cfi::puCorrPtSum, metsig::tau, and pat::Tau::tauID().

     {
         static constexpr bool check_all_set = false;
         static constexpr float default_value_for_set_check = -42;
 
         tensorflow::Tensor inputs(tensorflow::DT_FLOAT, { 1, dnn_inputs_2017v1::NumberOfInputs});
         const auto& get = [&](int var_index) -> float& { return inputs.matrix<float>()(0, var_index); };
         auto leadChargedHadrCand = dynamic_cast<const pat::PackedCandidate*>(tau.leadChargedHadrCand().get());
 
         if(check_all_set) {
             for(int var_index = 0; var_index < dnn::NumberOfInputs; ++var_index) {
                 get(var_index) = default_value_for_set_check;
             }
         }
 
         get(dnn::pt) = tau.p4().pt();
         get(dnn::eta) = tau.p4().eta();
         get(dnn::mass) = tau.p4().mass();
         get(dnn::decayMode) = tau.decayMode();
         get(dnn::chargedIsoPtSum) = tau.tauID("chargedIsoPtSum");
         get(dnn::neutralIsoPtSum) = tau.tauID("neutralIsoPtSum");
         get(dnn::neutralIsoPtSumWeight) = tau.tauID("neutralIsoPtSumWeight");
         get(dnn::photonPtSumOutsideSignalCone) = tau.tauID("photonPtSumOutsideSignalCone");
         get(dnn::puCorrPtSum) = tau.tauID("puCorrPtSum");
         get(dnn::dxy) = tau.dxy();
         get(dnn::dxy_sig) = tau.dxy_Sig();
         get(dnn::dz) = leadChargedHadrCand ? leadChargedHadrCand->dz() : default_value;
         get(dnn::ip3d) = tau.ip3d();
         get(dnn::ip3d_sig) = tau.ip3d_Sig();
         get(dnn::hasSecondaryVertex) = tau.hasSecondaryVertex();
         get(dnn::flightLength_r) = tau.flightLength().R();
         get(dnn::flightLength_dEta) = dEta(tau.flightLength(), tau.p4());
         get(dnn::flightLength_dPhi) = dPhi(tau.flightLength(), tau.p4());
         get(dnn::flightLength_sig) = tau.flightLengthSig();
         get(dnn::leadChargedHadrCand_pt) = leadChargedHadrCand ? leadChargedHadrCand->p4().Pt() : default_value;
         get(dnn::leadChargedHadrCand_dEta) = leadChargedHadrCand
                 ? dEta(leadChargedHadrCand->p4(), tau.p4()) : default_value;
         get(dnn::leadChargedHadrCand_dPhi) = leadChargedHadrCand
                 ? dPhi(leadChargedHadrCand->p4(), tau.p4()) : default_value;
         get(dnn::leadChargedHadrCand_mass) = leadChargedHadrCand
                 ? leadChargedHadrCand->p4().mass() : default_value;
         get(dnn::pt_weighted_deta_strip) = reco::tau::pt_weighted_deta_strip(tau, tau.decayMode());
         get(dnn::pt_weighted_dphi_strip) = reco::tau::pt_weighted_dphi_strip(tau, tau.decayMode());
         get(dnn::pt_weighted_dr_signal) = reco::tau::pt_weighted_dr_signal(tau, tau.decayMode());
         get(dnn::pt_weighted_dr_iso) = reco::tau::pt_weighted_dr_iso(tau, tau.decayMode());
         get(dnn::leadingTrackNormChi2) = tau.leadingTrackNormChi2();
         get(dnn::e_ratio) = reco::tau::eratio(tau);
         get(dnn::gj_angle_diff) = calculateGottfriedJacksonAngleDifference(tau);
         get(dnn::n_photons) = reco::tau::n_photons_total(tau);
         get(dnn::emFraction) = tau.emFraction_MVA();
         get(dnn::has_gsf_track) = leadChargedHadrCand && std::abs(leadChargedHadrCand->pdgId()) == 11;
         get(dnn::inside_ecal_crack) = isInEcalCrack(tau.p4().Eta());
         auto gsf_ele = findMatchedElectron(tau, electrons, 0.3);
         get(dnn::gsf_ele_matched) = gsf_ele != nullptr;
         get(dnn::gsf_ele_pt) = gsf_ele != nullptr ? gsf_ele->p4().Pt() : default_value;
         get(dnn::gsf_ele_dEta) = gsf_ele != nullptr ? dEta(gsf_ele->p4(), tau.p4()) : default_value;
         get(dnn::gsf_ele_dPhi) = gsf_ele != nullptr ? dPhi(gsf_ele->p4(), tau.p4()) : default_value;
         get(dnn::gsf_ele_mass) = gsf_ele != nullptr ? gsf_ele->p4().mass() : default_value;
         calculateElectronClusterVars(gsf_ele, get(dnn::gsf_ele_Ee), get(dnn::gsf_ele_Egamma));
         get(dnn::gsf_ele_Pin) = gsf_ele != nullptr ? gsf_ele->trackMomentumAtVtx().R() : default_value;
         get(dnn::gsf_ele_Pout) = gsf_ele != nullptr ? gsf_ele->trackMomentumOut().R() : default_value;
         get(dnn::gsf_ele_EtotOverPin) = get(dnn::gsf_ele_Pin) > 0
                 ? (get(dnn::gsf_ele_Ee) + get(dnn::gsf_ele_Egamma)) / get(dnn::gsf_ele_Pin)
                 : default_value;
         get(dnn::gsf_ele_Eecal) = gsf_ele != nullptr ? gsf_ele->ecalEnergy() : default_value;
         get(dnn::gsf_ele_dEta_SeedClusterTrackAtCalo) = gsf_ele != nullptr
                 ? gsf_ele->deltaEtaSeedClusterTrackAtCalo() : default_value;
         get(dnn::gsf_ele_dPhi_SeedClusterTrackAtCalo) = gsf_ele != nullptr
                 ? gsf_ele->deltaPhiSeedClusterTrackAtCalo() : default_value;
         get(dnn::gsf_ele_mvaIn_sigmaEtaEta) = gsf_ele != nullptr
                 ? gsf_ele->mvaInput().sigmaEtaEta : default_value;
         get(dnn::gsf_ele_mvaIn_hadEnergy) = gsf_ele != nullptr ? gsf_ele->mvaInput().hadEnergy : default_value;
         get(dnn::gsf_ele_mvaIn_deltaEta) = gsf_ele != nullptr ? gsf_ele->mvaInput().deltaEta : default_value;
 
         get(dnn::gsf_ele_Chi2NormGSF) = default_value;
         get(dnn::gsf_ele_GSFNumHits) = default_value;
         get(dnn::gsf_ele_GSFTrackResol) = default_value;
         get(dnn::gsf_ele_GSFTracklnPt) = default_value;
         if(gsf_ele != nullptr && gsf_ele->gsfTrack().isNonnull()) {
             get(dnn::gsf_ele_Chi2NormGSF) = gsf_ele->gsfTrack()->normalizedChi2();
             get(dnn::gsf_ele_GSFNumHits) = gsf_ele->gsfTrack()->numberOfValidHits();
             if(gsf_ele->gsfTrack()->pt() > 0) {
                 get(dnn::gsf_ele_GSFTrackResol) = gsf_ele->gsfTrack()->ptError() / gsf_ele->gsfTrack()->pt();
                 get(dnn::gsf_ele_GSFTracklnPt) = std::log10(gsf_ele->gsfTrack()->pt());
             }
         }
 
