Stage 1 sorting and truncation per r/z bin and FPGA

Co-authored-by: uttiyasarkar

L1Trigger/L1THGCal/interface/backend/HGCalBackendStage1Processor.h

@@ -0,0 +1,25 @@
+#ifndef __L1Trigger_L1THGCal_HGCalBackendStage1Processor_h__
+#define __L1Trigger_L1THGCal_HGCalBackendStage1Processor_h__
+
+#include "L1Trigger/L1THGCal/interface/HGCalProcessorBase.h"
+
+#include "DataFormats/L1THGCal/interface/HGCalTriggerCell.h"
+#include "DataFormats/L1THGCal/interface/HGCalCluster.h"
+
+#include "L1Trigger/L1THGCal/interface/backend/HGCalStage1TruncationImpl.h"
+#include "L1Trigger/L1THGCal/interface/backend/HGCalClusteringDummyImpl.h"
+
+class HGCalBackendStage1Processor : public HGCalBackendLayer1ProcessorBase {
+public:
+  HGCalBackendStage1Processor(const edm::ParameterSet& conf);
+
+  void run(const edm::Handle<l1t::HGCalTriggerCellBxCollection>& collHandle,
+           l1t::HGCalClusterBxCollection& collCluster2D,
+           const edm::EventSetup& es) override;
+
+private:
+  std::unique_ptr<HGCalClusteringDummyImpl> clusteringDummy_;
+  std::unique_ptr<HGCalStage1TruncationImpl> truncation_;
+};
+
+#endif

L1Trigger/L1THGCal/interface/backend/HGCalStage1TruncationImpl.h

@@ -0,0 +1,40 @@
+#ifndef __L1Trigger_L1THGCal_HGCalStage1TruncationImpl_h__
+#define __L1Trigger_L1THGCal_HGCalStage1TruncationImpl_h__
+
+#include "FWCore/ParameterSet/interface/ParameterSet.h"
+#include "DataFormats/L1THGCal/interface/HGCalTriggerCell.h"
+#include "L1Trigger/L1THGCal/interface/HGCalTriggerTools.h"
+
+#include <vector>
+
+class HGCalStage1TruncationImpl {
+public:
+  HGCalStage1TruncationImpl(const edm::ParameterSet& conf);
+
+  void eventSetup(const edm::EventSetup& es) { triggerTools_.eventSetup(es); }
+
+  void run(uint32_t fpga_id,
+           const std::vector<edm::Ptr<l1t::HGCalTriggerCell>>& tcs_in,
+           std::vector<edm::Ptr<l1t::HGCalTriggerCell>>& tcs_out);
+
+private:
+  HGCalTriggerTools triggerTools_;
+
+  static constexpr unsigned offset_roz_ = 1;
+  static constexpr unsigned mask_roz_ = 0x3f;  // 6 bits, max 64 bins
+  static constexpr unsigned mask_phi_ = 1;
+
+  double roz_min_ = 0.;
+  double roz_max_ = 0.;
+  unsigned roz_bins_ = 42;
+  std::vector<unsigned> max_tcs_per_bin_;
+  std::vector<double> phi_edges_;
+  double roz_bin_size_ = 0.;
+
+  uint32_t packBin(unsigned roverzbin, unsigned phibin) const;
+  void unpackBin(unsigned packedbin, unsigned& roverzbin, unsigned& phibin) const;
+  unsigned phiBin(unsigned roverzbin, double phi) const;
+  double rotatedphi(double x, double y, double z, int sector) const;
+};
+
+#endif

