Some modifications

limits/LimitsLauncher.py

@@ -22,18 +22,18 @@
 #"----------------User's decision board-----------------"
 
 # choose which scripts to run
-do_combineDatacards = False
-do_produceLimits = False
-do_producePlots = True   #note that this flag can be turned to true only when the limit results have been produced
+do_combineDatacards = True
+do_produceLimits = True
+do_producePlots = False   #note that this flag can be turned to true only when the limit results have been produced
 
 
-version_label = 'test3'
+version_label = 'test00'
 
 # path to datacards # Note that it will get /<year>/<channel> appended
-path_to_datacards = './datacards'
+path_to_datacards = '/work/manzoni/hnl_limits/CMSSW_10_2_13/src/HiggsAnalysis/CombinedLimit/WHNL_code/limits/datacards_july_talk/'
 
 # which years to process
-years = ['2016', '2017', '2018'] 
+years = ['2018'] 
 
 # which channels
 channels = ['mmm', 'mem_os', 'mem_ss']
@@ -50,7 +50,7 @@
 signal_type = 'majorana'
 
 # you may want to select the mass values you would like to run on
-mass_whiteList = ['10.0', '6.0'] #['10.0']
+mass_whiteList = [] #['10.0']
 
 # you may want to select the mass values you would like to ignore
 mass_blackList = [] 
@@ -100,6 +100,7 @@ def getStringParser(input):
   print 'Will run the limits production tool'
 
   os.system('mkdir -p results/limits/{}'.format(version_label)) 
+  os.system('mkdir -p logs/{}'.format(version_label)) 
 
   if run_blind: flag_blind = 1
   else: flag_blind = 0
@@ -140,7 +141,7 @@ def getStringParser(input):
 
   print 'will run the limit plotter'
 
-  command_plotter = 'python limit_plotter.py --version {ver} --signal {sig} --channels {ch}  --mass_whitelist {mwl} --mass_blacklist {mbl} --coupling_whitelist {cwl} --coupling_blacklist {cbl}'.format(ver=version_label, sig=signal_type, ch=getStringParser(channels), mwl=getStringParser(mass_whiteList), mbl=getStringParser(mass_blackList), cwl=getStringParser(coupling_whiteList), cbl=getStringParser(coupling_blackList))
+  command_plotter = 'python limit_plotter.py --version {ver} --signal {sig} --years {ys} --channels {ch}  --mass_whitelist {mwl} --mass_blacklist {mbl} --coupling_whitelist {cwl} --coupling_blacklist {cbl}'.format(ver=version_label, sig=signal_type, ys=getStringParser(years), ch=getStringParser(channels), mwl=getStringParser(mass_whiteList), mbl=getStringParser(mass_blackList), cwl=getStringParser(coupling_whiteList), cbl=getStringParser(coupling_blackList))
   if run_blind:
     command_plotter += ' --run_blind'
 

limits/README.md

@@ -1,6 +1,9 @@
 # Limits Analysis Tool
 
 ## Installation
+
+Note: it is preferable to use a bash environment. 
+
 ```
 cmsrel CMSSW_10_2_13
 cd CMSSW_10_2_13/src
@@ -15,7 +18,7 @@ make -j 8; make
 