         get(dnn::gsf_ele_Chi2NormKF) = default_value;
         get(dnn::gsf_ele_KFNumHits) = default_value;
         if(gsf_ele != nullptr && gsf_ele->closestCtfTrackRef().isNonnull()) {
             get(dnn::gsf_ele_Chi2NormKF) = gsf_ele->closestCtfTrackRef()->normalizedChi2();
             get(dnn::gsf_ele_KFNumHits) = gsf_ele->closestCtfTrackRef()->numberOfValidHits();
         }
         get(dnn::leadChargedCand_etaAtEcalEntrance) = tau.etaAtEcalEntranceLeadChargedCand();
         get(dnn::leadChargedCand_pt) = tau.ptLeadChargedCand();
 
         get(dnn::leadChargedHadrCand_HoP) = default_value;
         get(dnn::leadChargedHadrCand_EoP) = default_value;
         if(tau.leadChargedHadrCand()->pt() > 0) {
             get(dnn::leadChargedHadrCand_HoP) = tau.hcalEnergyLeadChargedHadrCand()
                                                     / tau.leadChargedHadrCand()->pt();
             get(dnn::leadChargedHadrCand_EoP) = tau.ecalEnergyLeadChargedHadrCand()
                                                     / tau.leadChargedHadrCand()->pt();
         }
 
         MuonHitMatchV1 muon_hit_match;
         if(tau.leadPFChargedHadrCand().isNonnull() && tau.leadPFChargedHadrCand()->muonRef().isNonnull())
             muon_hit_match.addMatchedMuon(*tau.leadPFChargedHadrCand()->muonRef(), tau);
 
         auto matched_muons = muon_hit_match.findMatchedMuons(tau, muons, 0.3, 5);
         for(auto muon : matched_muons)
             muon_hit_match.addMatchedMuon(*muon, tau);
         muon_hit_match.fillTensor<dnn>(get, tau, default_value);
 
         LorentzVectorXYZ signalChargedHadrCands_sumIn, signalChargedHadrCands_sumOut;
         processSignalPFComponents(tau, tau.signalChargedHadrCands(),
                                   signalChargedHadrCands_sumIn, signalChargedHadrCands_sumOut,
                                   get(dnn::signalChargedHadrCands_sum_innerSigCone_pt),
                                   get(dnn::signalChargedHadrCands_sum_innerSigCone_dEta),
                                   get(dnn::signalChargedHadrCands_sum_innerSigCone_dPhi),
                                   get(dnn::signalChargedHadrCands_sum_innerSigCone_mass),
                                   get(dnn::signalChargedHadrCands_sum_outerSigCone_pt),
                                   get(dnn::signalChargedHadrCands_sum_outerSigCone_dEta),
                                   get(dnn::signalChargedHadrCands_sum_outerSigCone_dPhi),
                                   get(dnn::signalChargedHadrCands_sum_outerSigCone_mass),
                                   get(dnn::signalChargedHadrCands_nTotal_innerSigCone),
                                   get(dnn::signalChargedHadrCands_nTotal_outerSigCone));
 
         LorentzVectorXYZ signalNeutrHadrCands_sumIn, signalNeutrHadrCands_sumOut;
         processSignalPFComponents(tau, tau.signalNeutrHadrCands(),
                                   signalNeutrHadrCands_sumIn, signalNeutrHadrCands_sumOut,
                                   get(dnn::signalNeutrHadrCands_sum_innerSigCone_pt),
                                   get(dnn::signalNeutrHadrCands_sum_innerSigCone_dEta),
                                   get(dnn::signalNeutrHadrCands_sum_innerSigCone_dPhi),
                                   get(dnn::signalNeutrHadrCands_sum_innerSigCone_mass),
                                   get(dnn::signalNeutrHadrCands_sum_outerSigCone_pt),
                                   get(dnn::signalNeutrHadrCands_sum_outerSigCone_dEta),
                                   get(dnn::signalNeutrHadrCands_sum_outerSigCone_dPhi),
                                   get(dnn::signalNeutrHadrCands_sum_outerSigCone_mass),
                                   get(dnn::signalNeutrHadrCands_nTotal_innerSigCone),
                                   get(dnn::signalNeutrHadrCands_nTotal_outerSigCone));
 
 
         LorentzVectorXYZ signalGammaCands_sumIn, signalGammaCands_sumOut;
         processSignalPFComponents(tau, tau.signalGammaCands(),
                                   signalGammaCands_sumIn, signalGammaCands_sumOut,
                                   get(dnn::signalGammaCands_sum_innerSigCone_pt),
                                   get(dnn::signalGammaCands_sum_innerSigCone_dEta),
                                   get(dnn::signalGammaCands_sum_innerSigCone_dPhi),
                                   get(dnn::signalGammaCands_sum_innerSigCone_mass),
                                   get(dnn::signalGammaCands_sum_outerSigCone_pt),
                                   get(dnn::signalGammaCands_sum_outerSigCone_dEta),
                                   get(dnn::signalGammaCands_sum_outerSigCone_dPhi),
                                   get(dnn::signalGammaCands_sum_outerSigCone_mass),
                                   get(dnn::signalGammaCands_nTotal_innerSigCone),
                                   get(dnn::signalGammaCands_nTotal_outerSigCone));
 
         LorentzVectorXYZ isolationChargedHadrCands_sum;
         processIsolationPFComponents(tau, tau.isolationChargedHadrCands(), isolationChargedHadrCands_sum,
                                      get(dnn::isolationChargedHadrCands_sum_pt),
                                      get(dnn::isolationChargedHadrCands_sum_dEta),
                                      get(dnn::isolationChargedHadrCands_sum_dPhi),
                                      get(dnn::isolationChargedHadrCands_sum_mass),
                                      get(dnn::isolationChargedHadrCands_nTotal));
 
         LorentzVectorXYZ isolationNeutrHadrCands_sum;
         processIsolationPFComponents(tau, tau.isolationNeutrHadrCands(), isolationNeutrHadrCands_sum,
                                      get(dnn::isolationNeutrHadrCands_sum_pt),
                                      get(dnn::isolationNeutrHadrCands_sum_dEta),
                                      get(dnn::isolationNeutrHadrCands_sum_dPhi),
                                      get(dnn::isolationNeutrHadrCands_sum_mass),
                                      get(dnn::isolationNeutrHadrCands_nTotal));
 
         LorentzVectorXYZ isolationGammaCands_sum;
         processIsolationPFComponents(tau, tau.isolationGammaCands(), isolationGammaCands_sum,
                                      get(dnn::isolationGammaCands_sum_pt),
                                      get(dnn::isolationGammaCands_sum_dEta),
                                      get(dnn::isolationGammaCands_sum_dPhi),
                                      get(dnn::isolationGammaCands_sum_mass),
                                      get(dnn::isolationGammaCands_nTotal));
 
         get(dnn::tau_visMass_innerSigCone) = (signalGammaCands_sumIn + signalChargedHadrCands_sumIn).mass();
 
         if(check_all_set) {
             for(int var_index = 0; var_index < dnn::NumberOfInputs; ++var_index) {
                 if(get(var_index) == default_value_for_set_check)
                     throw cms::Exception("DeepTauId: variable with index = ") << var_index << " is not set.";
             }
         }
 
         return inputs;
     }

void DeepTauId::createMuonBlockInputs	(	const TauType &	tau,
		const reco::Vertex &	pv,
		double	rho,
		const pat::MuonCollection &	muons,
		const pat::PackedCandidateCollection &	pfCands,
		const Cell &	cell_map,
		bool	is_inner
	)

inlineprivate

Definition at line 1155 of file DeepTauId.cc.

