#include <DAClusterizerInZ_vect.h>

Inheritance diagram for DAClusterizerInZ_vect:

Classes
struct	track_t

struct	vertex_t

Public Member Functions
double	beta0 (const double betamax, track_t const &tks, vertex_t const &y) const

void	clear_vtx_range (track_t &gtracks, vertex_t &gvertices) const

std::vector< std::vector< reco::TransientTrack > >	clusterize (const std::vector< reco::TransientTrack > &tracks) const override

	DAClusterizerInZ_vect (const edm::ParameterSet &conf)

void	dump (const double beta, const vertex_t &y, const track_t &tks, const int verbosity=0, const double rho0=0.) const

track_t	fill (const std::vector< reco::TransientTrack > &tracks) const

bool	merge (vertex_t &y, track_t &tks, double &beta) const

bool	purge (vertex_t &, track_t &, double &, const double) const

void	set_vtx_range (double beta, track_t &gtracks, vertex_t &gvertices) const

bool	split (const double beta, track_t &t, vertex_t &y, double threshold=1.) const

unsigned int	thermalize (double beta, track_t &gtracks, vertex_t &gvertices, const double delta_max, const double rho0=0.) const

double	update (double beta, track_t &gtracks, vertex_t &gvertices, const double rho0=0, const bool updateTc=false) const

void	verify (const vertex_t &v, const track_t &tks, unsigned int nv=999999, unsigned int nt=999999) const

std::vector< TransientVertex >	vertices (const std::vector< reco::TransientTrack > &tracks) const

Public Member Functions inherited from TrackClusterizerInZ
	TrackClusterizerInZ ()

	TrackClusterizerInZ (const edm::ParameterSet &conf)

virtual	~TrackClusterizerInZ ()

Static Public Member Functions
static void	fillPSetDescription (edm::ParameterSetDescription &desc)

Private Attributes
double	betamax_

double	betapurge_

double	betastop_

unsigned int	convergence_mode_

double	coolingFactor_

double	d0CutOff_

double	delta_highT_

double	delta_lowT_

double	dzCutOff_

unsigned int	maxIterations_

double	mintrkweight_

double	sel_zrange_

double	uniquetrkminp_

double	uniquetrkweight_

double	vertexSize_

double	zdumpcenter_

double	zdumpwidth_

double	zmerge_

const double	zrange_min_ = 0.1

Detailed Description

Description: separates event tracks into clusters along the beam line

Version which auto-vectorizes with gcc 4.6 or newer

Definition at line 21 of file DAClusterizerInZ_vect.h.

Constructor & Destructor Documentation

◆ DAClusterizerInZ_vect()

DAClusterizerInZ_vect::DAClusterizerInZ_vect ( const edm::ParameterSet & conf )

Definition at line 20 of file DAClusterizerInZ_vect.cc.

References gather_cfg::cout, edm::ParameterSet::getParameter(), edm::ParameterSet::getUntrackedParameter(), SiStripPI::max, HLT_2022v12_cff::Tmin, HLT_2022v12_cff::Tpurge, and HLT_2022v12_cff::Tstop.

                                                                         {
   // hardcoded parameters
   maxIterations_ = 1000;
   mintrkweight_ = 0.5;
 
   // configurable debug output
 #ifdef DEBUG
   zdumpcenter_ = conf.getUntrackedParameter<double>("zdumpcenter", 0.);
   zdumpwidth_ = conf.getUntrackedParameter<double>("zdumpwidth", 20.);
 #endif
 
   // configurable parameters
   double Tmin = conf.getParameter<double>("Tmin");
   double Tpurge = conf.getParameter<double>("Tpurge");
   double Tstop = conf.getParameter<double>("Tstop");
   vertexSize_ = conf.getParameter<double>("vertexSize");
   coolingFactor_ = conf.getParameter<double>("coolingFactor");
   d0CutOff_ = conf.getParameter<double>("d0CutOff");
   dzCutOff_ = conf.getParameter<double>("dzCutOff");
   uniquetrkweight_ = conf.getParameter<double>("uniquetrkweight");
   uniquetrkminp_ = conf.getParameter<double>("uniquetrkminp");
   zmerge_ = conf.getParameter<double>("zmerge");
   sel_zrange_ = conf.getParameter<double>("zrange");
   convergence_mode_ = conf.getParameter<int>("convergence_mode");
   delta_lowT_ = conf.getParameter<double>("delta_lowT");
   delta_highT_ = conf.getParameter<double>("delta_highT");
 
 #ifdef DEBUG
   std::cout << "DAClusterizerinZ_vect: mintrkweight = " << mintrkweight_ << std::endl;
   std::cout << "DAClusterizerinZ_vect: uniquetrkweight = " << uniquetrkweight_ << std::endl;
   std::cout << "DAClusterizerInZ_vect: uniquetrkminp = " << uniquetrkminp_ << std::endl;
   std::cout << "DAClusterizerinZ_vect: zmerge = " << zmerge_ << std::endl;
   std::cout << "DAClusterizerinZ_vect: Tmin = " << Tmin << std::endl;
   std::cout << "DAClusterizerinZ_vect: Tpurge = " << Tpurge << std::endl;
   std::cout << "DAClusterizerinZ_vect: Tstop = " << Tstop << std::endl;
   std::cout << "DAClusterizerinZ_vect: vertexSize = " << vertexSize_ << std::endl;
   std::cout << "DAClusterizerinZ_vect: coolingFactor = " << coolingFactor_ << std::endl;
   std::cout << "DAClusterizerinZ_vect: d0CutOff = " << d0CutOff_ << std::endl;
   std::cout << "DAClusterizerinZ_vect: dzCutOff = " << dzCutOff_ << std::endl;
   std::cout << "DAClusterizerInZ_vect: zrange = " << sel_zrange_ << std::endl;
   std::cout << "DAClusterizerinZ_vect: convergence mode = " << convergence_mode_ << std::endl;
   std::cout << "DAClusterizerinZ_vect: delta_highT = " << delta_highT_ << std::endl;
   std::cout << "DAClusterizerinZ_vect: delta_lowT = " << delta_lowT_ << std::endl;
   std::cout << "DAClusterizerinZ_vect: DEBUGLEVEL " << DEBUGLEVEL << std::endl;
 #endif
 
   if (convergence_mode_ > 1) {
     edm::LogWarning("DAClusterizerinZ_vect")
         << "DAClusterizerInZ_vect: invalid convergence_mode " << convergence_mode_ << "  reset to default " << 0;
     convergence_mode_ = 0;
   }
 
   if (Tmin == 0) {
     betamax_ = 1.0;
     edm::LogWarning("DAClusterizerinZ_vect")
         << "DAClusterizerInZ_vect: invalid Tmin " << Tmin << "  reset to default " << 1. / betamax_;
   } else {
     betamax_ = 1. / Tmin;
   }
 
   if ((Tpurge > Tmin) || (Tpurge == 0)) {
     edm::LogWarning("DAClusterizerinZ_vect")
         << "DAClusterizerInZ_vect: invalid Tpurge " << Tpurge << "  set to " << Tmin;
     Tpurge = Tmin;
   }
   betapurge_ = 1. / Tpurge;
 
   if ((Tstop > Tpurge) || (Tstop == 0)) {
     edm::LogWarning("DAClusterizerinZ_vect")
         << "DAClusterizerInZ_vect: invalid Tstop " << Tstop << "  set to  " << max(1., Tpurge);
     Tstop = max(1., Tpurge);
   }
   betastop_ = 1. / Tstop;
 }

Member Function Documentation

◆ beta0()

double DAClusterizerInZ_vect::beta0	(	const double	betamax,
		track_t const &	tks,
		vertex_t const &	y
	)		const

Definition at line 590 of file DAClusterizerInZ_vect.cc.

