Merge pull request #23033 from jmduarte/double-b-rebased

TensorFlow-based DeepDoubleB tagger integration
cms-sw · Jun 2, 2018 · 7f00bb3 · 7f00bb3
2 parents 714c402 + a3fceea
commit 7f00bb3
Show file tree

Hide file tree

Showing 52 changed files with 1,934 additions and 561 deletions.
diff --git a/DataFormats/BTauReco/interface/BoostedDoubleSVTagInfoFeatures.h b/DataFormats/BTauReco/interface/BoostedDoubleSVTagInfoFeatures.h
@@ -0,0 +1,44 @@
+#ifndef DataFormats_BTauReco_BoostedDoubleSVTagInfoFeatures_h
+#define DataFormats_BTauReco_BoostedDoubleSVTagInfoFeatures_h
+
+namespace btagbtvdeep {
+
+class BoostedDoubleSVTagInfoFeatures {
+
+  // Note: these variables are intended to match the variables defined in DataFormats/BTauReco/interface/TaggingVariable.h
+
+  public:
+    float jetNTracks; // tracks associated to jet
+    float jetNSecondaryVertices; // number of reconstructed possible secondary vertices in jet
+    float trackSip3dSig_0; // 1st largest track 3D signed impact parameter significance
+    float trackSip3dSig_1; // 2nd largest track 3D signed impact parameter significance
+    float trackSip3dSig_2; // 3rd largest track 3D signed impact parameter significance
+    float trackSip3dSig_3; // 4th largest track 3D signed impact parameter significance
+    float tau1_trackSip3dSig_0; // 1st largest track 3D signed impact parameter significance associated to the 1st N-subjettiness axis
+    float tau1_trackSip3dSig_1; // 2nd largest track 3D signed impact parameter significance associated to the 1st N-subjettiness axis
+    float tau2_trackSip3dSig_0; // 1st largest track 3D signed impact parameter significance associated to the 2nd N-subjettiness axis
+    float tau2_trackSip3dSig_1; // 2nd largest track 3D signed impact parameter significance associated to the 2nd N-subjettiness axis
+    float trackSip2dSigAboveBottom_0;// track 2D signed impact parameter significance of 1st track lifting mass above bottom
+    float trackSip2dSigAboveBottom_1;// track 2D signed impact parameter significance of 2nd track lifting mass above bottom
+    float trackSip2dSigAboveCharm; // track 2D signed impact parameter significance of first track lifting mass above charm
+    float tau1_trackEtaRel_0; // 1st smallest track pseudorapidity, relative to the jet axis, associated to the 1st N-subjettiness axis
+    float tau1_trackEtaRel_1; // 2nd smallest track pseudorapidity, relative to the jet axis, associated to the 1st N-subjettiness axis
+    float tau1_trackEtaRel_2; // 3rd smallest track pseudorapidity, relative to the jet axis, associated to the 1st N-subjettiness axis
+    float tau2_trackEtaRel_0; // 1st smallest track pseudorapidity, relative to the jet axis, associated to the 2nd N-subjettiness axis
+    float tau2_trackEtaRel_1; // 2nd smallest track pseudorapidity, relative to the jet axis, associated to the 2nd N-subjettiness axis
+    float tau2_trackEtaRel_2; // 3rd smallest track pseudorapidity, relative to the jet axis, associated to the 2nd N-subjettiness axis
+    float tau1_vertexMass; // mass of track sum at secondary vertex associated to the 1st N-subjettiness axis
+    float tau1_vertexEnergyRatio; // ratio of energy at secondary vertex over total energy associated to the 1st N-subjettiness axis
+    float tau1_flightDistance2dSig; // transverse distance significance between primary and secondary vertex associated to the 1st N-subjettiness axis
+    float tau1_vertexDeltaR; // pseudoangular distance between the 1st N-subjettiness axis and secondary vertex direction
+    float tau2_vertexMass; // mass of track sum at secondary vertex associated to the 2nd N-subjettiness axis
+    float tau2_vertexEnergyRatio; // ratio of energy at secondary vertex over total energy associated to the 2nd N-subjettiness axis
+    float tau2_flightDistance2dSig; // transverse distance significance between primary and secondary vertex associated to the 2nd N-subjettiness axis
+    float tau2_vertexDeltaR; // pseudoangular distance between the 2nd N-subjettiness axis and secondary vertex direction (NOT USED!)
+    float z_ratio; // z ratio
+
+};
+
+}
+
+#endif //DataFormats_BTauReco_BoostedDoubleSVTagInfoFeatures_h
diff --git a/DataFormats/BTauReco/interface/ChargedCandidateFeatures.h b/DataFormats/BTauReco/interface/ChargedCandidateFeatures.h
@@ -13,6 +13,7 @@ class ChargedCandidateFeatures {
 
