Merge pull request #46009 from nurfikri89/from1420pre1_btvChangePtEtaSel_AddpuppiVM

cmsbuild · web-flow · commit e9399399d9cf · 2024-09-18T15:06:17.000+02:00
[BTV] Disable pT cut for AK4 Puppi ParticleNet and UParT taggers inference
diff --git a/PhysicsTools/PatAlgos/python/tools/jetTools.py b/PhysicsTools/PatAlgos/python/tools/jetTools.py
@@ -407,6 +407,11 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
                             process, task)
 
 
+    # Setup the PUPPI ValueMap that will consumed by the TagInfo producers.
+    puppi_value_map = "puppi"
+    if pfCandidates.value() == 'packedPFCandidates':
+        puppi_value_map = setupPuppiForPackedPF(process)[0]
+
     acceptedTagInfos = list()
     for btagInfo in requiredTagInfos:
         if hasattr(btag,btagInfo):
@@ -637,14 +642,9 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
                                     process, task)
 
 
-
             if 'pfBoostedDouble' in btagInfo or 'SecondaryVertex' in btagInfo:
-              _btagInfo = getattr(process, btagPrefix+btagInfo+labelName+postfix)
-              if pfCandidates.value() == 'packedPFCandidates':
-                packedPFPuppiLabel = setupPuppiForPackedPF(process)[0]
-                _btagInfo.weights = cms.InputTag(packedPFPuppiLabel)
-              else:
-                _btagInfo.weights = cms.InputTag("puppi")
+                _btagInfo = getattr(process, btagPrefix+btagInfo+labelName+postfix)
+                _btagInfo.weights = cms.InputTag(puppi_value_map)
 
             if 'DeepFlavourTagInfos' in btagInfo:
                 svUsed = svSource
@@ -659,10 +659,8 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
                 # use right input tags when running with RECO PF candidates, which actually
                 # depends of whether jets use "particleFlow"
                 if pfCandidates.value() == 'packedPFCandidates':
-                    puppi_value_map = setupPuppiForPackedPF(process)[0]
                     vertex_associator = cms.InputTag("")
                 else:
-                    puppi_value_map = cms.InputTag("puppi")
                     vertex_associator = cms.InputTag("primaryVertexAssociation","original")
 
                 # If this jet is a puppi jet, then set is_weighted_jet to true.
@@ -680,7 +678,7 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
                                       is_weighted_jet = is_weighted_jet,
                                       flip = flip),
                                     process, task)
-            
+
             if ('ParticleTransformerAK4TagInfos' in btagInfo) and ('UnifiedParticleTransformerAK4TagInfos' not in btagInfo): #We also have to veto UParT is we select ParT
                 svUsed = svSource
                 if btagInfo == 'pfNegativeParticleTransformerAK4TagInfos':
@@ -690,10 +688,8 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
                 # use right input tags when running with RECO PF candidates, which actually
                 # depends of whether jets use "particleFlow"
                 if pfCandidates.value() == 'packedPFCandidates':
-                    puppi_value_map = setupPuppiForPackedPF(process)[0]
                     vertex_associator = cms.InputTag("")
                 else:
-                    puppi_value_map = cms.InputTag("puppi")
                     vertex_associator = cms.InputTag("primaryVertexAssociation","original")
 
                 # If this jet is a puppi jet, then set is_weighted_jet to true.
@@ -722,10 +718,8 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
                 # use right input tags when running with RECO PF candidates, which actually
                 # depends of whether jets use "particleFlow"
                 if pfCandidates.value() == 'packedPFCandidates':
-                    puppi_value_map = setupPuppiForPackedPF(process)[0]
                     vertex_associator = cms.InputTag("")
                 else:
-                    puppi_value_map = cms.InputTag("puppi")
                     vertex_associator = cms.InputTag("primaryVertexAssociation","original")
 
                 # If this jet is a puppi jet, then set is_weighted_jet to true.
@@ -747,8 +741,6 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
                 # can only run on PAT jets, so the updater needs to be used
                 if 'updated' not in jetSource.value().lower():
                     raise ValueError("Invalid jet collection: %s. pfDeepDoubleXTagInfos only supports running via updateJetCollection." % jetSource.value())
-                packedPFPuppiLabel = setupPuppiForPackedPF(process)[0]
-                puppi_value_map = cms.InputTag(packedPFPuppiLabel)
                 addToProcessAndTask(btagPrefix+btagInfo+labelName+postfix,
                                     btag.pfDeepDoubleXTagInfos.clone(
                                       jets = jetSource,
@@ -759,8 +751,6 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
                                       ),
                                     process, task)
             if btagInfo == 'pfHiggsInteractionNetTagInfos':
-                packedPFPuppiLabel = setupPuppiForPackedPF(process)[0]
-                puppi_value_map = cms.InputTag(packedPFPuppiLabel)
                 addToProcessAndTask(btagPrefix+btagInfo+labelName+postfix,
                                     btag.pfHiggsInteractionNetTagInfos.clone(
                                       jets = jetSource,
@@ -776,14 +766,12 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
                     # case 1: running over jets whose daughters are PackedCandidates (only via updateJetCollection for now)
                     if 'updated' not in jetSource.value().lower():
                         raise ValueError("Invalid jet collection: %s. pfDeepBoostedJetTagInfos only supports running via updateJetCollection." % jetSource.value())
-                    puppi_value_map = setupPuppiForPackedPF(process)[0]
                     vertex_associator = ""
                 elif pfCandidates.value() == 'particleFlow':
                     raise ValueError("Running pfDeepBoostedJetTagInfos with reco::PFCandidates is currently not supported.")
                     # case 2: running on new jet collection whose daughters are PFCandidates (e.g., cluster jets in RECO/AOD)
                     # daughters are the particles used in jet clustering, so already scaled by their puppi weights
                     # Uncomment the lines below after running pfDeepBoostedJetTagInfos with reco::PFCandidates becomes supported
-#                     puppi_value_map = "puppi"
 #                     vertex_associator = "primaryVertexAssociation:original"
                 else:
                     raise ValueError("Invalid pfCandidates collection: %s." % pfCandidates.value())
@@ -801,12 +789,10 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
             if btagInfo == 'pfParticleNetTagInfos':
                 if pfCandidates.value() == 'packedPFCandidates':
                     # case 1: running over jets whose daughters are PackedCandidates (only via updateJetCollection for now)
-                    puppi_value_map = setupPuppiForPackedPF(process)[0]
                     vertex_associator = ""
                 elif pfCandidates.value() == 'particleFlow':
                     raise ValueError("Running pfDeepBoostedJetTagInfos with reco::PFCandidates is currently not supported.")
                     # case 2: running on new jet collection whose daughters are PFCandidates (e.g., cluster jets in RECO/AOD)
-                    puppi_value_map = "puppi"
                     vertex_associator = "primaryVertexAssociation:original"
                 else:
                     raise ValueError("Invalid pfCandidates collection: %s." % pfCandidates.value())
@@ -833,12 +819,10 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
                     sip3dSigMax = -1
                 if pfCandidates.value() == 'packedPFCandidates':
                     # case 1: running over jets whose daughters are PackedCandidates (only via updateJetCollection for now)
-                    puppi_value_map = setupPuppiForPackedPF(process)[0]
                     vertex_associator = ""
                 elif pfCandidates.value() == 'particleFlow':
                     raise ValueError("Running pfDeepBoostedJetTagInfos with reco::PFCandidates is currently not supported.")
                     # case 2: running on new jet collection whose daughters are PFCandidates (e.g., cluster jets in RECO/AOD)
-                    puppi_value_map = "puppi"
                     vertex_associator = "primaryVertexAssociation:original"
                 else:
                     raise ValueError("Invalid pfCandidates collection: %s." % pfCandidates.value())
@@ -877,6 +861,7 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
                                   vertices = pvSource,
                                   secondary_vertices = svUsed,
                                   pf_candidates = pfCandidates,
+                                  puppi_value_map = puppi_value_map,
                                   flip_ip_sign = flip_ip_sign,
                                   max_sip3dsig_for_flip = max_sip3dsig_for_flip,
                                   ),
@@ -892,12 +877,13 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
                                   vertices = pvSource,
                                   secondary_vertices = svSource,
                                   pf_candidates = pfCandidates,
+                                  puppi_value_map = puppi_value_map
                                   ),
                                 process, task)
             acceptedTagInfos.append(btagInfo)
-        elif btagInfo == 'pfParticleNetFromMiniAODAK4CHSCentralTagInfos' or btagInfo == 'pfNegativeParticleNetFromMiniAODAK4PuppiForwardTagInfos':
+        elif btagInfo == 'pfParticleNetFromMiniAODAK4CHSCentralTagInfos' or btagInfo == 'pfNegativeParticleNetFromMiniAODAK4CHSCentralTagInfos':
             # ParticleNetFromMiniAOD cannot be run on RECO inputs, so need a workaround
-            if btagInfo == 'pfNegativeParticleNetFromMiniAODAK4PuppiForwardTagInfos':
+            if btagInfo == 'pfNegativeParticleNetFromMiniAODAK4CHSCentralTagInfos':
                 svUsed, flip_ip_sign, max_sip3dsig_for_flip = cms.InputTag(btagPrefix+'inclusiveCandidateNegativeSecondaryVertices'+labelName+postfix), True, 10.
             else:
                 svUsed, flip_ip_sign, max_sip3dsig_for_flip = svSource, False, -1.
@@ -909,6 +895,7 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
                                   vertices = pvSource,
                                   secondary_vertices = svUsed,
                                   pf_candidates = pfCandidates,
+                                  puppi_value_map = puppi_value_map,
                                   flip_ip_sign = flip_ip_sign,
                                   max_sip3dsig_for_flip = max_sip3dsig_for_flip,
                                   ),
@@ -924,6 +911,7 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
                                   vertices = pvSource,
                                   secondary_vertices = svSource,
                                   pf_candidates = pfCandidates,
+                                  puppi_value_map = puppi_value_map,
                                   ),
                                 process, task)
             acceptedTagInfos.append(btagInfo)
@@ -937,14 +925,15 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
                                   vertices = pvSource,
                                   secondary_vertices = svSource,
                                   pf_candidates = pfCandidates,
+                                  puppi_value_map = puppi_value_map,
                                   ),
                                 process, task)
             acceptedTagInfos.append(btagInfo)
         else:
             print('  --> %s ignored, since not available via RecoBTag.Configuration.RecoBTag_cff!'%(btagInfo))
     # setup all required btagDiscriminators
     acceptedBtagDiscriminators = list()
-    for discriminator_name in btagDiscriminators :			
+    for discriminator_name in btagDiscriminators :
         btagDiscr = discriminator_name.split(':')[0] #split input tag to get the producer label
         #print discriminator_name, '-->', btagDiscr
         newDiscr = btagPrefix+btagDiscr+labelName+postfix #new discriminator name
diff --git a/RecoBTag/FeatureTools/plugins/ParticleNetFeatureEvaluator.cc b/RecoBTag/FeatureTools/plugins/ParticleNetFeatureEvaluator.cc
@@ -64,6 +64,8 @@ class ParticleNetFeatureEvaluator : public edm::stream::EDProducer<> {
   const double max_eta_for_taus_;
   const bool include_neutrals_;
   const bool flip_ip_sign_;
+  const bool fallback_puppi_weight_;
+  bool use_puppi_value_map_;
   const double max_sip3dsig_for_flip_;
 
