#include <HGCalImagingAlgo.h>

Classes
struct	Hexel

Public Types
typedef math::XYZPoint	Point
	point in the space More...

enum	VerbosityLevel { pDEBUG = 0, pWARNING = 1, pINFO = 2, pERROR = 3 }

Public Member Functions
void	computeThreshold ()

std::vector< reco::BasicCluster >	getClusters (bool)

void	getEventSetup (const edm::EventSetup &es)

	HGCalImagingAlgo ()

	HGCalImagingAlgo (const std::vector< double > &vecDeltas_in, double kappa_in, double ecut_in, reco::CaloCluster::AlgoId algoId_in, bool dependSensor_in, const std::vector< double > &dEdXweights_in, const std::vector< double > &thicknessCorrection_in, const std::vector< double > &fcPerMip_in, double fcPerEle_in, const std::vector< double > &nonAgedNoises_in, double noiseMip_in, VerbosityLevel the_verbosity=pERROR)

	HGCalImagingAlgo (const std::vector< double > &vecDeltas_in, double kappa_in, double ecut_in, double showerSigma, reco::CaloCluster::AlgoId algoId_in, bool dependSensor_in, const std::vector< double > &dEdXweights_in, const std::vector< double > &thicknessCorrection_in, const std::vector< double > &fcPerMip_in, double fcPerEle_in, const std::vector< double > &nonAgedNoises_in, double noiseMip_in, VerbosityLevel the_verbosity=pERROR)

void	makeClusters ()

void	populate (const HGCRecHitCollection &hits)

void	reset ()

void	setVerbosity (VerbosityLevel the_verbosity)

virtual	~HGCalImagingAlgo ()

Static Public Attributes
static const unsigned int	maxlayer = 52

Private Types
typedef KDTreeNodeInfoT< Hexel, 2 >	KDNode

typedef KDTreeLinkerAlgo< Hexel, 2 >	KDTree

Private Member Functions
double	calculateDistanceToHigher (std::vector< KDNode > &) const

double	calculateEnergyWithFraction (const std::vector< KDNode > &, const std::vector< double > &)

double	calculateLocalDensity (std::vector< KDNode > &, KDTree &, const unsigned int) const

math::XYZPoint	calculatePosition (std::vector< KDNode > &) const

math::XYZPoint	calculatePositionWithFraction (const std::vector< KDNode > &, const std::vector< double > &)

double	distance (const Hexel &pt1, const Hexel &pt2) const

double	distance2 (const Hexel &pt1, const Hexel &pt2) const

int	findAndAssignClusters (std::vector< KDNode > &, KDTree &, double, KDTreeBox &, const unsigned int, std::vector< std::vector< KDNode > > &) const

std::vector< unsigned >	findLocalMaximaInCluster (const std::vector< KDNode > &)

void	shareEnergy (const std::vector< KDNode > &, const std::vector< unsigned > &, std::vector< std::vector< double > > &)

std::vector< size_t >	sort_by_delta (const std::vector< KDNode > &v) const

Private Attributes
reco::CaloCluster::AlgoId	algoId

std::vector< reco::BasicCluster >	clusters_v

std::vector< double >	dEdXweights

bool	dependSensor

double	ecut

double	fcPerEle

std::vector< double >	fcPerMip

bool	initialized

double	kappa

std::vector< std::vector< std::vector< KDNode > > >	layerClustersPerLayer

std::vector< std::array< float, 2 > >	maxpos

std::vector< std::array< float, 2 > >	minpos

double	noiseMip

std::vector< double >	nonAgedNoises

std::vector< std::vector< KDNode > >	points

hgcal::RecHitTools	rhtools_

double	sigma2

std::vector< double >	thicknessCorrection

std::vector< std::vector< double > >	thresholds

std::vector< std::vector< double > >	v_sigmaNoise

std::vector< double >	vecDeltas

VerbosityLevel	verbosity

Static Private Attributes
static const unsigned int	lastLayerEE = 28

static const unsigned int	lastLayerFH = 40

Detailed Description

Definition at line 44 of file HGCalImagingAlgo.h.

Member Typedef Documentation

typedef KDTreeNodeInfoT<Hexel,2> HGCalImagingAlgo::KDNode

private

Definition at line 252 of file HGCalImagingAlgo.h.

typedef KDTreeLinkerAlgo<Hexel,2> HGCalImagingAlgo::KDTree

private

Definition at line 251 of file HGCalImagingAlgo.h.

typedef math::XYZPoint HGCalImagingAlgo::Point

point in the space

Definition at line 157 of file HGCalImagingAlgo.h.

Member Enumeration Documentation

enum HGCalImagingAlgo::VerbosityLevel

Enumerator
pDEBUG
pWARNING
pINFO
pERROR

Definition at line 50 of file HGCalImagingAlgo.h.

50 { pDEBUG = 0, pWARNING = 1, pINFO = 2, pERROR = 3 };

HGCalImagingAlgo::pDEBUG

Definition: HGCalImagingAlgo.h:50

HGCalImagingAlgo::pERROR

Definition: HGCalImagingAlgo.h:50

HGCalImagingAlgo::pWARNING

Definition: HGCalImagingAlgo.h:50

HGCalImagingAlgo::pINFO

Definition: HGCalImagingAlgo.h:50

Constructor & Destructor Documentation

HGCalImagingAlgo::HGCalImagingAlgo ( )

inline

Definition at line 52 of file HGCalImagingAlgo.h.

