---

L1MuGMTMerger.cc

Go to the documentation of this file.

//-------------------------------------------------
//
//   Class: L1MuGMTMerger
//
//   Description:  GMT Merger
//
//
//   $Date $
//   $Revision $
//
//   Author :
//   H. Sakulin                   HEPHY Vienna
//
//   Migrated to CMSSW:
//   I. Mikulec
//
//--------------------------------------------------

//-----------------------
// This Class's Header --
//-----------------------

#include "L1Trigger/GlobalMuonTrigger/src/L1MuGMTMerger.h"

//---------------
// C++ Headers --
//---------------

#include <iostream>
#include <vector>
#include <cmath>

//-------------------------------
// Collaborating Class Headers --
//-------------------------------

#include "L1Trigger/GlobalMuonTrigger/src/L1MuGMTConfig.h"
#include "L1Trigger/GlobalMuonTrigger/interface/L1MuGlobalMuonTrigger.h"
#include "L1Trigger/GlobalMuonTrigger/src/L1MuGMTPSB.h"
#include "L1Trigger/GlobalMuonTrigger/src/L1MuGMTSortRankUnit.h"
#include "L1Trigger/GlobalMuonTrigger/src/L1MuGMTMatcher.h"
#include "L1Trigger/GlobalMuonTrigger/src/L1MuGMTCancelOutUnit.h"
#include "L1Trigger/GlobalMuonTrigger/src/L1MuGMTMipIsoAU.h"
#include "DataFormats/L1GlobalMuonTrigger/interface/L1MuGMTExtendedCand.h"


#include "L1Trigger/GlobalMuonTrigger/src/L1MuGMTLFMergeRankEtaQLUT.h"
#include "L1Trigger/GlobalMuonTrigger/src/L1MuGMTLFMergeRankPtQLUT.h"
#include "L1Trigger/GlobalMuonTrigger/src/L1MuGMTLFMergeRankEtaPhiLUT.h"
#include "L1Trigger/GlobalMuonTrigger/src/L1MuGMTLFMergeRankCombineLUT.h"

#include "L1Trigger/GlobalMuonTrigger/src/L1MuGMTLFEtaConvLUT.h"

#include "L1Trigger/GlobalMuonTrigger/src/L1MuGMTLFPhiProEtaConvLUT.h"
#include "L1Trigger/GlobalMuonTrigger/src/L1MuGMTLFPhiProLUT.h"

#include "L1Trigger/GlobalMuonTrigger/src/L1MuGMTLFPtMixLUT.h"

#include "CondFormats/L1TObjects/interface/L1MuTriggerScales.h"
#include "CondFormats/L1TObjects/interface/L1MuTriggerPtScale.h"

#include "L1Trigger/GlobalMuonTrigger/src/L1MuGMTReg.h"
  
#include "FWCore/MessageLogger/interface/MessageLogger.h"

//---------------------------------
//       class L1MuGMTMerger
//---------------------------------

//----------------
// Constructors --
//----------------
L1MuGMTMerger::L1MuGMTMerger(const L1MuGlobalMuonTrigger& gmt, int id) : 
  m_gmt(gmt), m_id(id), dtcsc_mu(4), rpc_mu(4) {

  dtcsc_mu.reserve(4);
  rpc_mu.reserve(4);
   
}

//--------------
// Destructor --
//--------------
L1MuGMTMerger::~L1MuGMTMerger() { 
  reset();  
}

//--------------
// Operations --
//--------------

//
// run Merger
//
void L1MuGMTMerger::run() {
  load();
  merge();  
}


//
// reset Merger
//
void L1MuGMTMerger::reset() {

  for ( int i = 0; i < 4; i++ ) {  
    dtcsc_mu[i] = 0;
    rpc_mu[i] = 0;
  }
  
  std::vector<L1MuGMTExtendedCand*>::iterator iter;
  for ( iter = m_MuonCands.begin(); iter != m_MuonCands.end(); iter++ ) {
    if ( *iter ) delete (*iter); 
    *iter = 0;
  }

  m_MuonCands.clear();

