cross sections, make names unique

Tools/interface/splitPandaExpress.h

@@ -0,0 +1,11 @@
+#include <TROOT.h>
+#include <TFile.h>
+#include <TTree.h>
+#include <TKey.h>
+#include <TString.h>
+#include <iostream>
+#include <string>
+#include <cassert>
+
+void splitPandaExpress(std::string inputFile, Long64_t nEvtsPerFile);
+

Tools/python/processes/SM.py

@@ -72,7 +72,7 @@
         'W2JetsToLNu_LHEWpT_50-150_TuneCP5_13TeV-amcnloFXFX-pythia8':('W2JetsToLNu_WpT50to150_CP5','MC',1275),
         'W2JetsToLNu_LHEWpT_100-150_TuneCP5_13TeV-amcnloFXFX-pythia8':('W2JetsToLNu_WpT100to150_CP5','MC',303.7),
         'W2JetsToLNu_LHEWpT_150-250_TuneCP5_13TeV-amcnloFXFX-pythia8':('W2JetsToLNu_WpT150to250_CP5','MC',106.4),
-        'W2JetsToLNu_LHEWpT_250-400_TuneCP5_13TeV-amcnloFXFX-pythia8':('W2JetsToLNu_WpT250to400_CP5','MC',10.98),
+        'W2JetsToLNu_LHEWpT_250-400_TuneCP5_13TeV-amcnloFXFX-pythia8':('W2JetsToLNu_WpT250to400_CP5','MC',17.93),
         'W2JetsToLNu_LHEWpT_400-inf_TuneCP5_13TeV-amcnloFXFX-pythia8':('W2JetsToLNu_WpT400toinf_CP5','MC',3.19),
 
         # NLO Z->ll
@@ -160,20 +160,16 @@
         #'TT_TuneCUETP8M2T4_13TeV-powheg-isrup-pythia8':                     ('TTbar_PowhegISRUp','MC',831.76),
         #'TT_TuneCUETP8M2T4down_13TeV-powheg-pythia8':                       ('TTbar_PowhegTuneDown','MC',831.76),
         #'TT_TuneCUETP8M2T4up_13TeV-powheg-pythia8':                         ('TTbar_PowhegTuneUp','MC',831.76),
-        'TTTo2L2Nu_TuneCUETP8M2_ttHtranche3_13TeV-powheg-pythia8':('TTTo2L2Nu','MC',88.288),
+        'TTTo2L2Nu_TuneCUETP8M2_ttHtranche3_13TeV-powheg-pythia8':('TTTo2L2Nu','MC',88.667),
         'TTGJets_TuneCUETP8M1_13TeV-amcatnloFXFX-madspin-pythia8':('TTGJets_TuneCUETP8M1_13TeV-amcatnloFXFX-madspin-pythia8','MC',1.444*3.2),
         'TTZToLLNuNu_M-10_TuneCUETP8M1_13TeV-amcatnlo-pythia8':('TTZToLLNuNu_M-10_TuneCUETP8M1_13TeV-amcatnlo-pythia8','MC',0.2529),
         'TTZToQQ_TuneCUETP8M1_13TeV-amcatnlo-pythia8':('TTZToQQ_TuneCUETP8M1_13TeV-amcatnlo-pythia8','MC',0.5297),
         'TTWJetsToLNu_TuneCUETP8M1_13TeV-amcatnloFXFX-madspin-pythia8':('TTWJetsToLNu_TuneCUETP8M1_13TeV-amcatnloFXFX-madspin-pythia8','MC',0.2043),
         'TTWJetsToQQ_TuneCUETP8M1_13TeV-amcatnloFXFX-madspin-pythia8':('TTWJetsToQQ_TuneCUETP8M1_13TeV-amcatnloFXFX-madspin-pythia8','MC',0.4062),
 