                    : vecDeltas(), kappa(1.), ecut(0.),
         sigma2(1.0),
         algoId(reco::CaloCluster::undefined),
         verbosity(pERROR),initialized(false){
 }

HGCalImagingAlgo::HGCalImagingAlgo	(	const std::vector< double > &	vecDeltas_in,
		double	kappa_in,
		double	ecut_in,
		reco::CaloCluster::AlgoId	algoId_in,
		bool	dependSensor_in,
		const std::vector< double > &	dEdXweights_in,
		const std::vector< double > &	thicknessCorrection_in,
		const std::vector< double > &	fcPerMip_in,
		double	fcPerEle_in,
		const std::vector< double > &	nonAgedNoises_in,
		double	noiseMip_in,
		VerbosityLevel	the_verbosity = `pERROR`
	)

inline

Definition at line 58 of file HGCalImagingAlgo.h.

References maxlayer, and maxpos.

                                                         :
         vecDeltas(vecDeltas_in), kappa(kappa_in),
         ecut(ecut_in),
         sigma2(1.0),
         algoId(algoId_in),
         dependSensor(dependSensor_in),
         dEdXweights(dEdXweights_in),
         thicknessCorrection(thicknessCorrection_in),
         fcPerMip(fcPerMip_in),
         fcPerEle(fcPerEle_in),
         nonAgedNoises(nonAgedNoises_in),
         noiseMip(noiseMip_in),
         verbosity(the_verbosity),
         initialized(false),
         points(2*(maxlayer+1)),
         minpos(2*(maxlayer+1),{
                 {0.0f,0.0f}
         }),
         maxpos(2*(maxlayer+1),{ {0.0f,0.0f} })
 {
 }

HGCalImagingAlgo::HGCalImagingAlgo	(	const std::vector< double > &	vecDeltas_in,
		double	kappa_in,
		double	ecut_in,
		double	showerSigma,
		reco::CaloCluster::AlgoId	algoId_in,
		bool	dependSensor_in,
		const std::vector< double > &	dEdXweights_in,
		const std::vector< double > &	thicknessCorrection_in,
		const std::vector< double > &	fcPerMip_in,
		double	fcPerEle_in,
		const std::vector< double > &	nonAgedNoises_in,
		double	noiseMip_in,
		VerbosityLevel	the_verbosity = `pERROR`
	)

inline

Definition at line 89 of file HGCalImagingAlgo.h.

References maxlayer, and maxpos.

                                                         : vecDeltas(vecDeltas_in), kappa(kappa_in),
         ecut(ecut_in),
         sigma2(std::pow(showerSigma,2.0)),
         algoId(algoId_in),
         dependSensor(dependSensor_in),
         dEdXweights(dEdXweights_in),
         thicknessCorrection(thicknessCorrection_in),
         fcPerMip(fcPerMip_in),
         fcPerEle(fcPerEle_in),
         nonAgedNoises(nonAgedNoises_in),
         noiseMip(noiseMip_in),
         verbosity(the_verbosity),
         initialized(false),
         points(2*(maxlayer+1)),
     minpos(2*(maxlayer+1),{
                 {0.0f,0.0f}
         }),
     maxpos(2*(maxlayer+1),{ {0.0f,0.0f} })
 {
 }

virtual HGCalImagingAlgo::~HGCalImagingAlgo ( )

inlinevirtual

Definition at line 120 of file HGCalImagingAlgo.h.

121 {

122 }

Member Function Documentation

double HGCalImagingAlgo::calculateDistanceToHigher ( std::vector< KDNode > & nd ) const

private

Definition at line 266 of file HGCalImagingAlgo.cc.

References data, distance2(), mps_fire::i, sorted_indices(), mathSSE::sqrt(), and tmp.

Referenced by distance(), and makeClusters().

                                                                        {
 
   // sort vector of Hexels by decreasing local density
   std::vector<size_t> rs = sorted_indices(nd);
 
   double maxdensity = 0.0;
   int nearestHigher = -1;
 
   if (!rs.empty())
     maxdensity = nd[rs[0]].data.rho;
   else
     return maxdensity; // there are no hits
   double dist2 = 0.;
   // start by setting delta for the highest density hit to
   // the most distant hit - this is a convention
 
   for (auto &j : nd) {
     double tmp = distance2(nd[rs[0]].data, j.data);
     if (tmp > dist2)
       dist2 = tmp;
   }
   nd[rs[0]].data.delta = std::sqrt(dist2);
   nd[rs[0]].data.nearestHigher = nearestHigher;
 
   // now we save the largest distance as a starting point
   const double max_dist2 = dist2;
   const unsigned int nd_size = nd.size();
 
   for (unsigned int oi = 1; oi < nd_size;
        ++oi) { // start from second-highest density
     dist2 = max_dist2;
     unsigned int i = rs[oi];
     // we only need to check up to oi since hits
     // are ordered by decreasing density
     // and all points coming BEFORE oi are guaranteed to have higher rho
     // and the ones AFTER to have lower rho
     for (unsigned int oj = 0; oj < oi; ++oj) {
       unsigned int j = rs[oj];
       double tmp = distance2(nd[i].data, nd[j].data);
       if (tmp <= dist2) { // this "<=" instead of "<" addresses the (rare) case
                           // when there are only two hits
         dist2 = tmp;
         nearestHigher = j;
       }
     }
     nd[i].data.delta = std::sqrt(dist2);
     nd[i].data.nearestHigher =
         nearestHigher; // this uses the original unsorted hitlist
   }
   return maxdensity;
 }

double HGCalImagingAlgo::calculateEnergyWithFraction	(	const std::vector< KDNode > &	hits,
		const std::vector< double > &	fractions
	)

private

Definition at line 505 of file HGCalImagingAlgo.cc.