}


//
// print selection results
//
void L1MuGMTMerger::print() const {

  edm::LogVerbatim("GMT_Merger_info") << " ";

  std::vector<L1MuGMTExtendedCand*>::const_iterator iter;
  for ( iter = m_MuonCands.begin(); iter != m_MuonCands.end(); iter++ ) {
    if ( *iter && !(*iter)->empty() ) (*iter)->print();
  }
  
  edm::LogVerbatim("GMT_Merger_info") << " ";

}


//
// load Merger (get data from PSB)
//
void L1MuGMTMerger::load() {

  // barrel Merger gets DTBX and barrel RPC muons
  if ( m_id == 0 ) {
    for ( unsigned idt = 0; idt < L1MuGMTConfig::MAXDTBX; idt++ ) {
      dtcsc_mu[idt] = m_gmt.Data()->DTBXMuon(idt);
    }
    for ( unsigned irpc = 0; irpc < L1MuGMTConfig::MAXRPCbarrel; irpc++ ) {
      rpc_mu[irpc] = m_gmt.Data()->RPCMuon(irpc);
    }
  }
  
  // endcap Merger gets CSC and endcap RPC muons
  if ( m_id == 1 ) {  
    for ( unsigned icsc = 0; icsc < L1MuGMTConfig::MAXCSC; icsc++ ) {
      dtcsc_mu[icsc] = m_gmt.Data()->CSCMuon(icsc);
    }
    for ( unsigned irpc = 0; irpc < L1MuGMTConfig::MAXRPCendcap; irpc++ ) {
      rpc_mu[irpc] = m_gmt.Data()->RPCMuon(irpc+4);
    }
  }
 
}


void L1MuGMTMerger::merge() {

  const L1MuGMTMatrix<bool>& pairM = m_gmt.Matcher(m_id)->pairM();

  // Handling of cancel-out and empty muons is different in software and hardware
  //
  // - in hardware, a candidate may be empty (pt_code == 0) and an empty bit
  //   is passed to the sorter in order to suppress the candidate. In the
  //   software no candidate is created in this case.
  //
  // - in hardware, RPC candidates are passed to the sorter even if they are 
  //   also used in a matched pair. They are then suppressed in the sorter. In
  //   software no RPC candidate is created if the muon is used in a pair.
  //
  // - in hardware, cancel-out signals from the cancel-out units in the own and   
  //   other Logic FPGA are passed to the sorter in order to cancel out muons.
  //   In software the cancel-out signals are alrady checked here in the merger 
  //   and no candidates are created for cancelled muons.
  //
  // There may therefore be less muons passed to the sorter in software than 
  // in Hardware. At the output of the first sorter stage the results should 
  // be comparable, again.
  
  unsigned HaloOverwritesMatched = 1;
  
  // loop over DT/CSC muons
  for (int i=0; i<4; i++) {
    if (dtcsc_mu[i] != 0) {

      int match_idx = pairM.rowAny(i);

      int csc_is_halo = (m_id==1) && (dtcsc_mu[i]->finehalo_packed() == 1);

      if ( (match_idx != -1) && // is it matched?
           (! (csc_is_halo && HaloOverwritesMatched) ) )
        createMergedCand(i, match_idx);         
      else {
        // check my first and the other chip's second cancel-out units
        if ((! m_gmt.CancelOutUnit(m_id)->cancelMyChipMuon(i) ) &&
            (! m_gmt.CancelOutUnit(3-m_id)->cancelOtherChipMuon(i) ) &&
            (! L1MuGMTSortRankUnit::isDisabled(dtcsc_mu[i]) ) )
          createDTCSCCand(i);
      }
    }
  }

  // additionally loop over RPC muons
  for (int j=0; j<4; j++) {
    if (rpc_mu[j] != 0) {      
      
      int match_idx = pairM.colAny(j);
      
      if (match_idx == -1) { // is it unmatched?
        if ((! m_gmt.CancelOutUnit(m_id+2)->cancelMyChipMuon(j) ) &&
            (! L1MuGMTSortRankUnit::isDisabled(rpc_mu[j]) ) )
          createRPCCand(j);
      }
    }
  }