L1Trigger/L1THGCal/plugins/backend/HGCalBackendStage1Processor.cc

@@ -0,0 +1,33 @@
+#include "L1Trigger/L1THGCal/interface/backend/HGCalBackendStage1Processor.h"
+
+DEFINE_EDM_PLUGIN(HGCalBackendLayer1Factory, HGCalBackendStage1Processor, "HGCalBackendStage1Processor");
+
+HGCalBackendStage1Processor::HGCalBackendStage1Processor(const edm::ParameterSet& conf)
+    : HGCalBackendLayer1ProcessorBase(conf) {
+  clusteringDummy_ = std::make_unique<HGCalClusteringDummyImpl>(conf.getParameterSet("C2d_parameters"));
+  truncation_ = std::make_unique<HGCalStage1TruncationImpl>(conf.getParameterSet("truncation_parameters"));
+}
+
+void HGCalBackendStage1Processor::run(const edm::Handle<l1t::HGCalTriggerCellBxCollection>& collHandle,
+                                      l1t::HGCalClusterBxCollection& collCluster2D,
+                                      const edm::EventSetup& es) {
+  if (clusteringDummy_)
+    clusteringDummy_->eventSetup(es);
+  if (truncation_)
+    truncation_->eventSetup(es);
+
+  std::unordered_map<uint32_t, std::vector<edm::Ptr<l1t::HGCalTriggerCell>>> tcs_per_fpga;
+
+  for (unsigned i = 0; i < collHandle->size(); ++i) {
+    edm::Ptr<l1t::HGCalTriggerCell> tc_ptr(collHandle, i);
+    uint32_t module = geometry_->getModuleFromTriggerCell(tc_ptr->detId());
+    uint32_t fpga = geometry_->getStage1FpgaFromModule(module);
+    tcs_per_fpga[fpga].push_back(tc_ptr);
+  }
+
+  std::vector<edm::Ptr<l1t::HGCalTriggerCell>> truncated_tcs;
+  for (auto& fpga_tcs : tcs_per_fpga) {
+    truncation_->run(fpga_tcs.first, fpga_tcs.second, truncated_tcs);
+  }
+  clusteringDummy_->clusterizeDummy(truncated_tcs, collCluster2D);
+}

L1Trigger/L1THGCal/python/customNewProcessors.py

@@ -1,4 +1,13 @@
 import FWCore.ParameterSet.Config as cms
+from L1Trigger.L1THGCal.hgcalBackEndLayer1Producer_cfi import stage1truncation_proc
+
+
+def custom_stage1_truncation(process):
+    parameters = stage1truncation_proc.clone()
+    process.hgcalBackEndLayer1Producer.ProcessorParameters = parameters
+    process.hgcalBackEndLayer2Producer.InputCluster = cms.InputTag('hgcalBackEndLayer1Producer:HGCalBackendStage1Processor')
+    process.hgcalTowerProducer.InputTriggerCells = cms.InputTag('hgcalBackEndLayer1Producer:HGCalBackendStage1Processor')
+    return process
 
 def custom_clustering_standalone(process):
     process.hgcalBackEndLayer2Producer.ProcessorParameters.ProcessorName = cms.string('HGCalBackendLayer2Processor3DClusteringSA')

L1Trigger/L1THGCal/python/hgcalBackEndLayer1Producer_cfi.py

@@ -57,6 +57,29 @@
                                         dR_cluster=cms.double(6.),
                                         )
 
+ntcs_72links = [ 1,  4, 13, 13, 10, 10,  8,  8,  8,  7,  7,  6,  6,  6,  6,  6,  5,  5,  5,  5,  5,  5,  5,  4,  4,  4,  4,  4,  4,  4,  4,  4,  3,  2,  2,  2,  2,  2,  2,  2,  2,  1]
+ntcs_120links = [ 2,  7, 27, 24, 19, 17, 16, 15, 14, 14, 13, 13, 13, 12, 12, 12, 11, 11, 11, 10, 10, 10, 10, 10,  9,  9, 10,  9,  9,  9,  8,  8,  7,  5,  3,  3,  3,  3,  3,  3,  3,  3]
+
+phi_edges = [0.98901991,0.72722052,0.6981317,0.87266463,0.93084227,0.90175345,
+0.87266463,0.90175345,0.95993109,0.95993109,0.93084227,0.93084227,
+0.95993109,0.98901991,0.95993109,0.95993109,0.95993109,0.98901991,
+0.98901991,0.95993109,0.95993109,0.98901991,0.98901991,0.98901991,
+0.98901991,0.98901991,1.01810873,0.98901991,0.98901991,0.98901991,
+0.98901991,0.98901991,0.98901991,0.98901991,1.04719755,1.04719755,
+1.04719755,1.04719755,1.01810873,1.04719755,1.01810873,1.01810873]
+
+truncation_params = cms.PSet(rozMin=cms.double(0.07587128),
+        rozMax=cms.double(0.55508006),
+        rozBins=cms.uint32(42),
+        maxTcsPerBin=cms.vuint32(ntcs_72links),
+        phiSectorEdges=cms.vdouble(phi_edges)
+        )
+
+
+stage1truncation_proc = cms.PSet(ProcessorName  = cms.string('HGCalBackendStage1Processor'),
+                   C2d_parameters = dummy_C2d_params.clone(),
+                   truncation_parameters = truncation_params.clone()
+                   )
 