References funct::abs(), MuonSubdetId::CSC, particleFlow_cfi::dPhi, MuonSubdetId::DT, f, pat::Tau::flightLength(), objects.autophobj::float, reco::JetExtendedAssociation::getValue(), PatBasicFWLiteJetAnalyzer_Selector_cfg::inputs, nanoDQM_cff::Muon, reco::LeafCandidate::polarP4(), reco::Vertex::position(), pat::Muon::PV2D, lowPtElectrons_cff::rho, MuonSubdetId::RPC, relativeConstraints::station, runTauDisplay::tau_eta, and runTauDisplay::tau_pt.

     {
         namespace dnn = dnn_inputs_2017_v2::MuonBlockInputs;
 
         tensorflow::Tensor& inputs = *muonTensor_.at(is_inner);
         inputs.flat<float>().setZero();
 
         const auto& get = [&](int var_index) -> float& {
             return inputs.tensor<float,4>()(0, 0, 0, var_index);
         };
 
         const bool valid_index_pf_muon = cell_map.count(CellObjectType::PfCand_muon);
         const bool valid_index_muon = cell_map.count(CellObjectType::Muon);
 
         if(!cell_map.empty()){
             get(dnn::rho) = getValueNorm(rho, 21.49f, 9.713f);
             get(dnn::tau_pt) =  getValueLinear(tau.polarP4().pt(), 20.f, 1000.f, true);
             get(dnn::tau_eta) = getValueLinear(tau.polarP4().eta(), -2.3f, 2.3f, false);
             get(dnn::tau_inside_ecal_crack) = getValue(isInEcalCrack(tau.polarP4().eta()));
         }
         if(valid_index_pf_muon){
             size_t index_pf_muon = cell_map.at(CellObjectType::PfCand_muon);
 
             get(dnn::pfCand_muon_valid) = valid_index_pf_muon;
             get(dnn::pfCand_muon_rel_pt) = getValueNorm(pfCands.at(index_pf_muon).polarP4().pt() / tau.polarP4().pt(),
                 is_inner ?  0.9509f : 0.0861f, is_inner ? 0.4294f : 0.4065f);
             get(dnn::pfCand_muon_deta) = getValueLinear(pfCands.at(index_pf_muon).polarP4().eta() - tau.polarP4().eta(),
                 is_inner ? -0.1f : -0.5f, is_inner ? 0.1f : 0.5f, false);
             get(dnn::pfCand_muon_dphi) = getValueLinear(dPhi(tau.polarP4(), pfCands.at(index_pf_muon).polarP4()),
                 is_inner ? -0.1f : -0.5f, is_inner ? 0.1f : 0.5f, false);
             get(dnn::pfCand_muon_pvAssociationQuality) = getValueLinear<int>(pfCands.at(index_pf_muon).pvAssociationQuality(), 0, 7, true);
             get(dnn::pfCand_muon_fromPV) = getValueLinear<int>(pfCands.at(index_pf_muon).fromPV(), 0, 3, true);
             get(dnn::pfCand_muon_puppiWeight) = getValue(pfCands.at(index_pf_muon).puppiWeight());
             get(dnn::pfCand_muon_charge) = getValue(pfCands.at(index_pf_muon).charge());
             get(dnn::pfCand_muon_lostInnerHits) = getValue<int>(pfCands.at(index_pf_muon).lostInnerHits());
             get(dnn::pfCand_muon_numberOfPixelHits) = getValueLinear(pfCands.at(index_pf_muon).numberOfPixelHits(), 0, 11, true);
             get(dnn::pfCand_muon_vertex_dx) = getValueNorm(pfCands.at(index_pf_muon).vertex().x() - pv.position().x(), -0.0007f, 0.6869f);
             get(dnn::pfCand_muon_vertex_dy) = getValueNorm(pfCands.at(index_pf_muon).vertex().y() - pv.position().y(), 0.0001f, 0.6784f);
             get(dnn::pfCand_muon_vertex_dz) = getValueNorm(pfCands.at(index_pf_muon).vertex().z() - pv.position().z(), -0.0117f, 4.097f);
             get(dnn::pfCand_muon_vertex_dx_tauFL) = getValueNorm(pfCands.at(index_pf_muon).vertex().x() -
                 pv.position().x() - tau.flightLength().x(), -0.0001f, 0.8642f);
             get(dnn::pfCand_muon_vertex_dy_tauFL) = getValueNorm(pfCands.at(index_pf_muon).vertex().y() -
                 pv.position().y() - tau.flightLength().y(), 0.0004f, 0.8561f);
             get(dnn::pfCand_muon_vertex_dz_tauFL) = getValueNorm(pfCands.at(index_pf_muon).vertex().z() -
                 pv.position().z() - tau.flightLength().z(), -0.0118f, 4.405f);
 
             const bool hasTrackDetails = pfCands.at(index_pf_muon).hasTrackDetails();
             if(hasTrackDetails){
                 get(dnn::pfCand_muon_hasTrackDetails) = hasTrackDetails;
                 get(dnn::pfCand_muon_dxy) = getValueNorm(pfCands.at(index_pf_muon).dxy(), -0.0045f, 0.9655f);
                 get(dnn::pfCand_muon_dxy_sig) = getValueNorm(std::abs(pfCands.at(index_pf_muon).dxy()) /
                     pfCands.at(index_pf_muon).dxyError(), 4.575f, 42.36f);
                 get(dnn::pfCand_muon_dz) =  getValueNorm(pfCands.at(index_pf_muon).dz(), -0.0117f, 4.097f);
                 get(dnn::pfCand_muon_dz_sig) =  getValueNorm(std::abs(pfCands.at(index_pf_muon).dz()) /
                     pfCands.at(index_pf_muon).dzError(), 80.37f, 343.3f);
                 get(dnn::pfCand_muon_track_chi2_ndof) =  getValueNorm(pfCands.at(index_pf_muon).pseudoTrack().chi2() /
                     pfCands.at(index_pf_muon).pseudoTrack().ndof(), 0.69f, 1.711f);
                 get(dnn::pfCand_muon_track_ndof) =  getValueNorm(pfCands.at(index_pf_muon).pseudoTrack().ndof(), 17.5f, 5.11f);
             }
         }
         if(valid_index_muon){
             size_t index_muon = cell_map.at(CellObjectType::Muon);
 