References mps_fire::i, and mps_fire::result.

Referenced by distance(), getClusters(), and shareEnergy().

                                                                        {
   double result = 0.0;
   for (unsigned i = 0; i < hits.size(); ++i) {
     result += fractions[i] * hits[i].data.weight;
   }
   return result;
 }

double HGCalImagingAlgo::calculateLocalDensity	(	std::vector< KDNode > &	nd,
		KDTree &	lp,
		const unsigned int	layer
	)		const

private

Definition at line 233 of file HGCalImagingAlgo.cc.

References data, distance(), runEdmFileComparison::found, mps_fire::i, lastLayerEE, lastLayerFH, SiStripPI::max, KDTreeLinkerAlgo< DATA, DIM >::search(), and vecDeltas.

Referenced by distance(), and makeClusters().

                                                                                {
 
   double maxdensity = 0.;
   float delta_c; // maximum search distance (critical distance) for local
                  // density calculation
   if (layer <= lastLayerEE)
     delta_c = vecDeltas[0];
   else if (layer <= lastLayerFH)
     delta_c = vecDeltas[1];
   else
     delta_c = vecDeltas[2];
 
   // for each node calculate local density rho and store it
   for (unsigned int i = 0; i < nd.size(); ++i) {
     // speec up search by looking within +/- delta_c window only
     KDTreeBox search_box(nd[i].dims[0] - delta_c, nd[i].dims[0] + delta_c,
                          nd[i].dims[1] - delta_c, nd[i].dims[1] + delta_c);
     std::vector<KDNode> found;
     lp.search(search_box, found);
     const unsigned int found_size = found.size();
     for (unsigned int j = 0; j < found_size; j++) {
       if (distance(nd[i].data, found[j].data) < delta_c) {
         nd[i].data.rho += found[j].data.weight;
         maxdensity = std::max(maxdensity, nd[i].data.rho);
       }
     } // end loop found
   }   // end loop nodes
   return maxdensity;
 }

math::XYZPoint HGCalImagingAlgo::calculatePosition ( std::vector< KDNode > & v ) const

private

Definition at line 191 of file HGCalImagingAlgo.cc.

References data, mps_fire::i, x, y, and z.

Referenced by distance(), and getClusters().

                                                               {
   float total_weight = 0.f;
   float x = 0.f;
   float y = 0.f;
   float z = 0.f;
   unsigned int v_size = v.size();
   unsigned int maxEnergyIndex = 0;
   float maxEnergyValue = 0;
   bool haloOnlyCluster = true;
 
   // loop over hits in cluster candidate building up weight for
   // energy-weighted position calculation and determining the maximum
   // energy hit in case this is a halo-only cluster
   for (unsigned int i = 0; i < v_size; i++) {
     if (!v[i].data.isHalo) {
       haloOnlyCluster = false;
       total_weight += v[i].data.weight;
       x += v[i].data.x * v[i].data.weight;
       y += v[i].data.y * v[i].data.weight;
       z += v[i].data.z * v[i].data.weight;
     } else {
       if (v[i].data.weight > maxEnergyValue) {
         maxEnergyValue = v[i].data.weight;
         maxEnergyIndex = i;
       }
     }
   }
 
   if (!haloOnlyCluster) {
     if (total_weight != 0) {
       auto inv_tot_weight = 1. / total_weight;
       return math::XYZPoint(x * inv_tot_weight, y * inv_tot_weight,
                             z * inv_tot_weight);
     }
   } else if (v_size > 0) {
     // return position of hit with maximum energy
     return math::XYZPoint(v[maxEnergyIndex].data.x, v[maxEnergyIndex].data.y,
                           v[maxEnergyIndex].data.z);
   }
   return math::XYZPoint(0, 0, 0);
 }

math::XYZPoint HGCalImagingAlgo::calculatePositionWithFraction	(	const std::vector< KDNode > &	hits,
		const std::vector< double > &	fractions
	)

private

Definition at line 490 of file HGCalImagingAlgo.cc.

References mps_fire::i, mps_fire::result, mps_merge::weight, x, y, and z.

Referenced by distance(), getClusters(), and shareEnergy().

                                                                        {
   double norm(0.0), x(0.0), y(0.0), z(0.0);
   for (unsigned i = 0; i < hits.size(); ++i) {
     const double weight = fractions[i] * hits[i].data.weight;
     norm += weight;
     x += weight * hits[i].data.x;
     y += weight * hits[i].data.y;
     z += weight * hits[i].data.z;
   }
   math::XYZPoint result(x, y, z);
   result /= norm;
   return result;
 }

void HGCalImagingAlgo::computeThreshold ( )

Definition at line 614 of file HGCalImagingAlgo.cc.