 be_proc = cms.PSet(ProcessorName  = cms.string('HGCalBackendLayer1Processor2DClustering'),
                    C2d_parameters = dummy_C2d_params.clone()

L1Trigger/L1THGCal/src/backend/HGCalStage1TruncationImpl.cc

@@ -0,0 +1,102 @@
+#include "L1Trigger/L1THGCal/interface/backend/HGCalStage1TruncationImpl.h"
+#include "L1Trigger/L1THGCal/interface/HGCalBackendDetId.h"
+#include <cmath>
+
+HGCalStage1TruncationImpl::HGCalStage1TruncationImpl(const edm::ParameterSet& conf)
+    : roz_min_(conf.getParameter<double>("rozMin")),
+      roz_max_(conf.getParameter<double>("rozMax")),
+      roz_bins_(conf.getParameter<unsigned>("rozBins")),
+      max_tcs_per_bin_(conf.getParameter<std::vector<unsigned>>("maxTcsPerBin")),
+      phi_edges_(conf.getParameter<std::vector<double>>("phiSectorEdges")) {
+  if (max_tcs_per_bin_.size() != roz_bins_)
+    throw cms::Exception("HGCalStage1TruncationImpl::BadConfig") << "Inconsistent sizes of maxTcsPerBin and rozBins";
+  if (phi_edges_.size() != roz_bins_)
+    throw cms::Exception("HGCalStage1TruncationImpl::BadConfig") << "Inconsistent sizes of phiSectorEdges and rozBins";
+
+  constexpr double margin = 1.001;
+  roz_bin_size_ = (roz_bins_ > 0 ? (roz_max_ - roz_min_) * margin / double(roz_bins_) : 0.);
+}
+
+void HGCalStage1TruncationImpl::run(uint32_t fpga_id,
+                                    const std::vector<edm::Ptr<l1t::HGCalTriggerCell>>& tcs_in,
+                                    std::vector<edm::Ptr<l1t::HGCalTriggerCell>>& tcs_out) {
+  unsigned sector120 = HGCalBackendDetId(fpga_id).sector();
+  std::unordered_map<unsigned, std::vector<edm::Ptr<l1t::HGCalTriggerCell>>> tcs_per_bin;
+
+  // group TCs per (r/z, phi) bins
+  for (const auto& tc : tcs_in) {
+    const GlobalPoint& position = tc->position();
+    double x = position.x();
+    double y = position.y();
+    double z = position.z();
+    double roverz = std::sqrt(x * x + y * y) / std::abs(z);
+    roverz = (roverz < roz_min_ ? roz_min_ : roverz);
+    roverz = (roverz > roz_max_ ? roz_max_ : roverz);
+    unsigned roverzbin = (roz_bin_size_ > 0. ? unsigned((roverz - roz_min_) / roz_bin_size_) : 0);
+    double phi = rotatedphi(x, y, z, sector120);
+    unsigned phibin = phiBin(roverzbin, phi);
+    unsigned packed_bin = packBin(roverzbin, phibin);
+
+    tcs_per_bin[packed_bin].push_back(tc);
+  }
+  // apply sorting and trunction in each (r/z, phi) bin
+  for (auto& bin_tcs : tcs_per_bin) {
+    std::sort(bin_tcs.second.begin(),
+              bin_tcs.second.end(),
+              [](const edm::Ptr<l1t::HGCalTriggerCell>& a, const edm::Ptr<l1t::HGCalTriggerCell>& b) -> bool {
+                return a->mipPt() > b->mipPt();
+              });
+
+    unsigned roverzbin = 0;
+    unsigned phibin = 0;
+    unpackBin(bin_tcs.first, roverzbin, phibin);
+    if (roverzbin >= max_tcs_per_bin_.size())
+      throw cms::Exception("HGCalStage1TruncationImpl::OutOfRange")
+          << "roverzbin index " << roverzbin << "out of range";
+    unsigned max_tc = max_tcs_per_bin_[roverzbin];
+    if (bin_tcs.second.size() > max_tc) {
+      bin_tcs.second.resize(max_tc);
+    }
+    for (const auto& tc : bin_tcs.second) {
+      tcs_out.push_back(tc);
+    }
+  }
+}
+
+unsigned HGCalStage1TruncationImpl::packBin(unsigned roverzbin, unsigned phibin) const {
+  unsigned packed_bin = 0;
+  packed_bin |= ((roverzbin & mask_roz_) << offset_roz_);
+  packed_bin |= (phibin & mask_phi_);
+  return packed_bin;
+}
+
+void HGCalStage1TruncationImpl::unpackBin(unsigned packedbin, unsigned& roverzbin, unsigned& phibin) const {
+  roverzbin = ((packedbin >> offset_roz_) & mask_roz_);
+  phibin = (packedbin & mask_phi_);
+}
+
+unsigned HGCalStage1TruncationImpl::phiBin(unsigned roverzbin, double phi) const {
+  unsigned phi_bin = 0;
+  if (roverzbin >= phi_edges_.size())
+    throw cms::Exception("HGCalStage1TruncationImpl::OutOfRange") << "roverzbin index " << roverzbin << "out of range";
+  double phi_edge = phi_edges_[roverzbin];
+  if (phi > phi_edge)
+    phi_bin = 1;
+  return phi_bin;
+}
+
+double HGCalStage1TruncationImpl::rotatedphi(double x, double y, double z, int sector) const {
+  if (z > 0)
+    x = -x;
+  double phi = std::atan2(y, x);
+
+  if (sector == 1) {
+    if (phi < M_PI and phi > 0)
+      phi = phi - (2. * M_PI / 3.);
+    else
+      phi = phi + (4. * M_PI / 3.);
+  } else if (sector == 2) {
+    phi = phi + (2. * M_PI / 3.);
+  }
+  return phi;
+}

