Merge pull request #397 from danielwinterbottom/2022_ditau_UROPs

2022 ditau uro ps

Analysis/HiggsTauTauRun2/scripts/HiggsTauTauPlot.py

@@ -1573,7 +1573,7 @@
 
 if options.era in ['22']:
 
-    ztt_samples = ['DYJetsToLL-LO_summer_preEE','DYJetsToLL-LO_postEE']
+    ztt_samples = ['DYJetsToLL-LO_preEE','DYJetsToLL-LO_postEE']
     top_samples = ['TTTo2L2Nu_preEE','TTto4Q_preEE','TTtoLNu2Q_preEE','TTTo2L2Nu_postEE','TTto4Q_postEE','TTtoLNu2Q_postEE']
     vv_samples = ['WW_preEE','WZ_preEE','ZZ_preEE','TWminusto2L2Nu_preEE',
     'TWminustoLNu2Q_preEE','TbarWplusto2L2Nu_preEE',

Analysis/HiggsTauTauRun2/scripts/TnP.py

@@ -974,15 +974,23 @@ def FitWorkspace(name,infile,outfile,sig_model='DoubleVCorr',bkg_model='Exponent
     elif options.era == 'UL_18':
         data_samples = ['SingleMuonA','SingleMuonB','SingleMuonC','SingleMuonD']
     elif options.era == '22_preEE':
-        data_samples = ['SingleMuonB_rereco_preEE','SingleMuonC_rereco_preEE','MuonC_rereco_preEE','MuonD_rereco_preEE'] # check this as Endre's code did not have single muon for Run C but I think it exists 
+        data_samples = ['SingleMuonC_rereco','MuonC_rereco','MuonD_rereco']
     elif options.era == '22_postEE':
-        data_samples = ['MuonE_rereco_postEE','MuonF_postEE','MuonG_postEE']       
+        data_samples = ['MuonE_rereco','MuonF','MuonG']       
 
 if  options.channel == 'tpzee': 
     if options.era == 'UL_16preVFP': data_samples = ['SingleElectronB','SingleElectronC','SingleElectronD','SingleElectronE','SingleElectronF']
     elif options.era == 'UL_16postVFP': data_samples = ['SingleElectronF','SingleElectronG','SingleElectronH']    
     elif options.era == 'UL_17': data_samples = ['SingleElectronB','SingleElectronC','SingleElectronD','SingleElectronE','SingleElectronF']
     elif options.era == 'UL_18': data_samples = ['EGammaA','EGammaB','EGammaC','EGammaD']
+   # elif options.era == '22_preEE':
+   #     data_samples = ['EGammaC_rereco','EGammaD_rereco']
+   # elif options.era == '22_postEE':
+   #     data_samples = ['EGammaE_rereco','EGammaF','EGammaG']
+    elif options.era == '22_preEE':
+        data_samples = ['EGammaC_rereco'] # change to above after ntuples have finished
+    elif options.era == '22_postEE':
+        data_samples = ['EGammaE_rereco']
 
 embed_samples = []
 if options.channel == "tpzmm":
@@ -1000,8 +1008,8 @@ def FitWorkspace(name,infile,outfile,sig_model='DoubleVCorr',bkg_model='Exponent
 elif options.era == 'UL_16postVFP': ztt_samples = ['DYJetsToLL-LO']
 elif options.era == 'UL_17':  ztt_samples = ['DYJetsToLL-LO','DYJetsToLL-LO-ext1']
 elif options.era == 'UL_18': ztt_samples = ['DYJetsToLL-LO','DYJetsToLL-LO-ext1']
-elif options.era == '22_preEE': ztt_samples = ['DYJetsToLL-LO_summer_preEE']
-elif options.era == '22_postEE': ztt_samples = ['DYJetsToLL-LO_postEE']
+elif options.era == '22_preEE': ztt_samples = ['DYJetsToLL-LO']
+elif options.era == '22_postEE': ztt_samples = ['DYJetsToLL-LO']
 