References dEdXweights, ecut, fcPerEle, fcPerMip, initialized, lastLayerFH, maxlayer, noiseMip, nonAgedNoises, thicknessCorrection, thresholds, and v_sigmaNoise.

Referenced by populate(), and reset().

                                         {
 
   if (initialized)
     return; // only need to calculate thresholds once
 
   std::vector<double> dummy;
   const unsigned maxNumberOfThickIndices = 3;
   dummy.resize(maxNumberOfThickIndices, 0);
   thresholds.resize(maxlayer, dummy);
   v_sigmaNoise.resize(maxlayer, dummy);
 
   for (unsigned ilayer = 1; ilayer <= maxlayer; ++ilayer) {
     for (unsigned ithick = 0; ithick < maxNumberOfThickIndices; ++ithick) {
       float sigmaNoise =
           0.001f * fcPerEle * nonAgedNoises[ithick] * dEdXweights[ilayer] /
           (fcPerMip[ithick] * thicknessCorrection[ithick]);
       thresholds[ilayer - 1][ithick] = sigmaNoise * ecut;
       v_sigmaNoise[ilayer - 1][ithick] = sigmaNoise;
     }
   }
 
   // now BH, much faster
   for (unsigned ilayer = lastLayerFH + 1; ilayer <= maxlayer; ++ilayer) {
     float sigmaNoise = 0.001f * noiseMip * dEdXweights[ilayer];
     std::vector<double> bhDummy_thresholds;
     std::vector<double> bhDummy_sigmaNoise;
     bhDummy_thresholds.push_back(sigmaNoise * ecut);
     bhDummy_sigmaNoise.push_back(sigmaNoise);
     thresholds[ilayer - 1] = bhDummy_thresholds;
     v_sigmaNoise[ilayer - 1] = bhDummy_sigmaNoise;
   }
   initialized = true;
 }

double HGCalImagingAlgo::distance	(	const Hexel &	pt1,
		const Hexel &	pt2
	)		const

inlineprivate

Definition at line 280 of file HGCalImagingAlgo.h.

References calculateDistanceToHigher(), calculateEnergyWithFraction(), calculateLocalDensity(), calculatePosition(), calculatePositionWithFraction(), distance2(), findAndAssignClusters(), findLocalMaximaInCluster(), shareEnergy(), and mathSSE::sqrt().

Referenced by calculateLocalDensity(), findAndAssignClusters(), and findLocalMaximaInCluster().

                                                                 {   //2-d distance on the layer (x-y)
         return std::sqrt(distance2(pt1,pt2));
 }

double HGCalImagingAlgo::distance2	(	const Hexel &	pt1,
		const Hexel &	pt2
	)		const

inlineprivate

Definition at line 275 of file HGCalImagingAlgo.h.

References PVValHelper::dx, PVValHelper::dy, HGCalImagingAlgo::Hexel::x, and HGCalImagingAlgo::Hexel::y.

Referenced by calculateDistanceToHigher(), and distance().

                                                                  {   //distance squared
         const double dx = pt1.x - pt2.x;
         const double dy = pt1.y - pt2.y;
         return (dx*dx + dy*dy);
 }   //distance squaredq

int HGCalImagingAlgo::findAndAssignClusters	(	std::vector< KDNode > &	,
		KDTree &	,
		double	,
		KDTreeBox &	,
		const unsigned int	,
		std::vector< std::vector< KDNode > > &
	)		const

private

Definition at line 317 of file HGCalImagingAlgo.cc.

References KDTreeLinkerAlgo< DATA, DIM >::build(), KDTreeLinkerAlgo< DATA, DIM >::clear(), gather_cfg::cout, data, dependSensor, distance(), runEdmFileComparison::found, mps_fire::i, kappa, lastLayerEE, lastLayerFH, pINFO, KDTreeLinkerAlgo< DATA, DIM >::search(), sort_by_delta(), sorted_indices(), vecDeltas, and verbosity.

Referenced by distance(), and makeClusters().

                                                          {
 
   // this is called once per layer and endcap...
   // so when filling the cluster temporary vector of Hexels we resize each time
   // by the number  of clusters found. This is always equal to the number of
   // cluster centers...
 
   unsigned int nClustersOnLayer = 0;
   float delta_c; // critical distance
   if (layer <= lastLayerEE)
     delta_c = vecDeltas[0];
   else if (layer <= lastLayerFH)
     delta_c = vecDeltas[1];
   else
     delta_c = vecDeltas[2];
 
   std::vector<size_t> rs =
       sorted_indices(nd); // indices sorted by decreasing rho
   std::vector<size_t> ds =
       sort_by_delta(nd); // sort in decreasing distance to higher
 
   const unsigned int nd_size = nd.size();
   for (unsigned int i = 0; i < nd_size; ++i) {
 
     if (nd[ds[i]].data.delta < delta_c)
       break; // no more cluster centers to be looked at
     if (dependSensor) {
 
       float rho_c = kappa * nd[ds[i]].data.sigmaNoise;
       if (nd[ds[i]].data.rho < rho_c)
         continue; // set equal to kappa times noise threshold
 
