Merge pull request #7 from Sam-Harper/GSIssueNoThrow

stopping throws on duplicate superclusters in the event & other debugging info

RecoEcal/EgammaClusterProducers/python/gsFixedRefinedSuperClusters_cfi.py → RecoEcal/EgammaClusterProducers/python/particleFlowEGammaGSFixed_cfi.py

@@ -1,8 +1,11 @@
 import FWCore.ParameterSet.Config as cms
 
-gsFixedRefinedSuperClusters = cms.EDProducer("EGRefinedSCFixer",
+particleFlowEGammaGSFixed = cms.EDProducer("EGRefinedSCFixer",
     orgRefinedSC = cms.InputTag("particleFlowEGamma", '', cms.InputTag.skipCurrentProcess()),
     orgSC = cms.InputTag("particleFlowSuperClusterECAL", '', cms.InputTag.skipCurrentProcess()),
     fixedSC = cms.InputTag("particleFlowSuperClusterECALGSFixed"),
     fixedPFClusters = cms.InputTag("particleFlowClusterECALGSFixed"),
+    throwOnDupECALClustersInEvent = cms.bool(False),
+    throwOnDupESClustersInEvent = cms.bool(True)
+
 )

RecoEcal/EgammaClusterProducers/src/EGRefinedSCFixer.cc

@@ -19,13 +19,11 @@
 #include <unordered_set>
 #include <iostream>
 
-//work in progress (emergancy fix!)
-//issue: we need to re-make the refined superclusters
-//we only really need to do this for barrel super with a gain switch
-//but usual to remake without gain switch for debugging
-//however it is barrel only
-//everything else is copied or relinked to produce the same set of outputs
-//as particleFlowEGamma
+
+//issue: we need to re-make the refined superclusters in the AOD
+//problem: to do this properly, we need the outer position of the track, not stored in AOD
+//         so we cant re-run PFEGammaProducer which would normally do this
+//solution: we fudge it in the following way
 
 //how it works:
 //  a refined supercluster will have subclusters added or removed w.r.t to its parent SC
@@ -35,14 +33,18 @@
 //  we id clusters via seed crystal, this shouldnt change for any cluster without a gain switch, 
 //  its still a local maximum
 //  
+//  it is important to know that for the digis to be saved in miniAOD, it must be within the 3x3
+//  of a hybrid (or multi 5x5 in endcap) supercluster seed crystal. The hybrid supercluster seed
+//  crystals should map well to the PF supercluster seed crystal, particularly for isolated high 
+//  energy e/gamma objects
+//
 //  for matching the new fixed superclusters vs old superclusters, the seed crystal may gain
 //  as the gain switched crystal will now have a larger energy
 //  but it really should be in the 3x3 of the orginal SC (the gain switch crysal has to be here to be redone)
 //  so we do a dIR = dIEta^2 + dIPhi^2 match of <=2 (ie be in the 3x3)
 //  but take the smallest dIR
 
 //  issues: sub cluster ordering may not be correct (its sorted by decreasing energy)
-//  issues: when it assigns a sub cluster to a refined SC, it doesnt remove it from others
 
 class EGRefinedSCFixer : public edm::stream::EDProducer<> {
 public:
@@ -104,14 +106,23 @@ class EGRefinedSCFixer : public edm::stream::EDProducer<> {
   edm::EDGetTokenT<reco::PFClusterCollection> fixedPFClustersToken_;
 
   // output instance name of EB/EE BC collection
-  std::string ebeeClustersCollection_;
+  const std::string ebeeClustersCollection_;
   // output instance name of ES BC collection
-  std::string esClustersCollection_;
+  const std::string esClustersCollection_;
+
+  //if this is true, throw on  duplicate clusters entering the event
+  //if false, we flags the event and move on
+  const bool throwOnDupECALClustersInEvent_;
+  const bool throwOnDupESClustersInEvent_;
+
 };
 