             get(dnn::muon_valid) = valid_index_muon;
             get(dnn::muon_rel_pt) = getValueNorm(muons.at(index_muon).polarP4().pt() / tau.polarP4().pt(),
                 is_inner ?  0.7966f : 0.2678f, is_inner ? 3.402f : 3.592f);
             get(dnn::muon_deta) = getValueLinear(muons.at(index_muon).polarP4().eta() - tau.polarP4().eta(),
                 is_inner ? -0.1f : -0.5f, is_inner ? 0.1f : 0.5f, false);
             get(dnn::muon_dphi) = getValueLinear(dPhi(tau.polarP4(), muons.at(index_muon).polarP4()),
                 is_inner ? -0.1f : -0.5f, is_inner ? 0.1f : 0.5f, false);
             get(dnn::muon_dxy) = getValueNorm(muons.at(index_muon).dB(pat::Muon::PV2D), 0.0019f, 1.039f);
             get(dnn::muon_dxy_sig) = getValueNorm(std::abs(muons.at(index_muon).dB(pat::Muon::PV2D)) /
                 muons.at(index_muon).edB(pat::Muon::PV2D), 8.98f, 71.17f);
 
             const bool normalizedChi2_valid = muons.at(index_muon).globalTrack().isNonnull() && muons.at(index_muon).normChi2() >= 0;
             if(normalizedChi2_valid){
                 get(dnn::muon_normalizedChi2_valid) = normalizedChi2_valid;
                 get(dnn::muon_normalizedChi2) = getValueNorm(muons.at(index_muon).normChi2(), 21.52f, 265.8f);
                 if(muons.at(index_muon).innerTrack().isNonnull())
                     get(dnn::muon_numberOfValidHits) = getValueNorm(muons.at(index_muon).numberOfValidHits(), 21.84f, 10.59f);
             }
             get(dnn::muon_segmentCompatibility) = getValue(muons.at(index_muon).segmentCompatibility());
             get(dnn::muon_caloCompatibility) = getValue(muons.at(index_muon).caloCompatibility());
 
             const bool pfEcalEnergy_valid = muons.at(index_muon).pfEcalEnergy() >= 0;
             if(pfEcalEnergy_valid){
                 get(dnn::muon_pfEcalEnergy_valid) = pfEcalEnergy_valid;
                 get(dnn::muon_rel_pfEcalEnergy) = getValueNorm(muons.at(index_muon).pfEcalEnergy() /
                     muons.at(index_muon).polarP4().pt(), 0.2273f, 0.4865f);
             }
 
             MuonHitMatchV2 hit_match(muons.at(index_muon));
             static const std::map<int, std::pair<int, int>> muonMatchHitVars = {
                 { MuonSubdetId::DT, { dnn::muon_n_matches_DT_1, dnn::muon_n_hits_DT_1 } },
                 { MuonSubdetId::CSC, { dnn::muon_n_matches_CSC_1, dnn::muon_n_hits_CSC_1 } },
                 { MuonSubdetId::RPC, { dnn::muon_n_matches_RPC_1, dnn::muon_n_hits_RPC_1 } }
             };
 
             static const std::map<int, std::vector<float>> muonMatchVarLimits = {
                 { MuonSubdetId::DT, { 2, 2, 2, 2 } },
                 { MuonSubdetId::CSC, { 6, 2, 2, 2 } },
                 { MuonSubdetId::RPC, { 7, 6, 4, 4 } }
             };
 
             static const std::map<int, std::vector<float>> muonHitVarLimits = {
                 { MuonSubdetId::DT, { 12, 12, 12, 8 } },
                 { MuonSubdetId::CSC, { 24, 12, 12, 12 } },
                 { MuonSubdetId::RPC, { 4, 4, 2, 2 } }
             };
 
             for(int subdet : hit_match.MuonHitMatchV2::consideredSubdets()) {
                 const auto& matchHitVar = muonMatchHitVars.at(subdet);
                 const auto& matchLimits = muonMatchVarLimits.at(subdet);
                 const auto& hitLimits = muonHitVarLimits.at(subdet);
                 for(int station = MuonHitMatchV2::first_station_id; station <= MuonHitMatchV2::last_station_id; ++station) {
                     const unsigned n_matches = hit_match.nMatches(subdet, station);
                     const unsigned n_hits = hit_match.nHits(subdet, station);
                     get(matchHitVar.first + station - 1) = getValueLinear(n_matches, 0, matchLimits.at(station - 1), true);
                     get(matchHitVar.second + station - 1) = getValueLinear(n_hits, 0, hitLimits.at(station - 1), true);
                 }
             }
         }
         checkInputs(inputs, is_inner ? "muon_inner_block" : "muon_outer_block", dnn::NumberOfInputs);
     }

void DeepTauId::createTauBlockInputs	(	const TauType &	tau,
		const reco::Vertex &	pv,
		double	rho
	)

inlineprivate

Definition at line 881 of file DeepTauId.cc.

References funct::abs(), reco::LeafCandidate::charge(), tauProducer_cfi::chargedIsoPtSum, pat::Tau::decayMode(), pat::Tau::dxy(), pat::Tau::dxy_error(), pat::Tau::dxy_PCA(), pat::Tau::emFraction_MVA(), reco::tau::eratio(), pat::Tau::etaAtEcalEntranceLeadChargedCand(), f, pat::Tau::flightLength(), tauProducer_cfi::footprintCorrection, edm::Ptr< T >::get(), reco::JetExtendedAssociation::getValue(), PatBasicFWLiteJetAnalyzer_Selector_cfg::inputs, pat::Tau::ip3d(), pat::Tau::ip3d_error(), pat::Tau::leadChargedHadrCand(), pat::Tau::leadingTrackNormChi2(), reco::tau::n_photons_total(), tauProducer_cfi::neutralIsoPtSum, reco::LeafCandidate::p4(), tauProducer_cfi::photonPtSumOutsideSignalCone, pi, reco::LeafCandidate::polarP4(), reco::tau::pt_weighted_deta_strip(), reco::tau::pt_weighted_dphi_strip(), reco::tau::pt_weighted_dr_iso(), reco::tau::pt_weighted_dr_signal(), tauProducer_cfi::puCorrPtSum, lowPtElectrons_cff::rho, runTauDisplay::tau_eta, runTauDisplay::tau_mass, runTauDisplay::tau_phi, runTauDisplay::tau_pt, and pat::Tau::tauID().