  // set physical values in the GMT candidates for use in the analysis
  const L1MuTriggerScales* theTriggerScales = L1MuGMTConfig::getTriggerScales();
  const L1MuTriggerPtScale* theTriggerPtScale = L1MuGMTConfig::getTriggerPtScale();
  
  std::vector<L1MuGMTExtendedCand*>::const_iterator icand;
  for(icand=m_MuonCands.begin();icand!=m_MuonCands.end();icand++) {
    L1MuGMTExtendedCand* cand = (*icand);
    cand->setPhiValue( theTriggerScales->getPhiScale()->getLowEdge( cand->phiIndex() ));
    cand->setEtaValue( theTriggerScales->getGMTEtaScale()->getCenter( cand->etaIndex() ));
    cand->setPtValue( theTriggerPtScale->getPtScale()->getLowEdge( cand->ptIndex() ));
    // cand->setPtValue( theTriggerScales->getPtScale()->getLowEdge( cand->ptIndex() ));
  }

}

void L1MuGMTMerger::createDTCSCCand(int idx_dtcsc) {
  L1MuGMTExtendedCand* tmpmuon = new L1MuGMTExtendedCand();
   
  tmpmuon->setBx(dtcsc_mu[idx_dtcsc]->bx());
  tmpmuon->setPhiPacked( projectedPhi(dtcsc_mu[idx_dtcsc]) ); 
  tmpmuon->setEtaPacked( convertedEta(dtcsc_mu[idx_dtcsc]) ); 
  tmpmuon->setPtPacked( dtcsc_mu[idx_dtcsc]->pt_packed() ); 
  tmpmuon->setChargePacked( sysign(dtcsc_mu[idx_dtcsc]) ); 
  tmpmuon->setMIP( m_gmt.MipIsoAU(m_id)->MIP(idx_dtcsc) );    
  tmpmuon->setIsolation ( m_gmt.MipIsoAU(m_id)->ISO(idx_dtcsc) );    
  tmpmuon->setRank( L1MuGMTSortRankUnit::sort_rank(dtcsc_mu[idx_dtcsc]) );

  unsigned quality = 0;
  switch ( L1MuGMTSortRankUnit::getVeryLowQualityLevel(dtcsc_mu[idx_dtcsc]) ) {
  case 0: quality = 6; break; //DT/CSC
  case 1: quality = 2; break; //VERY LOW QUALITY LEVEL 1
  case 2: quality = 3; break; //VERY LOW QUALITY LEVEL 2
  case 3: quality = 4; break; //VERY LOW QUALITY LEVEL 3
  }

  if ( (m_id==1) && (dtcsc_mu[idx_dtcsc]->finehalo_packed() == 1) ) 
    quality = 1; // HALO quality

  tmpmuon->setQuality( quality ); // RPC
  tmpmuon->setDTCSCIndex(idx_dtcsc);
  tmpmuon->setRPCIndex(0);
  tmpmuon->setFwdBit( m_id );
  tmpmuon->setRPCBit( 0 );

  m_MuonCands.push_back(tmpmuon);
}


void L1MuGMTMerger::createRPCCand(int idx_rpc) {
  L1MuGMTExtendedCand* tmpmuon = new L1MuGMTExtendedCand();
   
  tmpmuon->setBx(rpc_mu[idx_rpc]->bx()); 
  tmpmuon->setPhiPacked( projectedPhi(rpc_mu[idx_rpc]) ); 
  tmpmuon->setEtaPacked( convertedEta(rpc_mu[idx_rpc]) ); 
  tmpmuon->setPtPacked( rpc_mu[idx_rpc]->pt_packed() ); 
  tmpmuon->setChargePacked( sysign( rpc_mu[idx_rpc] ) ); 
  tmpmuon->setMIP( m_gmt.MipIsoAU(m_id)->MIP(idx_rpc+4) );    
  tmpmuon->setIsolation ( m_gmt.MipIsoAU(m_id)->ISO(idx_rpc+4) );    
  tmpmuon->setRank( L1MuGMTSortRankUnit::sort_rank(rpc_mu[idx_rpc]) );