 #Formula:  abs(eta_2),pt_2,m_vis[0,0.1,0.3,0.8,1.0,1.2,1.6,1.8,2.1,2.4],[10,15,17,19,21,23,24,25,26,27,28,31,34,37,40,45,50,60,70,100,1000],(40,70,110) (wt)*((m_vis>50&&m_vis<150&&pt_1>28&&abs(eta_1)<2.1&&iso_1<0.15&&id_tag_1&&trg_tag_1&&os)*(1&&id_probe_1))*(os)*(1)*(!(trg_probe_2))
 ROOT.TH1.SetDefaultSumw2(True)
@@ -1021,7 +1029,7 @@ def FitWorkspace(name,infile,outfile,sig_model='DoubleVCorr',bkg_model='Exponent
   iso_cut_1='iso_1<0.15'
   iso_cut_2='iso_2<0.15'
 
-  if options.era in ['UL_17','UL_18']:
+  if options.era in ['UL_17','UL_18','22_preEE','22_postEE']:
     baseline_tag1 = '(m_vis>50&&m_vis<150&&pt_1>28&&abs(eta_1)<2.1&&iso_1<0.15&&id_tag_1&&trg_tag_1&&os)'
     baseline_tag2 = '(m_vis>50&&m_vis<150&&pt_2>28&&abs(eta_2)<2.1&&iso_2<0.15&&id_tag_2&&trg_tag_2&&os)'
   else:
@@ -1033,11 +1041,11 @@ def FitWorkspace(name,infile,outfile,sig_model='DoubleVCorr',bkg_model='Exponent
     iso_cut_2="1"
 
 if options.channel == 'tpzee':
-  if options.era in ['UL_16preVFP','UL_16postVFP','UL_17','UL_18']:
+  if options.era in ['UL_16preVFP','UL_16postVFP','UL_17','UL_18','22_preEE','22_postEE']:
     iso_cut_1='iso_1<0.15'
     iso_cut_2='iso_2<0.15'
 
-  if options.era in ['UL_17','UL_18']:
+  if options.era in ['UL_17','UL_18','22_preEE','22_postEE']:
     if options.tree_name == 'tagandprobe_ET':
       baseline_tag1 = '(m_vis>50&&m_vis<150&&pt_1>36&&abs(eta_1)<2.1&&iso_1<0.1&&id_tag_1&&trg_tag_1&&trg_tag_extra_1&&os)'
       baseline_tag2 = '(m_vis>50&&m_vis<150&&pt_2>36&&abs(eta_2)<2.1&&iso_2<0.1&&id_tag_2&&trg_tag_2&&trg_tag_extra_2&&os)'
@@ -1101,7 +1109,7 @@ def FitWorkspace(name,infile,outfile,sig_model='DoubleVCorr',bkg_model='Exponent
     ana.remaps = {}
     if options.channel =='tpzmm': ana.remaps['SingleMuon'] = 'data_obs'
     elif options.channel == 'tpzee':
-        if options.era in ['UL_18']: ana.remaps['EGamma'] = 'data_obs'
+        if options.era in ['UL_18','22_preEE','22_postEE']: ana.remaps['EGamma'] = 'data_obs'
         else: ana.remaps['SingleElectron'] = 'data_obs'
 
     # Add all data files
@@ -1178,7 +1186,7 @@ def FitWorkspace(name,infile,outfile,sig_model='DoubleVCorr',bkg_model='Exponent
                if "iso" in name: sig_model = "DoubleVUncorr_elec_TwoPeaks"
                if "trg" in name: sig_model = "DoubleVUncorr_elec"
 