     {
         namespace dnn = dnn_inputs_2017_v2::TauBlockInputs;
 
         tensorflow::Tensor& inputs = *tauBlockTensor_;
         inputs.flat<float>().setZero();
 
         const auto& get = [&](int var_index) -> float& { return inputs.matrix<float>()(0, var_index); };
 
         auto leadChargedHadrCand = dynamic_cast<const pat::PackedCandidate*>(tau.leadChargedHadrCand().get());
 
         get(dnn::rho) = getValueNorm(rho, 21.49f, 9.713f);
         get(dnn::tau_pt) =  getValueLinear(tau.polarP4().pt(), 20.f, 1000.f, true);
         get(dnn::tau_eta) = getValueLinear(tau.polarP4().eta(), -2.3f, 2.3f, false);
         get(dnn::tau_phi) = getValueLinear(tau.polarP4().phi(), -pi, pi, false);
         get(dnn::tau_mass) = getValueNorm(tau.polarP4().mass(), 0.6669f, 0.6553f);
         get(dnn::tau_E_over_pt) = getValueLinear(tau.p4().energy() / tau.p4().pt(), 1.f, 5.2f, true);
         get(dnn::tau_charge) = getValue(tau.charge());
         get(dnn::tau_n_charged_prongs) = getValueLinear(tau.decayMode() / 5 + 1, 1, 3, true);
         get(dnn::tau_n_neutral_prongs) = getValueLinear(tau.decayMode() % 5, 0, 2, true);
         get(dnn::chargedIsoPtSum) = getValueNorm(tau.tauID("chargedIsoPtSum"), 47.78f, 123.5f);
         get(dnn::chargedIsoPtSumdR03_over_dR05) = getValue(tau.tauID("chargedIsoPtSumdR03") / tau.tauID("chargedIsoPtSum"));
         get(dnn::footprintCorrection) = getValueNorm(tau.tauID("footprintCorrectiondR03"), 9.029f, 26.42f);
         get(dnn::neutralIsoPtSum) = getValueNorm(tau.tauID("neutralIsoPtSum"), 57.59f, 155.3f);
         get(dnn::neutralIsoPtSumWeight_over_neutralIsoPtSum) =
             getValue(tau.tauID("neutralIsoPtSumWeight") / tau.tauID("neutralIsoPtSum"));
         get(dnn::neutralIsoPtSumWeightdR03_over_neutralIsoPtSum) =
             getValue(tau.tauID("neutralIsoPtSumWeightdR03") / tau.tauID("neutralIsoPtSum"));
         get(dnn::neutralIsoPtSumdR03_over_dR05) = getValue(tau.tauID("neutralIsoPtSumdR03") / tau.tauID("neutralIsoPtSum"));
         get(dnn::photonPtSumOutsideSignalCone) = getValueNorm(tau.tauID("photonPtSumOutsideSignalConedR03"), 1.731f, 6.846f);
         get(dnn::puCorrPtSum) = getValueNorm(tau.tauID("puCorrPtSum"), 22.38f, 16.34f);
     // The global PCA coordinates were used as inputs during the NN training, but it was decided to disable
     // them for the inference, because modeling of dxy_PCA in MC poorly describes the data, and x and y coordinates
     // in data results outside of the expected 5 std. dev. input validity range. On the other hand,
     // these coordinates are strongly era-dependent. Kept as comment to document what NN expects.
     if(!disable_dxy_pca_) {
       get(dnn::tau_dxy_pca_x) = getValueNorm(tau.dxy_PCA().x(), -0.0241f, 0.0074f);
       get(dnn::tau_dxy_pca_y) = getValueNorm(tau.dxy_PCA().y(),0.0675f, 0.0128f);
       get(dnn::tau_dxy_pca_z) = getValueNorm(tau.dxy_PCA().z(), 0.7973f, 3.456f);
     } else {
       get(dnn::tau_dxy_pca_x) = 0;
       get(dnn::tau_dxy_pca_y) = 0;
       get(dnn::tau_dxy_pca_z) = 0;
     }
         const bool tau_dxy_valid = std::isnormal(tau.dxy()) && tau.dxy() > - 10 && std::isnormal(tau.dxy_error())
             && tau.dxy_error() > 0;
         if(tau_dxy_valid){
             get(dnn::tau_dxy_valid) = tau_dxy_valid;
             get(dnn::tau_dxy) = getValueNorm(tau.dxy(), 0.0018f, 0.0085f);
             get(dnn::tau_dxy_sig) = getValueNorm(std::abs(tau.dxy())/tau.dxy_error(), 2.26f, 4.191f);
         }
         const bool tau_ip3d_valid = std::isnormal(tau.ip3d()) && tau.ip3d() > - 10 && std::isnormal(tau.ip3d_error())
             && tau.ip3d_error() > 0;
         if(tau_ip3d_valid){
             get(dnn::tau_ip3d_valid) = tau_ip3d_valid;
             get(dnn::tau_ip3d) = getValueNorm(tau.ip3d(), 0.0026f, 0.0114f);
             get(dnn::tau_ip3d_sig) = getValueNorm(std::abs(tau.ip3d()) / tau.ip3d_error(), 2.928f, 4.466f);
         }
         if(leadChargedHadrCand){
             get(dnn::tau_dz) = getValueNorm(leadChargedHadrCand->dz(), 0.f, 0.0190f);
             const bool tau_dz_sig_valid = leadChargedHadrCand->hasTrackDetails() && std::isnormal(leadChargedHadrCand->dz())
                 && std::isnormal(leadChargedHadrCand->dzError()) && leadChargedHadrCand->dzError() > 0;
             get(dnn::tau_dz_sig_valid) = tau_dz_sig_valid;
             const double dzError = leadChargedHadrCand->hasTrackDetails() ? leadChargedHadrCand->dzError() : default_value;
             get(dnn::tau_dz_sig) = getValueNorm(std::abs(leadChargedHadrCand->dz()) / dzError, 4.717f, 11.78f);
         }
         get(dnn::tau_flightLength_x) = getValueNorm(tau.flightLength().x(), -0.0003f, 0.7362f);
         get(dnn::tau_flightLength_y) = getValueNorm(tau.flightLength().y(), -0.0009f, 0.7354f);
         get(dnn::tau_flightLength_z) = getValueNorm(tau.flightLength().z(), -0.0022f, 1.993f);
         // get(dnn::tau_flightLength_sig) = getValueNorm(tau.flightLengthSig(), -4.78f, 9.573f);
         get(dnn::tau_flightLength_sig) = 0.55756444; //This value is set due to a bug in the training
         get(dnn::tau_pt_weighted_deta_strip) = getValueLinear(reco::tau::pt_weighted_deta_strip(tau, tau.decayMode()), 0, 1, true);
 
