Merge pull request #258 from gpetruc/heppyOptimizations

Heppy optimizations and an improved time report

PhysicsTools/Heppy/python/analyzers/core/autovars.py

@@ -24,6 +24,8 @@ def makeBranch(self,treeNumpy,isMC):
     def fillBranch(self,treeNumpy,object,isMC):
         if self.mcOnly and not isMC: return
         treeNumpy.fill(self.name, self(object))
+    def __repr__(self):
+        return "<NTupleVariable[%s]>" % self.name
 
 class NTupleObjectType:
     def __init__(self,name,baseObjectTypes=[],mcOnly=[],variables=[]):
@@ -60,6 +62,8 @@ def allBases(self):
         return ret
     def removeVariable(self,name):
         self.variables = [ v for v in self.variables if v.name != name]
+    def __repr__(self):
+        return "<NTupleObjectType[%s]>" % self.name
 
 class NTupleObject:
     def __init__(self, name, objectType, help="", mcOnly=False):
@@ -79,6 +83,8 @@ def fillBranches(self,treeNumpy,object,isMC):
         allvars = self.objectType.allVars(isMC)
         for v in allvars:
             treeNumpy.fill("%s_%s" % (self.name, v.name), v(object))
+    def __repr__(self):
+        return "<NTupleObject[%s]>" % self.name
 
 
 class NTupleCollection:
@@ -137,5 +143,7 @@ def fillBranchesVector(self,treeNumpy,collection,isMC):
         for v in allvars:
             name="%s_%s" % (self.name, v.name) if v.name != "" else self.name
             treeNumpy.vfill(name, [ v(collection[i]) for i in xrange(num) ])
+    def __repr__(self):
+        return "<NTupleCollection[%s]>" % self.name
 
 

PhysicsTools/Heppy/python/analyzers/gen/GeneratorAnalyzer.py

@@ -64,7 +64,9 @@ def makeMCInfo(self, event):
         verbose = getattr(self.cfg_ana, 'verbose', False)
         rawGenParticles = self.mchandles['genParticles'].product() 
         good = []; keymap = {};
+        allGenParticles = []
         for rawIndex,p in enumerate(rawGenParticles):
+            if self.makeAllGenParticles: allGenParticles.append(p)
             id     = abs(p.pdgId())
             status = p.status()
             # particles must be status > 2, except for prompt leptons, photons, neutralinos
@@ -134,15 +136,15 @@ def makeMCInfo(self, event):
                         print "Error keying %d: motherIndex %d, ancestor.pdgId %d, good[gp.motherIndex].pdgId() %d " % (igp, gp.motherIndex, ancestor.pdgId(),  good[gp.motherIndex].pdgId())
                     break
                 ancestor = None if ancestor.numberOfMothers() == 0 else ancestor.motherRef(0)
-            if abs(gp.pdgId()) not in {1,2,3,4,5,11,12,13,14,15,16,21}:
+            if abs(gp.pdgId()) not in [1,2,3,4,5,11,12,13,14,15,16,21]:
                 gp.sourceId = gp.pdgId()
             if gp.motherIndex != -1:
                 ancestor = good[gp.motherIndex]
                 if ancestor.sourceId != 99 and (ancestor.mass() > gp.mass() or gp.sourceId == 99):
                     gp.sourceId = ancestor.sourceId
         event.generatorSummary = good
         # add the ID of the mother to be able to recreate last decay chains
-        for ip,p in enumerate(event.generatorSummary):
+        for ip,p in enumerate(good):
             moms = realGenMothers(p)
             if len(moms)==0:
                 p.motherId = 0
@@ -164,15 +166,8 @@ def makeMCInfo(self, event):
             print "\n\n"
 
         if self.makeAllGenParticles:
-            event.genParticles = []
-            for rawIndex,p in enumerate(rawGenParticles):
-                if rawIndex in keymap:
-                    gp = event.generatorSummary[keymap[rawIndex]]
-                else:
-                    gp = p
-                    gp.rawIndex = rawIndex 
-                    gp.genSummaryIndex = -1
-                event.genParticles.append(gp)
+            event.genParticles = allGenParticles
+
         if self.makeSplittedGenLists:
             event.genHiggsBosons = []
             event.genVBosons     = []
@@ -225,11 +220,11 @@ def process(self, event):
 setattr(GeneratorAnalyzer,"defaultConfig",
     cfg.Analyzer(GeneratorAnalyzer,
         # BSM particles that can appear with status <= 2 and should be kept
-        stableBSMParticleIds = { 1000022 }, 
+        stableBSMParticleIds = [ 1000022 ], 
         # Particles of which we want to save the pre-FSR momentum (a la status 3).
         # Note that for quarks and gluons the post-FSR doesn't make sense,
         # so those should always be in the list
-        savePreFSRParticleIds = { 1,2,3,4,5, 11,12,13,14,15,16, 21 },
+        savePreFSRParticleIds = [ 1,2,3,4,5, 11,12,13,14,15,16, 21 ],
         # Make also the list of all genParticles, for other analyzers to handle
         makeAllGenParticles = True,
         # Make also the splitted lists