-         if options.era in ["UL_18"]:
+         if options.era in ["UL_18",'22_preEE','22_postEE']:
             if i == "signal": sig_model = "BWCBConvCorr"
             else:
                if "id" in name: sig_model = "DoubleVUncorr_elec_18"
@@ -1244,7 +1252,7 @@ def FitWorkspace(name,infile,outfile,sig_model='DoubleVCorr',bkg_model='Exponent
          if 'trg' in sf: label = 'Electron Trigger, %s, %.1f < |#eta| < %.1f' % (options.era,ymin,ymax)
          elif 'iso' in sf: label = 'Electron Iso, %s, %.1f < |#eta| < %.1f' % (options.era,ymin,ymax)
          elif 'id' in sf: label = 'Electron ID, %s, %.1f < |#eta| < %.1f' % (options.era,ymin,ymax)
-       plotting.TagAndProbePlot(graphs,leg_labels,"",True,False,options.era=='UL_18',ratio_range,True,200,10,False,0,1.5,x_title, "Efficiency",0,options.outputfolder+'/'+plot_name+sf+'_eta_%.1f_to_%.1f_%s'%(ymin,ymax,i),label)
+       plotting.TagAndProbePlot(graphs,leg_labels,"",True,False,options.era in ['UL_18','22_preEE','22_postEE'],ratio_range,True,200,10,False,0,1.5,x_title, "Efficiency",0,options.outputfolder+'/'+plot_name+sf+'_eta_%.1f_to_%.1f_%s'%(ymin,ymax,i),label)
 
    sffile.Close()
 

Analysis/HiggsTauTauRun2/scripts/config_2022_postEE.json

@@ -6,7 +6,7 @@
     "timings":      true,
    //"channels":     ["mt","et","tt","zmm","zee","em"],
    // "channels":     ["mt","zmm","tt","et"],
-    "channels":     ["mt"],
+    "channels":     ["tpzee","tpzmm"],
 //   "channels":     ["tt","mt","et","em","zmm","zee"],
     "sequences": {
       "all":  ["default"]
@@ -66,8 +66,7 @@
     "do_ff_systematics": false,
     "do_pdf_wts": false,
     "do_qcd_scale_wts": false,
-    //"output_folder" : "/vols/cms/dw515/outputs/MSSM/UL_Mar31_2018_sync_v2/",
-    "output_folder" : "/vols/cms/eb921/test_output/output_mt_2022postEE",
+    "output_folder" : "/vols/cms/dw515/outputs/MSSM/2022-postEE_031023_v2/",
     "svfit_from_grid" : true,
     "read_all_svfit_files" : true,
     "svfit_folder"  : "/vols/cms/dw515/outputs/MSSM/mssm_2018_tres1_v2_svfit/",

Analysis/HiggsTauTauRun2/scripts/config_2022_preEE.json

@@ -6,7 +6,7 @@
     "timings":      true,
    //"channels":     ["mt","et","tt","zmm","zee","em"],
    // "channels":     ["mt","zmm","tt","et"],
-    "channels":     ["mt"],
+    "channels":     ["tpzmm","tpzee"],
 //   "channels":     ["tt","mt","et","em","zmm","zee"],
     "sequences": {
       "all":  ["default"]
@@ -66,8 +66,7 @@
     "do_ff_systematics": false,
     "do_pdf_wts": false,
     "do_qcd_scale_wts": false,
-    //"output_folder" : "/vols/cms/dw515/outputs/MSSM/UL_Mar31_2018_sync_v2/",
-    "output_folder" : "/vols/cms/eb921/output/output_mt_2022",
+    "output_folder" : "/vols/cms/dw515/outputs/MSSM/2022-preEE_031023_v2/",
     "svfit_from_grid" : true,
     "read_all_svfit_files" : true,
     "svfit_folder"  : "/vols/cms/dw515/outputs/MSSM/mssm_2018_tres1_v2_svfit/",

Analysis/HiggsTauTauRun2/scripts/hadd_22_postEE_output.py

@@ -43,46 +43,36 @@ def list_paths(path):
 sample_list = [
 
           # Tau
-         'TauA_preEE',
-         'TauB_preEE',
-         'TauB_rereco_preEE',
-         'TauC_preEE',
-         'TauC_rereco_preEE',
-         'TauD_preEE',
-         'TauD_rereco_preEE',
-         'TauE_postEE',
-         'TauE_rereco_postEE',
-         'TauF_postEE',
-         'TauG_postEE',
-         'MuonE_rereco_postEE',
-         'MuonE_postEE',
-         'MuonF_postEE',
-         'MuonG_postEE',
+         'TauE_rereco',
+         'TauF',
+         'TauG',
+
+         'MuonE_rereco',
+         'MuonF',
+         'MuonG',
 
