Merge pull request cms-analysis#16 from JaLuka98/dev_higgsdnafinalfit_law_test_2023

Dev higgsdnafinalfit law test 2023

Author: TheRealLoliges486

Trees2WS/law_trees2ws.py

@@ -65,6 +65,17 @@ def create_branch_map(self):
             for input_path in glob.glob(f"{self.input_paths}/{process}_M-{mass}_{self.era}/*.root")
         ]
         branch_map = {i: mode_proc_mass for i, mode_proc_mass in enumerate(mode_proc_mass_list)}
+        # Test if empty, then try naming convention with year in front
+        # Note here: self.year = real year + era string (maybe should be improved)
+        if not branch_map:
+            print("branch_map is empty, trying with year+era at the end.")
+            mode_proc_mass_list = [
+            (mode, mass, input_path)
+            for mode, process in production_modes
+            for mass in input_masses
+            for input_path in glob.glob(f"{self.input_paths}/{process}_M-{mass}_{self.year}/*.root")
+            ]
+        branch_map = {i: mode_proc_mass for i, mode_proc_mass in enumerate(mode_proc_mass_list)}
         return branch_map
 
     def output(self):
@@ -587,8 +598,8 @@ def requires(self):
 
             tasks.append(Trees2WSSingleProcess(input_paths=path_to_root_files, era=era, apply_mass_cut=mass_cut, mass_cut_range=mass_cut_r, year=f"{self.year}{era}", doSystematics=doSystematics, doDiffSplitting=doDiffSplitting, doSTXSSplitting=doSTXSSplitting, doInOutSplitting=doInOutSplitting, output_dir=current_output_path, variable=var, version=f"v{i}", workflow=config['execution']))
             i += 1
-        
-        return tasks    
+        return tasks
+
     def output(self):
 
         if self.variable == '':
@@ -618,7 +629,7 @@ def output(self):
                 current_output_path = output_dir + "/input_output_{}_{}{}".format(var, self.year, era)
 
             outputFolders.append(law.LocalFileTarget(current_output_path + '/ws_signal'))
-        
+
         return outputFolders
 
     def run(self):

commonTools/commonObjects.py

@@ -25,7 +25,10 @@
     'merged':137.65,
     '2022preEE':8.00,
     '2022postEE':26.70,
-    '2022': 34.7
+    '2022': 34.7,
+    '2023preBPix': 17.8,
+    '2023postBPix': 9.5,
+    '2023': 27.3
 }
 
 def CreateVariableParameters(gen_variable, reco_variable, bins, year, BMW):
@@ -59,7 +62,7 @@ def CreateVariableParameters(gen_variable, reco_variable, bins, year, BMW):
 BR_W_qq = 67.41*0.01
 
 # List of years
-years_to_process = ['2016','2017','2018','2022preEE','2022postEE']
+years_to_process = ['2016','2017','2018','2022preEE','2022postEE','2022preBPix','2022postBPix']
 # Production modes and decay channel: for extract XS from combine
 productionModes = ['ggH','qqH','ttH','tHq','tHW','ggZH', 'WH','ZH','bbH']
 decayMode = 'hgg'
@@ -81,25 +84,38 @@ def CreateVariableParameters(gen_variable, reco_variable, bins, year, BMW):
 input_masses = [120, 125, 130]
 
 # Define an array of production modes and corresponding process strings
+# JLS 23th Jan 2025, also adding 2G naming conventions
 production_modes = [
     ("ggh", "GluGluHtoGG"),
+    ("ggh", "GluGluHto2G"),
     ("vbf", "VBFHtoGG"),
+    ("vbf", "VBFHto2G"),
     ("vh", "VHtoGG"),
-    ("tth", "ttHtoGG")
+    ("vh", "VHto2G"),
+    ("tth", "ttHtoGG"),
+    ("tth", "ttHto2G"),
 ]
 
 # Define an array of eras
+# JLS 22th of Jan 2025: This syntax looks pretty criminal and should be improved at some point
 TwentyTwentyTwoEras = ["preEE", "postEE"]
+TwentyTwentyThreeEras = ["preBPix", "postBPix"]
+
 
 allErasMap = {
-    '2022': TwentyTwentyTwoEras
+    '2022': TwentyTwentyTwoEras,
+    '2023': TwentyTwentyThreeEras
 }
 
 conversionTable_ = {
     "GluGluHtoGG": "ggh",
+    "GluGluHto2G": "ggh",
     "ttHtoGG": "tth",
+    "ttHto2G": "tth",
     "VBFHtoGG": "vbf",
+    "VBFHto2G": "vbf",
     "VHtoGG": "vh",
+    "VHto2G": "vh",
     }
 
 # List of all jet-related variables. Variables listed here will get the CMS_scale_j and CMS_res_j uncertainty in the datacard step.