 EGRefinedSCFixer::EGRefinedSCFixer(const edm::ParameterSet& iConfig) :
   ebeeClustersCollection_("EBEEClusters"),
-  esClustersCollection_("ESClusters")
+  esClustersCollection_("ESClusters"),
+  throwOnDupECALClustersInEvent_(iConfig.getParameter<bool>("throwOnDupECALClustersInEvent")),
+  throwOnDupESClustersInEvent_(iConfig.getParameter<bool>("throwOnDupESClustersInEvent"))
+
 {
   getToken(orgRefinedSCToken_, iConfig, "orgRefinedSC");
   getToken(orgBCToken_, iConfig, "orgRefinedSC", "EBEEClusters");
@@ -132,6 +143,9 @@ EGRefinedSCFixer::EGRefinedSCFixer(const edm::ParameterSet& iConfig) :
   produces<SCRefMap>();
   produces<SCRefMap>("parentSCsEB");
   produces<SCRefMap>("parentSCsEE");
+  //flags to tell us if we have duplicate clusters
+  produces<bool>("dupECALClusters");
+  produces<bool>("dupESClusters");
 }
 
 namespace {
@@ -145,38 +159,6 @@ namespace {
   }
 }
 
-namespace{
-  class SCPrinter {
-    const reco::SuperCluster& sc_;
-  public:
-    SCPrinter(const reco::SuperCluster& sc):sc_(sc){}
-    std::ostream& operator()(std::ostream& out)const{
-      out <<"E "<<sc_.energy()<<" raw E "<<sc_.rawEnergy()<<" eta "<<sc_.eta()<<" phi "<<sc_.phi()<<" seedId "<<sc_.seed()->seed().rawId()<<" nrclus "<<sc_.clustersSize();
-      return out;
-    }
-
-  };
-  // std::ostream& operator<<(std::ostream& out,const SCPrinter& obj){return obj(out);}
-
-}
-
-namespace{
-  class SCDeepPrinter {
-    const reco::SuperCluster& sc_;
-  public:
-    SCDeepPrinter(const reco::SuperCluster& sc):sc_(sc){}
-    std::ostream& operator()(std::ostream& out)const{
-      out <<"E "<<sc_.energy()<<" raw E "<<sc_.rawEnergy()<<" eta "<<sc_.eta()<<" phi "<<sc_.phi()<<" seedId "<<sc_.seed()->seed().rawId()<<" nrclus "<<sc_.clustersSize() <<std::endl;
-      for(auto& clus : sc_.clusters()){
-	out <<" clus E "<<clus->energy()<<" Ecorr "<<clus->correctedEnergy()<<" seed "<<clus->seed().rawId()<<" nrHits "<<clus->hitsAndFractions().size()<<std::endl;
-      }
-      return out;
-    }
-  };
-  //  std::ostream& operator<<(std::ostream& out,const SCDeepPrinter& obj){return obj(out);}
-
-} 
-
 void EGRefinedSCFixer::produce(edm::Event & iEvent, const edm::EventSetup & iSetup)
 {
   auto const& orgRefinedSCs = getHandle(iEvent, orgRefinedSCToken_);
@@ -193,6 +175,9 @@ void EGRefinedSCFixer::produce(edm::Event & iEvent, const edm::EventSetup & iSet
   // EB clusters are fixed, EE are direct translation from the original
   auto fixedBCs = std::make_unique<reco::CaloClusterCollection>();
   // direct translation of the original ES collection - there is nothing "fixed" about this
+  // however they may be slightly different as we are reclustering from reduced rec-hit collections
+  // this will mainly effect non-photon or non-electron superclusters as those superclusters
+  // save less rec-hits surrounding them
   auto fixedESs = std::make_unique<reco::CaloClusterCollection>();
   auto fixedConvs = std::make_unique<reco::ConversionCollection>();
 