   edm::EDGetTokenT<pat::MuonCollection> muon_token_;
@@ -76,11 +78,13 @@ class ParticleNetFeatureEvaluator : public edm::stream::EDProducer<> {
   edm::EDGetTokenT<reco::VertexCompositePtrCandidateCollection> sv_token_;
   edm::EDGetTokenT<edm::View<reco::Candidate>> pfcand_token_;
   edm::ESGetToken<TransientTrackBuilder, TransientTrackRecord> track_builder_token_;
+  edm::EDGetTokenT<edm::ValueMap<float>> puppi_value_map_token_;
 
   edm::Handle<reco::VertexCollection> vtxs_;
   edm::Handle<reco::VertexCompositePtrCandidateCollection> svs_;
   edm::Handle<edm::View<reco::Candidate>> pfcands_;
   edm::Handle<pat::PackedCandidateCollection> losttracks_;
+  edm::Handle<edm::ValueMap<float>> puppi_value_map_;
   edm::ESHandle<TransientTrackBuilder> track_builder_;
 
   const static std::vector<std::string> particle_features_;
@@ -264,6 +268,8 @@ ParticleNetFeatureEvaluator::ParticleNetFeatureEvaluator(const edm::ParameterSet
       max_eta_for_taus_(iConfig.getParameter<double>("max_eta_for_taus")),
       include_neutrals_(iConfig.getParameter<bool>("include_neutrals")),
       flip_ip_sign_(iConfig.getParameter<bool>("flip_ip_sign")),
+      fallback_puppi_weight_(iConfig.getParameter<bool>("fallback_puppi_weight")),
+      use_puppi_value_map_(false),
       max_sip3dsig_for_flip_(iConfig.getParameter<double>("max_sip3dsig_for_flip")),
       muon_token_(consumes<pat::MuonCollection>(iConfig.getParameter<edm::InputTag>("muons"))),
       electron_token_(consumes<pat::ElectronCollection>(iConfig.getParameter<edm::InputTag>("electrons"))),
@@ -277,6 +283,11 @@ ParticleNetFeatureEvaluator::ParticleNetFeatureEvaluator(const edm::ParameterSet
       pfcand_token_(consumes<edm::View<reco::Candidate>>(iConfig.getParameter<edm::InputTag>("pf_candidates"))),
       track_builder_token_(
           esConsumes<TransientTrackBuilder, TransientTrackRecord>(edm::ESInputTag("", "TransientTrackBuilder"))) {
+  const auto &puppi_value_map_tag = iConfig.getParameter<edm::InputTag>("puppi_value_map");
+  if (!puppi_value_map_tag.label().empty()) {
+    puppi_value_map_token_ = consumes<edm::ValueMap<float>>(puppi_value_map_tag);
+    use_puppi_value_map_ = true;
+  }
   produces<std::vector<reco::DeepBoostedJetTagInfo>>();
 }
 
@@ -297,10 +308,12 @@ void ParticleNetFeatureEvaluator::fillDescriptions(edm::ConfigurationDescription
   desc.add<double>("max_eta_for_taus", 2.5);
   desc.add<bool>("include_neutrals", true);
   desc.add<bool>("flip_ip_sign", false);
+  desc.add<bool>("fallback_puppi_weight", false);
   desc.add<double>("max_sip3dsig_for_flip", 99999);
   desc.add<edm::InputTag>("vertices", edm::InputTag("offlineSlimmedPrimaryVertices"));
   desc.add<edm::InputTag>("secondary_vertices", edm::InputTag("slimmedSecondaryVertices"));
   desc.add<edm::InputTag>("pf_candidates", edm::InputTag("packedPFCandidates"));
+  desc.add<edm::InputTag>("puppi_value_map", edm::InputTag("puppi"));
   desc.add<edm::InputTag>("losttracks", edm::InputTag("lostTracks"));
   desc.add<edm::InputTag>("jets", edm::InputTag("slimmedJetsAK8"));
   desc.add<edm::InputTag>("muons", edm::InputTag("slimmedMuons"));
@@ -325,6 +338,11 @@ void ParticleNetFeatureEvaluator::produce(edm::Event &iEvent, const edm::EventSe
   auto photons = iEvent.getHandle(photon_token_);
   // Input lost tracks
   iEvent.getByToken(losttrack_token_, losttracks_);
+  // Get puuppi value map
+  if (use_puppi_value_map_) {
+    iEvent.getByToken(puppi_value_map_token_, puppi_value_map_);
+  }
+
   // Primary vertexes
   iEvent.getByToken(vtx_token_, vtxs_);
   if (vtxs_->empty()) {
@@ -413,6 +431,9 @@ void ParticleNetFeatureEvaluator::fillParticleFeatures(DeepBoostedJetFeatures &f
   // track builder
   TrackInfoBuilder trackinfo(track_builder_);
 