References a, b, gather_cfg::cout, PVValHelper::dx, DAClusterizerInZ_vect::track_t::dz2, DAClusterizerInZ_vect::track_t::getSize(), mps_fire::i, createfilelist::int, dqmdumpme::k, dqm-mbProfile::log, nt, funct::pow(), DAClusterizerInZ_vect::track_t::tkwt, w(), and DAClusterizerInZ_vect::track_t::zpca.

                                                                                                {
   double T0 = 0;  // max Tc for beta=0
   // estimate critical temperature from beta=0 (T=inf)
   const unsigned int nt = tks.getSize();
   const unsigned int nv = y.getSize();
 
   for (unsigned int k = 0; k < nv; k++) {
     // vertex fit at T=inf
     double sumwz = 0;
     double sumw = 0;
     for (unsigned int i = 0; i < nt; i++) {
       double w = tks.tkwt[i] * tks.dz2[i];
       sumwz += w * tks.zpca[i];
       sumw += w;
     }
 
     y.zvtx[k] = sumwz / sumw;
 
     // estimate Tcrit
     double a = 0, b = 0;
     for (unsigned int i = 0; i < nt; i++) {
       double dx = tks.zpca[i] - y.zvtx[k];
       double w = tks.tkwt[i] * tks.dz2[i];
       a += w * std::pow(dx, 2) * tks.dz2[i];
       b += w;
     }
     double Tc = 2. * a / b;  // the critical temperature of this vertex
 
     if (Tc > T0)
       T0 = Tc;
 
   }  // vertex loop (normally there should be only one vertex at beta=0)
 
 #ifdef DEBUG
   if (DEBUGLEVEL > 0) {
     std::cout << "DAClusterizerInZ_vect.beta0:   Tc = " << T0 << std::endl;
     int coolingsteps = 1 - int(std::log(T0 * betamax) / std::log(coolingFactor_));
     std::cout << "DAClusterizerInZ_vect.beta0:   nstep = " << coolingsteps << std::endl;
   }
 #endif
 
   if (T0 > 1. / betamax) {
     int coolingsteps = 1 - int(std::log(T0 * betamax) / std::log(coolingFactor_));
 
     return betamax * std::pow(coolingFactor_, coolingsteps);
   } else {
     // ensure at least one annealing step
     return betamax * coolingFactor_;
   }
 }

◆ clear_vtx_range()

void DAClusterizerInZ_vect::clear_vtx_range	(	track_t &	gtracks,
		vertex_t &	gvertices
	)		const

Definition at line 269 of file DAClusterizerInZ_vect.cc.

References DAClusterizerInZ_vect::track_t::getSize(), DAClusterizerInZ_vect::vertex_t::getSize(), DAClusterizerInZ_vect::track_t::kmax, DAClusterizerInZ_vect::track_t::kmin, nt, and mitigatedMETSequence_cff::U.

                                                                                        {
   const unsigned int nt = gtracks.getSize();
   const unsigned int nv = gvertices.getSize();
   for (auto itrack = 0U; itrack < nt; ++itrack) {
     gtracks.kmin[itrack] = 0;
     gtracks.kmax[itrack] = nv;
   }
 }

◆ clusterize()

vector< vector< reco::TransientTrack > > DAClusterizerInZ_vect::clusterize ( const std::vector< reco::TransientTrack > & tracks ) const

overridevirtual

Implements TrackClusterizerInZ.

Definition at line 974 of file DAClusterizerInZ_vect.cc.

References funct::abs(), bsc_activity_cfg::clusters, gather_cfg::cout, mps_fire::i, dqmdumpme::k, eostools::move(), position, MetAnalyzer::pv(), tracks, and AlignmentTracksFromVertexSelector_cfi::vertices.

                                                     {
   vector<vector<reco::TransientTrack> > clusters;
   vector<TransientVertex>&& pv = vertices(tracks);
 
 #ifdef DEBUG
   if (DEBUGLEVEL > 0) {
     std::cout << "###################################################" << endl;
     std::cout << "# vectorized DAClusterizerInZ_vect::clusterize   nt=" << tracks.size() << endl;
     std::cout << "# DAClusterizerInZ_vect::clusterize   pv.size=" << pv.size() << endl;
     std::cout << "###################################################" << endl;
   }
 #endif
 
   if (pv.empty()) {
     return clusters;
   }
 
   // fill into clusters and merge
   vector<reco::TransientTrack> aCluster = pv.begin()->originalTracks();
 
   for (auto k = pv.begin() + 1; k != pv.end(); k++) {
     if (std::abs(k->position().z() - (k - 1)->position().z()) > (2 * vertexSize_)) {
       // close a cluster
       if (aCluster.size() > 1) {
         clusters.push_back(aCluster);
       }
 #ifdef DEBUG
       else {
         std::cout << " one track cluster at " << k->position().z() << "  suppressed" << std::endl;
       }
 #endif
       aCluster.clear();
     }
     for (unsigned int i = 0; i < k->originalTracks().size(); i++) {
       aCluster.push_back(k->originalTracks()[i]);
     }
   }
   clusters.emplace_back(std::move(aCluster));
 
   return clusters;
 }

◆ dump()

void DAClusterizerInZ_vect::dump	(	const double	beta,
		const vertex_t &	y,
		const track_t &	tks,
		const int	verbosity = `0`,
		const double	rho0 = `0.`
	)		const

Definition at line 1017 of file DAClusterizerInZ_vect.cc.

                                                                                                             {
 #ifdef DEBUG
   const unsigned int nv = y.getSize();
   const unsigned int nt = tks.getSize();
   // make a local copies of clusters and tracks to update Tc  [ = y_local.swE / y_local.sw ]
   vertex_t y_local = y;
   track_t tks_local = tks;
   update(beta, tks_local, y_local, rho0, true);
 
   std::vector<unsigned int> iz;
   for (unsigned int j = 0; j < nt; j++) {
     iz.push_back(j);
   }
   std::sort(iz.begin(), iz.end(), [tks](unsigned int a, unsigned int b) { return tks.zpca[a] < tks.zpca[b]; });
   std::cout << std::endl;
   std::cout << "-----DAClusterizerInZ::dump ----" << nv << "  clusters " << std::endl;
   std::cout << "                                                                   ";
   for (unsigned int ivertex = 0; ivertex < nv; ++ivertex) {
     if (std::fabs(y.zvtx[ivertex] - zdumpcenter_) < zdumpwidth_) {
       std::cout << "   " << setw(3) << ivertex << "  ";
     }
   }
   std::cout << endl;
   std::cout << "                                                                z= ";
   std::cout << setprecision(4);
   for (unsigned int ivertex = 0; ivertex < nv; ++ivertex) {
     if (std::fabs(y.zvtx[ivertex] - zdumpcenter_) < zdumpwidth_) {
       std::cout << setw(8) << fixed << y.zvtx[ivertex];
     }
   }
   std::cout << endl
             << "T=" << setw(15) << 1. / beta << " Tmin =" << setw(10) << 1. / betamax_
             << "                             Tc= ";
   for (unsigned int ivertex = 0; ivertex < nv; ++ivertex) {
     if (std::fabs(y.zvtx[ivertex] - zdumpcenter_) < zdumpwidth_) {
       double Tc = 2 * y_local.swE[ivertex] / y_local.sw[ivertex];
       std::cout << setw(8) << fixed << setprecision(1) << Tc;
     }
   }
   std::cout << endl;
 
   std::cout << "                                                               pk= ";
   double sumpk = 0;
   for (unsigned int ivertex = 0; ivertex < nv; ++ivertex) {
     sumpk += y.rho[ivertex];
     if (std::fabs(y.zvtx[ivertex] - zdumpcenter_) > zdumpwidth_)
       continue;
     std::cout << setw(8) << setprecision(4) << fixed << y.rho[ivertex];
   }
   std::cout << endl;
 
   std::cout << "                                                               nt= ";
   for (unsigned int ivertex = 0; ivertex < nv; ++ivertex) {
     if (std::fabs(y.zvtx[ivertex] - zdumpcenter_) > zdumpwidth_)
       continue;
     std::cout << setw(8) << setprecision(1) << fixed << y.rho[ivertex] * nt;
   }
   std::cout << endl;
 
   if (verbosity > 0) {
     double E = 0, F = 0;
     std::cout << endl;
     std::cout << "----        z +/- dz                ip +/-dip       pt    phi  eta    weights  ----" << endl;
     std::cout << setprecision(4);
     for (unsigned int i0 = 0; i0 < nt; i0++) {
       unsigned int i = iz[i0];
       if (tks.sum_Z[i] > 0) {
         F -= std::log(tks.sum_Z[i]) / beta;
       }
       double tz = tks.zpca[i];
 
       if (std::fabs(tz - zdumpcenter_) > zdumpwidth_)
         continue;
       std::cout << setw(4) << i << ")" << setw(8) << fixed << setprecision(4) << tz << " +/-" << setw(6)
                 << sqrt(1. / tks.dz2[i]);
       if ((tks.tt[i] == nullptr)) {
         std::cout << "          effective track                             ";
       } else {
         if (tks.tt[i]->track().quality(reco::TrackBase::highPurity)) {
           std::cout << " *";
         } else {
           std::cout << "  ";
         }
         if (tks.tt[i]->track().hitPattern().hasValidHitInPixelLayer(PixelSubdetector::SubDetector::PixelBarrel, 1)) {
           std::cout << "+";
         } else {
           std::cout << "-";
         }
         std::cout << setw(1)
                   << tks.tt[i]
                          ->track()
                          .hitPattern()
                          .pixelBarrelLayersWithMeasurement();  // see DataFormats/TrackReco/interface/HitPattern.h
         std::cout << setw(1) << tks.tt[i]->track().hitPattern().pixelEndcapLayersWithMeasurement();
         std::cout << setw(1) << hex
                   << tks.tt[i]->track().hitPattern().trackerLayersWithMeasurement() -
                          tks.tt[i]->track().hitPattern().pixelLayersWithMeasurement()
                   << dec;
         std::cout << "=" << setw(1) << hex
                   << tks.tt[i]->track().hitPattern().numberOfLostHits(reco::HitPattern::MISSING_OUTER_HITS) << dec;
 
         Measurement1D IP = tks.tt[i]->stateAtBeamLine().transverseImpactParameter();
         std::cout << setw(8) << IP.value() << "+/-" << setw(6) << IP.error();
         std::cout << " " << setw(6) << setprecision(2) << tks.tt[i]->track().pt() * tks.tt[i]->track().charge();
         std::cout << " " << setw(5) << setprecision(2) << tks.tt[i]->track().phi() << " " << setw(5) << setprecision(2)
                   << tks.tt[i]->track().eta();
       }  // not a dummy track
 
       double sump = 0.;
       for (unsigned int ivertex = 0; ivertex < nv; ++ivertex) {
         if (std::fabs(y.zvtx[ivertex] - zdumpcenter_) > zdumpwidth_)
           continue;
 
         if ((tks.tkwt[i] > 0) && (tks.sum_Z[i] > 0)) {
           //double p=pik(beta,tks[i],*k);
           double p = y.rho[ivertex] * local_exp(-beta * Eik(tks.zpca[i], y.zvtx[ivertex], tks.dz2[i])) / tks.sum_Z[i];
           if (p > 0.0001) {
             std::cout << setw(8) << setprecision(3) << p;
           } else {
             std::cout << "    .   ";
           }
           E += p * Eik(tks.zpca[i], y.zvtx[ivertex], tks.dz2[i]);
           sump += p;
         } else {
           std::cout << "        ";
         }
       }
       std::cout << "  ( " << std::setw(3) << tks.kmin[i] << "," << std::setw(3) << tks.kmax[i] - 1 << " ) ";
       std::cout << endl;
     }
     std::cout << "                                                                   ";
     for (unsigned int ivertex = 0; ivertex < nv; ++ivertex) {
       if (std::fabs(y.zvtx[ivertex] - zdumpcenter_) < zdumpwidth_) {
         std::cout << "   " << setw(3) << ivertex << "  ";
       }
     }
     std::cout << endl;
     std::cout << "                                                                z= ";
     std::cout << setprecision(4);
     for (unsigned int ivertex = 0; ivertex < nv; ++ivertex) {
       if (std::fabs(y.zvtx[ivertex] - zdumpcenter_) < zdumpwidth_) {
         std::cout << setw(8) << fixed << y.zvtx[ivertex];
       }
     }
     std::cout << endl;
     std::cout << endl
               << "T=" << 1 / beta << " E=" << E << " n=" << y.getSize() << "  F= " << F << endl
               << "----------" << endl;
   }
 #endif
 }

◆ fill()

DAClusterizerInZ_vect::track_t DAClusterizerInZ_vect::fill ( const std::vector< reco::TransientTrack > & tracks ) const

Definition at line 170 of file DAClusterizerInZ_vect.cc.

References DAClusterizerInZ_vect::track_t::addItemSorted(), HLT_2022v12_cff::atIP, pwdgSkimBPark_cfi::beamSpot, gather_cfg::cout, geometryDiff::epsilon, DAClusterizerInZ_vect::track_t::extractRaw(), DAClusterizerInZ_vect::track_t::getSize(), edm::isNotFinite(), SiStripPI::min, DAClusterizerInZ_vect::track_t::osumtkwt, position, funct::pow(), AlignmentPI::t_z, and tracks.

                                                                                                        {
   // prepare track data for clustering
   track_t tks;
   double sumtkwt = 0.;
   for (auto it = tracks.begin(); it != tracks.end(); it++) {
     if (!(*it).isValid())
       continue;
     double t_tkwt = 1.;
     double t_z = ((*it).stateAtBeamLine().trackStateAtPCA()).position().z();
     if (std::fabs(t_z) > 1000.)
       continue;
     auto const& t_mom = (*it).stateAtBeamLine().trackStateAtPCA().momentum();
     //  get the beam-spot
     reco::BeamSpot beamspot = (it->stateAtBeamLine()).beamSpot();
     double t_dz2 = std::pow((*it).track().dzError(), 2)  // track errror
                    + (std::pow(beamspot.BeamWidthX() * t_mom.x(), 2) + std::pow(beamspot.BeamWidthY() * t_mom.y(), 2)) *
                          std::pow(t_mom.z(), 2) / std::pow(t_mom.perp2(), 2)  // beam spot width
                    + std::pow(vertexSize_, 2);  // intrinsic vertex size, safer for outliers and short lived decays
     t_dz2 = 1. / t_dz2;
     if (edm::isNotFinite(t_dz2) || t_dz2 < std::numeric_limits<double>::min())
       continue;
     if (d0CutOff_ > 0) {
       Measurement1D atIP = (*it).stateAtBeamLine().transverseImpactParameter();  // error contains beamspot
       t_tkwt = 1. / (1. + local_exp(std::pow(atIP.value() / atIP.error(), 2) -
                                     std::pow(d0CutOff_, 2)));  // reduce weight for high ip tracks
       if (edm::isNotFinite(t_tkwt) || t_tkwt < std::numeric_limits<double>::epsilon()) {
         edm::LogWarning("DAClusterizerinZ_vect") << "rejected track t_tkwt " << t_tkwt;
         continue;  // usually is > 0.99
       }
     }
     tks.addItemSorted(t_z, t_dz2, &(*it), t_tkwt);
     sumtkwt += t_tkwt;
   }
 
   tks.extractRaw();
   tks.osumtkwt = sumtkwt > 0 ? 1. / sumtkwt : 0.;
 
 #ifdef DEBUG
   if (DEBUGLEVEL > 0) {
     std::cout << "Track count (Z) " << tks.getSize() << std::endl;
   }
 #endif
 
   return tks;
 }

◆ fillPSetDescription()

void DAClusterizerInZ_vect::fillPSetDescription ( edm::ParameterSetDescription & desc )

static

Definition at line 1170 of file DAClusterizerInZ_vect.cc.