  unsigned quality = 0;
  switch ( L1MuGMTSortRankUnit::getVeryLowQualityLevel(rpc_mu[idx_rpc]) ) {
  case 0: quality = 5; break; //RPC
  case 1: quality = 2; break; //VERY LOW QUALITY LEVEL1
  case 2: quality = 3; break; //VERY LOW QUALITY LEVEL2
  case 3: quality = 4; break; //VERY LOW QUALITY LEVEL3
  }

  tmpmuon->setQuality( quality ); // RPC
  tmpmuon->setDTCSCIndex(0);
  tmpmuon->setRPCIndex(idx_rpc);
  tmpmuon->setFwdBit( m_id );
  tmpmuon->setRPCBit( 1 );
    
  m_MuonCands.push_back(tmpmuon);
}

int L1MuGMTMerger::selectDTCSC(unsigned MMconfig, int by_rank, int by_pt, int by_combi) const {
  return 
    (   (MMconfig & 32) == 32 ) ||
    ( ( (MMconfig &  8) ==  8 ) && by_rank ) ||
    ( ( (MMconfig &  4) ==  4 ) && by_pt   ) ||
    ( ( (MMconfig &  2) ==  2 ) && by_combi);

} 

int L1MuGMTMerger::doSpecialMerge(unsigned MMconfig) const {
  return (MMconfig & 1) == 1;
}

int L1MuGMTMerger::doANDMerge(unsigned MMconfig) const {
  return (MMconfig & 64) == 64;
}

unsigned L1MuGMTMerger::convertedEta(const L1MuRegionalCand* mu) const {
  L1MuGMTLFEtaConvLUT* etaconv_lut = L1MuGMTConfig::getLFEtaConvLUT();
  return etaconv_lut->SpecificLookup_eta_gmt (mu->type_idx(), mu->eta_packed() );
}

unsigned L1MuGMTMerger::projectedPhi(const L1MuRegionalCand* mu) const {

  // convert eta
  L1MuGMTLFPhiProEtaConvLUT* phiproetaconv_lut = L1MuGMTConfig::getLFPhiProEtaConvLUT();
  unsigned eta4 = phiproetaconv_lut->SpecificLookup_eta_out (mu->type_idx(), mu->eta_packed() );
  
  // look up delta-phi 9 bit signed
  L1MuGMTLFPhiProLUT* phipro_lut = L1MuGMTConfig::getLFPhiProLUT();
  unsigned dphi9 = phipro_lut->SpecificLookup_dphi (mu->type_idx(), eta4, mu->pt_packed(), mu->charge_packed());
  
  // sign extend
  L1MuSignedPacking<9> DPhiPacking;
  int dphi = DPhiPacking.idxFromPacked( dphi9 );
  
  // add modulo 144
  int newphi = mu->phi_packed() + dphi; 
  if (newphi < 0) newphi += 144;
  if (newphi >= 144) newphi -= 144;

  return (unsigned) newphi;  
}


unsigned L1MuGMTMerger::sysign(const L1MuRegionalCand* mu) const {

  unsigned sysign = mu->charge_packed();

  if ( mu->charge_valid_packed() == 0 )
    sysign = 2; // undefined charge

  return sysign;
}

void L1MuGMTMerger::createMergedCand(int idx_dtcsc, int idx_rpc) {

  // In the hardware matrices of select_bits are calculated for all
  // possible pairings. 
  // In ORCA we only calculate selec-bits for the actual 
  // pairs to save time.

  // look up merge ranks
  int merge_rank_dtcsc = merge_rank(dtcsc_mu[idx_dtcsc]);
  int merge_rank_rpc   = merge_rank(rpc_mu[idx_rpc]);

  // calculate select-bits (1: take DT/CSC, 0: take RPC)
  int selected_by_rank = abs(merge_rank_dtcsc) > abs(merge_rank_rpc);
  int selected_by_pt   = dtcsc_mu[idx_dtcsc]->pt_packed() <= rpc_mu[idx_rpc]->pt_packed();

  // Selection by combination of min pt and higher rank
  // select by rank if both flags are set, otherwise by min pt
  // in other words: select by minpt if one flag is not set
  int selected_by_combi = (merge_rank_dtcsc<0 && merge_rank_rpc<0) ? selected_by_rank : selected_by_pt;

  
  L1MuGMTExtendedCand* tmpmuon = new L1MuGMTExtendedCand();
   
  tmpmuon->setBx(dtcsc_mu[idx_dtcsc]->bx());