+  //Save puppi weight for selected constituents in this map
+  std::map<const pat::PackedCandidate *, float> map_pc2puppiweight;
+
   // make list of pf-candidates to be considered
   std::vector<const pat::PackedCandidate *> daughters;
   for (const auto &dau : jet.daughterPtrVector()) {
@@ -428,6 +449,15 @@ void ParticleNetFeatureEvaluator::fillParticleFeatures(DeepBoostedJetFeatures &f
       continue;
     // filling daughters
     daughters.push_back(cand);
+
+    float puppiw = 1.;
+    if (use_puppi_value_map_) {
+      puppiw = (*puppi_value_map_)[dau];
+    } else if (!fallback_puppi_weight_) {
+      throw edm::Exception(edm::errors::InvalidReference, "PUPPI value map missing")
+          << "use fallback_puppi_weight option to use " << puppiw << " for cand as default";
+    }
+    map_pc2puppiweight[cand] = puppiw;
   }
 
   // sort by original pt (not Puppi-weighted)
@@ -486,7 +516,7 @@ void ParticleNetFeatureEvaluator::fillParticleFeatures(DeepBoostedJetFeatures &f
     fts.fill("jet_pfcand_frompv", cand->fromPV());
     fts.fill("jet_pfcand_dz", ip_sign * (std::isnan(cand->dz(pv_ass_pos)) ? 0 : cand->dz(pv_ass_pos)));
     fts.fill("jet_pfcand_dxy", ip_sign * (std::isnan(cand->dxy(pv_ass_pos)) ? 0 : cand->dxy(pv_ass_pos)));
-    fts.fill("jet_pfcand_puppiw", cand->puppiWeight());
+    fts.fill("jet_pfcand_puppiw", map_pc2puppiweight[cand]);
     fts.fill("jet_pfcand_nlostinnerhits", cand->lostInnerHits());
     fts.fill("jet_pfcand_nhits", cand->numberOfHits());
     fts.fill("jet_pfcand_npixhits", cand->numberOfPixelHits());
diff --git a/RecoBTag/FeatureTools/plugins/UnifiedParticleTransformerAK4TagInfoProducer.cc b/RecoBTag/FeatureTools/plugins/UnifiedParticleTransformerAK4TagInfoProducer.cc
@@ -171,7 +171,7 @@ void UnifiedParticleTransformerAK4TagInfoProducer::fillDescriptions(edm::Configu
   desc.add<bool>("fallback_puppi_weight", false);
   desc.add<bool>("fallback_vertex_association", false);
   desc.add<bool>("is_weighted_jet", false);
-  desc.add<double>("min_jet_pt", 15.0);
+  desc.add<double>("min_jet_pt", 0.0);
   desc.add<double>("max_jet_eta", 2.5);
   descriptions.add("pfUnifiedParticleTransformerAK4TagInfos", desc);
 }
@@ -225,10 +225,10 @@ void UnifiedParticleTransformerAK4TagInfoProducer::produce(edm::Event& iEvent, c
 
     // reco jet reference (use as much as possible)
     const auto& jet = jets->at(jet_n);
-    if (jet.pt() < 15.0) {
+    if (jet.pt() < min_jet_pt_) {
       features.is_filled = false;
     }
-    if (std::abs(jet.eta()) > 2.5) {
+    if (std::abs(jet.eta()) > max_jet_eta_) {
       features.is_filled = false;
     }
     // dynamical casting to pointers, null if not possible
diff --git a/RecoBTag/ONNXRuntime/python/pfParticleNetFromMiniAODAK4_cff.py b/RecoBTag/ONNXRuntime/python/pfParticleNetFromMiniAODAK4_cff.py
@@ -26,17 +26,17 @@
 pfParticleNetFromMiniAODAK4PuppiCentralTagInfos = ParticleNetFeatureEvaluator.clone(
     jets = "slimmedJetsPuppi",
     jet_radius = 0.4,
-    min_jet_pt = 15,
+    min_jet_pt = 0.,
     min_jet_eta = 0.,
     max_jet_eta = 2.5,
 )
 
 pfParticleNetFromMiniAODAK4PuppiForwardTagInfos = ParticleNetFeatureEvaluator.clone(
     jets = "slimmedJetsPuppi",
     jet_radius = 0.4,
-    min_jet_pt = 15,
+    min_jet_pt = 0.,
     min_jet_eta = 2.5,
-    max_jet_eta = 4.7,
+    max_jet_eta = 5.0,
 )
 
 