     } else if (nd[ds[i]].data.rho * kappa < maxdensity)
       continue;
 
     nd[ds[i]].data.clusterIndex = nClustersOnLayer;
     if (verbosity < pINFO) {
       std::cout << "Adding new cluster with index " << nClustersOnLayer
                 << std::endl;
       std::cout << "Cluster center is hit " << ds[i] << std::endl;
     }
     nClustersOnLayer++;
   }
 
   // at this point nClustersOnLayer is equal to the number of cluster centers -
   // if it is zero we are  done
   if (nClustersOnLayer == 0)
     return nClustersOnLayer;
 
   // assign remaining points to clusters, using the nearestHigher set from
   // previous step (always set except
   // for top density hit that is skipped...)
   for (unsigned int oi = 1; oi < nd_size; ++oi) {
     unsigned int i = rs[oi];
     int ci = nd[i].data.clusterIndex;
     if (ci ==
         -1) { // clusterIndex is initialised with -1 if not yet used in cluster
       nd[i].data.clusterIndex = nd[nd[i].data.nearestHigher].data.clusterIndex;
     }
   }
 
   // make room in the temporary cluster vector for the additional clusterIndex
   // clusters
   // from this layer
   if (verbosity < pINFO) {
     std::cout << "resizing cluster vector by " << nClustersOnLayer << std::endl;
   }
   clustersOnLayer.resize(nClustersOnLayer);
 
   // assign points closer than dc to other clusters to border region
   // and find critical border density
   std::vector<double> rho_b(nClustersOnLayer, 0.);
   lp.clear();
   lp.build(nd, bounds);
   // now loop on all hits again :( and check: if there are hits from another
   // cluster within d_c -> flag as border hit
   for (unsigned int i = 0; i < nd_size; ++i) {
     int ci = nd[i].data.clusterIndex;
     bool flag_isolated = true;
     if (ci != -1) {
       KDTreeBox search_box(nd[i].dims[0] - delta_c, nd[i].dims[0] + delta_c,
                            nd[i].dims[1] - delta_c, nd[i].dims[1] + delta_c);
       std::vector<KDNode> found;
       lp.search(search_box, found);
 
       const unsigned int found_size = found.size();
       for (unsigned int j = 0; j < found_size;
            j++) { // start from 0 here instead of 1
         // check if the hit is not within d_c of another cluster
         if (found[j].data.clusterIndex != -1) {
           float dist = distance(found[j].data, nd[i].data);
           if (dist < delta_c && found[j].data.clusterIndex != ci) {
             // in which case we assign it to the border
             nd[i].data.isBorder = true;
             break;
           }
           // because we are using two different containers, we have to make sure
           // that we don't unflag the
           // hit when it finds *itself* closer than delta_c
           if (dist < delta_c && dist != 0. &&
               found[j].data.clusterIndex == ci) {
             // in this case it is not an isolated hit
             // the dist!=0 is because the hit being looked at is also inside the
             // search box and at dist==0
             flag_isolated = false;
           }
         }
       }
       if (flag_isolated)
         nd[i].data.isBorder =
             true; // the hit is more than delta_c from any of its brethren
     }
     // check if this border hit has density larger than the current rho_b and
     // update
     if (nd[i].data.isBorder && rho_b[ci] < nd[i].data.rho)
       rho_b[ci] = nd[i].data.rho;
   } // end loop all hits
 
   // flag points in cluster with density < rho_b as halo points, then fill the
   // cluster vector
   for (unsigned int i = 0; i < nd_size; ++i) {
     int ci = nd[i].data.clusterIndex;
     if (ci != -1) {
       if (nd[i].data.rho <= rho_b[ci])
         nd[i].data.isHalo = true;
       clustersOnLayer[ci].push_back(nd[i]);
       if (verbosity < pINFO) {
         std::cout << "Pushing hit " << i << " into cluster with index " << ci
                   << std::endl;
       }
     }
   }
 
   // prepare the offset for the next layer if there is one
   if (verbosity < pINFO) {
     std::cout << "moving cluster offset by " << nClustersOnLayer << std::endl;
   }
   return nClustersOnLayer;
 }

std::vector< unsigned > HGCalImagingAlgo::findLocalMaximaInCluster ( const std::vector< KDNode > & cluster )

private

Definition at line 462 of file HGCalImagingAlgo.cc.

References data, distance(), MillePedeFileConverter_cfg::e, mps_fire::i, mps_fire::result, and SurveyInfoScenario_cff::seed.

Referenced by distance(), and getClusters().

                                                                            {
   std::vector<unsigned> result;
   std::vector<bool> seed(cluster.size(), true);
   float delta_c = 2.;
 
   for (unsigned i = 0; i < cluster.size(); ++i) {
     for (unsigned j = 0; j < cluster.size(); ++j) {
       if (i != j and distance(cluster[i].data, cluster[j].data) < delta_c) {
         if (cluster[i].data.weight < cluster[j].data.weight) {
           seed[i] = false;
           break;
         }
       }
     }
   }
 
   for (unsigned i = 0; i < cluster.size(); ++i) {
     if (seed[i] && cluster[i].data.weight > 5e-4) {
       // seed at i with energy cluster[i].weight
       result.push_back(i);
     }
   }
 
   // Found result.size() sub-clusters in input cluster of length cluster.size()
 
   return result;
 }

std::vector< reco::BasicCluster > HGCalImagingAlgo::getClusters ( bool doSharing )

Definition at line 112 of file HGCalImagingAlgo.cc.