@@ -224,31 +209,26 @@ void EGRefinedSCFixer::produce(edm::Event & iEvent, const edm::EventSetup & iSet
       // particleFlowEGamma can create superclusters directly out of PFClusters too
       // -> there is not always a matching EB SC
       if (orgEBSC.isNonnull()) {
-	//changing the matching to be in  3x3 now (had issues with suprious matches)
         auto fixedEBSC(GainSwitchTools::matchSCBySeedCrys(*orgEBSC, fixedEBSCs, 1, 1));
-
-        // here we may genuinely miss a mapping, if the seed position moves too much by re-reconstruction
-	// Sam: its very unlikely, if not impossible. To be replaced the gain switched crystal must be within +/-1 crystal a hybrid supercluster 
-	// seed crystal because of ecalSelectedDigis. 
-	// You could only get a shift larger than 1 if you had two supercluster seed crystals very close together and even then I'm not sure its possible. 
         if (fixedEBSC.isNonnull()) {
           mappedSCsEB[fixedEBSC.key()] = orgEBSC;
 
           auto fixedRefinedSC(makeFixedRefinedBarrelSC(orgRefinedSC, *orgEBSC, *fixedEBSC, fixedPFClusters,clusterAddedToAnyEBSC));
           fixedRefinedSCs->push_back(fixedRefinedSC);
 
-
           continue;
         }
       }
     }
     else {
       auto orgEESC(GainSwitchTools::matchSCBySeedCrys(orgRefinedSC, orgEESCs));
       if (orgEESC.isNonnull()) {
-        // there is nothing "fixed" here - two clusters are identical
+	//there is nothing "fixed" here - two clusters are in theory identical but
+	//there could be mild differences given we are clustering from the 
+	//reduced rec-hit collecitons 
+	//for example this can lead to small energy changes leading a cluster to be 
+	//now below the 4 GeV Et threshold to make a supercluster and therefore disappear	 
         auto fixedEESC(GainSwitchTools::matchSCBySeedCrys(*orgEESC, fixedEESCs));
-
-        // fixedEESC has to be nonnull
 	if(fixedEESC.isNonnull()) mappedSCsEE[fixedEESC.key()] = orgEESC;
       }
     }
@@ -275,6 +255,8 @@ void EGRefinedSCFixer::produce(edm::Event & iEvent, const edm::EventSetup & iSet
   std::map<reco::CaloClusterPtr, unsigned int> pfClusterMapEBEE; //maps of pfclusters to caloclusters
   std::map<reco::CaloClusterPtr, unsigned int> pfClusterMapES;
 
+  bool duplicateECALClusters=false;
+  bool duplicateESClusters=false; 
   for (auto& sc : *fixedRefinedSCs) {
     // The cluster ref in fixed EB and the other superclusters point to different collections (former to gs-fixed, latter to original)
     // but we are copying the basic clusters by value and remaking yet another collection here -> no need to distinguish
@@ -284,9 +266,13 @@ void EGRefinedSCFixer::produce(edm::Event & iEvent, const edm::EventSetup & iSet
         pfClusterMapEBEE[ptr] = fixedBCs->size();
         fixedBCs->emplace_back(*ptr);
       }
-      else
-        throw cms::Exception("EGRefinedSCFixer::produce")
-          << "Found an EB/EE pfcluster matched to more than one supercluster!";
+      else{
+	duplicateECALClusters=true;
+	if(throwOnDupECALClustersInEvent_){
+	  throw cms::Exception("EGRefinedSCFixer::produce")
+	    << "Found an EB/EE pfcluster matched to more than one supercluster!";
+	}
+      }
     }
 
     for (auto&& cItr = sc.preshowerClustersBegin(); cItr!=sc.preshowerClustersEnd(); ++cItr) {
@@ -295,12 +281,18 @@ void EGRefinedSCFixer::produce(edm::Event & iEvent, const edm::EventSetup & iSet
         pfClusterMapES[ptr] = fixedESs->size();
         fixedESs->emplace_back(*ptr);
       }
-      else
-        throw cms::Exception("PFEgammaProducer::produce")
-          << "Found an ES pfcluster matched to more than one supercluster!";
+      else{
+	duplicateESClusters=true;
+	if(throwOnDupESClustersInEvent_){  
+	  throw cms::Exception("PFEgammaProducer::produce")
+	    << "Found an ES pfcluster matched to more than one supercluster!";
+	}
+      }
     }
   }
-
+  iEvent.put(std::make_unique<bool>(duplicateECALClusters),"dupECALClusters");
+  iEvent.put(std::make_unique<bool>(duplicateESClusters),"dupESClusters");
+
   //put calocluster output collections in event and get orphan handles to create ptrs
   auto caloClusHandleEBEE(iEvent.put(std::move(fixedBCs), ebeeClustersCollection_));
   auto caloClusHandleES(iEvent.put(std::move(fixedESs), esClustersCollection_));

