kaulmer / l1t-macros

#l1t-macros

How to run

1. Setup the dataset config file

Move into the Config folder and change the relevant config files.

ntuple_cfg.h contains the information on the dataset and where to get them. To add a dataset simply:

• Create a new function which returns a ntuple_cfg object and set the following variables
  • sampleName for a shortened version of the sample's name, e.g. "Data"
  • sampleTitle for a longer version of the sample's name, e.g. "2016 Data"
  • triggerName for data, e.g. "SingleMu" or "ZeroBias"
  • triggerTitle for data, e.g. "Single Muon" or "Zero Bias"
  • puFilename for the pu weights for reweighting the PU distribution of MC samples (this file is produced using MakePlots/generatePUWeights.cxx
  • run for the run number of data
  • outDirBase for the base directory of all output files
  • doFit to add a relevent fit to the plots. If there is not relevant fit, then no fit is done
  • puType is a vector of strings in the format of {"0PU20","20PU40","40PU"}
  • puBins is a vector of ints and should match puType, e.g. {0,20,40,999}. The final bin is significantly large to be effectively inifinite
  • inFiles is a vector of strings for the location of the ntuples. I tend to use the EOS locations (examples are shown in ntuple_cfg.h). Also replace the ntuple number with a wildcard, e.g. "L1Ntuple_*.root"
  • baseOWdir is the directory inside outDirBase that contains all the hadded files from a set of jobs.
  • outDir is the directory inside outDirBase where the output files will go. This will include the date in it's name
• Then return the ntuple_cfg object
• Return your newly created function in the GetNtuple_cfg() function to use your dataset

2. Setup the plot config file

I will be using the jetTurnons_cfg.h as an example here (similar things apply to the other files).

The function jetTurnons returns a map of strings to TL1Turnon* (the class to produce turnon curves). This contains all turnons for jets (e.g. barrel, endcap, central and HF turnons).

To add your own turnon:

• Create a new function that returns a TL1Turnon and emplace it in the jetTurnons function
• The function should take a ntuple_cfg as the argument to get details of the dataset
• Set the following variables:
  • seed is the name of the variable that is the seed for the turnon
  • xparam is the x-axis parameter of the turnon (y-axis is efficiency)
  • outName is the output name of the histograms (follow similar naming schemes as with the other turnon functions)
  • SetOverwriteNames takes 2 arguments. The first is the root-file associated with this turnon in the hadded directory of a set of jobs. The second is the name of the histogram inside the root-file for this particular turnon (with the seed value and PU value suppressed)
  • SetSeed takes 2 arguments. The first is seed and the second is a longer version of it
  • SetSeeds takes a vector of doubles which will be the set of seed values used. The first must always be 0.0 (i.e. no seed)
  • SetX takes 2 arguments. The first is xparam and the second is a longer version of xparam
  • SetXBins takes a vector of doubles representing all edges of the histogram's bins. I typically use separate functions to create these vectors
  • SetOutName takes outName as an argument
  • SetFit takes dataset->doFit as an argument
  • SetAddMark places additional text on the final canvas
• Return the TL1Turnon

We now need to add the logic to fill your turnon:

• Open the file MakePlots/makeJetTurnons.cxx and go inside the for-loop after the // Loop comment
• Use continue; to skip an event that is not desired (e.g. if it doesn't pass some filter defined in Events/TL1Events.h)
• Get the relevant event parameters from the TL1Event object (named event) and store them
• Fill the turnons object like so:

turnons[turnonName]->Fill(xparam, seed, pu);

where turnonName is a string representing the name chosen in the config file, xparam is the value of the x-axis parameter, seed is the value of the seed for that particular event and pu is the number of PU vertices (the default value for pu is zero which is understood as the PU is not give).

3. Run the submission script on a batch system

To submit jobs to the CERN batch (simple to extend to other batch systems) using the following command:

export L1T=/afs/cern.ch/work/s/sbreeze/l1tClasses # Make this point to your dir
./batchSubmit.sh submitJetTurnons.sh $NJOBS
bjobs # check the status of your jobs

where $NJOBS is the number of jobs to submit.

Now the histograms must be hadded together using the script Scripts/haddBatchFiles.py like so:

export CHUNKDIR=/path/to/output_CHUNK*/Turnons
python Scripts/haddBatchFiles.py -i "${CHUNKDIR}"

You must have a wildcard so that all CHUNKS (individual jobs) are hadded. The quotes in "${CHUNKDIR}" must be present to allow the wildcard to work properly. This script will delete the individual CHUNKS and leave a HADDED directory.

Now we create the plots from the hadded root-files. Double-check the config file Config/jetTurnons_cfg.h" and make sure that the SetOverwriteNames` takes the correct arguments. Once certain that they do run the following:

./endJob.sh submitJetTurnons.sh

Now the hadded output directory should contain all turnon plots and root-files.

You can create scripts to mix and match histograms from the root-files.

How to get variables from the ntuples

1. Create an object (named event for example) of class TL1EventClass with a single input argument, a vector of strings representing the directories where the root-files are located
2. Call the Next() function of class TL1EventClass. This returns a boolean, true if there is a next event, false if not. Hence this can be placed in the argument of a while statement
3. If we want to access the sums tree from l1RecoJetTree we can get to this like so:

       event->GetPEvent()->fSums   

4. The names for the ntuple trees are in Event/TL1PrimitiveEvent.h. GetPEvent() returns a constant to an object of class TL1PrimitiveEventClass. The trees from TL1PrimitiveEventClass are public, so caution must be taken if creating an object of TL1PrimitiveEventClass
5. To get the level-1 upgrade tree parameters (since these are stored in arrays) they are accessed from TL1EventClass and are public (due to laziness), e.g. fL1Met, fL1EmuMetHF, ...