L1Trigger/L1THGCalUtilities/python/customNtuples.py

@@ -8,6 +8,13 @@ def custom_ntuples_V9(process):
             ntuple.bhSimHits = cms.InputTag('g4SimHits:HGCHitsHEback')
     return process
 
+def custom_ntuples_stage1_truncation(process):
+    ntuples = process.hgcalTriggerNtuplizer.Ntuples
+    for ntuple in ntuples:
+        if ntuple.NtupleName=='HGCalTriggerNtupleHGCClusters':
+            ntuple.Clusters = cms.InputTag('hgcalBackEndLayer1Producer:HGCalBackendStage1Processor')
+    return process
+
 def custom_ntuples_standalone_clustering(process):
     ntuples = process.hgcalTriggerNtuplizer.Ntuples
     for ntuple in ntuples:

L1Trigger/L1THGCalUtilities/test/testHGCalL1T_stage1Truncation_V11_cfg.py

@@ -0,0 +1,96 @@
+import FWCore.ParameterSet.Config as cms 
+
+from Configuration.Eras.Era_Phase2C9_cff import Phase2C9
+process = cms.Process('DIGI',Phase2C9)
+
+# import of standard configurations
+process.load('Configuration.StandardSequences.Services_cff')
+process.load('SimGeneral.HepPDTESSource.pythiapdt_cfi')
+process.load('FWCore.MessageService.MessageLogger_cfi')
+process.load('Configuration.EventContent.EventContent_cff')
+process.load('SimGeneral.MixingModule.mixNoPU_cfi')
+process.load('Configuration.Geometry.GeometryExtended2026D49Reco_cff')
+process.load('Configuration.Geometry.GeometryExtended2026D49_cff')
+process.load('Configuration.StandardSequences.MagneticField_cff')
+process.load('Configuration.StandardSequences.Generator_cff')
+process.load('IOMC.EventVertexGenerators.VtxSmearedHLLHC14TeV_cfi')
+process.load('GeneratorInterface.Core.genFilterSummary_cff')
+process.load('Configuration.StandardSequences.SimIdeal_cff')
+process.load('Configuration.StandardSequences.Digi_cff')
+process.load('Configuration.StandardSequences.SimL1Emulator_cff')
+process.load('Configuration.StandardSequences.DigiToRaw_cff')
+process.load('Configuration.StandardSequences.EndOfProcess_cff')
+process.load('Configuration.StandardSequences.FrontierConditions_GlobalTag_cff')
+
+
+process.maxEvents = cms.untracked.PSet(
+    input = cms.untracked.int32(5)
+)
+
+# Input source
+process.source = cms.Source("PoolSource",
+       fileNames = cms.untracked.vstring('/store/mc/Phase2HLTTDRWinter20DIGI/SingleElectron_PT2to200/GEN-SIM-DIGI-RAW/PU200_110X_mcRun4_realistic_v3_ext2-v2/40000/00582F93-5A2A-5847-8162-D81EE503500F.root'),
+       inputCommands=cms.untracked.vstring(
+           'keep *',
+           'drop l1tEMTFHit2016Extras_simEmtfDigis_CSC_HLT',