RecoEgamma/EgammaElectronProducers/python/gsFixedGsfElectronCores_cfi.py → RecoEgamma/EgammaElectronProducers/python/gedGsfElectronCoresGSFixed_cfi.py

@@ -1,7 +1,7 @@
 import FWCore.ParameterSet.Config as cms
 
-gsFixedGsfElectronCores = cms.EDProducer("GsfElectronCoreGSCrysFixer",
+gedGsfElectronCoresGSFixed = cms.EDProducer("GsfElectronCoreGSCrysFixer",
     orgCores=cms.InputTag("gedGsfElectronCores", '', cms.InputTag.skipCurrentProcess()),
-    refinedSCs = cms.InputTag('gsFixedRefinedSuperClusters'),
+    refinedSCs = cms.InputTag('particleFlowEGammaGSFixed'),
     scs = cms.InputTag('particleFlowSuperClusterECALGSFixed'),
 )

RecoEgamma/EgammaElectronProducers/python/gsFixedGsfElectrons_cfi.py → RecoEgamma/EgammaElectronProducers/python/gedGsfElectronsGSFixed_cfi.py

@@ -1,8 +1,8 @@
 import FWCore.ParameterSet.Config as cms
 from RecoEgamma.EgammaTools.regressionModifier_cfi import regressionModifier
 
-gsFixedGsfElectrons = cms.EDProducer("GsfElectronGSCrysFixer",
-    newCores=cms.InputTag("gsFixedGsfElectronCores"),
+gedGsfElectronsGSFixed = cms.EDProducer("GsfElectronGSCrysFixer",
+    newCores=cms.InputTag("gedGsfElectronCoresGSFixed"),
     oldEles=cms.InputTag("gedGsfElectrons", '', cms.InputTag.skipCurrentProcess()),
     ebRecHits=cms.InputTag("ecalMultiAndGSGlobalRecHitEB"),
     regressionConfig = regressionModifier.clone(rhoCollection=cms.InputTag("fixedGridRhoFastjetAllTmp")),

RecoEgamma/EgammaPhotonProducers/python/gedPhotonCoreGSFixed_cfi.py

@@ -0,0 +1,9 @@
+import FWCore.ParameterSet.Config as cms
+
+gedPhotonCoreGSFixed = cms.EDProducer("GEDPhotonCoreGSCrysFixer",
+    photonCores = cms.InputTag("gedPhotonCore", '', cms.InputTag.skipCurrentProcess()),
+    refinedSCs = cms.InputTag('particleFlowEGammaGSFixed'),
+    scs = cms.InputTag('particleFlowSuperClusterECALGSFixed'),
+    conversions = cms.InputTag('allConversions'),
+    singleconversions = cms.InputTag('particleFlowEGamma')
+)

RecoEgamma/EgammaPhotonProducers/python/gsFixedGEDPhotons_cfi.py → RecoEgamma/EgammaPhotonProducers/python/gedPhotonsGSFixed_cfi.py

@@ -2,9 +2,9 @@
 
 from RecoEgamma.EgammaTools.regressionModifier_cfi import *
 
-gsFixedGEDPhotons = cms.EDProducer('GEDPhotonGSCrysFixer',
+gedPhotonsGSFixed = cms.EDProducer('GEDPhotonGSCrysFixer',
     photons = cms.InputTag('gedPhotons', '', cms.InputTag.skipCurrentProcess()),
-    newCores = cms.InputTag('gsFixedGEDPhotonCores'),
+    newCores = cms.InputTag('gedPhotonCoreGSFixed'),
     barrelEcalHits=cms.InputTag("ecalMultiAndGSGlobalRecHitEB"),
     primaryVertexProducer = cms.InputTag('offlinePrimaryVerticesWithBS'),
     # rest for regression