This page will document the main tools available in the framework, in roughly the order they would be used in an analysis workflow.

An example cmsRun config can be found in NTupler/test/wgamma_cfg.py.

Most object collection plugins are dervied from a common EDProducer: AcornBaseProducer. This is a template class, where the template parameter should be the type of collection our derived producer will make:

        class AcornMuonProducer : public AcornBaseProducer<std::vector<ac::Muon>>

The AcornBaseProducer handles a few common tasks automatically:

• Provides a branch setting in the PSet to specify the branch name in the output TTree, e.g. branch=cms.string('photons').
• Takes care of creating a pointer to the output collection, and connecting this to the output TTree (which is itself created by the AcornEventProducer).
• Provides a select setting that can be used to specify exactly which variables should be saved in the output object, and optionally allows the numerical precision to be truncated (inspired by miniAOD/nanoAOD). The syntax is similar to that used in the GenParticlePruner: a series of keep and drop rules are given in order, which is then checked for each variable. The basic pattern is keep/drop regex[=N], where the =N is optional, and specifies the number of bits on the mantissa to keep for floating point variables.
• Works correctly as part of the CMSSW multi-threaded framework (if enabled in the config via process.options.numberOfThreads = cms.untracked.uint32(N) and process.options.numberOfStreams = cms.untracked.uint32(N))

Can use a json file to keep track of all the samples to be processed, and what cmsRun options should be used with which sample. See NTupler/test/samples/wgamma_2016.json for an example. The basic format is:

{
  "samples": {
    "SingleMuon-2016G": {
      "dataset": "/SingleMuon/Run2016G-03Feb2017-v1/MINIAOD",
      "attributes": ["submit"],
      "events": 149916849,
      "config": "data_reproc"
    },
    "SingleMuon-2016H-ver2": {
      "dataset": "/SingleMuon/Run2016H-03Feb2017_ver2-v1/MINIAOD",
      "attributes": ["submit"],
      "events": 169642135,
      "config": "data_prompt"
    }
  },
  "configs": {
    "data_reproc": ["year=2016", "isData=1", "globalTag=80X_dataRun2_2016SeptRepro_v7", "cores=1"],
    "data_prompt": ["year=2016", "isData=1", "globalTag=80X_dataRun2_Prompt_v16", "cores=1"]
  }
}

Each sample gives the dataset name and a config label, which is mapped to the set of options at the bottom. It is not essential to fill the number of events, but could be useful for book-keeping later. The list of attributes are to be used by other scripts which might want to select some subset of the entries in the json to act on.

Example usage:

python test/cmsRunTest.py test/samples/wgamma_2016.json --samples ZZTo4L-amcatnloFXFX test/wgamma_cfg.py

The first option is the chosen json file, the --samples option is a comma separated list of samples to check from that file. All remaining arguments will be passed through to cmsRun (here we just give the cmsRun config). This script will query DBS to find a file from the sample, and will automatically add an option input=[filename] to the cmsRun config - so important to make sure your config will accept this option.

A barebones cmsRun config that just runs the GenXSecAnalyzer is available in test/GenXSecAnalyzer_cfg.py. Can be used in conjunction with cmsRunTest.py:

python test/cmsRunTest.py test/samples/wgamma_2016.json --samples WZTo1L3Nu-amcatnloFXFX --no-cfg --nfiles 3 test/GenXSecAnalyzer_cfg.py maxEvents=-1

Here the --no-cfg option is added to suppress the automatic addition of the cmsRun command line options from the json file, and --nfiles X lets us process more files from the sample to increase the stats for the cross section info.

A script for submitting jobs with crab:

python test/crabSubmit.py test/samples/wgamma_2016.json -p test/wgamma_cfg.py -l wgamma_2016_v1 -u 250000 --attribute submitgenonly -v 1 [--submit]

Where the first argument is the json file, -p is the cmsRun config, -l is a label for the production (config.General.workArea and config.Data.outLFNDirBase will use this), -u sets the unitsPerJob (i.e. number of events per job) and -v is the verbosity of the script. The --attribute option can be used to submit only the samples that have that particular attribute assigned in the json file.

FIXME: the output path is current hard-coded - needs to be an option!

The script only needs the crab directory, from this it will check all expected output files are present and if so will produce the filelist:

python ../test/crabFilelist.py crab_wgamma_2016_v1_WGToLNuG-EFT-madgraphMLM

Additional options can be specified: --replace-xrootd can be used to set the local xrootd server, otherwise a global redirector will be used, making access slightly slower. The default is to replace with the CERN EOS server eoscms.cern.ch, so should override this at other sites. The --prefix option adds the given string to the beginning of the filelist (i.e. [prefix]wgamma_2016_v1_WGToLNuG-EFT-madgraphMLM.txt for the above example).