          #MC postEE
-
-          'DYJetsToLL-LO_postEE',
-          'DYto2TautoMuTauh_M50_postEE',
-          'GluGluHToTauTau_M125_v2_postEE',
-          'GluGluHToTauTau_M125_v3_postEE',
-          'TBbarQ_t-channel_4FS_postEE',
-          'TTTo2L2Nu_postEE',
-          'TTto4Q_postEE',
-          'TTtoLNu2Q_postEE',
-          'TWminusto2L2Nu_postEE',
-          'TWminustoLNu2Q_postEE',
-          'TbarBQ_t-channel_4FS_postEE',
-          'TbarWplusto2L2Nu_postEE',
-          'TbarWplustoLNu2Q_postEE',
-          'VBFHToTauTau_M125_Poisson60KeepRAW_postEE',
-          'VBFHToTauTau_M125_v2_Poisson70KeepRAW_postEE',
-          'W1JetsToLNu-LO_postEE',
-          'W2JetsToLNu-LO_postEE',
-          'WJetsToLNu-LO_postEE',
-          'WW_postEE',
-          'WZ_postEE',
-          'ZZ_postEE',
-
+
+         'DYJetsToLL-LO',
+         # missing exclusive jet samples
+         'DYto2TautoMuTauh_M50',
+         'TTTo2L2Nu',
+         'TTto4Q',
+         'TTtoLNu2Q',
+         'TBbarQ_t-channel_4FS',
+         'TWminusto2L2Nu',
+         'TWminustoLNu2Q',
+         'TbarBQ_t-channel_4FS',
+         'TbarWplusto2L2Nu',
+         'TbarWplustoLNu2Q',
+         'WW',
+         'WZ',
+         'ZZ',
+         'W1JetsToLNu-LO',
+         'W2JetsToLNu-LO',
+         'WJetsToLNu-LO',
+         #missing W+jets 3 and 4 samples
+
 	]
 
 
@@ -113,14 +103,14 @@ def list_paths(path):
 
 
 def FindMissingFiles(outf, d, samp, chan, infiles):
-  files=fnmatch.filter(infiles,'%(samp)s_2022_%(chan)s_*'%vars())
+  files=fnmatch.filter(infiles,'%(samp)s_2022_postEE_%(chan)s_*'%vars())
   nums = [int(x.split('_')[-1].replace('.root','')) for x in files]
   nums.sort()
   res = [ele for ele in range(max(nums)+1) if ele not in nums]
 