References submitPVResolutionJobs::desc.

Referenced by PrimaryVertexValidation::fillDescriptions(), and DAClusterizerInZT_vect::fillPSetDescription().

                                                                                 {
   desc.addUntracked<double>("zdumpcenter", 0.);
   desc.addUntracked<double>("zdumpwidth", 20.);
   desc.add<double>("d0CutOff", 3.0);
   desc.add<double>("Tmin", 2.0);
   desc.add<double>("delta_lowT", 0.001);
   desc.add<double>("zmerge", 0.01);
   desc.add<double>("dzCutOff", 3.0);
   desc.add<double>("Tpurge", 2.0);
   desc.add<int>("convergence_mode", 0);
   desc.add<double>("delta_highT", 0.01);
   desc.add<double>("Tstop", 0.5);
   desc.add<double>("coolingFactor", 0.6);
   desc.add<double>("vertexSize", 0.006);
   desc.add<double>("uniquetrkweight", 0.8);
   desc.add<double>("uniquetrkminp", 0.0);
   desc.add<double>("zrange", 4.0);
 }

◆ merge()

bool DAClusterizerInZ_vect::merge	(	vertex_t &	y,
		track_t &	tks,
		double &	beta
	)		const

Definition at line 462 of file DAClusterizerInZ_vect.cc.

References cms::cuda::assert(), HLT_2022v12_cff::beta, gather_cfg::cout, hlt_jetmet_dqm_QT_fromfile_cfg::critical, alignBH_cfg::fixed, and dqmdumpme::k.

                                                                                {
   // merge clusters that collapsed or never separated,
   // only merge if the estimated critical temperature of the merged vertex is below the current temperature
   // return true if vertices were merged, false otherwise
   const unsigned int nv = y.getSize();
 
   if (nv < 2)
     return false;
 
   // merge the smallest distance clusters first
   std::vector<std::pair<double, unsigned int> > critical;
   for (unsigned int k = 0; (k + 1) < nv; k++) {
     if (std::fabs(y.zvtx[k + 1] - y.zvtx[k]) < zmerge_) {
       critical.push_back(make_pair(std::fabs(y.zvtx[k + 1] - y.zvtx[k]), k));
     }
   }
   if (critical.empty())
     return false;
 
   std::stable_sort(critical.begin(), critical.end(), std::less<std::pair<double, unsigned int> >());
 
   for (unsigned int ik = 0; ik < critical.size(); ik++) {
     unsigned int k = critical[ik].second;
     double rho = y.rho[k] + y.rho[k + 1];
 
 #ifdef DEBUG
     assert((k + 1) < nv);
     if (DEBUGLEVEL > 1) {
       std::cout << "merging " << fixed << setprecision(4) << y.zvtx[k + 1] << " and " << y.zvtx[k]
                 << "  sw = " << y.sw[k] + y.sw[k + 1] << std::endl;
     }
 #endif
 
     if (rho > 0) {
       y.zvtx[k] = (y.rho[k] * y.zvtx[k] + y.rho[k + 1] * y.zvtx[k + 1]) / rho;
     } else {
       y.zvtx[k] = 0.5 * (y.zvtx[k] + y.zvtx[k + 1]);
     }
     y.rho[k] = rho;
     y.sw[k] += y.sw[k + 1];
     y.removeItem(k + 1, tks);
     set_vtx_range(beta, tks, y);
     y.extractRaw();
     return true;
   }
 
   return false;
 }

◆ purge()

bool DAClusterizerInZ_vect::purge	(	vertex_t &	y,
		track_t &	tks,
		double &	rho0,
		const double	beta
	)		const

Definition at line 511 of file DAClusterizerInZ_vect.cc.

References cms::cuda::assert(), HLT_2022v12_cff::beta, gather_cfg::cout, DAClusterizerInZ_vect::track_t::dz2, HLT_2022v12_cff::eps, cms::cuda::for(), DAClusterizerInZ_vect::track_t::getSize(), mps_fire::i, dqmdumpme::k, reco::ParticleMasses::k0, DAClusterizerInZ_vect::track_t::kmax, DAClusterizerInZ_vect::track_t::kmin, nt, AlCaHLTBitMon_ParallelJobs::p, DAClusterizerInZ_vect::track_t::sum_Z, DAClusterizerInZ_vect::track_t::tkwt, and DAClusterizerInZ_vect::track_t::zpca.

                                                                                                   {
   constexpr double eps = 1.e-100;
   // eliminate clusters with only one significant/unique track
   const unsigned int nv = y.getSize();
   const unsigned int nt = tks.getSize();
 
   if (nv < 2)
     return false;
 
   std::vector<double> sump_v(nv), arg_cache_v(nv), exp_cache_v(nv), pcut_cache_v(nv);
   std::vector<int> nUnique_v(nv);
   double* __restrict__ parg_cache;
   double* __restrict__ pexp_cache;
   double* __restrict__ ppcut_cache;
   double* __restrict__ psump;
   int* __restrict__ pnUnique;
   int constexpr nunique_min_ = 2;
 
   set_vtx_range(beta, tks, y);
 
   parg_cache = arg_cache_v.data();
   pexp_cache = exp_cache_v.data();
   ppcut_cache = pcut_cache_v.data();
   psump = sump_v.data();
   pnUnique = nUnique_v.data();
 
   const auto rhoconst = rho0 * local_exp(-beta * dzCutOff_ * dzCutOff_);
   for (unsigned int k = 0; k < nv; k++) {
     const double pmax = y.rho[k] / (y.rho[k] + rhoconst);
     ppcut_cache[k] = uniquetrkweight_ * pmax;
   }
 
   for (unsigned int i = 0; i < nt; i++) {
     const auto invZ = ((tks.sum_Z[i] > eps) && (tks.tkwt[i] > uniquetrkminp_)) ? 1. / tks.sum_Z[i] : 0.;
     const auto track_z = tks.zpca[i];
     const auto botrack_dz2 = -beta * tks.dz2[i];
     const auto kmin = tks.kmin[i];
     const auto kmax = tks.kmax[i];
 
     for (unsigned int k = kmin; k < kmax; k++) {
       const auto mult_resz = track_z - y.zvtx[k];
       parg_cache[k] = botrack_dz2 * (mult_resz * mult_resz);
     }
 
     local_exp_list_range(parg_cache, pexp_cache, kmin, kmax);
 
     for (unsigned int k = kmin; k < kmax; k++) {
       const double p = y.rho[k] * pexp_cache[k] * invZ;
       psump[k] += p;
       pnUnique[k] += (p > ppcut_cache[k]) ? 1 : 0;
     }
   }
 
   double sumpmin = nt;
   unsigned int k0 = nv;
   for (unsigned k = 0; k < nv; k++) {
     if ((pnUnique[k] < nunique_min_) && (psump[k] < sumpmin)) {
       sumpmin = psump[k];
       k0 = k;
     }
   }
 
   if (k0 != nv) {
 #ifdef DEBUG
     assert(k0 < y.getSize());
     if (DEBUGLEVEL > 1) {
       std::cout << "eliminating prototype at " << std::setw(10) << std::setprecision(4) << y.zvtx[k0]
                 << " with sump=" << sumpmin << "  rho*nt =" << y.rho[k0] * nt << endl;
     }
 #endif
 
     y.removeItem(k0, tks);
     set_vtx_range(beta, tks, y);
     return true;
   } else {
     return false;
   }
 }

◆ set_vtx_range()

void DAClusterizerInZ_vect::set_vtx_range	(	double	beta,
		track_t &	gtracks,
		vertex_t &	gvertices
	)		const

Definition at line 220 of file DAClusterizerInZ_vect.cc.

References HLT_2022v12_cff::beta, DAClusterizerInZ_vect::track_t::dz2, DAClusterizerInZ_vect::track_t::getSize(), DAClusterizerInZ_vect::vertex_t::getSize(), DAClusterizerInZ_vect::track_t::kmax, DAClusterizerInZ_vect::track_t::kmin, SiStripPI::max, SiStripPI::min, nt, mathSSE::sqrt(), mitigatedMETSequence_cff::U, SiStripMonitorCluster_cfi::zmax, SiStripMonitorCluster_cfi::zmin, DAClusterizerInZ_vect::track_t::zpca, TkClusParameters_cff::zrange, and DAClusterizerInZ_vect::vertex_t::zvtx.

                                                                                                   {
   const unsigned int nv = gvertices.getSize();
   const unsigned int nt = gtracks.getSize();
 
   if (nv == 0) {
     edm::LogWarning("DAClusterizerinZ_vect") << "empty cluster list in set_vtx_range";
     return;
   }
 
   for (auto itrack = 0U; itrack < nt; ++itrack) {
     double zrange = max(sel_zrange_ / sqrt(beta * gtracks.dz2[itrack]), zrange_min_);
 
     double zmin = gtracks.zpca[itrack] - zrange;
     unsigned int kmin = min(nv - 1, gtracks.kmin[itrack]);
     // find the smallest vertex_z that is larger than zmin
     if (gvertices.zvtx[kmin] > zmin) {
       while ((kmin > 0) && (gvertices.zvtx[kmin - 1] > zmin)) {
         kmin--;
       }
     } else {
       while ((kmin < (nv - 1)) && (gvertices.zvtx[kmin] < zmin)) {
         kmin++;
       }
     }
 
     double zmax = gtracks.zpca[itrack] + zrange;
     unsigned int kmax = min(nv - 1, gtracks.kmax[itrack] - 1);
     // note: kmax points to the last vertex in the range, while gtracks.kmax points to the entry BEHIND the last vertex
     // find the largest vertex_z that is smaller than zmax
     if (gvertices.zvtx[kmax] < zmax) {
       while ((kmax < (nv - 1)) && (gvertices.zvtx[kmax + 1] < zmax)) {
         kmax++;
       }
     } else {
       while ((kmax > 0) && (gvertices.zvtx[kmax] > zmax)) {
         kmax--;
       }
     }
 
     if (kmin <= kmax) {
       gtracks.kmin[itrack] = kmin;
       gtracks.kmax[itrack] = kmax + 1;
     } else {
       gtracks.kmin[itrack] = max(0U, min(kmin, kmax));
       gtracks.kmax[itrack] = min(nv, max(kmin, kmax) + 1);
     }
   }
 }

◆ split()

bool DAClusterizerInZ_vect::split	(	const double	beta,
		track_t &	t,
		vertex_t &	y,
		double	threshold = `1.`
	)		const

Definition at line 641 of file DAClusterizerInZ_vect.cc.

References cms::cuda::assert(), HLT_2022v12_cff::beta, gather_cfg::cout, hlt_jetmet_dqm_QT_fromfile_cfg::critical, DAClusterizerInZ_vect::track_t::dz2, MillePedeFileConverter_cfg::e, geometryDiff::epsilon, alignBH_cfg::fixed, DAClusterizerInZ_vect::track_t::getSize(), mps_fire::i, dqmdumpme::k, nt, AlCaHLTBitMon_ParallelJobs::p, LaserDQM_cfg::p1, SiStripOfflineCRack_cfg::p2, edm::second(), submitPVValidationJobs::split(), mathSSE::sqrt(), DAClusterizerInZ_vect::track_t::sum_Z, submitPVValidationJobs::t, remoteMonitoring_LASER_era2018_cfg::threshold, DAClusterizerInZ_vect::track_t::tkwt, update, w(), w2, testProducerWithPsetDescEmpty_cfi::z2, and DAClusterizerInZ_vect::track_t::zpca.

                                                                                                       {
   // split only critical vertices (Tc >~ T=1/beta   <==>   beta*Tc>~1)
   // returns true if at least one cluster was split
 
   // update the sums needed for Tc, rho0 is never non-zero while splitting is still active
   update(beta, tks, y, 0., true);  // the "true" flag enables Tc evaluation
 
   constexpr double epsilon = 1e-3;  // minimum split size
   unsigned int nv = y.getSize();
 
   // avoid left-right biases by splitting highest Tc first
 
   std::vector<std::pair<double, unsigned int> > critical;
   for (unsigned int k = 0; k < nv; k++) {
     double Tc = 2 * y.swE[k] / y.sw[k];
     if (beta * Tc > threshold) {
       critical.push_back(make_pair(Tc, k));
     }
   }
   if (critical.empty())
     return false;
 
   std::stable_sort(critical.begin(), critical.end(), std::greater<std::pair<double, unsigned int> >());
 
   bool split = false;
   const unsigned int nt = tks.getSize();
 
   for (unsigned int ic = 0; ic < critical.size(); ic++) {
     unsigned int k = critical[ic].second;
 
     // estimate subcluster positions and weight
     double p1 = 0, z1 = 0, w1 = 0;
     double p2 = 0, z2 = 0, w2 = 0;
     for (unsigned int i = 0; i < nt; i++) {
       if (tks.sum_Z[i] > 1.e-100) {
         // winner-takes-all, usually overestimates splitting
         double tl = tks.zpca[i] < y.zvtx[k] ? 1. : 0.;
         double tr = 1. - tl;
 
         // soften it, especially at low T
         double arg = (tks.zpca[i] - y.zvtx[k]) * sqrt(beta * tks.dz2[i]);
         if (std::fabs(arg) < 20) {
           double t = local_exp(-arg);
           tl = t / (t + 1.);
           tr = 1 / (t + 1.);
         }
 
         double p = y.rho[k] * tks.tkwt[i] * local_exp(-beta * Eik(tks.zpca[i], y.zvtx[k], tks.dz2[i])) / tks.sum_Z[i];
         double w = p * tks.dz2[i];
         p1 += p * tl;
         z1 += w * tl * tks.zpca[i];
         w1 += w * tl;
         p2 += p * tr;
         z2 += w * tr * tks.zpca[i];
         w2 += w * tr;
       }
     }
 
     if (w1 > 0) {
       z1 = z1 / w1;
     } else {
       z1 = y.zvtx[k] - epsilon;
     }
     if (w2 > 0) {
       z2 = z2 / w2;
     } else {
       z2 = y.zvtx[k] + epsilon;
     }
 
     // reduce split size if there is not enough room
     if ((k > 0) && (z1 < (0.6 * y.zvtx[k] + 0.4 * y.zvtx[k - 1]))) {
       z1 = 0.6 * y.zvtx[k] + 0.4 * y.zvtx[k - 1];
     }
     if ((k + 1 < nv) && (z2 > (0.6 * y.zvtx[k] + 0.4 * y.zvtx[k + 1]))) {
       z2 = 0.6 * y.zvtx[k] + 0.4 * y.zvtx[k + 1];
     }
 
 #ifdef DEBUG
     assert(k < nv);
     if (DEBUGLEVEL > 1) {
       if (std::fabs(y.zvtx[k] - zdumpcenter_) < zdumpwidth_) {
         std::cout << " T= " << std::setw(8) << 1. / beta << " Tc= " << critical[ic].first << "    splitting "
                   << std::fixed << std::setprecision(4) << y.zvtx[k] << " --> " << z1 << "," << z2 << "     [" << p1
                   << "," << p2 << "]";
         if (std::fabs(z2 - z1) > epsilon) {
           std::cout << std::endl;
         } else {
           std::cout << "  rejected " << std::endl;
         }
       }
     }
 #endif
 
     // split if the new subclusters are significantly separated
     if ((z2 - z1) > epsilon) {
       split = true;
       double pk1 = p1 * y.rho[k] / (p1 + p2);
       double pk2 = p2 * y.rho[k] / (p1 + p2);
       y.zvtx[k] = z2;
       y.rho[k] = pk2;
       y.insertItem(k, z1, pk1, tks);
       if (k == 0)
         y.extractRaw();
 
       nv++;
 
       // adjust remaining pointers
       for (unsigned int jc = ic; jc < critical.size(); jc++) {
         if (critical[jc].second >= k) {
           critical[jc].second++;
         }
       }
     } else {
 #ifdef DEBUG
       std::cout << "warning ! split rejected, too small." << endl;
 #endif
     }
   }
 
   return split;
 }

◆ thermalize()

unsigned int DAClusterizerInZ_vect::thermalize	(	double	beta,
		track_t &	gtracks,
		vertex_t &	gvertices,
		const double	delta_max,
		const double	rho0 = `0.`
	)		const

Definition at line 414 of file DAClusterizerInZ_vect.cc.

References funct::abs(), HLT_2022v12_cff::beta, gather_cfg::cout, dumpMFGeometry_cfg::delta, submitPVValidationJobs::dump, dqmdumpme::k, SiStripPI::max, mathSSE::sqrt(), update, findQualityFiles::v, and HLTMuonOfflineAnalyzer_cfi::z0.

                                                                                               {
   unsigned int niter = 0;
   double delta = 0;
   double delta_max = delta_lowT_;
 
   if (convergence_mode_ == 0) {
     delta_max = delta_max0;
   } else if (convergence_mode_ == 1) {
     delta_max = delta_lowT_ / sqrt(std::max(beta, 1.0));
   }
 
   set_vtx_range(beta, tks, v);
   double delta_sum_range = 0;  // accumulate max(|delta-z|) as a lower bound
   std::vector<double> z0 = v.zvtx_vec;
 
   while (niter++ < maxIterations_) {
     delta = update(beta, tks, v, rho0);
     delta_sum_range += delta;
 
     if (delta_sum_range > zrange_min_) {
       for (unsigned int k = 0; k < v.getSize(); k++) {
         if (std::abs(v.zvtx_vec[k] - z0[k]) > zrange_min_) {
           set_vtx_range(beta, tks, v);
           delta_sum_range = 0;
           z0 = v.zvtx_vec;
           break;
         }
       }
     }
 
     if (delta < delta_max) {
       break;
     }
   }
 
 #ifdef DEBUG
   if (DEBUGLEVEL > 0) {
     std::cout << "DAClusterizerInZ_vect.thermalize niter = " << niter << " at T = " << 1 / beta
               << "  nv = " << v.getSize() << std::endl;
     if (DEBUGLEVEL > 2)
       dump(beta, v, tks, 0, rho0);
   }
 #endif
 
   return niter;
 }

◆ update()

double DAClusterizerInZ_vect::update	(	double	beta,
		track_t &	gtracks,
		vertex_t &	gvertices,
		const double	rho0 = `0`,
		const bool	updateTc = `false`
	)		const

Definition at line 278 of file DAClusterizerInZ_vect.cc.

Referenced by progressbar.ProgressBar::__next__(), MatrixUtil.Matrix::__setitem__(), MatrixUtil.Steps::__setitem__(), progressbar.ProgressBar::finish(), and MatrixUtil.Steps::overwrite().

                                                                                                       {
   // update weights and vertex positions
   // returns the maximum of changes of vertex positions
   // sums needed for Tc are only updated if updateTC == true
 
   const unsigned int nt = gtracks.getSize();
   const unsigned int nv = gvertices.getSize();
   auto osumtkwt = gtracks.osumtkwt;
 
   double Z_init = 0;
   if (rho0 > 0) {
     Z_init = rho0 * local_exp(-beta * dzCutOff_ * dzCutOff_);  // cut-off
   }
 
   // define kernels
   auto kernel_calc_exp_arg_range = [beta](const unsigned int itrack,
                                           track_t const& tracks,
                                           vertex_t const& vertices,
                                           const unsigned int kmin,
                                           const unsigned int kmax) {
     const double track_z = tracks.zpca[itrack];
     const double botrack_dz2 = -beta * tracks.dz2[itrack];
 
     // auto-vectorized
     for (unsigned int ivertex = kmin; ivertex < kmax; ++ivertex) {
       auto mult_res = track_z - vertices.zvtx[ivertex];
       vertices.exp_arg[ivertex] = botrack_dz2 * (mult_res * mult_res);
     }
   };
 
   auto kernel_add_Z_range = [Z_init](
                                 vertex_t const& vertices, const unsigned int kmin, const unsigned int kmax) -> double {
     double ZTemp = Z_init;
     for (unsigned int ivertex = kmin; ivertex < kmax; ++ivertex) {
       ZTemp += vertices.rho[ivertex] * vertices.exp[ivertex];
     }
     return ZTemp;
   };
 
   auto kernel_calc_normalization_range = [updateTc](const unsigned int track_num,
                                                     track_t& tracks,
                                                     vertex_t& vertices,
                                                     const unsigned int kmin,
                                                     const unsigned int kmax) {
     auto o_trk_sum_Z = tracks.tkwt[track_num] / tracks.sum_Z[track_num];
     auto o_trk_dz2 = tracks.dz2[track_num];
     auto tmp_trk_z = tracks.zpca[track_num];
 
     // auto-vectorized
     if (updateTc) {
 #pragma GCC ivdep
       for (unsigned int k = kmin; k < kmax; ++k) {
         vertices.se[k] += vertices.exp[k] * o_trk_sum_Z;
         auto w = vertices.rho[k] * vertices.exp[k] * (o_trk_sum_Z * o_trk_dz2);
         vertices.sw[k] += w;
         vertices.swz[k] += w * tmp_trk_z;
         vertices.swE[k] += w * vertices.exp_arg[k];
       }
     } else {
       // same loop but without updating sWE
 #pragma GCC ivdep
       for (unsigned int k = kmin; k < kmax; ++k) {
         vertices.se[k] += vertices.exp[k] * o_trk_sum_Z;
         auto w = vertices.rho[k] * vertices.exp[k] * (o_trk_sum_Z * o_trk_dz2);
         vertices.sw[k] += w;
         vertices.swz[k] += w * tmp_trk_z;
       }
     }
   };
 
   if (updateTc) {
     for (auto ivertex = 0U; ivertex < nv; ++ivertex) {
       gvertices.se[ivertex] = 0.0;
       gvertices.sw[ivertex] = 0.0;
       gvertices.swz[ivertex] = 0.0;
       gvertices.swE[ivertex] = 0.0;
     }
   } else {
     for (auto ivertex = 0U; ivertex < nv; ++ivertex) {
       gvertices.se[ivertex] = 0.0;
       gvertices.sw[ivertex] = 0.0;
       gvertices.swz[ivertex] = 0.0;
     }
   }
 
   // loop over tracks
   for (auto itrack = 0U; itrack < nt; ++itrack) {
     const unsigned int kmin = gtracks.kmin[itrack];
     const unsigned int kmax = gtracks.kmax[itrack];
 
     kernel_calc_exp_arg_range(itrack, gtracks, gvertices, kmin, kmax);
     local_exp_list_range(gvertices.exp_arg, gvertices.exp, kmin, kmax);
     gtracks.sum_Z[itrack] = kernel_add_Z_range(gvertices, kmin, kmax);
 
     if (edm::isNotFinite(gtracks.sum_Z[itrack]))
       gtracks.sum_Z[itrack] = 0.0;
 
     if (gtracks.sum_Z[itrack] > 1.e-100) {
       kernel_calc_normalization_range(itrack, gtracks, gvertices, kmin, kmax);
     }
   }
 
   // (un-)apply the factor -beta which is needed in exp_arg, but not in swE
   if (updateTc) {
     auto obeta = -1. / beta;
     for (auto ivertex = 0U; ivertex < nv; ++ivertex) {
       gvertices.swE[ivertex] *= obeta;
     }
   }
 
   // now update z and rho
   auto kernel_calc_z = [osumtkwt, nv](vertex_t& vertices) -> double {
     double delta = 0;
     // does not vectorize
     for (unsigned int ivertex = 0; ivertex < nv; ++ivertex) {
       if (vertices.sw[ivertex] > 0) {
         auto znew = vertices.swz[ivertex] / vertices.sw[ivertex];
         // prevents from vectorizing if
         delta = max(std::abs(vertices.zvtx[ivertex] - znew), delta);
         vertices.zvtx[ivertex] = znew;
       }
     }
 
     for (unsigned int ivertex = 0; ivertex < nv; ++ivertex)
       vertices.rho[ivertex] = vertices.rho[ivertex] * vertices.se[ivertex] * osumtkwt;
 
     return delta;
   };
 
   double delta = kernel_calc_z(gvertices);
 
   // return how much the prototypes moved
   return delta;
 }

◆ verify()

void DAClusterizerInZ_vect::verify	(	const vertex_t &	v,
		const track_t &	tks,
		unsigned int	nv = `999999`,
		unsigned int	nt = `999999`
	)		const

Definition at line 108 of file DAClusterizerInZ_vect.cc.

                                                                                                                 {
   if (!(nv == 999999)) {
     assert(nv == v.getSize());
   } else {
     nv = v.getSize();
   }
 
   if (!(nt == 999999)) {
     assert(nt == tks.getSize());
   } else {
     nt = tks.getSize();
   }
 
   assert(v.zvtx_vec.size() == nv);
   assert(v.rho_vec.size() == nv);
   assert(v.swz_vec.size() == nv);
   assert(v.exp_arg_vec.size() == nv);
   assert(v.exp_vec.size() == nv);
   assert(v.se_vec.size() == nv);
   assert(v.swz_vec.size() == nv);
   assert(v.swE_vec.size() == nv);
 
   assert(v.zvtx == &v.zvtx_vec.front());
   assert(v.rho == &v.rho_vec.front());
   assert(v.exp_arg == &v.exp_arg_vec.front());
   assert(v.sw == &v.sw_vec.front());
   assert(v.swz == &v.swz_vec.front());
   assert(v.se == &v.se_vec.front());
   assert(v.swE == &v.swE_vec.front());
 
   for (unsigned int k = 0; k < nv - 1; k++) {
     if (v.zvtx_vec[k] <= v.zvtx_vec[k + 1])
       continue;
     cout << " Z, cluster z-ordering assertion failure   z[" << k << "] =" << v.zvtx_vec[k] << "    z[" << k + 1
          << "] =" << v.zvtx_vec[k + 1] << endl;
   }
 
   assert(nt == tks.zpca_vec.size());
   assert(nt == tks.dz2_vec.size());
   assert(nt == tks.tt.size());
   assert(nt == tks.tkwt_vec.size());
   assert(nt == tks.sum_Z_vec.size());
   assert(nt == tks.kmin.size());
   assert(nt == tks.kmax.size());
 
   assert(tks.zpca == &tks.zpca_vec.front());
   assert(tks.dz2 == &tks.dz2_vec.front());
   assert(tks.tkwt == &tks.tkwt_vec.front());
   assert(tks.sum_Z == &tks.sum_Z_vec.front());
 
   for (unsigned int i = 0; i < nt; i++) {
     if ((tks.kmin[i] < tks.kmax[i]) && (tks.kmax[i] <= nv))
       continue;
     cout << "track vertex range assertion failure" << i << "/" << nt << "   kmin,kmax=" << tks.kmin[i] << ", "
          << tks.kmax[i] << "  nv=" << nv << endl;
   }
 
   for (unsigned int i = 0; i < nt; i++) {
     assert((tks.kmin[i] < tks.kmax[i]) && (tks.kmax[i] <= nv));
   }
 }

◆ vertices()

vector< TransientVertex > DAClusterizerInZ_vect::vertices ( const std::vector< reco::TransientTrack > & tracks ) const

Definition at line 763 of file DAClusterizerInZ_vect.cc.

                                                                                                       {
   track_t&& tks = fill(tracks);
   tks.extractRaw();
 
   unsigned int nt = tks.getSize();
   double rho0 = 0.0;  // start with no outlier rejection
 
   vector<TransientVertex> clusters;
   if (tks.getSize() == 0)
     return clusters;
 
   vertex_t y;  // the vertex prototypes
 
   // initialize:single vertex at infinite temperature
   y.addItem(0, 1.0);
   clear_vtx_range(tks, y);
 
   // estimate first critical temperature
   double beta = beta0(betamax_, tks, y);
 #ifdef DEBUG
   if (DEBUGLEVEL > 0)
     std::cout << "Beta0 is " << beta << std::endl;
 #endif
 
   thermalize(beta, tks, y, delta_highT_);
 
   // annealing loop, stop when T<Tmin  (i.e. beta>1/Tmin)
 
   double betafreeze = betamax_ * sqrt(coolingFactor_);
 
   while (beta < betafreeze) {
     while (merge(y, tks, beta)) {
       update(beta, tks, y, rho0, false);
     }
     split(beta, tks, y);
 
     beta = beta / coolingFactor_;
     thermalize(beta, tks, y, delta_highT_);
   }
 
 #ifdef DEBUG
   verify(y, tks);
 
   if (DEBUGLEVEL > 0) {
     std::cout << "DAClusterizerInZ_vect::vertices :"
               << "last round of splitting" << std::endl;
   }
 #endif
 
   set_vtx_range(beta, tks, y);
   update(beta, tks, y, rho0, false);
 
   while (merge(y, tks, beta)) {
     set_vtx_range(beta, tks, y);
     update(beta, tks, y, rho0, false);
   }
 
   unsigned int ntry = 0;
   double threshold = 1.0;
   while (split(beta, tks, y, threshold) && (ntry++ < 10)) {
     thermalize(beta, tks, y, delta_highT_, rho0);  // rho0 = 0. here
     while (merge(y, tks, beta)) {
       update(beta, tks, y, rho0, false);
     }
 
     // relax splitting a bit to reduce multiple split-merge cycles of the same cluster
     threshold *= 1.1;
   }
 
 #ifdef DEBUG
   verify(y, tks);
   if (DEBUGLEVEL > 0) {
     std::cout << "DAClusterizerInZ_vect::vertices :"
               << "turning on outlier rejection at T=" << 1 / beta << std::endl;
   }
 #endif
 
   // switch on outlier rejection at T=Tmin
   if (dzCutOff_ > 0) {
     rho0 = y.getSize() > 1 ? 1. / y.getSize() : 1.;
     for (unsigned int a = 0; a < 5; a++) {
       update(beta, tks, y, a * rho0 / 5.);  // adiabatic turn-on
     }
   }
 
   thermalize(beta, tks, y, delta_lowT_, rho0);
 
 #ifdef DEBUG
   verify(y, tks);
   if (DEBUGLEVEL > 0) {
     std::cout << "DAClusterizerInZ_vect::vertices :"
               << "merging with outlier rejection at T=" << 1 / beta << std::endl;
   }
   if (DEBUGLEVEL > 2)
     dump(beta, y, tks, 2, rho0);
 #endif
 
   // merge again  (some cluster split by outliers collapse here)
   while (merge(y, tks, beta)) {
     set_vtx_range(beta, tks, y);
     update(beta, tks, y, rho0, false);
   }
 
 #ifdef DEBUG
   verify(y, tks);
   if (DEBUGLEVEL > 0) {
     std::cout << "DAClusterizerInZ_vect::vertices :"
               << "after merging with outlier rejection at T=" << 1 / beta << std::endl;
   }
   if (DEBUGLEVEL > 2)
     dump(beta, y, tks, 2, rho0);
 #endif
 
   // go down to the purging temperature (if it is lower than tmin)
   while (beta < betapurge_) {
     beta = min(beta / coolingFactor_, betapurge_);
     thermalize(beta, tks, y, delta_lowT_, rho0);
   }
 
 #ifdef DEBUG
   verify(y, tks);
   if (DEBUGLEVEL > 0) {
     std::cout << "DAClusterizerInZ_vect::vertices :"
               << "purging at T=" << 1 / beta << std::endl;
   }
 #endif
 
   // eliminate insignificant vertices, this is more restrictive at higher T
   while (purge(y, tks, rho0, beta)) {
     thermalize(beta, tks, y, delta_lowT_, rho0);
   }
 
 #ifdef DEBUG
   verify(y, tks);
   if (DEBUGLEVEL > 0) {
     std::cout << "DAClusterizerInZ_vect::vertices :"
               << "last cooling T=" << 1 / beta << std::endl;
   }
 #endif
 
   // optionally cool some more without doing anything, to make the assignment harder
   while (beta < betastop_) {
     beta = min(beta / coolingFactor_, betastop_);
     thermalize(beta, tks, y, delta_lowT_, rho0);
   }
 
 #ifdef DEBUG
   verify(y, tks);
   if (DEBUGLEVEL > 0) {
     std::cout << "DAClusterizerInZ_vect::vertices :"
               << "stop cooling at T=" << 1 / beta << std::endl;
   }
   if (DEBUGLEVEL > 2)
     dump(beta, y, tks, 2, rho0);
 #endif
 
   // select significant tracks and use a TransientVertex as a container
 
   set_vtx_range(beta, tks, y);
   const unsigned int nv = y.getSize();
   for (unsigned int k = 0; k < nv; k++) {
     if (edm::isNotFinite(y.rho[k]) || edm::isNotFinite(y.zvtx[k])) {
       y.rho[k] = 0;
       y.zvtx[k] = 0;
     }
   }
 
   const auto z_sum_init = rho0 * local_exp(-beta * dzCutOff_ * dzCutOff_);
   std::vector<std::vector<unsigned int> > vtx_track_indices(nv);
   for (unsigned int i = 0; i < nt; i++) {
     const auto kmin = tks.kmin[i];
     const auto kmax = tks.kmax[i];
     for (auto k = kmin; k < kmax; k++) {
       y.exp_arg[k] = -beta * Eik(tks.zpca[i], y.zvtx[k], tks.dz2[i]);
     }
 
     local_exp_list_range(y.exp_arg, y.exp, kmin, kmax);
 
     tks.sum_Z[i] = z_sum_init;
     for (auto k = kmin; k < kmax; k++) {
       tks.sum_Z[i] += y.rho[k] * y.exp[k];
     }
     const double invZ = tks.sum_Z[i] > 1e-100 ? 1. / tks.sum_Z[i] : 0.0;
 
     for (auto k = kmin; k < kmax; k++) {
       double p = y.rho[k] * y.exp[k] * invZ;
       if (p > mintrkweight_) {
         // assign  track i -> vertex k (hard, mintrkweight_ should be >= 0.5 here
         vtx_track_indices[k].push_back(i);
         break;
       }
     }
 
   }  // track loop
 
   GlobalError dummyError(0.01, 0, 0.01, 0., 0., 0.01);
   for (unsigned int k = 0; k < nv; k++) {
     if (!vtx_track_indices[k].empty()) {
       GlobalPoint pos(0, 0, y.zvtx[k]);
       vector<reco::TransientTrack> vertexTracks;
       for (auto i : vtx_track_indices[k]) {
         vertexTracks.push_back(*(tks.tt[i]));
       }
       TransientVertex v(pos, dummyError, vertexTracks, 0);
       clusters.push_back(v);
     }
   }
 
   return clusters;
 }

Member Data Documentation

◆ betamax_

double DAClusterizerInZ_vect::betamax_

private

Definition at line 208 of file DAClusterizerInZ_vect.h.

◆ betapurge_

double DAClusterizerInZ_vect::betapurge_

private

Definition at line 217 of file DAClusterizerInZ_vect.h.

◆ betastop_

double DAClusterizerInZ_vect::betastop_

private

Definition at line 209 of file DAClusterizerInZ_vect.h.

◆ convergence_mode_

unsigned int DAClusterizerInZ_vect::convergence_mode_

private

Definition at line 219 of file DAClusterizerInZ_vect.h.

◆ coolingFactor_

double DAClusterizerInZ_vect::coolingFactor_

private

Definition at line 207 of file DAClusterizerInZ_vect.h.

◆ d0CutOff_

double DAClusterizerInZ_vect::d0CutOff_

private

Definition at line 211 of file DAClusterizerInZ_vect.h.

◆ delta_highT_

double DAClusterizerInZ_vect::delta_highT_

private

Definition at line 220 of file DAClusterizerInZ_vect.h.

◆ delta_lowT_

double DAClusterizerInZ_vect::delta_lowT_

private

Definition at line 221 of file DAClusterizerInZ_vect.h.

◆ dzCutOff_

double DAClusterizerInZ_vect::dzCutOff_

private

Definition at line 210 of file DAClusterizerInZ_vect.h.

◆ maxIterations_

unsigned int DAClusterizerInZ_vect::maxIterations_

private

Definition at line 206 of file DAClusterizerInZ_vect.h.

◆ mintrkweight_

double DAClusterizerInZ_vect::mintrkweight_

private

Definition at line 213 of file DAClusterizerInZ_vect.h.

◆ sel_zrange_

double DAClusterizerInZ_vect::sel_zrange_

private

Definition at line 223 of file DAClusterizerInZ_vect.h.

◆ uniquetrkminp_

double DAClusterizerInZ_vect::uniquetrkminp_

private

Definition at line 215 of file DAClusterizerInZ_vect.h.

◆ uniquetrkweight_

double DAClusterizerInZ_vect::uniquetrkweight_

private

Definition at line 214 of file DAClusterizerInZ_vect.h.

◆ vertexSize_

double DAClusterizerInZ_vect::vertexSize_

private

Definition at line 205 of file DAClusterizerInZ_vect.h.

◆ zdumpcenter_

double DAClusterizerInZ_vect::zdumpcenter_

private

Definition at line 202 of file DAClusterizerInZ_vect.h.

◆ zdumpwidth_

double DAClusterizerInZ_vect::zdumpwidth_

private

Definition at line 203 of file DAClusterizerInZ_vect.h.

◆ zmerge_

double DAClusterizerInZ_vect::zmerge_

private

Definition at line 216 of file DAClusterizerInZ_vect.h.

◆ zrange_min_

const double DAClusterizerInZ_vect::zrange_min_ = 0.1

private

Definition at line 224 of file DAClusterizerInZ_vect.h.

Classes

Public Member Functions

Static Public Member Functions

Private Attributes

Detailed Description

Constructor & Destructor Documentation

◆ DAClusterizerInZ_vect()

Member Function Documentation

◆ beta0()

◆ clear_vtx_range()

◆ clusterize()

◆ dump()

◆ fill()

◆ fillPSetDescription()

◆ merge()

◆ purge()

◆ set_vtx_range()

◆ split()

◆ thermalize()

◆ update()

◆ verify()

◆ vertices()

Member Data Documentation

◆ betamax_

◆ betapurge_

◆ betastop_

◆ convergence_mode_

◆ coolingFactor_

◆ d0CutOff_

◆ delta_highT_

◆ delta_lowT_

◆ dzCutOff_

◆ maxIterations_

◆ mintrkweight_

◆ sel_zrange_

◆ uniquetrkminp_

◆ uniquetrkweight_

◆ vertexSize_

◆ zdumpcenter_

◆ zdumpwidth_

◆ zmerge_

◆ zrange_min_