         get(dnn::tau_pt_weighted_dphi_strip) = getValueLinear(reco::tau::pt_weighted_dphi_strip(tau, tau.decayMode()), 0, 1, true);
         get(dnn::tau_pt_weighted_dr_signal) = getValueNorm(reco::tau::pt_weighted_dr_signal(tau, tau.decayMode()), 0.0052f, 0.01433f);
         get(dnn::tau_pt_weighted_dr_iso) = getValueLinear(reco::tau::pt_weighted_dr_iso(tau,tau.decayMode()), 0, 1, true);
         get(dnn::tau_leadingTrackNormChi2) = getValueNorm(tau.leadingTrackNormChi2(), 1.538f, 4.401f);
         const auto eratio = reco::tau::eratio(tau);
         const bool tau_e_ratio_valid = std::isnormal(eratio) && eratio > 0.f;
         get(dnn::tau_e_ratio_valid) = tau_e_ratio_valid;
         get(dnn::tau_e_ratio) = tau_e_ratio_valid ? getValueLinear(eratio, 0, 1, true) : 0.f;
         const double gj_angle_diff = calculateGottfriedJacksonAngleDifference(tau);
         const bool tau_gj_angle_diff_valid = (std::isnormal(gj_angle_diff) || gj_angle_diff == 0) && gj_angle_diff >= 0;
         get(dnn::tau_gj_angle_diff_valid) = tau_gj_angle_diff_valid;
         get(dnn::tau_gj_angle_diff) = tau_gj_angle_diff_valid ? getValueLinear(gj_angle_diff, 0, pi, true) : 0;
         get(dnn::tau_n_photons) = getValueNorm(reco::tau::n_photons_total(tau), 2.95f, 3.927f);
         get(dnn::tau_emFraction) = getValueLinear(tau.emFraction_MVA(), -1, 1, false);
         get(dnn::tau_inside_ecal_crack) = getValue(isInEcalCrack(tau.p4().eta()));
         get(dnn::leadChargedCand_etaAtEcalEntrance_minus_tau_eta) =
             getValueNorm(tau.etaAtEcalEntranceLeadChargedCand() - tau.p4().eta(), 0.0042f, 0.0323f);
 
         checkInputs(inputs, "tau_block",  dnn::NumberOfInputs);
     }

static void DeepTauId::fillDescriptions ( edm::ConfigurationDescriptions & descriptions )

inlinestatic

Definition at line 567 of file DeepTauId.cc.

References edm::ConfigurationDescriptions::add(), edm::ParameterSetDescription::add(), and AlCaHLTBitMon_QueryRunRegistry::string.

     {
         edm::ParameterSetDescription desc;
         desc.add<edm::InputTag>("electrons", edm::InputTag("slimmedElectrons"));
         desc.add<edm::InputTag>("muons", edm::InputTag("slimmedMuons"));
         desc.add<edm::InputTag>("taus", edm::InputTag("slimmedTaus"));
         desc.add<edm::InputTag>("pfcands", edm::InputTag("packedPFCandidates"));
         desc.add<edm::InputTag>("vertices", edm::InputTag("offlineSlimmedPrimaryVertices"));
         desc.add<edm::InputTag>("rho", edm::InputTag("fixedGridRhoAll"));
         desc.add<std::vector<std::string>>("graph_file", {"RecoTauTag/TrainingFiles/data/DeepTauId/deepTau_2017v2p6_e6.pb"});
         desc.add<bool>("mem_mapped", false);
         desc.add<unsigned>("version", 2);
         desc.add<int>("debug_level", 0);
         desc.add<bool>("disable_dxy_pca", false);
 
         edm::ParameterSetDescription descWP;
         descWP.add<std::string>("VVVLoose", "0");
         descWP.add<std::string>("VVLoose", "0");
         descWP.add<std::string>("VLoose", "0");
         descWP.add<std::string>("Loose", "0");
         descWP.add<std::string>("Medium", "0");
         descWP.add<std::string>("Tight", "0");
         descWP.add<std::string>("VTight", "0");
         descWP.add<std::string>("VVTight", "0");
         descWP.add<std::string>("VVVTight", "0");
         desc.add<edm::ParameterSetDescription>("VSeWP", descWP);
         desc.add<edm::ParameterSetDescription>("VSmuWP", descWP);
         desc.add<edm::ParameterSetDescription>("VSjetWP", descWP);
         descriptions.add("DeepTau", desc);
     }

template<typename Collection >

void DeepTauId::fillGrids	(	const TauType &	tau,
		const Collection &	objects,
		CellGrid &	inner_grid,
		CellGrid &	outer_grid
	)

inlineprivate

Definition at line 794 of file DeepTauId.cc.

References constexpr, reco::deltaPhi(), listHistos::grid, gen::n, hgcalPlots::obj, reco::LeafCandidate::polarP4(), and funct::pow().

     {
         static constexpr double outer_dR2 = 0.25;//0.5^2
         const double inner_radius = getInnerSignalConeRadius(tau.polarP4().pt());
         const double inner_dR2 = std::pow(inner_radius, 2);
 
         const auto addObject = [&](size_t n, double deta, double dphi, CellGrid& grid) {
             const auto& obj = objects.at(n);
             const CellObjectType obj_type = GetCellObjectType(obj);
             if(obj_type  == CellObjectType::Other) return;
             CellIndex cell_index;
             if(grid.tryGetCellIndex(deta, dphi, cell_index)) {
                 Cell& cell = grid[cell_index];
                 auto iter = cell.find(obj_type);
                 if(iter != cell.end()) {
                      const auto& prev_obj = objects.at(iter->second);
                      if(obj.polarP4().pt() > prev_obj.polarP4().pt())
                         iter->second = n;
                 } else {
                     cell[obj_type] = n;
                 }
             }
         };
 
         for(size_t n = 0; n < objects.size(); ++n) {
             const auto& obj = objects.at(n);
             const double deta = obj.polarP4().eta() - tau.polarP4().eta();
             const double dphi = reco::deltaPhi(obj.polarP4().phi(), tau.polarP4().phi());
             const double dR2 = std::pow(deta, 2) + std::pow(dphi, 2);
             if(dR2 < inner_dR2)
                 addObject(n, deta, dphi, inner_grid);
             if(dR2 < outer_dR2)
                 addObject(n, deta, dphi, outer_grid);
         }
     }

static const pat::Electron* DeepTauId::findMatchedElectron	(	const pat::Tau &	tau,
		const pat::ElectronCollection &	electrons,
		double	deltaR
	)

inlinestaticprivate

Definition at line 1692 of file DeepTauId.cc.

References boostedElectronIsolation_cff::deltaR, reco::deltaR2(), reco::LeafCandidate::p4(), and reco::LeafCandidate::pt().

     {
         const double dR2 = deltaR*deltaR;
         const pat::Electron* matched_ele = nullptr;
         for(const auto& ele : electrons) {
             if(reco::deltaR2(tau.p4(), ele.p4()) < dR2 && (!matched_ele || matched_ele->pt() < ele.pt())) {
                 matched_ele = &ele;
             }
         }
         return matched_ele;
     }

static double DeepTauId::getInnerSignalConeRadius ( double pt )

inlinestaticprivate

Definition at line 1651 of file DeepTauId.cc.

References constexpr, SiStripPI::max, and cosmictrackSelector_cfi::min_pt.

     {
         static constexpr double min_pt = 30., min_radius = 0.05, cone_opening_coef = 3.;
         // This is equivalent of the original formula (std::max(std::min(0.1, 3.0/pt), 0.05)
         return std::max(cone_opening_coef / std::max(pt, min_pt), min_radius);
     }

static const OutputCollection& DeepTauId::GetOutputs ( )

inlinestatic

Definition at line 556 of file DeepTauId.cc.

References constexpr, and Output.

     {
         static constexpr size_t e_index = 0, mu_index = 1, tau_index = 2, jet_index = 3;
         static const OutputCollection outputs_ = {
             { "VSe", Output({tau_index}, {e_index, tau_index}) },
             { "VSmu", Output({tau_index}, {mu_index, tau_index}) },
             { "VSjet", Output({tau_index}, {jet_index, tau_index}) },
         };
         return outputs_;
     }

void DeepTauId::getPartialPredictions	(	tensorflow::Tensor &	convTensor,
		bool	is_inner,
		int	eta_index,
		int	phi_index
	)

inlineprivate

Definition at line 830 of file DeepTauId.cc.