     float btagPf_trackEtaRel;
     float btagPf_trackPtRel;
+    float btagPf_trackPtRatio;
     float btagPf_trackPPar;
     float btagPf_trackDeltaR;
     float btagPf_trackPParRatio;

diff --git a/DataFormats/BTauReco/interface/DeepDoubleBFeatures.h b/DataFormats/BTauReco/interface/DeepDoubleBFeatures.h
@@ -0,0 +1,29 @@
+#ifndef DataFormats_BTauReco_DeepDoubleBFeatures_h
+#define DataFormats_BTauReco_DeepDoubleBFeatures_h
+
+#include <vector>
+
+#include "DataFormats/BTauReco/interface/JetFeatures.h"
+#include "DataFormats/BTauReco/interface/SecondaryVertexFeatures.h"
+#include "DataFormats/BTauReco/interface/BoostedDoubleSVTagInfoFeatures.h"
+#include "DataFormats/BTauReco/interface/ChargedCandidateFeatures.h"
+
+namespace btagbtvdeep {
+
+class DeepDoubleBFeatures {
+
+  public:
+    JetFeatures jet_features;
+    BoostedDoubleSVTagInfoFeatures tag_info_features;
+
+    std::vector<SecondaryVertexFeatures> sv_features;
+
+    std::vector<ChargedCandidateFeatures> c_pf_features;
+
+    std::size_t npv; // used by deep flavour     
+};    
+
+
+}  
+
+#endif //DataFormats_BTauReco_DeepDoubleBFeatures_h
diff --git a/DataFormats/BTauReco/interface/DeepDoubleBTagInfo.h b/DataFormats/BTauReco/interface/DeepDoubleBTagInfo.h
@@ -0,0 +1,19 @@
+#ifndef DataFormats_BTauReco_DeepDoubleBTagInfo_h
+#define DataFormats_BTauReco_DeepDoubleBTagInfo_h
+
+#include "DataFormats/Common/interface/CMS_CLASS_VERSION.h"
+#include "DataFormats/BTauReco/interface/BaseTagInfo.h"
+#include "DataFormats/BTauReco/interface/DeepFlavourTagInfo.h"
+#include "DataFormats/BTauReco/interface/DeepDoubleBFeatures.h"
+
+#include "DataFormats/PatCandidates/interface/Jet.h"
+
+namespace reco {
+
+typedef  FeaturesTagInfo<btagbtvdeep::DeepDoubleBFeatures> DeepDoubleBTagInfo;
+
+DECLARE_EDM_REFS( DeepDoubleBTagInfo )
+
+}
+
+#endif // DataFormats_BTauReco_DeepDoubleBTagInfo_h
diff --git a/DataFormats/BTauReco/src/classes.h b/DataFormats/BTauReco/src/classes.h
@@ -33,6 +33,7 @@
 #include "DataFormats/BTauReco/interface/SecondaryVertexTagInfo.h"
 #include "DataFormats/BTauReco/interface/CandSecondaryVertexTagInfo.h"
 #include "DataFormats/BTauReco/interface/BoostedDoubleSVTagInfo.h"
+#include "DataFormats/BTauReco/interface/BoostedDoubleSVTagInfoFeatures.h"
 #include "DataFormats/BTauReco/interface/ShallowTagInfo.h"
 #include "DataFormats/BTauReco/interface/SoftLeptonTagInfo.h"
 #include "DataFormats/BTauReco/interface/CandSoftLeptonTagInfo.h"
@@ -53,6 +54,9 @@
 #include "DataFormats/BTauReco/interface/ChargedCandidateFeatures.h"
 #include "DataFormats/BTauReco/interface/DeepFlavourFeatures.h"
 #include "DataFormats/BTauReco/interface/DeepFlavourTagInfo.h"
+#include "DataFormats/BTauReco/interface/DeepDoubleBFeatures.h"
+#include "DataFormats/BTauReco/interface/DeepDoubleBTagInfo.h"
+
 