References algoId, calculateEnergyWithFraction(), calculatePosition(), calculatePositionWithFraction(), clusters_v, gather_cfg::cout, data, reco::CaloID::DET_HGCAL_ENDCAP, MillePedeFileConverter_cfg::e, findLocalMaximaInCluster(), dedxEstimators_cff::fraction, mps_fire::i, layerClustersPerLayer, pINFO, position, shareEnergy(), and verbosity.

Referenced by setVerbosity().

                                                                         {
 
   reco::CaloID caloID = reco::CaloID::DET_HGCAL_ENDCAP;
   std::vector<std::pair<DetId, float>> thisCluster;
   for (auto &clsOnLayer : layerClustersPerLayer) {
     for (unsigned int i = 0; i < clsOnLayer.size(); ++i) {
       double energy = 0;
       Point position;
 
       if (doSharing) {
 
         std::vector<unsigned> seeds = findLocalMaximaInCluster(clsOnLayer[i]);
         // sharing found seeds.size() sub-cluster seeds in cluster i
 
         std::vector<std::vector<double>> fractions;
         // first pass can have noise it in
         shareEnergy(clsOnLayer[i], seeds, fractions);
 
         // reset and run second pass after vetoing seeds
         // that result in trivial clusters (less than 2 effective cells)
 
         for (unsigned isub = 0; isub < fractions.size(); ++isub) {
           double effective_hits = 0.0;
           double energy =
               calculateEnergyWithFraction(clsOnLayer[i], fractions[isub]);
           Point position =
               calculatePositionWithFraction(clsOnLayer[i], fractions[isub]);
 
           for (unsigned ihit = 0; ihit < fractions[isub].size(); ++ihit) {
             const double fraction = fractions[isub][ihit];
             if (fraction > 1e-7) {
               effective_hits += fraction;
               thisCluster.emplace_back(clsOnLayer[i][ihit].data.detid,
                                        fraction);
             }
           }
 
           if (verbosity < pINFO) {
             std::cout << "\t******** NEW CLUSTER (SHARING) ********"
                       << std::endl;
             std::cout << "\tEff. No. of cells = " << effective_hits
                       << std::endl;
             std::cout << "\t     Energy       = " << energy << std::endl;
             std::cout << "\t     Phi          = " << position.phi()
                       << std::endl;
             std::cout << "\t     Eta          = " << position.eta()
                       << std::endl;
             std::cout << "\t*****************************" << std::endl;
           }
           clusters_v.emplace_back(energy, position, caloID, thisCluster,
                                   algoId);
           thisCluster.clear();
         }
       } else {
         position = calculatePosition(clsOnLayer[i]); // energy-weighted position
         //   std::vector< KDNode >::iterator it;
         for (auto &it : clsOnLayer[i]) {
           energy += it.data.isHalo ? 0. : it.data.weight;
           // use fraction to store whether this is a Halo hit or not
           thisCluster.emplace_back(it.data.detid, (it.data.isHalo ? 0.f : 1.f));
         }
         if (verbosity < pINFO) {
           std::cout << "******** NEW CLUSTER (HGCIA) ********" << std::endl;
           std::cout << "Index          " << i << std::endl;
           std::cout << "No. of cells = " << clsOnLayer[i].size() << std::endl;
           std::cout << "     Energy     = " << energy << std::endl;
           std::cout << "     Phi        = " << position.phi() << std::endl;
           std::cout << "     Eta        = " << position.eta() << std::endl;
           std::cout << "*****************************" << std::endl;
         }
         clusters_v.emplace_back(energy, position, caloID, thisCluster, algoId);
         thisCluster.clear();
       }
     }
   }
   return clusters_v;
 }

void HGCalImagingAlgo::getEventSetup ( const edm::EventSetup & es )

inline

Definition at line 136 of file HGCalImagingAlgo.h.

References hgcal::RecHitTools::getEventSetup(), and rhtools_.

                                            {
         rhtools_.getEventSetup(es);
 }

void HGCalImagingAlgo::makeClusters ( )

Definition at line 86 of file HGCalImagingAlgo.cc.

References KDTreeLinkerAlgo< DATA, DIM >::build(), calculateDistanceToHigher(), calculateLocalDensity(), findAndAssignClusters(), mps_fire::i, layerClustersPerLayer, maxlayer, maxpos, minpos, and points.

Referenced by setVerbosity().

                                     {
   layerClustersPerLayer.resize(2 * maxlayer + 2);
   // assign all hits in each layer to a cluster core or halo
   tbb::this_task_arena::isolate([&] {
     tbb::parallel_for(size_t(0), size_t(2 * maxlayer + 2), [&](size_t i) {
       KDTreeBox bounds(minpos[i][0], maxpos[i][0], minpos[i][1], maxpos[i][1]);
       KDTree hit_kdtree;
       hit_kdtree.build(points[i], bounds);
 
       unsigned int actualLayer =
           i > maxlayer
               ? (i - (maxlayer + 1))
               : i; // maps back from index used for KD trees to actual layer
 
       double maxdensity = calculateLocalDensity(
           points[i], hit_kdtree, actualLayer); // also stores rho (energy
                                                // density) for each point (node)
       // calculate distance to nearest point with higher density storing
       // distance (delta) and point's index
       calculateDistanceToHigher(points[i]);
       findAndAssignClusters(points[i], hit_kdtree, maxdensity, bounds,
                             actualLayer, layerClustersPerLayer[i]);
     });
   });
 }

void HGCalImagingAlgo::populate ( const HGCRecHitCollection & hits )

Definition at line 15 of file HGCalImagingAlgo.cc.

