Skip to content

Simplified ATLAS SUSY analysis framework

Holds a collections of SUSY analyses. These can be run over samples in different input formats:

  • DAOD_TRUTH (TRUTH1 and TRUTH3 tested)
  • xAOD (either truth-level and or reco-level - the latter with some constraints)
  • HepMC (uncompressed HepMC v2 tested)
  • slimmed ntuples (Reduced ntuples produced from above input)
  • DELPHES (Converter to slimmed ntuples provided)

It provides the analysis acceptance per control and signal region as well as optionally histograms or ntuples with event level objects.

Setting Up

First, you'll go inside of the docker image to set up the environment

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
$ docker pull gitlab-registry.cern.ch/atlas-sa/simple-analysis
$ docker run --rm -it gitlab-registry.cern.ch/atlas-sa/simple-analysis
             _ _____ _      _   ___
            /_\_   _| |    /_\ / __|
           / _ \| | | |__ / _ \\__ \
          /_/ \_\_| |____/_/ \_\___/

This is a self-contained ATLAS SimpleAnalysis image
based on AnalysisBase. To set up SimpleAnalysis and
the analysis release installed in the image, please
execute:

          source /release_setup.sh

For the latest documentation, please visit:

          https://simpleanalysis.docs.cern.ch/

Configured GCC from: /opt/lcg/gcc/8.3.0-cebb0/x86_64-centos7/bin/gcc
Configured AnalysisBase from: /usr/AnalysisBase/21.2.158/InstallArea/x86_64-centos7-gcc8-opt
Configured SimpleAnalysis from: /usr/SimpleAnalysis
[bash][atlas]:workdir >

at which point you can then run the code. All docker images can be found in the GitLab registry.

Release Setup

By default, the docker image will execute source /release_setup.sh for you.

Running

You can either run inside the docker image using the simpleAnalysis command

1
simpleAnalysis [-a listOfAnalysis] <inputFile1> [inputFile2]...

or outside of the docker image (mounting your current working directory in)

1
docker run --rm -it -v $HOME:$HOME -w $PWD gitlab-registry.cern.ch/atlas-sa/simple-analysis -a listOfAnalysis <inputFile1> [inputFile2]...

Docker Entrypoint

The docker file will only prefix simpleAnalysis to any commands you pass if you start with a dash (-). For example

1
docker run --rm -it -v $HOME:$HOME -w $PWD gitlab-registry.cern.ch/atlas-sa/simple-analysis inputFile1.root

will try to run inputFile1.root as a bash command instead. If you do not need any of the flags listed below, then

1
docker run --rm -it -v $HOME:$HOME -w $PWD gitlab-registry.cern.ch/atlas-sa/simple-analysis simpleAnalysis inputFile1.root

will work as expected.

This will run the analyses specified with -a option (or all if not given) over all of the input files and provide acceptances in a text file for each analysis (analysisName.txt) and histograms in a root file (analysisName.root).

The following additional commandline options that can be specified:

  • --nevents <num> limit the number of events processed
  • -l [--listanalysis] lists all available analyses
  • -n [ --ntuple] activates ntuple outputs
  • -o [--output] <name> merges the different analysis outputs into single text and root files
  • -w [--mcweight] <num> chooses which event weight to use in case of multiple. Default is 0 and weighting can be disabled by setting it to -1
  • -P [--pdfVariations] <initPDF> do PDF variations following U.L. method of https://arxiv.org/abs/1206.2892
  • -T [--useTrueTau] use p_T of true tau instead of visible p_T (not recommended, but for backward compatibility)

Outputs

SimpleAnalysis will produce a .txt and .root file, either one pair of files per analysis or a single pair if the -o option is used. In the latter case, everything gets prefixed with the analysis name to avoid name clashes.

The plain-text file contains the number of events accepted, the acceptance and its error for each signal region in the analysis in the form of a comma-separated table. The number of events is the actual number of MC events accepted, while the acceptance is calculated taking into account event weights (unless disabled with option -w -1). Besides the analysis defined signal regions, there is an _All line which gives the total number of events processed, the sum of all event weights and the sum of the square of the event weight, in that order. This can be used for normalization and merging results.

The root file(s) will contain all histograms defined in the analysis. Note all histogram filling is done with the event weight. If the -n option is used, the root file will also contain one ntuple per analysis with all the variables defined in the analysis code. The ntuple has one entry per event and in case the analysis code filling a variable was not reached in a given event, the value will either be 0 or an empty vector. This has to be accounted for in case 0 is a valid value for an ntuple variable. Besides the analysis defined variables, the ntuple has an Event number (counting from 1) branch, an event weight branch and a branch for each signal region. The latter will be 1 for events that passed that selection.

Slimming

For running over large input files more than once, it can be advantageous to first slim the files to an absolute minimum by making an ntuple with all the input objects. This can be done trivially with slimMaker. Minimum object pTs can be specified on the commandline, see slimMaker --help. The output can supplied to simpleAnalysis in the same way as DAOD or HepMC files and the program will automatically detect the type of input.

DELPHES input

SimpleAnalysis is not directly integrated with DELPHES. Instead a python script is provided which can be run in a DELPHES setup to convert the DELPHES output ROOT file into the simple ROOT TTree format which can be directly used by SimpleAnalysis

1
Delphes2SA.py <input Delphes file> <Output SA file>"
The DELPHES files should be produced with a detector card that matches the physics object performance of the specific analysis of interest.


Last update: February 16, 2022