 ```
 git clone git@github.com:rmanzoni/plotter.git WHNL_code
-cd plotter/limits
+cd WHNLcode/limits
 ```
 
 ## Running the limits

limits/combine_datacards.py

@@ -73,8 +73,8 @@ def getOptions():
 
 for channel in channels:
   for year in years:
-    all_datacards[channel][year] = glob('/'.join([path_year[year], path_channel['mmm'], datacard_wildcard]))
-    all_datacards[channel][year] = [dd for dd in all_datacards[channel][year] if 'coarse' not in dd and signal_type in dd]
+    all_datacards[channel][year] = glob('/'.join([path_year[year], path_channel[channels[0]], datacard_wildcard]))
+    all_datacards[channel][year] = [dd for dd in all_datacards[channel][year] if signal_type in dd]
     all_datacards[channel][year].sort()
     print '... datacards {a}_{b} loaded'.format(a=channel, b=year)
 

limits/limit_plotter.py

@@ -210,7 +210,6 @@ def sortList(input):
     limits2D[mass]['exp_plus_two' ] = x_plus_two 
     if not opt.run_blind:
         limits2D[mass]['obs'] = x_obs 
-
 
   print '\nwill plot 2D limits' 
   with open('{}/results.pck'.format(plotDir), 'w') as ff:
@@ -237,20 +236,18 @@ def sortList(input):
             obs.append( min(limits2D[mass]['obs']) )
             masses_obs.append(mass)
 
-      if len(limits2D[mass]['exp_minus_one'])==len(limits2D[mass]['exp_plus_one']) and len(limits2D[mass]['exp_minus_two'])==len(limits2D[mass]['exp_plus_two']):
-        if len(limits2D[mass]['exp_central'])>0:
+      if len(limits2D[mass]['exp_central'])>0 and len(limits2D[mass]['exp_minus_one'])>0 and len(limits2D[mass]['exp_plus_one' ])>0 and len(limits2D[mass]['exp_minus_two'])>0 and len(limits2D[mass]['exp_plus_two' ])>0:
           central.append( min(limits2D[mass]['exp_central']) )
           masses_central.append(float(mass))
 
-        if len(limits2D[mass]['exp_minus_one'])>0 and len(limits2D[mass]['exp_plus_one' ])>0: 
           minus_one.append( min(limits2D[mass]['exp_minus_one']) )
-          plus_one .append( min(limits2D[mass]['exp_plus_one' ]) )
+          plus_one.append( min(limits2D[mass]['exp_plus_one' ]) )
           masses_one_sigma.append(float(mass))
 
-        if len(limits2D[mass]['exp_minus_two'])>0 and len(limits2D[mass]['exp_plus_two' ])>0: 
           minus_two.append( min(limits2D[mass]['exp_minus_two']) )
-          plus_two .append( min(limits2D[mass]['exp_plus_two' ]) )
+          plus_two.append( min(limits2D[mass]['exp_plus_two' ]) )
           masses_two_sigma.append(float(mass))
+
 
   for mass in sorted(limits2D.keys(), key=sortList, reverse=True):
 

nn/cfg/train_all_years.cfg.py

@@ -107,7 +107,7 @@
 #         'l2_ptcone'      ,
         'log_hnl_2d_disp',
 #         'isdata'         ,
-        'log_hnl_2d_disp_sig_log',
+#        'log_hnl_2d_disp_sig_log',
 
 #         'l0_pt'          ,
 #         'abs_l0_eta'     ,

plotter/cfg/2017/plot_mem_ttbar.cfg.py

@@ -21,8 +21,10 @@
 selection_mc = selection # + [cuts.selections['is_prompt_lepton']]
 selection_tight = cuts.selections_pd['tight']
 
+
+training = 'run2/all_channels__200623_17h_51m'
 #training = '2018/all_channels__200528_23h_35m'
-training = 'all_channels_200213_14h_55m'
+#training = 'all_channels_200213_14h_55m'
 # training = 'all_channels_200526_12h_46m'
 # training = 'all_channels_200525_19h_38m'
 # training = 'all_channels_200525_18h_55m'
@@ -54,11 +56,13 @@
     plot_signals     = False, 
     blinded          = False,
 
-    datacards        = ['hnl_m_12_lxy_lt_0p5', 'hnl_m_12_lxy_0p5_to_1p5', 'hnl_m_12_lxy_1p5_to_4p0', 'hnl_m_12_lxy_mt_4p0'], # FIXME! improve this to accept wildcards / regex
+    #datacards        = ['hnl_m_12_lxy_lt_0p5', 'hnl_m_12_lxy_0p5_to_1p5', 'hnl_m_12_lxy_1p5_to_4p0', 'hnl_m_12_lxy_mt_4p0'], # FIXME! improve this to accept wildcards / regex