+           'drop l1tEMTFHit2016Extras_simEmtfDigis_RPC_HLT',
+           'drop l1tEMTFHit2016s_simEmtfDigis__HLT',
+           'drop l1tEMTFTrack2016Extras_simEmtfDigis__HLT',
+           'drop l1tEMTFTrack2016s_simEmtfDigis__HLT',
+           'drop FTLClusteredmNewDetSetVector_mtdClusters_FTLBarrel_RECO',
+           'drop FTLClusteredmNewDetSetVector_mtdClusters_FTLEndcap_RECO',
+           'drop MTDTrackingRecHitedmNewDetSetVector_mtdTrackingRecHits__RECO',
+           'drop BTLDetIdBTLSampleFTLDataFrameTsSorted_mix_FTLBarrel_HLT',
+           'drop ETLDetIdETLSampleFTLDataFrameTsSorted_mix_FTLEndcap_HLT',
+           )
+       )
+
+process.options = cms.untracked.PSet(
+
+)
+
+# Production Info
+process.configurationMetadata = cms.untracked.PSet(
+    version = cms.untracked.string('$Revision: 1.20 $'),
+    annotation = cms.untracked.string('SingleElectronPt10_cfi nevts:10'),
+    name = cms.untracked.string('Applications')
+)
+
+# Output definition
+process.TFileService = cms.Service(
+    "TFileService",
+    fileName = cms.string("ntuple.root")
+    )
+
+# Other statements
+from Configuration.AlCa.GlobalTag import GlobalTag
+process.GlobalTag = GlobalTag(process.GlobalTag, 'auto:phase2_realistic_T15', '')
+
+# load HGCAL TPG simulation
+process.load('L1Trigger.L1THGCal.hgcalTriggerPrimitives_cff')
+
+# Use new Stage 1 processor
+from L1Trigger.L1THGCal.customNewProcessors import custom_stage1_truncation
+process = custom_stage1_truncation(process)
+
+# Switch to latest trigger geometry containing information on links mapping
+from L1Trigger.L1THGCal.customTriggerGeometry import custom_geometry_decentralized_V11
+process = custom_geometry_decentralized_V11(process, links='signaldriven', implementation=2)
+
+process.hgcl1tpg_step = cms.Path(process.hgcalTriggerPrimitives)
+
+
+# load ntuplizer and custom to use collections from Stag1 truncation processor
+process.load('L1Trigger.L1THGCalUtilities.hgcalTriggerNtuples_cff')
+from L1Trigger.L1THGCalUtilities.customNtuples import custom_ntuples_stage1_truncation
+process = custom_ntuples_stage1_truncation(process)
+process.ntuple_step = cms.Path(process.hgcalTriggerNtuples)
+
+# Schedule definition
+process.schedule = cms.Schedule(process.hgcl1tpg_step, process.ntuple_step)
+
+# Add early deletion of temporary data products to reduce peak memory need
+from Configuration.StandardSequences.earlyDeleteSettings_cff import customiseEarlyDelete
+process = customiseEarlyDelete(process)
+# End adding early deletion
+