 namespace reco {
     typedef TrackTauImpactParameterAssociationCollection::map_type          TrackTauImpactParameterAssociationMapType;
@@ -412,5 +416,16 @@ namespace DataFormats_BTauReco {
     reco::DeepFlavourTagInfoRefProd deep_flavour_tag_info_collection_ref_prod;
     reco::DeepFlavourTagInfoRefVector deep_flavour_tag_info_collection_ref_vector;
     edm::Wrapper<reco::DeepFlavourTagInfoCollection> deep_flavour_tag_info_collection_edm_wrapper;
+
+    btagbtvdeep::BoostedDoubleSVTagInfoFeatures boosted_double_sv_tag_info_features;
+    btagbtvdeep::DeepDoubleBFeatures deep_doubleb_features;
+    reco::DeepDoubleBTagInfo deep_doubleb_tag_info;
+    reco::DeepDoubleBTagInfoCollection deep_doubleb_tag_info_collection;
+    reco::DeepDoubleBTagInfoRef deep_doubleb_tag_info_collection_ref;
+    reco::DeepDoubleBTagInfoFwdRef deep_doubleb_tag_info_collection_fwd_ref;
+    reco::DeepDoubleBTagInfoRefProd deep_doubleb_tag_info_collection_ref_prod;
+    reco::DeepDoubleBTagInfoRefVector deep_doubleb_tag_info_collection_ref_vector;
+    edm::Wrapper<reco::DeepDoubleBTagInfoCollection> deep_doubleb_tag_info_collection_edm_wrapper;
+
   };
 }
diff --git a/DataFormats/BTauReco/src/classes_def.xml b/DataFormats/BTauReco/src/classes_def.xml
@@ -70,6 +70,7 @@
   <class name="reco::BoostedDoubleSVTagInfoRefProd"/>
   <class name="reco::BoostedDoubleSVTagInfoRefVector"/>
   <class name="edm::Wrapper<reco::BoostedDoubleSVTagInfoCollection>"/>
+  <class name="edm::Wrapper<std::vector<reco::BoostedDoubleSVTagInfo> >"/>
 
   <class name="reco::ShallowTagInfo" ClassVersion="3">
 		<version ClassVersion="3" checksum="3904302444"/>
@@ -439,12 +440,20 @@
   <class name="btagbtvdeep::NeutralCandidateFeatures" ClassVersion="3">
     <version ClassVersion="3" checksum="938442516"/>
   </class>
-  <class name="btagbtvdeep::ChargedCandidateFeatures" ClassVersion="3">
+  <class name="btagbtvdeep::ChargedCandidateFeatures" ClassVersion="4">
+    <version ClassVersion="4" checksum="2173116472"/>
     <version ClassVersion="3" checksum="1302021141"/>
   </class>
   <class name="btagbtvdeep::DeepFlavourFeatures" ClassVersion="3">
     <version ClassVersion="3" checksum="2487956635"/>
   </class>
+  <class name="btagbtvdeep::DeepDoubleBFeatures" ClassVersion="3">
+    <version ClassVersion="3" checksum="2060552560"/>
+  </class>
+  <class name="btagbtvdeep::BoostedDoubleSVTagInfoFeatures" ClassVersion="3">
+    <version ClassVersion="3" checksum="683471882"/>
+  </class>
+
 
   <class name="reco::DeepFlavourTagInfo"/>
   <class name="reco::DeepFlavourTagInfoCollection"/>
@@ -454,4 +463,12 @@
   <class name="reco::DeepFlavourTagInfoRefVector"/>
   <class name="edm::Wrapper<reco::DeepFlavourTagInfoCollection>"/>
 