+    datacards        = ['hnl_m_12_lxy_lt_0p5', 'hnl_m_12_lxy_0p5_to_1p5', 'hnl_m_12_lxy_1p5_to_4p0', 'hnl_m_12_lxy_mt_4p0'],                    
 
     mc_subtraction   = True,
 
-    data_driven      = False,
+    #data_driven      = False,
+    data_driven      = True,
 )
 
 if __name__ == '__main__':

plotter/cfg/2017/plot_mmm.cfg.py

@@ -26,9 +26,10 @@
 selection_mc = selection + [cuts.selections['is_prompt_lepton']]
 selection_tight = cuts.selections_pd['tight']
 
+training = 'run2/all_channels__200623_17h_51m'
 #training = '2018/all_channels__200528_23h_35m'
 #training = 'all_2017_channels_200116_15h_15m'
-training = 'all_channels_200213_14h_55m'
+#training = 'all_channels_200213_14h_55m'
 # training = 'all_channels_200526_12h_46m'
 # training = 'all_channels_200525_19h_38m'
 # training = 'all_channels_200525_18h_55m'
@@ -61,12 +62,14 @@
     plot_signals     = False, 
     blinded          = False,
 
-    datacards        = ['log_hnl_2d_disp'        , 
-                        'hnl_m_12'               ,  
-                        'hnl_m_12_lxy_lt_0p5'    , 
-                        'hnl_m_12_lxy_0p5_to_1p5', 
-                        'hnl_m_12_lxy_1p5_to_4p0', 
-                        'hnl_m_12_lxy_mt_4p0'], # FIXME! improve this to accept wildcards / regex
+    #datacards        = ['log_hnl_2d_disp'        , 
+    #                    'hnl_m_12'               ,  
+    #                    'hnl_m_12_lxy_lt_0p5'    , 
+    #                    'hnl_m_12_lxy_0p5_to_1p5', 
+    #                    'hnl_m_12_lxy_1p5_to_4p0', 
+    #                    'hnl_m_12_lxy_mt_4p0'], # FIXME! improve this to accept wildcards / regex
+
+    datacards        = ['hnl_m_12_lxy_lt_0p5', 'hnl_m_12_lxy_0p5_to_1p5', 'hnl_m_12_lxy_1p5_to_4p0', 'hnl_m_12_lxy_mt_4p0'],                    
 
     mc_subtraction   = True,
 

plotter/cfg/2017/plot_mmm_zmm.cfg.py

@@ -26,8 +26,9 @@
 selection_mc = selection # + [cuts.selections['is_prompt_lepton']]
 selection_tight = cuts.selections_pd['tight']
 
+training = 'run2/all_channels__200623_17h_51m'
 #training = '2018/all_channels__200528_23h_35m'
-training = 'all_channels_200213_14h_55m'
+#training = 'all_channels_200213_14h_55m'
 # training = 'all_channels_200526_12h_46m'
 # training = 'all_channels_200525_19h_38m'
 # training = 'all_channels_200525_18h_55m'
@@ -60,11 +61,14 @@
     plot_signals     = False, 
     blinded          = False,
 
-    datacards        = ['hnl_m_12_lxy_lt_0p5', 'hnl_m_12_lxy_0p5_to_1p5', 'hnl_m_12_lxy_1p5_to_4p0', 'hnl_m_12_lxy_mt_4p0'], # FIXME! improve this to accept wildcards / regex
+    #datacards        = ['hnl_m_12_lxy_lt_0p5', 'hnl_m_12_lxy_0p5_to_1p5', 'hnl_m_12_lxy_1p5_to_4p0', 'hnl_m_12_lxy_mt_4p0'], # FIXME! improve this to accept wildcards / regex
+    datacards        = ['hnl_m_12_lxy_lt_0p5', 'hnl_m_12_lxy_0p5_to_1p5', 'hnl_m_12_lxy_1p5_to_4p0', 'hnl_m_12_lxy_mt_4p0'],                    
 
-    mc_subtraction   = False,
+    #mc_subtraction   = False,
+    mc_subtraction   = True,
 
-    data_driven      = False,
+    #data_driven      = False,
+    data_driven      = True,
 )                   
 
 if __name__ == '__main__':

plotter/cfg/2018/plot_mem_os.cfg.py

@@ -25,7 +25,8 @@
 selection_mc = selection + [cuts.selections['is_prompt_lepton']]
 selection_tight = cuts.selections_pd['tight']
 
-training = '2018/all_channels__200528_23h_35m'
+training = 'run2/all_channels__200623_17h_51m'
+# training = '2018/all_channels__200528_23h_35m'
 # training = 'all_channels_200526_12h_46m'
 # training = 'all_channels_200525_19h_38m'
 # training = 'all_channels_200525_18h_55m'

plotter/cfg/2018/plot_mem_os_loose.cfg.py

@@ -38,7 +38,8 @@
 selection_mc = selection + [cuts.selections['is_prompt_lepton']]
 selection_tight = cuts.selections_pd['tight']
 
-training = 'run2/all_channels__200604_15h_13m'
+training = 'run2/all_channels__200623_17h_51m'