  // merge phi
  //  unsigned MMConfig_phi = 32; // take DT
  unsigned MMConfig_phi = L1MuGMTConfig::getRegMMConfigPhi()->getValue(m_id);

  unsigned phi = 0;
  
  if (selectDTCSC(MMConfig_phi, selected_by_rank, selected_by_pt, selected_by_combi) )
    phi = projectedPhi( dtcsc_mu[idx_dtcsc] );
  else
    phi = projectedPhi( rpc_mu[idx_rpc] );

  tmpmuon->setPhiPacked( phi );
   
  // merge eta
  unsigned MMConfig_eta = L1MuGMTConfig::getRegMMConfigEta()->getValue(m_id);

  unsigned eta = 0;
  
  if (doSpecialMerge( MMConfig_eta ) ) {
    if ( (m_id == 1) || dtcsc_mu[idx_dtcsc]->finehalo_packed() ) 
      eta = convertedEta( dtcsc_mu[idx_dtcsc] );
    else
      eta = convertedEta( rpc_mu[idx_rpc] );
  } 
  else {
    if ( selectDTCSC(MMConfig_eta, selected_by_rank, selected_by_pt, selected_by_combi) )
      eta = convertedEta( dtcsc_mu[idx_dtcsc] );
    else
      eta = convertedEta( rpc_mu[idx_rpc] );    
  }
  tmpmuon->setEtaPacked( eta );

  // merge pt
  unsigned MMConfig_pt = L1MuGMTConfig::getRegMMConfigPt()->getValue(m_id);

  unsigned pt = 0;

  if (doSpecialMerge( MMConfig_pt ) ) { // mix pt
    L1MuGMTLFPtMixLUT* ptmix_lut = L1MuGMTConfig::getLFPtMixLUT();
    pt = ptmix_lut->SpecificLookup_pt_mixed (m_id, dtcsc_mu[idx_dtcsc]->pt_packed(), 
                                             rpc_mu[idx_rpc]->pt_packed());    
  } 
  else {
    if ( selectDTCSC(MMConfig_pt, selected_by_rank, selected_by_pt, selected_by_combi) )
      pt = dtcsc_mu[idx_dtcsc]->pt_packed();
    else
      pt = rpc_mu[idx_rpc]->pt_packed();
  }
  tmpmuon->setPtPacked( pt );

  // merge charge
  unsigned MMConfig_charge = L1MuGMTConfig::getRegMMConfigCharge()->getValue(m_id);

  unsigned sy_sign = 0;

  if (doSpecialMerge( MMConfig_charge ) ) {
    // based on charge valid bits
    if ( rpc_mu[idx_rpc]->charge_valid_packed()==1 && dtcsc_mu[idx_dtcsc]->charge_valid_packed()==0 )
      sy_sign = sysign(rpc_mu[idx_rpc]);
    else 
      sy_sign = sysign(dtcsc_mu[idx_dtcsc]);
  } 
  else {
    if ( selectDTCSC(MMConfig_charge, selected_by_rank, selected_by_pt, selected_by_combi) )
      sy_sign = sysign(dtcsc_mu[idx_dtcsc]);
    else
      sy_sign = sysign(rpc_mu[idx_rpc]);
  }
  tmpmuon->setChargePacked( sy_sign );
   
  // merge quality  

  
  // merge MIP
  unsigned MMConfig_MIP = L1MuGMTConfig::getRegMMConfigMIP()->getValue(m_id);

  bool mip_bit = false;

  bool mip_bit_dtcsc = m_gmt.MipIsoAU(m_id)->MIP(idx_dtcsc);
  bool mip_bit_rpc   = m_gmt.MipIsoAU(m_id)->MIP(idx_rpc+4);
 
  if ( doSpecialMerge(MMConfig_MIP) ) {
    if ( doANDMerge(MMConfig_MIP) )
      mip_bit =  mip_bit_dtcsc && mip_bit_rpc;
    else
      mip_bit =  mip_bit_dtcsc || mip_bit_rpc;
  }
  else {
    if ( selectDTCSC(MMConfig_MIP, selected_by_rank, selected_by_pt, selected_by_combi) )
      mip_bit = mip_bit_dtcsc;
    else
      mip_bit = mip_bit_rpc;
  }

  tmpmuon->setMIP(mip_bit);
 