config/2023_inclusive.yml

@@ -0,0 +1,195 @@
+# Config file: options for signal fitting
+
+outputFolder: '/.automount/net_rw/net__data_cms3a-1/spaeh/private/PhD/analyses/partial_Run3_differential/Hgg-PartialRun3-3A-ETH-Analysis/fitting/CMSSW_14_1_0_pre4/src/flashggFinalFit/test_law_branch_2023'
+
+inputFiles:
+  # Trees2WS
+  Trees2WSData: '/eos/cms/store/group/phys_higgs/cmshgg/2023_2022_Hgg_differentials/2025_01_20_intermediateNTuples_2023/data/root/Data/allData_2023.root'
+  Trees2WS: '/eos/cms/store/group/phys_higgs/cmshgg/2023_2022_Hgg_differentials/2025_01_20_intermediateNTuples_2023/signal/root'
+  # Also does not work if we substitute the 2022 paths...
+
+backgroundScriptCfg:
+  # Setup
+  cats: 'auto' # auto: automatically inferred from input ws
+  catOffset: 0 # add offset to category numbers (useful for categories from different allData.root files)
+  ext: 'backgroundPartialAnalysis' # extension to add to output directory
+  year: '2023' # Use combined when merging all years in category (for plots)
+  execution: 'local'
+
+trees2wsCfg:
+  inputTreeDir: 'DiphotonTree' # Name of RooDirectory storing input tree
+
+  apply_mass_cut: False # Apply cut on CMS_hgg_mass
+  mass_cut_range: '100,180' # CMS_hgg_mass cut range
+  doSTXSSplitting: False # Split output WS per STXS bin
+  doDiffSplitting: False # Split output WS per differential bin
+  doInOutSplitting: True # Split output WS into in/out fiducial based on some variable in the input trees (to be improved).
+  doSystematics: True # Add systematics datasets to output WS
+
+  # Variables to be added to dataframe: use wildcard * for common strings
+  mainVars:
+    - "CMS_hgg_mass"
+    - "weight"
+    - "weight_central"
+    - "dZ"
+    - "*Up"
+    - "*Down"
+    - "fiducialGeometricFlag"
+  
+  dataVars:
+    - "CMS_hgg_mass"
+    - "weight" # Vars to be added for data
+    
+  stxsVar: ''
+  diffVar: ''
+  notagVars: [] # Vars to add to NOTAG RooDataset
+  
+  systematicsVars:
+    - "CMS_hgg_mass"
+    - "weight"
+    - "fiducialGeometricFlag"
+  
+  # Do empty dict, only this way it works
+  theoryWeightContainers: {}
+    #weight_LHEPdf: 101
+    #weight_LHEScale: 9
+
+  # List of systematics: use string YEAR for year-dependent systematics
+  systematics:
+    - "ScaleEB"
+    - "ScaleEE"
+    - "Smearing"
+    #- "Material"
+    #- "FNUF"
+    - "energyErrShift"
+
+  cats: 'auto'
+
+  execution: 'local'
+
+signalScriptCfg_2023_preBPix:
+  # Setup
+  procs: 'auto' # if auto: inferred automatically from filenames (requires names to be of form *pythia8_{PROC}.root)
+  cats: 'auto' # if auto: inferred automatically from (0) workspace
+  ext: 'signalPartialAnalysis_2023preBPix' # output directory extension
+  analysis: 'earlyAnalysisInOut' # To specify replacement dataset and XS*BR mapping (defined in ./tools/replacementMap.py and ./tools/XSBRMap.py respectively)
+  year: '2023preBPix' # Use 'combined' if merging all years: not recommended
+  massPoints: '120,125,130' # You can now run with a single mass point if necessary
+
+  # Photon shape systematics  
+  scales: 'ScaleEE,ScaleEB' # separate nuisance per year
+  scalesCorr: ''
+  #scalesCorr: 'FNUF,Material' # correlated across years
+  scalesGlobal: 'NonLinearity,Geant4' # affect all processes equally, correlated across years
+  smears: 'Smearing' # separate nuisance per year
+
+  doPlots: True
+
+  beamspotWidthData: '3.2'
+  beamspotWidthMC: '3.732'
+
+
+  # Job submission options
+  execution: 'local'
+
+signalScriptCfg_2023_postBPix:
+  # Setup
+  procs: 'auto' # if auto: inferred automatically from filenames (requires names to be of form *pythia8_{PROC}.root)
+  cats: 'auto' # if auto: inferred automatically from (0) workspace
+  ext: 'signalPartialAnalysis_2023postBPix' # output directory extension
+  analysis: 'earlyAnalysisInOut' # To specify replacement dataset and XS*BR mapping (defined in ./tools/replacementMap.py and ./tools/XSBRMap.py respectively)
+  year: '2023postBPix' # Use 'combined' if merging all years: not recommended
+  massPoints: '120,125,130' # You can now run with a single mass point if necessary
+
+  # Photon shape systematics  
+  scales: 'ScaleEE,ScaleEB' # separate nuisance per year
+  scalesCorr: ''
+  #scalesCorr: 'FNUF,Material' # correlated across years
+  scalesGlobal: 'NonLinearity,Geant4' # affect all processes equally, correlated across years
+  smears: 'Smearing' # separate nuisance per year
+
+  doPlots: True
+
+  
+  beamspotWidthData: '3.5'
+  beamspotWidthMC: '3.732'
+
+  # Job submission options
+  execution: 'local'
+
+packaged_2023:
+  cats: 'auto' # if auto: inferred automatically from (0) workspace
+  massPoints: '120,125,130' # You can now run with a single mass point if necessary
+  mergeYears: True # Merges the eras
+  ext: ''
+  execution: 'local'
+
+datacard_yields:
+  cats: 'auto' # if auto: inferred automatically from (0) workspace
+  procs: 'auto' # if auto: inferred automatically from (0) workspace
+  ext: 'earlyAnalysis' # Extension for saving
+  # inputWSDirMap: '2023preBPix,2023postBPix'
+  sigModelWSDir: './Models/signal' # Input signal model WS directory
+  bkgModelWSDir: './Models/background' # Input background model WS directory
+  bkgModelExt: 'multipdf' # Extension used when saving background model
+  doSystematics: True # Include systematics calculations and add to datacard
+  mergeYears: True # Merge category across years
+  skipZeroes: True # Skip signal processes with 0 sum of weights
+  ignore_warnings: True # Skip errors for missing systematics. Instead output warning message
+  mass: '125' # Input workspace mass
+  skipBkg: False # Only add signal processes to datacard
+  bkgScaler: 1. # Add overall scale factor for background
+  skipCOWCorr: True # Skip centralObjectWeight correction for events in acceptance. Use if no centralObjectWeight in workspace
+  execution: 'local'
+
+datacard:
+  # For pruning processes
+  prune: True # Prune proc x cat which make up less than pruneThreshold (default 0.1%) of given total category
+  pruneThreshold: 0.001 # Threshold with which to prune proc x cat as fraction of total category yield (default=0.1%)
+  doTrueYield: False # For pruning: use true number of expected events for proc x cat i.e. Product(XS,BR,eff*acc,lumi). Use only if NOTAG dataset has been included. If false then will use nominal_yield (i.e. sumEntries)
+  analysis: 'earlyAnalysis' # Analysis extension: required for doTrueYield (see ./tools/XSBR.py for example)
+
+  # For yield/systematics:
+  skipCOWCorr: True # Skip centralObjectWeight correction for events in acceptance
+  doSystematics: True # Include systematics calculations and add to datacard
+  doMCStatUncertainty: True # Add uncertainty for MC stats
+  doSTXSMerging: False # Calculate additional migrations uncertainties for merged STXS bins (for 'mnorm' tier in systematics)
+  doSTXSScaleCorrelationScheme: False # Partially de-correlate scale uncertainties for different phase space regions
+
+  # For output:
+  saveDataFrame: False # Save final dataframe as pkl file
+  output: 'Datacard_2023' # Datacard name
+
+datacard_clean:
+  factor: 10000 # Factor beyond which uncertainty is considered incorrect and is removed
+  removeDoubleSided: False # Remove any nuisances which are listed as antisymmetric but both values point the same way
+  removeNonDiagonal: False # Remove any nuisances which are affect processes unimportant in that category
+  symmetrizeNuisance: 'CMS_hgg_SigmaEOverEShift,CMS_hgg_scale_0_shape,CMS_hgg_scale_1_shape,CMS_hgg_scale_3_shape' # Symmetrizes asymmetric nuisance. Enter in the format nuisance1,nuisance2,nuisance3,etc.
+  verbose: False # Spit out all the cleaning being done
+
+combine_fit:
+  asimov_numPoints: 30 # Number of points the Asimov generator should generate for the NLL scan
+  unblindedFit_numPoints: 100 
+  setParameterRange: "r=0,2"
+  cminApproxPreFitTolerance: 0.01
+  rMin: 0.7
+  rMax: 1.6
+  execution: 'local'
+
+combine_impacts:
+  exclude: "" # Nuisances that should be excluded from the impact computation
+  cminApproxPreFitTolerance: 0.01
+  setParameters: 1.000
+  stepSize: 0.05
+  setCrossingTolerance: 0.00005 
+  not_show_POI: False
+  execution: 'htcondor'
+
+combine_hesse:
+  execution: 'local'
+
+combine_mggToys:
+  nToys: 1000 # Number of toys = nToys * 10
+  loadSnapshot: 'MultiDimFit'
+  doBands: True
+  execution: 'htcondor'