References computeThreshold(), dependSensor, CaloRecHit::detid(), ecut, CaloRecHit::energy(), hgcal::RecHitTools::getLayerWithOffset(), hgcal::RecHitTools::getPosition(), hgcal::RecHitTools::getSiThickIndex(), hgcal::RecHitTools::getSiThickness(), mps_fire::i, createfilelist::int, hgcal::RecHitTools::isHalfCell(), lastLayerFH, SiStripPI::max, maxlayer, maxpos, min(), minpos, points, position, rhtools_, edm::SortedCollection< T, SORT >::size(), thresholds, v_sigmaNoise, and hgcal::RecHitTools::zside().

Referenced by setVerbosity().

                                                                {
   // loop over all hits and create the Hexel structure, skip energies below ecut
 
   if (dependSensor) {
     // for each layer and wafer calculate the thresholds (sigmaNoise and energy)
     // once
     computeThreshold();
   }
 
   std::vector<bool> firstHit(2 * (maxlayer + 1), true);
   for (unsigned int i = 0; i < hits.size(); ++i) {
 
     const HGCRecHit &hgrh = hits[i];
     DetId detid = hgrh.detid();
     unsigned int layer = rhtools_.getLayerWithOffset(detid);
     float thickness = 0.f;
     // set sigmaNoise default value 1 to use kappa value directly in case of
     // sensor-independent thresholds
     float sigmaNoise = 1.f;
     if (dependSensor) {
       if (layer <= lastLayerFH) // only EE and FH have silicon sensors
         thickness = rhtools_.getSiThickness(detid);
       double storedThreshold =
           thresholds[layer - 1]
                     [layer <= lastLayerFH ? rhtools_.getSiThickIndex(detid) : 0];
       sigmaNoise =
           v_sigmaNoise[layer - 1]
                       [layer <= lastLayerFH ? rhtools_.getSiThickIndex(detid) : 0];
 
       if (hgrh.energy() < storedThreshold)
         continue; // this sets the ZS threshold at ecut times the sigma noise
                   // for the sensor
     }
     if (!dependSensor && hgrh.energy() < ecut)
       continue;
 
     // map layers from positive endcap (z) to layer + maxlayer+1 to prevent
     // mixing up hits from different sides
     layer += int(rhtools_.zside(detid) > 0) * (maxlayer + 1);
 
     // determine whether this is a half-hexagon
     bool isHalf = rhtools_.isHalfCell(detid);
     const GlobalPoint position(rhtools_.getPosition(detid));
 
     // here's were the KDNode is passed its dims arguments - note that these are
     // *copied* from the Hexel
     points[layer].emplace_back(
         Hexel(hgrh, detid, isHalf, sigmaNoise, thickness, &rhtools_),
         position.x(), position.y());
 
     // for each layer, store the minimum and maximum x and y coordinates for the
     // KDTreeBox boundaries
     if (firstHit[layer]) {
       minpos[layer][0] = position.x();
       minpos[layer][1] = position.y();
       maxpos[layer][0] = position.x();
       maxpos[layer][1] = position.y();
       firstHit[layer] = false;
     } else {
       minpos[layer][0] = std::min((float)position.x(), minpos[layer][0]);
       minpos[layer][1] = std::min((float)position.y(), minpos[layer][1]);
       maxpos[layer][0] = std::max((float)position.x(), maxpos[layer][0]);
       maxpos[layer][1] = std::max((float)position.y(), maxpos[layer][1]);
     }
 
   } // end loop hits
 }

void HGCalImagingAlgo::reset ( void )

inline

Definition at line 140 of file HGCalImagingAlgo.h.

References clusters_v, computeThreshold(), mps_fire::i, layerClustersPerLayer, maxpos, minpos, points, and edm::swap().

             {
         clusters_v.clear();
         layerClustersPerLayer.clear();
         for( auto& it: points)
         {
                 it.clear();
                 std::vector<KDNode>().swap(it);
         }
         for(unsigned int i = 0; i < minpos.size(); i++)
         {
                 minpos[i][0]=0.; minpos[i][1]=0.;
                 maxpos[i][0]=0.; maxpos[i][1]=0.;
         }
 }

void HGCalImagingAlgo::setVerbosity ( VerbosityLevel the_verbosity )

inline

Definition at line 124 of file HGCalImagingAlgo.h.

References getClusters(), hfClusterShapes_cfi::hits, makeClusters(), populate(), and verbosity.

 {
         verbosity = the_verbosity;
 }

void HGCalImagingAlgo::shareEnergy	(	const std::vector< KDNode > &	,
		const std::vector< unsigned > &	,
		std::vector< std::vector< double > > &
	)

private

Definition at line 514 of file HGCalImagingAlgo.cc.