-        'TTTo2L2Nu_TuneCP5_13TeV-powheg-pythia8':('TTTo2L2Nu_CP5','MC',88.288),
-        'TTTo2L2Nu_TuneCP5_PSweights_13TeV-powheg-pythia8':('TTTo2L2Nu_CP5_PSweights','MC',88.288),
-        'TTToSemiLeptonic_TuneCP5_13TeV-powheg-pythia8':('TTToSemiLeptonic_CP5','MC',1),
-        'TTToSemiLeptonic_TuneCP5_PSweights_13TeV-powheg-pythia8':('TTToSemiLeptonic_CP5_PSweights','MC',1),
-        'TTToHadronic_TuneCP5_13TeV-powheg-pythia8':('TTToHadronic_CP5','MC',1),
-        'TTToHadronic_TuneCP5_PSweights_13TeV-powheg-pythia8':('TTToHadronic_CP5_PSweights','MC',1),
-
+        'TTTo2L2Nu_TuneCP5_13TeV-powheg-pythia8':('TTTo2L2Nu_CP5','MC',88.667),
+        'TTToSemiLeptonic_TuneCP5_13TeV-powheg-pythia8':('TTToSemiLeptonic_CP5','MC',365.80), #300.9 in MCM
+        'TTToHadronic_TuneCP5_13TeV-powheg-pythia8':('TTToHadronic_CP5','MC',377.29), #313.9 in MCM
 
         # exotic top
         'tZq_ll_4f_13TeV-amcatnlo-pythia8':('SingleTop_tZll','MC',0.0758),
@@ -197,6 +193,8 @@
 
         # fancy dibosons
         'WWTo2L2Nu_13TeV-powheg':('WWTo2L2Nu','MC',(118.7-3.974)*0.1086*0.1086*9), #12.178
+        'WWTo2L2Nu_NNPDF31_TuneCP5_PSweights_13TeV-powheg-pythia8':('WWTo2L2Nu_CP5','MC',(118.7-3.974)*0.1086*0.1086*9), #12.178
+        'WWTo1L1Nu2Q_13TeV_amcatnloFXFX_madspin_pythia8':('WWTo1L1Nu2Q_AMC','MC',40.58),
         'WWTo4Q_13TeV-powheg':('WWTo4Q','MC',51.723),
         'WWToLNuQQ_13TeV-powheg':('WWToLNuQQ','MC',49.997),
         'WZTo1L1Nu2Q_13TeV_amcatnloFXFX_madspin_pythia8':('WZTo1L1Nu2Q','MC',10.71),
@@ -234,8 +232,8 @@
         'ZH_HToBB_ZToLL_M125_13TeV_powheg_pythia8':('ZllHbb_mH125','MC',0.04865),
         'ggZH_HToBB_ZToNuNu_M125_13TeV_powheg_pythia8':('ggZvvHbb_mH125','MC',0.014366),
         'ggZH_HToBB_ZToLL_M125_13TeV_powheg_pythia8':('ggZllHbb_mH125','MC',0.007842),
-        'WminusH_HToBB_WToLNu_M125_13TeV_powheg_pythia8':('WmLNuHbb','MC',0.100),
-        'WplusH_HToBB_WToLNu_M125_13TeV_powheg_pythia8':('WpLNuHbb','MC',0.159),
+        'WminusH_HToBB_WToLNu_M125_13TeV_powheg_pythia8':('WmLNuHbb_mH125','MC',0.100),
+        'WplusH_HToBB_WToLNu_M125_13TeV_powheg_pythia8':('WpLNuHbb_mH125','MC',0.159),
         'WminusH_HToBB_WToLNu_M125_13TeV_powheg_herwigpp':('WmLNuHbb_herwigpp','MC',0.100),
         'WplusH_HToBB_WToLNu_M125_13TeV_powheg_herwigpp':('WpLNuHbb_herwigpp','MC',0.159),
         'WminusH_HToBB_WToLNu_M120_13TeV_powheg_pythia8':('WmLNuHbb_mH120','MC',0.100),
@@ -250,14 +248,14 @@
         'ZH_HToBB_ZToQQ_M125_13TeV_powheg_pythia8':('ZQQHbb_mH125','MC',0.04865 * 20.767 / 3), 
 