  // merge ISO
  unsigned MMConfig_ISO = L1MuGMTConfig::getRegMMConfigISO()->getValue(m_id);

  bool iso_bit = false;

  bool iso_bit_dtcsc = m_gmt.MipIsoAU(m_id)->ISO(idx_dtcsc);
  bool iso_bit_rpc   = m_gmt.MipIsoAU(m_id)->ISO(idx_rpc+4);
 
  if ( doSpecialMerge(MMConfig_ISO) ) {
    if ( doANDMerge(MMConfig_ISO) )
      iso_bit =  iso_bit_dtcsc && iso_bit_rpc;
    else
      iso_bit =  iso_bit_dtcsc || iso_bit_rpc;
  }
  else {
    if ( selectDTCSC(MMConfig_ISO, selected_by_rank, selected_by_pt, selected_by_combi) )
      iso_bit = iso_bit_dtcsc;
    else
      iso_bit = iso_bit_rpc;
  }

  tmpmuon->setIsolation(iso_bit);

  // merge sort rank
  unsigned MMConfig_SRK = L1MuGMTConfig::getRegMMConfigSRK()->getValue(m_id);

  unsigned rank_offset = L1MuGMTConfig::getRegSortRankOffset()->getValue(m_id);

  unsigned rank = 0;
  if ( selectDTCSC(MMConfig_SRK, selected_by_rank, selected_by_pt, selected_by_combi) )
    rank = L1MuGMTSortRankUnit::sort_rank(dtcsc_mu[idx_dtcsc]) + rank_offset; 
  else
    rank = L1MuGMTSortRankUnit::sort_rank(rpc_mu[idx_rpc])     + rank_offset;

  tmpmuon->setRank(rank); 

  // quality of merged candidate
  tmpmuon->setQuality( 7 ); // code for matched muons

  tmpmuon->setDTCSCIndex(idx_dtcsc);
  tmpmuon->setRPCIndex(idx_rpc);
  tmpmuon->setFwdBit( m_id );
  tmpmuon->setRPCBit( 0 );
   
  m_MuonCands.push_back(tmpmuon);
}

// calculate merge rank as in HW

int L1MuGMTMerger::merge_rank(const L1MuRegionalCand* muon) const {

  if ( muon == 0 || muon->empty() ) return 0;

  unsigned lut_idx= muon->type_idx();
  
  // obtain inputs as coded in HW
  unsigned eta = muon->eta_packed();
  unsigned q = muon->quality_packed();
  unsigned pt  = muon->pt_packed();
  unsigned phi = muon->phi_packed();

  // lookup eta-q
  L1MuGMTLFMergeRankEtaQLUT* etaq_lut = L1MuGMTConfig::getLFMergeRankEtaQLUT();
  unsigned rank_etaq = etaq_lut->SpecificLookup_rank_etaq (lut_idx, eta, q);
  unsigned flag = etaq_lut->SpecificLookup_flag (lut_idx, eta, q);
   
  // lookup pt-q
  L1MuGMTLFMergeRankPtQLUT* ptq_lut = L1MuGMTConfig::getLFMergeRankPtQLUT();
  unsigned rank_ptq = ptq_lut->SpecificLookup_rank_ptq (lut_idx, q, pt);
  
  // lookup etaphi
  L1MuGMTLFMergeRankEtaPhiLUT* etaphi_lut = L1MuGMTConfig::getLFMergeRankEtaPhiLUT();
  unsigned rank_etaphi = etaphi_lut->SpecificLookup_rank_etaphi (lut_idx, eta, phi);
  
  // combine
  L1MuGMTLFMergeRankCombineLUT* combine_lut = L1MuGMTConfig::getLFMergeRankCombineLUT();
  unsigned rank = combine_lut->SpecificLookup_merge_rank (lut_idx, rank_etaq, rank_ptq, rank_etaphi);

  int rank_signed = rank;

  if (flag == 1) rank_signed *= -1;
   
  return rank_signed;
}

---