PhysicsTools/HeppyCore/python/framework/looper.py

@@ -89,14 +89,7 @@ def __init__( self, name,
         if len(self.cfg_comp.files)==0:
             errmsg = 'please provide at least an input file in the files attribute of this component\n' + str(self.cfg_comp)
             raise ValueError( errmsg )
-	if hasattr(config,"preprocessor") and config.preprocessor is not None :
-		self.cfg_comp = config.preprocessor.run(self.cfg_comp,self.outDir,firstEvent,nEvents)
-	if hasattr(self.cfg_comp,"options"):
-                print self.cfg_comp.files,self.cfg_comp.options
-	        self.events = config.events_class(self.cfg_comp.files, tree_name,options=self.cfg_comp.options)
-	else :
-	        self.events = config.events_class(self.cfg_comp.files, tree_name)
-
+        self.events = config.events_class(self.cfg_comp.files, tree_name)
         if hasattr(self.cfg_comp, 'fineSplit'):
             fineSplitIndex, fineSplitFactor = self.cfg_comp.fineSplit
             if fineSplitFactor > 1:
@@ -192,10 +185,19 @@ def loop(self):
             allev = max([x['events'] for x in self.timeReport])
             warning = self.logger.warning
             warning("\n      ---- TimeReport (all times in ms; first evt is skipped) ---- ")
-            warning("%9s   %9s    %9s   %9s   %s" % ("processed","all evts","time/proc", " time/all", "analyer"))
-            warning("%9s   %9s    %9s   %9s   %s" % ("---------","--------","---------", "---------", "-------------"))
+            warning("%9s   %9s    %9s   %9s %6s   %s" % ("processed","all evts","time/proc", " time/all", "  [%] ", "analyer"))
+            warning("%9s   %9s    %9s   %9s %6s   %s" % ("---------","--------","---------", "---------", " -----", "-------------"))
+            sumtime = sum(rep['time'] for rep in self.timeReport)
+            passev  = self.timeReport[-1]['events']
             for ana,rep in zip(self.analyzers,self.timeReport):
-                warning( "%9d   %9d   %10.2f  %10.2f   %s" % ( rep['events'], allev, 1000*rep['time']/(rep['events']-1) if rep['events']>1 else 0, 1000*rep['time']/(allev-1) if allev > 1 else 0, ana.name))
+                timePerProcEv = rep['time']/(rep['events']-1) if rep['events'] > 1 else 0
+                timePerAllEv  = rep['time']/(allev-1)         if allev > 1         else 0
+                fracAllEv     = rep['time']/sumtime
+                warning( "%9d   %9d   %10.2f  %10.2f %5.1f%%   %s" % ( rep['events'], allev, 1000*timePerProcEv, 1000*timePerAllEv, 100.0*fracAllEv, ana.name))
+            totPerProcEv = sumtime/(passev-1) if passev > 1 else 0
+            totPerAllEv  = sumtime/(allev-1)  if allev > 1  else 0
+            warning("%9s   %9s    %9s   %9s   %s" % ("---------","--------","---------", "---------", "-------------"))
+            warning("%9d   %9d   %10.2f  %10.2f %5.1f%%   %s" % ( passev, allev, 1000*totPerProcEv, 1000*totPerAllEv, 100.0, "TOTAL"))
             warning("")
 
     def process(self, iEv ):
@@ -232,7 +234,6 @@ def write(self):
         self.setup.close() 
 
 
-
 if __name__ == '__main__':
 
     import pickle

PhysicsTools/HeppyCore/python/utils/deltar.py

@@ -49,7 +49,10 @@ def matchObjectCollection3 ( objects, matchCollection, deltaRMax = 0.3, filter =
             return dict( zip(objects, [None]*len(objects)) )
     # build all possible combinations
 
-    allPairs = [(deltaR2 (object.eta(), object.phi(), match.eta(), match.phi()), (object, match)) for object in objects for match in matchCollection if filter(object,match) ]
+    objectCoords = [ (o.eta(),o.phi(),o) for o in objects ]
+    matchdCoords = [ (o.eta(),o.phi(),o) for o in matchCollection ]
+    allPairs = [(deltaR2 (oeta, ophi, meta, mphi), (object, match)) for (oeta,ophi,object) in objectCoords for (meta,mphi,match) in matchdCoords if abs(oeta-meta)<=deltaRMax and filter(object,match) ]
+    #allPairs = [(deltaR2 (object.eta(), object.phi(), match.eta(), match.phi()), (object, match)) for object in objects for match in matchCollection if filter(object,match) ]
     allPairs.sort ()
     #
     # to flag already matched objects