         # B enriched V+jets samples
-        'DYBJetsToLL_M-50_TuneCUETP8M1_13TeV-madgraphMLM-pythia8':('ZJets_bQuarks','MC',71.77),
+        'DYBJetsToLL_M-50_TuneCUETP8M1_13TeV-madgraphMLM-pythia8':('ZJets_bQuarks_incl','MC',71.77),
         'DYBJetsToLL_M-50_Zpt-100to200_TuneCUETP8M1_13TeV-madgraphMLM-pythia8':('ZJets_bQuarks_pt100to200','MC',3.027),
         'DYBJetsToLL_M-50_Zpt-200toInf_TuneCUETP8M1_13TeV-madgraphMLM-pythia8':('ZJets_bQuarks_pt200toinf','MC',0.297),
         'WBJetsToLNu_Wpt-100to200_TuneCUETP8M1_13TeV-madgraphMLM-pythia8':('WJets_bQuarks_pt100to200','MC',6.004),
         'WBJetsToLNu_Wpt-200toInf_TuneCUETP8M1_13TeV-madgraphMLM-pythia8':('WJets_bQuarks_pt200toinf','MC',0.8524),
         'ZBJetsToNuNu_Zpt-100to200_TuneCUETP8M1_13TeV-madgraphMLM-pythia8':('ZtoNuNu_bQuarks_pt100to200','MC',6.03),
         'ZBJetsToNuNu_Zpt-200toInf_TuneCUETP8M1_13TeV-madgraphMLM-pythia8':('ZtoNuNu_bQuarks_pt200toinf','MC',0.5946),
-        'DYJetsToLL_BGenFilter_M-50_TuneCUETP8M1_13TeV-madgraphMLM-pythia8':('ZJets_bHadrons','MC',229.7),
+        'DYJetsToLL_BGenFilter_M-50_TuneCUETP8M1_13TeV-madgraphMLM-pythia8':('ZJets_bHadrons_incl','MC',229.7),
         'DYJetsToLL_BGenFilter_Zpt-100to200_M-50_TuneCUETP8M1_13TeV-madgraphMLM-pythia8':('ZJets_bHadrons_pt100to200','MC',3.506),
         'DYJetsToLL_BGenFilter_Zpt-200toInf_M-50_TuneCUETP8M1_13TeV-madgraphMLM-pythia8':('ZJets_bHadrons_pt200toinf','MC',0.5083),
         'WJetsToLNu_BGenFilter_Wpt-100to200_TuneCUETP8M1_13TeV-madgraphMLM-pythia8':('WJets_bHadrons_pt100to200','MC',26.1),

Tools/src/LinkDef.h

@@ -6,6 +6,7 @@
 #include "PandaCore/Tools/interface/Cutter.h"
 #include "PandaCore/Tools/interface/BranchAdder.h"
 #include "PandaCore/Tools/interface/EventSyncher.h"
+#include "PandaCore/Tools/interface/splitPandaExpress.h"
 
 #ifdef __CLING__
 #pragma link off all globals;
@@ -24,6 +25,7 @@
 #pragma link C++ class EventSyncher;
 #pragma link C++ class BranchAdder;
 #pragma link C++ class FormulaBranchAdder;
+#pragma link C++ class HBranchAdder;
 #pragma link C++ class H1BranchAdder;
 #pragma link C++ class H2BranchAdder;
 #pragma link C++ class ProgressReporter;
@@ -41,5 +43,6 @@
 #pragma link C++ function DeltaR2;
 #pragma link C++ function ExpErf;
 #pragma link C++ function insertion_sort;
+#pragma link C++ function splitPandaExpress;
 