+#training = 'run2/all_channels__200604_15h_13m'
 # training = 'run2/all_channels__200602_17h_26m'
 # training = 'run2/all_channels__200601_18h_20m'
 # training = '2018/all_channels__200528_23h_35m'
@@ -76,13 +77,15 @@
     plot_signals     = False, 
     blinded          = False,
 
-    datacards        = ['log_hnl_2d_disp'        , 
-                        'hnl_m_12'               ,  
-                        'hnl_m_12_lxy_lt_0p5'    , 
-                        'hnl_m_12_lxy_0p5_to_1p5', 
-                        'hnl_m_12_lxy_1p5_to_4p0', 
-                        'hnl_m_12_lxy_mt_4p0'], # FIXME! improve this to accept wildcards / regex
+    #datacards        = ['log_hnl_2d_disp'        , 
+    #                    'hnl_m_12'               ,  
+    #                    'hnl_m_12_lxy_lt_0p5'    , 
+    #                    'hnl_m_12_lxy_0p5_to_1p5', 
+    #                    'hnl_m_12_lxy_1p5_to_4p0', 
+    #                    'hnl_m_12_lxy_mt_4p0'], # FIXME! improve this to accept wildcards / regex
 
+    datacards        = ['hnl_m_12_lxy_lt_0p5', 'hnl_m_12_lxy_0p5_to_1p5', 'hnl_m_12_lxy_1p5_to_4p0', 'hnl_m_12_lxy_mt_4p0'],                    
+
     mc_subtraction   = False,
 
     data_driven      = True,

plotter/cfg/2018/plot_mem_ss.cfg.py

@@ -24,8 +24,9 @@
 selection_mc = selection + [cuts.selections['is_prompt_lepton']]
 selection_tight = cuts.selections_pd['tight']
 
+training = 'run2/all_channels__200623_17h_51m'
 #training = '2018/all_channels__200528_23h_35m'
-training = 'all_channels_200213_14h_55m'
+#training = 'all_channels_200213_14h_55m'
 # training = 'all_channels_200526_12h_46m'
 # training = 'all_channels_200525_19h_38m'
 # training = 'all_channels_200525_18h_55m'

plotter/cfg/2018/plot_mem_ss_loose.cfg.py

@@ -36,7 +36,8 @@
 selection_mc = selection + [cuts.selections['is_prompt_lepton']]
 selection_tight = cuts.selections_pd['tight']
 
-training = 'run2/all_channels__200604_15h_13m'
+training = 'run2/all_channels__200623_17h_51m'
+#training = 'run2/all_channels__200604_15h_13m'
 # training = 'run2/all_channels__200602_17h_26m'
 # training = 'run2/all_channels__200601_18h_20m'
 # training = '2018/all_channels__200528_23h_35m'

plotter/cfg/2018/plot_mmm.cfg.py

@@ -13,8 +13,8 @@
     cuts.selections['vetoes_12_OS'], 
     cuts.selections['vetoes_01_OS'], 
     cuts.selections['vetoes_02_OS'],
-#     cuts.selections['signal_region'], 
-    cuts.selections['sideband'], 
+    cuts.selections['signal_region'], 
+#     cuts.selections['sideband'], 
     'hlt_IsoMu24',
 ]
 
@@ -26,8 +26,11 @@
 selection_mc = selection + [cuts.selections['is_prompt_lepton']]
 selection_tight = cuts.selections_pd['tight']
 
-#training = 'run2/all_channels__200601_18h_20m'
-training = 'all_channels_200213_14h_55m'
+# training = 'run2/all_channels__200623_19h_2m'
+training = 'run2/all_channels__200623_16h_37m'
+# training = 'run2/all_channels__200604_15h_13m'
+# training = 'run2/all_channels__200602_17h_26m'
+# training = 'run2/all_channels__200601_18h_20m'
 # training = '2018/all_channels__200528_23h_35m'
 # training = 'all_channels_200526_12h_46m'
 # training = 'all_channels_200525_19h_38m'
@@ -43,8 +46,8 @@
     plot_dir         = '/'.join([env['BASE_DIR'], 'plotter', 'plots', '2018']), 
     base_dir         = '/'.join([env['BASE_DIR'], 'ntuples', 'may20', '2018']),
     post_fix         = 'HNLTreeProducer_%s/tree.root' %ch,
-#     dir_suffix       = 'signal_dd_datacards', #'signal',
-    dir_suffix       = 'sideband', 
+    dir_suffix       = 'signal_dd_datacards_blinded', #'signal',
+#     dir_suffix       = 'sideband', 
 
     selection_data   = selection,
     selection_mc     = selection_mc,
@@ -58,9 +61,9 @@
     features         = '/'.join([env['BASE_DIR'], 'nn', 'trainings', training, 'input_features.pck'              ]),
 