+  <class name="reco::DeepDoubleBTagInfo"/>
+  <class name="reco::DeepDoubleBTagInfoCollection"/>
+  <class name="reco::DeepDoubleBTagInfoRef"/>
+  <class name="reco::DeepDoubleBTagInfoFwdRef"/>
+  <class name="reco::DeepDoubleBTagInfoRefProd"/>
+  <class name="reco::DeepDoubleBTagInfoRefVector"/>
+  <class name="edm::Wrapper<reco::DeepDoubleBTagInfoCollection>"/>
+
 </lcgdict>
diff --git a/PhysicsTools/PatAlgos/python/recoLayer0/bTagging_cff.py b/PhysicsTools/PatAlgos/python/recoLayer0/bTagging_cff.py
@@ -41,6 +41,8 @@
     # DeepFlavour tag infos
   , 'pfDeepFlavourTagInfos'
   , 'pfNegativeDeepFlavourTagInfos'
+    # DeepDoubleB tag infos
+  , 'pfDeepDoubleBTagInfos'
   ]
 # extend for "internal use" in PAT/MINIAOD (renaming)
 supportedBtagInfos.append( 'caTopTagInfosPAT' )
@@ -185,6 +187,8 @@
   , 'pfNegativeDeepFlavourJetTags:probc'                            : [["pfNegativeDeepFlavourTagInfos"], ['pfDeepCSVNegativeTagInfos', "pfImpactParameterTagInfos", 'pfInclusiveSecondaryVertexFinderNegativeTagInfos']]
   , 'pfNegativeDeepFlavourJetTags:probuds'                          : [["pfNegativeDeepFlavourTagInfos"], ['pfDeepCSVNegativeTagInfos', "pfImpactParameterTagInfos", 'pfInclusiveSecondaryVertexFinderNegativeTagInfos']]
   , 'pfNegativeDeepFlavourJetTags:probg'                            : [["pfNegativeDeepFlavourTagInfos"], ['pfDeepCSVNegativeTagInfos', "pfImpactParameterTagInfos", 'pfInclusiveSecondaryVertexFinderNegativeTagInfos']]
+  , 'pfDeepDoubleBJetTags:probQ'                            : [["pfDeepDoubleBTagInfos"], ['pfBoostedDoubleSVAK8TagInfos', "pfImpactParameterAK8TagInfos", 'pfInclusiveSecondaryVertexFinderAK8TagInfos']]
+  , 'pfDeepDoubleBJetTags:probH'                            : [["pfDeepDoubleBTagInfos"], ['pfBoostedDoubleSVAK8TagInfos', "pfImpactParameterAK8TagInfos", 'pfInclusiveSecondaryVertexFinderAK8TagInfos']]
 }
 
 #meta-taggers are simple arithmetic on top of other taggers, they are stored here

diff --git a/PhysicsTools/PatAlgos/python/slimming/MicroEventContent_cff.py b/PhysicsTools/PatAlgos/python/slimming/MicroEventContent_cff.py
@@ -14,6 +14,8 @@
         # drop content created by MINIAOD DeepFlavour production
         'drop recoBaseTagInfosOwned_slimmedJets_*_*',
         'keep *_slimmedJetsAK8_*_*',
+        # drop content created by MINIAOD DeepDoubleB production
+        'drop recoBaseTagInfosOwned_slimmedJetsAK8_*_*',
         'keep *_slimmedJetsPuppi_*_*',
         'keep *_slimmedMETs_*_*',
         'keep *_slimmedMETsNoHF_*_*',

diff --git a/PhysicsTools/PatAlgos/python/slimming/applyDeepBtagging_cff.py b/PhysicsTools/PatAlgos/python/slimming/applyDeepBtagging_cff.py
@@ -0,0 +1,76 @@
+import FWCore.ParameterSet.Config as cms
+
+from PhysicsTools.PatAlgos.tools.helpers import getPatAlgosToolsTask, addToProcessAndTask
+
+def applyDeepBtagging( process, postfix="" ) :
+
+    task = getPatAlgosToolsTask(process)
+
+    from PhysicsTools.PatAlgos.tools.jetTools import updateJetCollection
+
+    process.load('PhysicsTools.PatAlgos.slimming.slimmedJets_cfi')
+
+    # update slimmed jets to include DeepFlavour (keep same name)
+    # make clone for DeepFlavour-less slimmed jets, so output name is preserved
+    addToProcessAndTask('slimmedJetsNoDeepFlavour', process.slimmedJets.clone(), process, task)
+    updateJetCollection(
+       process,
+       jetSource = cms.InputTag('slimmedJetsNoDeepFlavour'),
+       # updateJetCollection defaults to MiniAOD inputs but
+       # here it is made explicit (as in training or MINIAOD redoing)
+       pvSource = cms.InputTag('offlineSlimmedPrimaryVertices'),
+       pfCandidates = cms.InputTag('packedPFCandidates'),
+       svSource = cms.InputTag('slimmedSecondaryVertices'),
+       muSource = cms.InputTag('slimmedMuons'),
+       elSource = cms.InputTag('slimmedElectrons'),
+       jetCorrections = ('AK4PFchs', cms.vstring(['L1FastJet', 'L2Relative', 'L3Absolute']), 'None'),
+       btagDiscriminators = [
+          'pfDeepFlavourJetTags:probb',
+          'pfDeepFlavourJetTags:probbb',
+          'pfDeepFlavourJetTags:problepb',
+          'pfDeepFlavourJetTags:probc',
+          'pfDeepFlavourJetTags:probuds',
+          'pfDeepFlavourJetTags:probg',
+       ],
+       postfix = 'SlimmedDeepFlavour'+postfix,
+       printWarning = False
+    )
+
+    # slimmedJets with DeepFlavour (remove DeepFlavour-less)
+    delattr(process, 'slimmedJets')
+    addToProcessAndTask('slimmedJets', getattr(process,'selectedUpdatedPatJetsSlimmedDeepFlavour'+postfix).clone(), process, task)
+    # delete module not used anymore (slimmedJets substitutes)
+    delattr(process, 'selectedUpdatedPatJetsSlimmedDeepFlavour'+postfix)
+
+
+
+    # update slimmed jets to include DeepFlavour (keep same name)
+    # make clone for DeepDoubleB-less slimmed AK8 jets, so output name is preserved
+    addToProcessAndTask('slimmedJetsAK8NoDeepDoubleB', process.slimmedJetsAK8.clone(), process, task)
+    updateJetCollection(
+       process,
+       jetSource = cms.InputTag('slimmedJetsAK8NoDeepDoubleB'),
+       # updateJetCollection defaults to MiniAOD inputs but
+       # here it is made explicit (as in training or MINIAOD redoing)
+       pvSource = cms.InputTag('offlineSlimmedPrimaryVertices'),
+       pfCandidates = cms.InputTag('packedPFCandidates'),
+       svSource = cms.InputTag('slimmedSecondaryVertices'),
+       muSource = cms.InputTag('slimmedMuons'),
+       elSource = cms.InputTag('slimmedElectrons'),
+       rParam = 0.8,
+       jetCorrections = ('AK8PFPuppi', cms.vstring(['L1FastJet', 'L2Relative', 'L3Absolute']), 'None'),
+       btagDiscriminators = [
+          'pfDeepDoubleBJetTags:probQ',
+          'pfDeepDoubleBJetTags:probH',
+       ],
+       postfix = 'SlimmedAK8DeepDoubleB'+postfix,
+       printWarning = False
+    )
+
+    # slimmedJetsAK8 with DeepDoubleB (remove DeepDoubleB-less)
+    delattr(process, 'slimmedJetsAK8')
+    addToProcessAndTask('slimmedJetsAK8', getattr(process,'selectedUpdatedPatJetsSlimmedAK8DeepDoubleB'+postfix).clone(), process, task)
+    # delete module not used anymore (slimmedJetsAK8 substitutes)
+    delattr(process, 'selectedUpdatedPatJetsSlimmedAK8DeepDoubleB'+postfix)
+
+
diff --git a/PhysicsTools/PatAlgos/python/slimming/miniAOD_tools.py b/PhysicsTools/PatAlgos/python/slimming/miniAOD_tools.py
@@ -390,50 +390,15 @@ def miniAOD_customizeCommon(process):
 
     process.selectedPatJetsPuppi.cut = cms.string("pt > 15")
 
-    process.load('PhysicsTools.PatAlgos.slimming.slimmedJets_cfi')
-
-    # update slimmed jets to include DeepFlavour (keep same name)
-    from PhysicsTools.PatAlgos.tools.jetTools import updateJetCollection
-    # make clone for DeepFlavour-less slimmed jets, so output name is preserved
-    process.slimmedJetsNoDeepFlavour = process.slimmedJets.clone()
-    task.add(process.slimmedJetsNoDeepFlavour)
-    updateJetCollection(
-       process,
-       jetSource = cms.InputTag('slimmedJetsNoDeepFlavour'),
-       # updateJetCollection defaults to MiniAOD inputs but
-       # here it is made explicit (as in training or MINIAOD redoing)
-       pvSource = cms.InputTag('offlineSlimmedPrimaryVertices'),
-       pfCandidates = cms.InputTag('packedPFCandidates'),
-       svSource = cms.InputTag('slimmedSecondaryVertices'),
-       muSource = cms.InputTag('slimmedMuons'),
-       elSource = cms.InputTag('slimmedElectrons'),
-       jetCorrections = ('AK4PFchs', cms.vstring(['L1FastJet', 'L2Relative', 'L3Absolute']), 'None'),
-       btagDiscriminators = [
-          'pfDeepFlavourJetTags:probb',
-          'pfDeepFlavourJetTags:probbb',
-          'pfDeepFlavourJetTags:problepb',
-          'pfDeepFlavourJetTags:probc',
-          'pfDeepFlavourJetTags:probuds',
-          'pfDeepFlavourJetTags:probg',
-       ],
-       postfix = 'SlimmedDeepFlavour',
-       printWarning = False
-    )
-
-    # slimmedJets with DeepFlavour (remove DeepFlavour-less)
-    delattr(process, 'slimmedJets')
-    process.slimmedJets = process.selectedUpdatedPatJetsSlimmedDeepFlavour.clone()
-    # delete module not used anymore (slimmedJets substitutes)
-    delattr(process, 'selectedUpdatedPatJetsSlimmedDeepFlavour')
-
-    task.add(process.slimmedJets)
-    task.add(process.slimmedJetsAK8)
+    from PhysicsTools.PatAlgos.slimming.applyDeepBtagging_cff import applyDeepBtagging
+    applyDeepBtagging( process )
 
     addToProcessAndTask('slimmedJetsPuppiNoMultiplicities', process.slimmedJetsNoDeepFlavour.clone(), process, task)
     process.slimmedJetsPuppiNoMultiplicities.src = cms.InputTag("selectedPatJetsPuppi")
     process.slimmedJetsPuppiNoMultiplicities.packedPFCandidates = cms.InputTag("packedPFCandidates")
 
     from PhysicsTools.PatAlgos.patPuppiJetSpecificProducer_cfi import patPuppiJetSpecificProducer
+    from PhysicsTools.PatAlgos.tools.jetTools import updateJetCollection
     process.patPuppiJetSpecificProducer = patPuppiJetSpecificProducer.clone(
       src=cms.InputTag("slimmedJetsPuppiNoMultiplicities"),
     )

diff --git a/PhysicsTools/PatAlgos/python/tools/jetTools.py b/PhysicsTools/PatAlgos/python/tools/jetTools.py
@@ -619,6 +619,16 @@ def setupBTagging(process, jetSource, pfCandidates, explicitJTA, pvSource, svSou
                                       vertex_associator = vertex_associator,
                                       ),
                                     process, task)
+
+            if btagInfo == 'pfDeepDoubleBTagInfos':
+                addToProcessAndTask(btagPrefix+btagInfo+labelName+postfix,
+                                    btag.pfDeepDoubleBTagInfos.clone(
+                                      jets = jetSource,
+                                      vertices=pvSource,
+                                      secondary_vertices=svSource,
+                                      shallow_tag_infos = cms.InputTag(btagPrefix+'pfBoostedDoubleSVAK8TagInfos'+labelName+postfix),
+                                      ),
+                                    process, task)
 
             acceptedTagInfos.append(btagInfo)
         elif hasattr(toptag, btagInfo) :

diff --git a/RecoBTag/Configuration/python/RecoBTag_cff.py b/RecoBTag/Configuration/python/RecoBTag_cff.py
@@ -7,7 +7,7 @@
 from RecoBTag.Combined.combinedMVA_cff import *
 from RecoBTag.CTagging.RecoCTagging_cff import *
 from RecoBTag.Combined.deepFlavour_cff import *
-from RecoBTag.DeepFlavour.pfDeepFlavour_cff import *
+from RecoBTag.TensorFlow.pfDeepFlavour_cff import *
 from RecoVertex.AdaptiveVertexFinder.inclusiveVertexing_cff import *
 
 legacyBTaggingTask = cms.Task(