 #endif 

Tools/src/splitPandaExpress.C

@@ -0,0 +1,59 @@
+#include "../interface/splitPandaExpress.h"
+void splitPandaExpress(std::string inputFile, Long64_t nEvtsPerFile) {
+  printf("Opening file \"%s\" and splitting it into files with %lld events...\n", inputFile.c_str(), nEvtsPerFile);
+  TFile *f = TFile::Open(inputFile.c_str(),"update"); assert(f);
+  TTree *events=0, *treebuffer;
+
+  // Drop extra cycle numbers so rooteventselector can do its job
+  printf("\tLooking for extra cycle numbers to drop...\n");
+  TIter keyList(f->GetListOfKeys());
+  std::vector<short> extraCycleNumbers; short highestCycleNumber=-1;
+  TKey *key;
+  while ((key = (TKey*)keyList())) {
+    TClass *cl = gROOT->GetClass(key->GetClassName());
+    if (!cl->InheritsFrom("TTree")) continue;
+    treebuffer=(TTree*)key->ReadObj();
+    if(strcmp(treebuffer->GetName(),"events")!=0) continue;
+    short cycleNumber=key->GetCycle();
+    printf("\t%s;%d\n",treebuffer->GetName(),cycleNumber);
+    if(cycleNumber > highestCycleNumber) {
+      // replace the highest cycle number, putting the previous highest in the list of extras if real
+      if(highestCycleNumber>0) extraCycleNumbers.push_back(highestCycleNumber);
+      highestCycleNumber=cycleNumber;
+    } else {
+      // this is an extra cycle number
+      extraCycleNumbers.push_back(cycleNumber);
+    }
+  }
+  std::cout << "Extra cycle numbers: ";
+  for (std::vector<short>::const_iterator i = extraCycleNumbers.begin(); i != extraCycleNumbers.end(); ++i)
+    std::cout << *i << ", ";
+  std::cout << std::endl;
+  for (std::vector<short>::const_iterator i = extraCycleNumbers.begin(); i != extraCycleNumbers.end(); ++i)
+    f->Delete(Form("events;%d",*i));
+  // Done dropping extra cycle numbers
+
+
+  events = (TTree*)f->Get(Form("events;%d",highestCycleNumber)); assert(events);
+  Long64_t nEntries = events->GetEntries();
+  unsigned nSplit = ceil( float(nEntries) / nEvtsPerFile);
+  f->Close();
+  printf("The file had %lld entries, splitting into %d subfiles\n", nEntries, nSplit);
+  size_t lastDot = inputFile.find_last_of(".");
+  size_t lastSlash = inputFile.find_last_of("/");
+  std::string splitDir = inputFile.substr(0, lastSlash) + std::string("/split/");
+  std::string rawName = splitDir + inputFile.substr(lastSlash+1, lastDot-lastSlash-1);
+  system(Form("mkdir -p %s",splitDir.c_str()));
+  for(unsigned i=0; i<nSplit; i++) {
+    std::string splitName;
+    if(lastDot!=std::string::npos) splitName = rawName + std::string(Form("_%d.root", i));
+    else                           splitName = rawName + std::string(Form("_%d", i));
+    Long64_t firstEvt = i*nEvtsPerFile;
+    Long64_t lastEvt  = (i+1)*nEvtsPerFile - 1;
+    const char * command = Form("rooteventselector --recreate -f %lld -l %lld %s:events %s", firstEvt, lastEvt, inputFile.c_str(), splitName.c_str());
+    printf("\tSubfile #%d: writing event # %lld-%lld to \"%s\"\n", i+1, firstEvt, lastEvt, splitName.c_str());
+    system(command);
+    //printf("\t%s\n",command);
+  }
+  printf("Done splitting file \"%s\"\n", inputFile.c_str());
+}

Utils/python/load.py

@@ -24,6 +24,7 @@ def __init__(self,name,objects,deps=[]):
                                                     'Cutter',
                                                     'BranchAdder',
                                                     'EventSyncher',
+                                                    'splitPandaExpress'
                                                   ]
            ),
     Library(name='PandaCoreLearning',   objects = [ 'TMVATrainer',