     process_signals  = True, # switch off for control regions
-    mini_signals     = True, # process only the signals that you'll plot
+    mini_signals     = False, # process only the signals that you'll plot
     plot_signals     = True, 
-    blinded          = False,
+    blinded          = True,
 
     datacards        = ['log_hnl_2d_disp'        , 
                         'hnl_m_12'               ,  
@@ -77,6 +80,7 @@
 if __name__ == '__main__':
     plotter.plot()
     # save the plotter and all
+    #save_plotter_and_selections(plotter, selection, selection_mc, selection_tight, training)
     save_plotter_and_selections(plotter, selection, selection_mc, selection_tight)
     pass
 

plotter/cfg/2018/plot_mmm_loose.cfg.py

@@ -38,7 +38,8 @@
 selection_mc = selection + [cuts.selections['is_prompt_lepton']]
 selection_tight = cuts.selections_pd['tight']
 
-training = 'run2/all_channels__200604_15h_13m'
+training = 'run2/all_channels__200623_17h_51m'
+#training = 'run2/all_channels__200604_15h_13m'
 # training = 'run2/all_channels__200602_17h_26m'
 # training = 'run2/all_channels__200601_18h_20m'
 # training = '2018/all_channels__200528_23h_35m'
@@ -76,12 +77,14 @@
     plot_signals     = False, 
     blinded          = False,
 
-    datacards        = ['log_hnl_2d_disp'        , 
-                        'hnl_m_12'               ,  
-                        'hnl_m_12_lxy_lt_0p5'    , 
-                        'hnl_m_12_lxy_0p5_to_1p5', 
-                        'hnl_m_12_lxy_1p5_to_4p0', 
-                        'hnl_m_12_lxy_mt_4p0'], # FIXME! improve this to accept wildcards / regex
+    #datacards        = ['log_hnl_2d_disp'        , 
+    #                    'hnl_m_12'               ,  
+    #                    'hnl_m_12_lxy_lt_0p5'    , 
+    #                    'hnl_m_12_lxy_0p5_to_1p5', 
+    #                    'hnl_m_12_lxy_1p5_to_4p0', 
+    #                    'hnl_m_12_lxy_mt_4p0'], # FIXME! improve this to accept wildcards / regex
+
+    datacards        = ['hnl_m_12_lxy_lt_0p5', 'hnl_m_12_lxy_0p5_to_1p5', 'hnl_m_12_lxy_1p5_to_4p0', 'hnl_m_12_lxy_mt_4p0'],                    
 
     mc_subtraction   = False,
 

plotter/cfg/2018/plot_mmm_signal.cfg.py

@@ -26,8 +26,9 @@
 selection_mc = selection + [cuts.selections['is_prompt_lepton']]
 selection_tight = cuts.selections_pd['tight']
 
+training = 'run2/all_channels__200623_17h_51m'
 #training = 'all_channels_200526_12h_14m'
-training = 'all_channels_200213_14h_55m'
+#training = 'all_channels_200213_14h_55m'
 # training = 'all_channels_200525_18h_55m'
 # training = 'all_channels_200523_22h_39m' #<==== GOOD
 # training = 'all_channels_200523_15h_53m'

plotter/cfg/2018/plot_mmm_zmm.cfg.py

@@ -25,8 +25,9 @@
 selection_mc = selection # + [cuts.selections['is_prompt_lepton']]
 selection_tight = cuts.selections_pd['tight']
 
+training = 'run2/all_channels__200623_17h_51m'
 #training = '2018/all_channels__200528_18h_42m'
-training = 'all_channels_200213_14h_55m'
+#training = 'all_channels_200213_14h_55m'
 # training = 'all_channels_200525_19h_38m'
 # training = 'all_channels_200525_18h_55m'
 # training = 'all_channels_200523_22h_39m' #<==== GOOD

plotter/objects/plotter.py

@@ -664,7 +664,9 @@ def plot(self):
 
                     # data in tight
                     all_obs_prompt_norm = copy(all_obs_prompt)
-                    all_obs_prompt_norm.Scale(np.nan_to_num(np.divide(1., all_obs_prompt_norm.integral())))
+                    if all_obs_prompt_norm.integral() != 0:
+                      all_obs_prompt_norm.Scale(np.nan_to_num(np.divide(1., all_obs_prompt_norm.integral())))
+                    #import pdb; pdb.set_trace()
                     all_obs_prompt_norm.drawstyle = 'hist e'
                     all_obs_prompt_norm.linecolor = 'black'
                     all_obs_prompt_norm.markersize = 0