References calculateEnergyWithFraction(), calculatePositionWithFraction(), data, diffTreeTool::diff, MillePedeFileConverter_cfg::e, JetChargeProducer_cfi::exp, cropTnPTrees::frac, dedxEstimators_cff::fraction, mps_fire::i, SiStripPI::max, particleFlowClusterECALUncorrected_cfi::minFracTot, eostools::move(), perp2(), funct::pow(), sigma2, mathSSE::sqrt(), particleFlowClusterECALUncorrected_cfi::stoppingTolerance, HGCalImagingAlgo::Hexel::x, HGCalImagingAlgo::Hexel::y, and HGCalImagingAlgo::Hexel::z.

Referenced by distance(), and getClusters().

                                                {
   std::vector<bool> isaseed(incluster.size(), false);
   outclusters.clear();
   outclusters.resize(seeds.size());
   std::vector<Point> centroids(seeds.size());
   std::vector<double> energies(seeds.size());
 
   if (seeds.size() == 1) { // short circuit the case of a lone cluster
     outclusters[0].clear();
     outclusters[0].resize(incluster.size(), 1.0);
     return;
   }
 
   // saving seeds
 
   // create quick seed lookup
   for (unsigned i = 0; i < seeds.size(); ++i) {
     isaseed[seeds[i]] = true;
   }
 
   // initialize clusters to be shared
   // centroids start off at seed positions
   // seeds always have fraction 1.0, to stabilize fit
   // initializing fit
   for (unsigned i = 0; i < seeds.size(); ++i) {
     outclusters[i].resize(incluster.size(), 0.0);
     for (unsigned j = 0; j < incluster.size(); ++j) {
       if (j == seeds[i]) {
         outclusters[i][j] = 1.0;
         centroids[i] = math::XYZPoint(incluster[j].data.x, incluster[j].data.y,
                                       incluster[j].data.z);
         energies[i] = incluster[j].data.weight;
       }
     }
   }
 
   // run the fit while we are less than max iterations, and clusters are still
   // moving
   const double minFracTot = 1e-20;
   unsigned iter = 0;
   const unsigned iterMax = 50;
   double diff = std::numeric_limits<double>::max();
   const double stoppingTolerance = 1e-8;
   const auto numberOfSeeds = seeds.size();
   auto toleranceScaling =
       numberOfSeeds > 2 ? (numberOfSeeds - 1) * (numberOfSeeds - 1) : 1;
   std::vector<Point> prevCentroids;
   std::vector<double> frac(numberOfSeeds), dist2(numberOfSeeds);
   while (iter++ < iterMax && diff > stoppingTolerance * toleranceScaling) {
     for (unsigned i = 0; i < incluster.size(); ++i) {
       const Hexel &ihit = incluster[i].data;
       double fracTot(0.0);
       for (unsigned j = 0; j < numberOfSeeds; ++j) {
         double fraction = 0.0;
         double d2 = (std::pow(ihit.x - centroids[j].x(), 2.0) +
                      std::pow(ihit.y - centroids[j].y(), 2.0) +
                      std::pow(ihit.z - centroids[j].z(), 2.0)) /
                     sigma2;
         dist2[j] = d2;
         // now we set the fractions up based on hit type
         if (i == seeds[j]) { // this cluster's seed
           fraction = 1.0;
         } else if (isaseed[i]) {
           fraction = 0.0;
         } else {
           fraction = energies[j] * std::exp(-0.5 * d2);
         }
         fracTot += fraction;
         frac[j] = fraction;
       }
       // now that we have calculated all fractions for all hits
       // assign the new fractions
       for (unsigned j = 0; j < numberOfSeeds; ++j) {
         if (fracTot > minFracTot || (i == seeds[j] && fracTot > 0.0)) {
           outclusters[j][i] = frac[j] / fracTot;
         } else {
           outclusters[j][i] = 0.0;
         }
       }
     }
 
     // save previous centroids
     prevCentroids = std::move(centroids);
     // finally update the position of the centroids from the last iteration
     centroids.resize(numberOfSeeds);
     double diff2 = 0.0;
     for (unsigned i = 0; i < numberOfSeeds; ++i) {
       centroids[i] = calculatePositionWithFraction(incluster, outclusters[i]);
       energies[i] = calculateEnergyWithFraction(incluster, outclusters[i]);
       // calculate convergence parameters
       const double delta2 = (prevCentroids[i] - centroids[i]).perp2();
       diff2 = std::max(delta2, diff2);
     }
     // update convergance parameter outside loop
     diff = std::sqrt(diff2);
   }
 }

std::vector<size_t> HGCalImagingAlgo::sort_by_delta ( const std::vector< KDNode > & v ) const

inlineprivate

Definition at line 257 of file HGCalImagingAlgo.h.

References begin, end, training_settings::idx, and findQualityFiles::v.

Referenced by findAndAssignClusters().

                                                                   {
         std::vector<size_t> idx(v.size());
         std::iota (std::begin(idx), std::end(idx), 0);
         sort(idx.begin(), idx.end(),
              [&v](size_t i1, size_t i2) {
                         return v[i1].data.delta > v[i2].data.delta;
                 });
         return idx;
 }