   if len(res) !=0:
-    print "Some files are missing for sample %(samp)s_2022_%(chan)s! in %(d)s:"%vars()
-    for x in res: print '%(samp)s_2022_%(chan)s_%(x)i.root' % vars()
+    print "Some files are missing for sample %(samp)s_2022_postEE_%(chan)s! in %(d)s:"%vars()
+    for x in res: print '%(samp)s_2022_postEE_%(chan)s_%(x)i.root' % vars()
     return False
   else:
     return True
@@ -133,35 +123,35 @@ def FindMissingFiles(outf, d, samp, chan, infiles):
   hadd_dirs=[]
   command=''
   if batch:
-    JOB='jobs/hadd_%s_2022.sh' % sa
+    JOB='jobs/hadd_%s_2022_postEE.sh' % sa
     os.system('%(JOBWRAPPER)s "" %(JOB)s' %vars())
   for ch in channel:
     for jsdir in subdirs:
       sdir = jsdir[0]
       infiles=jsdir[1]
-      if os.path.isfile('%(outputf)s/%(sdir)s/%(sa)s_2022_%(ch)s_0.root'%vars()):
-        if "%(sa)s_2022"%vars() in nfiles or ignore==True:
-#          files=glob.glob('%(outputf)s/%(sdir)s/%(sa)s_2022_%(ch)s_*.root'%vars())
+      if os.path.isfile('%(outputf)s/%(sdir)s/%(sa)s_2022_postEE_%(ch)s_0.root'%vars()):
+        if "%(sa)s_2022_postEE"%vars() in nfiles or ignore==True:
+#          files=glob.glob('%(outputf)s/%(sdir)s/%(sa)s_2022_postEE_%(ch)s_*.root'%vars())
           no_missing_files = FindMissingFiles(outputf, sdir, sa, ch,infiles) 
-          if no_missing_files and (ignore ==True or len(fnmatch.filter(infiles,'%(sa)s_2022_%(ch)s_*'%vars())) == nfiles["%(sa)s_2022"%vars()]):
+          if no_missing_files and (ignore ==True or len(fnmatch.filter(infiles,'%(sa)s_2022_postEE_%(ch)s_*'%vars())) == nfiles["%(sa)s_2022_postEE"%vars()]):
             if not batch:  
               print "Hadding in subdir %(sdir)s"%vars()
               print "Hadding %(sa)s_%(ch)s in %(sdir)s"%vars()
-              os.system('hadd -f %(outputf)s/%(sdir)s/%(sa)s_%(ch)s_2022.root %(outputf)s/%(sdir)s/%(sa)s_2022_%(ch)s_* &> ./haddout.txt'% vars()) 
+              os.system('hadd -f %(outputf)s/%(sdir)s/%(sa)s_%(ch)s_2022_postEE.root %(outputf)s/%(sdir)s/%(sa)s_2022_postEE_%(ch)s_* &> ./haddout.txt'% vars()) 
               os.system("sed -i '/Warning in <TInterpreter::ReadRootmapFile>/d' ./haddout.txt")
               filetext = open("./haddout.txt").read()
               if 'Warning' in filetext or 'Error' in filetext:
                 print "Hadd had a problem:"
                 print filetext 
                 remove=False
               else :
-                to_remove.append('rm %(outputf)s/%(sdir)s/%(sa)s_2022_%(ch)s_*' %vars())
+                to_remove.append('rm %(outputf)s/%(sdir)s/%(sa)s_2022_postEE_%(ch)s_*' %vars())
             else:
               haddout='haddout_%s_%s_%s.txt' % (sa,ch,sdir)  
-              hadd_dirs.append((haddout, 'rm %(outputf)s/%(sdir)s/%(sa)s_2022_%(ch)s_*' %vars())) 
-              command+="echo \"Hadding %(sa)s_%(ch)s in %(sdir)s\"\necho \"Hadding %(sa)s_%(ch)s\"\nhadd -f %(outputf)s/%(sdir)s/%(sa)s_%(ch)s_2022.root %(outputf)s/%(sdir)s/%(sa)s_2022_%(ch)s_* &> ./%(haddout)s\nsed -i '/Warning in <TInterpreter::ReadRootmapFile>/d' ./%(haddout)s\n" % vars()     
+              hadd_dirs.append((haddout, 'rm %(outputf)s/%(sdir)s/%(sa)s_2022_postEE_%(ch)s_*' %vars())) 
+              command+="echo \"Hadding %(sa)s_%(ch)s in %(sdir)s\"\necho \"Hadding %(sa)s_%(ch)s\"\nhadd -f %(outputf)s/%(sdir)s/%(sa)s_%(ch)s_2022_postEE.root %(outputf)s/%(sdir)s/%(sa)s_2022_postEE_%(ch)s_* &> ./%(haddout)s\nsed -i '/Warning in <TInterpreter::ReadRootmapFile>/d' ./%(haddout)s\n" % vars()     
           else :
-            print "Incorrect number of files for sample %(sa)s_2022_%(ch)s! in %(sdir)s"%vars()
+            print "Incorrect number of files for sample %(sa)s_2022_postEE_%(ch)s! in %(sdir)s"%vars()
             remove=False
 
   if not batch and remove: