Skip to content

Latest commit

 

History

History
100 lines (83 loc) · 5.08 KB

README.md

File metadata and controls

100 lines (83 loc) · 5.08 KB
Raw

RateEstimate

Python script used to estimate the rate of HLT

To get steam rates, can use this script and follow the step:

  1. make trigger map.

  2. set RateEstimate.py.

  3. submit jobs to bqueues.

  4. merge banch results together and calculate rates.

  5. make trigger map. make a .py script, which will be imported into RateEstimate.py in step 2. This .py can provide group, dataset, stream, prescale information.

script: triggersGroupMap/makeTriggerMap.py parameters: from line 1 to line 10. file_menu = 'HLT_Menu_v4.2_v6.tsv' # a .tsv table include menu information, download from google doc, ask it from Nadir currently. file_sample = 'config/triggerDatasetMapTail.py' # no need to modify file_output = 'HLT_Menu_v4p2_v6.py' # name of output .py file triggerName = 'tttt' # just a label column_stream = 0 # column of stream, i.e: column "A" is 0, column "AB" is 27 column_dataset = 1 column_trigger = 2 column_group = 4 column_type = -1 # these columns can be set to -1, if no related column in .tsv column_ps = 10

  1. set RateEstimate.py

script: RateEstimate.py this script can read Ntuples and output fired count. This script can also be submitted to bqueues, one root file one job. parameters: from line 1 to line 60 from triggersGroupMap.HLT_Menu_v4p2_v6 import * # trigger map, made in step 1 from datasetCrossSections.datasetCrossSectionsHLTPhysics import * # provide the name of dataset

batchSplit = False		# if batch jobs and submit to bqueues, set True, recommanded.
batchSplit = True
looping = False			# there are two method to get fired count, tree.Draw() or looping over events. Draw() is fast but less features, looping is slow but more features, for data, looping is recommanded.
looping = True

folder = '/store/group/dpg_trigger/comm_trigger/TriggerStudiesGroup/STEAM/Run2016H/HLTPhysics_2016H_menu4p2/HLTPhysics2016H_DB' # directory to Ntuples
localdir = '/afs/cern.ch/user/x/xgao/eos/cms'		# in case test locally, will not use if batchSplit = True
lumi = 1              # luminosity [s-1cm-2]		# just a label
if (batchSplit): multiprocess = 1           # number of processes
else: multiprocess = 1 # 8 multiprocessing disbaled for now because of incompatibilities with the way the files are accessed. Need some development.
pileupMAX = 100			# For MC
pileupMIN = 0			# For MC
pileupFilter = False        	# For MC use pile-up filter?
pileupFilterGen = False    	# For MC use pile-up filter gen or L1?
useEMEnriched = False       	# For MC use plain QCD mu-enriched samples (Pt30to170)?
useMuEnriched = False       	# For MC use plain QCD EM-enriched samples (Pt30to170)?
evalL1 = False              	# For MC evaluate L1 triggers rates? code can't work for 2016 L1 rates. set False
evalHLTpaths = True        # evaluate HLT triggers rates?
evalHLTgroups = True       # evaluate HLT triggers groups rates and global HLT and L1 rates
evalHLTprimaryDatasets = True # evaluate HLT triggers primary datasets rates and global HLT and L1 rates
evalHLTprimaryDatasets_core = True # evaluate HLT triggers primary datasets rates and global HLT and L1 rates
evalHLTTrigger_primaryDatasets_core = True # evaluate HLT triggers primary datasets rates and global HLT and L1 rates
evalHLTstream = True # evaluate HLT triggers primary datasets rates and global HLT and L1 rates
#evalHLTtwopaths = True    # evaluate the coreelation among the HLT trigger paths rates?
evalHLTtwogroups = False   # evaluate the coreelation among the HLT trigger groups rates?
evalPureRate_Group = True
evalPureRate_Dataset = True
evalPureRate_Stream = True
evalExclusive_Trigger = True
evalExclusive_group = True
evalExclusive_dataset = False
evalExclusive_dataset = False
use_json = True			# if use json to filter event
json_file_name = '/afs/cern.ch/user/x/xgao/work/RateEstimate_31_08_2016/test_4.2/1p15/columns_2016H/columns_1p15e34.txt'
label = "test"         		# a label
runNo = "0"           		# if runNo='0' and use_json = False, means code will run for all Run. 
LS_min = '0'			# default is 0
LS_max = '9999'            	# default is 9999

isData = True
  1. submit jobs to bqueues.

script: scripts/RatesBatchScript.py this script can scan Ntuples directory, create job script for each root file and submit them to bqueues 8nh. parameters: from line 8 to line 12 note that the modules and Ntuples directory have to be same as one in RateEstimate.py

  1. merge banch results together and calculate rates.

script: scripts/mergeRates_path_v2.py this script for individual trigger from triggersGroupMap.HLT_Menu_v4p2_v6 import * # should be same in RateEstimate.py from datasetCrossSections.datasetCrossSectionsHLTPhysics import * # should be same in RateEstimate.py

Method = 1 #0: rate = count ; 1:HLT, rate = psNorm*count / LS*nLS ; 2:Zerobias, rate = 11245Hz * target nBunchs * nCount/total Event
LS = 23.31
PsNorm = 107*4.i	# prescale of dataset
nLS = 246-43+1		
nLS = 0			# recommand set 0, then code will calculate how many lumi section has been run over during read ntuples automatically.
ps_const = 11245.0*2200.0

script: scripts/mergeRates_GroupDataset_v2.py this script for group, dataset, stream