References tensorflow::run().

     {
         std::vector<tensorflow::Tensor> pred_vector;
         if(is_inner) {
             tensorflow::run(&(cache_->getSession("inner")),
                 { { "input_inner_egamma", *eGammaTensor_.at(is_inner) },
                   { "input_inner_muon", *muonTensor_.at(is_inner) },
                   { "input_inner_hadrons", *hadronsTensor_.at(is_inner) }, },
                 { "inner_all_dropout_4/Identity" }, &pred_vector);
         } else {
             tensorflow::run(&(cache_->getSession("outer")),
                 { { "input_outer_egamma", *eGammaTensor_.at(is_inner) },
                   { "input_outer_muon", *muonTensor_.at(is_inner) },
                   { "input_outer_hadrons", *hadronsTensor_.at(is_inner) }, },
                 { "outer_all_dropout_4/Identity" }, &pred_vector);
         }
         setCellConvFeatures(convTensor, pred_vector.at(0), eta_index, phi_index);
     }

tensorflow::Tensor DeepTauId::getPredictions	(	edm::Event &	event,
		const edm::EventSetup &	es,
		edm::Handle< TauCollection >	taus
	)

inlineoverrideprivatevirtual

Implements deep_tau::DeepTauBase.

Definition at line 725 of file DeepTauId.cc.

References lowPtElectrons_cff::electrons, Exception, gen::k, extraflags_cff::muons, deep_tau::NumberOfOutputs, jets_cff::version, and electrons_cff::vertices.

     {
         edm::Handle<pat::ElectronCollection> electrons;
         event.getByToken(electrons_token_, electrons);
 
         edm::Handle<pat::MuonCollection> muons;
         event.getByToken(muons_token_, muons);
 
         edm::Handle<pat::PackedCandidateCollection> pfCands;
         event.getByToken(pfcandToken_, pfCands);
 
         edm::Handle<reco::VertexCollection> vertices;
         event.getByToken(vtxToken_, vertices);
 
         edm::Handle<double> rho;
         event.getByToken(rho_token_, rho);
 
         tensorflow::Tensor predictions(tensorflow::DT_FLOAT, { static_cast<int>(taus->size()),
                                        deep_tau::NumberOfOutputs});
         for(size_t tau_index = 0; tau_index < taus->size(); ++tau_index) {
             std::vector<tensorflow::Tensor> pred_vector;
             if(version == 1)
                 getPredictionsV1(taus->at(tau_index), *electrons, *muons, pred_vector);
             else if(version == 2)
                 getPredictionsV2(taus->at(tau_index), *electrons, *muons, *pfCands, vertices->at(0), *rho, pred_vector);
             else
                 throw cms::Exception("DeepTauId") << "version " << version << " is not supported.";
             for(int k = 0; k < deep_tau::NumberOfOutputs; ++k) {
                 const float pred = pred_vector[0].flat<float>()(k);
                 if(!(pred >= 0 && pred <= 1))
                     throw cms::Exception("DeepTauId") << "invalid prediction = " << pred << " for tau_index = "
                                                       << tau_index << ", pred_index = " << k;
                 predictions.matrix<float>()(tau_index, k) = pred;
             }
         }
         return predictions;
     }

void DeepTauId::getPredictionsV1	(	const TauType &	tau,
		const pat::ElectronCollection &	electrons,
		const pat::MuonCollection &	muons,
		std::vector< tensorflow::Tensor > &	pred_vector
	)

inlineprivate

Definition at line 764 of file DeepTauId.cc.

References lowPtElectrons_cff::electrons, PatBasicFWLiteJetAnalyzer_Selector_cfg::inputs, extraflags_cff::muons, tensorflow::run(), and metsig::tau.

     {
         const tensorflow::Tensor& inputs = createInputsV1<dnn_inputs_2017v1>(tau, electrons, muons);
         tensorflow::run(&(cache_->getSession()), { { input_layer_, inputs } }, { output_layer_ }, &pred_vector);
     }

void DeepTauId::getPredictionsV2	(	const TauType &	tau,
		const pat::ElectronCollection &	electrons,
		const pat::MuonCollection &	muons,
		const pat::PackedCandidateCollection &	pfCands,
		const reco::Vertex &	pv,
		double	rho,
		std::vector< tensorflow::Tensor > &	pred_vector
	)

inlineprivate

Definition at line 771 of file DeepTauId.cc.

References tensorflow::run().

     {
         CellGrid inner_grid(dnn_inputs_2017_v2::number_of_inner_cell, dnn_inputs_2017_v2::number_of_inner_cell,
                             0.02, 0.02);
         CellGrid outer_grid(dnn_inputs_2017_v2::number_of_outer_cell, dnn_inputs_2017_v2::number_of_outer_cell,
                             0.05, 0.05);
         fillGrids(tau, electrons, inner_grid, outer_grid);
         fillGrids(tau, muons, inner_grid, outer_grid);
         fillGrids(tau, pfCands, inner_grid, outer_grid);
 
         createTauBlockInputs(tau, pv, rho);
         createConvFeatures(tau, pv, rho, electrons, muons, pfCands, inner_grid, true);
         createConvFeatures(tau, pv, rho, electrons, muons, pfCands, outer_grid, false);
 
         tensorflow::run(&(cache_->getSession("core")),
             { { "input_tau", *tauBlockTensor_ },
               { "input_inner", *convTensor_.at(true)}, { "input_outer", *convTensor_.at(false) } },
             { "main_output/Softmax" }, &pred_vector);
     }

template<typename T >

static float DeepTauId::getValue ( T value )

inlinestaticprivate

Definition at line 658 of file DeepTauId.cc.

References f, and relativeConstraints::value.

     {
         return std::isnormal(value) ? static_cast<float>(value) : 0.f;
     }

template<typename T >

static float DeepTauId::getValueLinear	(	T	value,
		float	min_value,
		float	max_value,
		bool	positive
	)

inlinestaticprivate

Definition at line 664 of file DeepTauId.cc.

References reco::JetExtendedAssociation::getValue(), and hcaldqm::quantity::min_value.

     {
         const float fixed_value = getValue(value);
         const float clamped_value = std::clamp(fixed_value, min_value, max_value);
         float transformed_value = (clamped_value - min_value) / (max_value - min_value);
         if(!positive)
             transformed_value = transformed_value * 2 - 1;
         return transformed_value;
     }

template<typename T >

static float DeepTauId::getValueNorm	(	T	value,
		float	mean,
		float	sigma,
		float	n_sigmas_max = `5`
	)

inlinestaticprivate

Definition at line 675 of file DeepTauId.cc.

References reco::JetExtendedAssociation::getValue(), and SiStripPI::mean.

     {
         const float fixed_value = getValue(value);
         const float norm_value = (fixed_value - mean) / sigma;
         return std::clamp(norm_value, -n_sigmas_max, n_sigmas_max);
     }

static void DeepTauId::globalEndJob ( const deep_tau::DeepTauCache * cache_ )

inlinestatic

Definition at line 650 of file DeepTauId.cc.

References deep_tau::DeepTauBase::globalEndJob().

     {
         return DeepTauBase::globalEndJob(cache_);
     }

static std::unique_ptr<deep_tau::DeepTauCache> DeepTauId::initializeGlobalCache ( const edm::ParameterSet & cfg )

inlinestatic

Definition at line 645 of file DeepTauId.cc.

References deep_tau::DeepTauBase::initializeGlobalCache().

     {
         return DeepTauBase::initializeGlobalCache(cfg);
     }

static bool DeepTauId::isInEcalCrack ( double eta )

inlinestaticprivate

Definition at line 1686 of file DeepTauId.cc.

References funct::abs().

     {
         const double abs_eta = std::abs(eta);
         return abs_eta > 1.46 && abs_eta < 1.558;
     }

template<typename CandidateCollection >

static void DeepTauId::processIsolationPFComponents	(	const pat::Tau &	tau,
		const CandidateCollection &	candidates,
		LorentzVectorXYZ &	p4,
		float &	pt,
		float &	d_eta,
		float &	d_phi,
		float &	m,
		float &	n
	)

inlinestaticprivate

Definition at line 1633 of file DeepTauId.cc.

References particleFlow_cfi::dEta, particleFlow_cfi::dPhi, gen::n, and reco::LeafCandidate::p4().

     {
         p4 = LorentzVectorXYZ(0, 0, 0, 0);
         n = 0;
 
         for(const auto& cand : candidates) {
             p4 += cand->p4();
             ++n;
         }
 
         pt = n != 0 ? p4.Pt() : default_value;
         d_eta = n != 0 ? dEta(p4, tau.p4()) : default_value;
         d_phi = n != 0 ? dPhi(p4, tau.p4()) : default_value;
         m = n != 0 ? p4.mass() : default_value;
     }

template<typename CandidateCollection >

static void DeepTauId::processSignalPFComponents	(	const pat::Tau &	tau,
		const CandidateCollection &	candidates,
		LorentzVectorXYZ &	p4_inner,
		LorentzVectorXYZ &	p4_outer,
		float &	pt_inner,
		float &	dEta_inner,
		float &	dPhi_inner,
		float &	m_inner,
		float &	pt_outer,
		float &	dEta_outer,
		float &	dPhi_outer,
		float &	m_outer,
		float &	n_inner,
		float &	n_outer
	)

inlinestaticprivate

Definition at line 1597 of file DeepTauId.cc.

References reco::deltaR(), particleFlow_cfi::dEta, particleFlow_cfi::dPhi, pat::Tau::leadChargedHadrCand(), reco::LeafCandidate::p4(), and reco::LeafCandidate::pt().

     {
         p4_inner = LorentzVectorXYZ(0, 0, 0, 0);
         p4_outer = LorentzVectorXYZ(0, 0, 0, 0);
         n_inner = 0;
         n_outer = 0;
 
         const double innerSigCone_radius = getInnerSignalConeRadius(tau.pt());
         for(const auto& cand : candidates) {
             const double dR = reco::deltaR(cand->p4(), tau.leadChargedHadrCand()->p4());
             const bool isInside_innerSigCone = dR < innerSigCone_radius;
             if(isInside_innerSigCone) {
                 p4_inner += cand->p4();
                 ++n_inner;
             } else {
                 p4_outer += cand->p4();
                 ++n_outer;
             }
         }
 
         pt_inner = n_inner != 0 ? p4_inner.Pt() : default_value;
         dEta_inner = n_inner != 0 ? dEta(p4_inner, tau.p4()) : default_value;
         dPhi_inner = n_inner != 0 ? dPhi(p4_inner, tau.p4()) : default_value;
         m_inner = n_inner != 0 ? p4_inner.mass() : default_value;
 
         pt_outer = n_outer != 0 ? p4_outer.Pt() : default_value;
         dEta_outer = n_outer != 0 ? dEta(p4_outer, tau.p4()) : default_value;
         dPhi_outer = n_outer != 0 ? dPhi(p4_outer, tau.p4()) : default_value;
         m_outer = n_outer != 0 ? p4_outer.mass() : default_value;
     }

void DeepTauId::setCellConvFeatures	(	tensorflow::Tensor &	convTensor,
		const tensorflow::Tensor &	features,
		int	eta_index,
		int	phi_index
	)

inlineprivate

Definition at line 874 of file DeepTauId.cc.

References gen::n.

     {
         for(int n = 0; n < dnn_inputs_2017_v2::number_of_conv_features; ++n)
             convTensor.tensor<float, 4>()(0, eta_index, phi_index, n) = features.tensor<float, 4>()(0, 0, 0, n);
     }

Member Data Documentation

std::array<std::shared_ptr<tensorflow::Tensor>, 2> DeepTauId::convTensor_

private

Definition at line 1714 of file DeepTauId.cc.

const int DeepTauId::debug_level

private

Definition at line 1711 of file DeepTauId.cc.

float DeepTauId::default_value = -999.

static

Definition at line 554 of file DeepTauId.cc.

const bool DeepTauId::disable_dxy_pca_

private

Definition at line 1712 of file DeepTauId.cc.

std::array<std::shared_ptr<tensorflow::Tensor>, 2> DeepTauId::eGammaTensor_

private

Definition at line 1714 of file DeepTauId.cc.

edm::EDGetTokenT<ElectronCollection> DeepTauId::electrons_token_

private

Definition at line 1706 of file DeepTauId.cc.

std::array<std::shared_ptr<tensorflow::Tensor>, 2> DeepTauId::hadronsTensor_

private

Definition at line 1714 of file DeepTauId.cc.

std::string DeepTauId::input_layer_

private

Definition at line 1709 of file DeepTauId.cc.

edm::EDGetTokenT<MuonCollection> DeepTauId::muons_token_

private

Definition at line 1707 of file DeepTauId.cc.

std::array<std::shared_ptr<tensorflow::Tensor>, 2> DeepTauId::muonTensor_

private

Definition at line 1714 of file DeepTauId.cc.

std::string DeepTauId::output_layer_

private

Definition at line 1709 of file DeepTauId.cc.

float DeepTauId::pi = M_PI

staticprivate

Definition at line 655 of file DeepTauId.cc.

edm::EDGetTokenT<double> DeepTauId::rho_token_

private

Definition at line 1708 of file DeepTauId.cc.

std::shared_ptr<tensorflow::Tensor> DeepTauId::tauBlockTensor_

private

Definition at line 1713 of file DeepTauId.cc.

const unsigned DeepTauId::version

private

Definition at line 1710 of file DeepTauId.cc.

Referenced by validation.Sample::datasetpattern(), and validation.Sample::filename().

std::array<std::shared_ptr<tensorflow::Tensor>, 2> DeepTauId::zeroOutputTensor_

private

Definition at line 1714 of file DeepTauId.cc.

Public Member Functions

Static Public Member Functions

Static Public Attributes

Private Member Functions

Static Private Member Functions

Private Attributes

Static Private Attributes

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation

Member Data Documentation