README.md

Example CAFCore Analysis

This repository is meant as an example for how to construct an analysis using the CAFCore framework. In order to begin a new analysis, simply fork this project and modify the scripts and configs which are available here.

Cloning the project

# Use -c el9+batch if you're not on a el9 machine. Otherwise, you may use setupATLAS as normal.
setupATLAS -c el9+batch
lsetup git
mkdir AnalysisExample
cd AnalysisExample

# Note that you should `setupATLAS` in a container that allows AnalysisBase access.
# There are a few different protocol options for cloning the project, which are all provided at the top of the main page of the repository.
# Kerberos is typically recommended if it is available (e.g. lxplus) since it does not require a username or password when interacting with remote repositories.
# ssh acts similarly, but requires a password and a bit of initial setup
# https is usually the most robust, but always requres a username and password

# ssh
git clone --recursive ssh://git@gitlab.cern.ch:7999/atlas-caf/CAFExample.git
# Kerberos
# git clone --recursive https://:@gitlab.cern.ch:8443/atlas-caf/CAFExample.git
# https
# git clone --recursive https://gitlab.cern.ch/atlas-caf/CAFExample.git

Building the project

mkdir build
cd build
asetup AnalysisBase,25.2.27
cmake ../CAFExample
source setupAnalysis.sh
cafbuild # build the code (check details by typing "type cafbuild")

Running a minimal example

The following commands will reproduce (in seconds) a minimal example, showcasing the least amount of configuration necessary to produce results by taking a VBF signal MC sample and passing it through a single cut on Mjj while producing one histogram.

cd ../CAFExample/share
./prepare.py minimal/config/master/prepare-Minimal-Example.cfg
./initialize.py minimal/config/master/initialize-Minimal-Example.cfg
./analyze.py minimal/config/master/analyze-Minimal-Example.cfg
./visualize.py minimal/config/master/visualize-Minimal-Example.cfg

Running an example analysis on flat nTuples

The following commands will (very quickly) run over a selection of flat nTuples to reproduce the cutflow and some visualized results for a Zjets Fake Factor analysis. It is not meant to describe the bare minimum configuration needed for producing results (that's what the minimal example above is for). Similarly, a full-blown analysis is likely to include a host of custom observables which are calculated on-the-fly during runtime and for many more events (all quantities in this case have been pre-computed). Rather, it is simply meant to showcase how fast results can be obtained once the uninteresting events have been skimmed away and the necessary quantities are already available directly in the TTree.

cd ../CAFExample/share
./prepare.py flatNTuple/config/master/prepare-flatNTuple-Example.cfg
./initialize.py flatNTuple/config/master/initialize-flatNTuple-Example.cfg
./analyze.py flatNTuple/config/master/analyze-flatNTuple-Example.cfg
./visualize.py flatNTuple/config/master/visualize-flatNTuple-Example.cfg

Running an example analysis on HHML PxAOD inputs

The following commands will run over a selection of PxAOD inputs (mc23a and data 2022) to produce different kinematic distributions in WZ CR for 3L channel.

cd ../CAFExample/share
./prepare.py HHML_example/config/master/prepare-HHML_3L.cfg
./initialize.py HHML_example/config/master/initialize-HHML_3L.cfg
./analyze.py HHML_example/config/master/analyze-HHML_3L.cfg
./visualize.py HHML_example/config/master/visualize-HHML_3L.cfg

Running an example analysis using easyjet ntuple output

The following commands will run over a test ntuple and produce the leading electron pt in 2LSC channel

./prepare.py easyjet/config/master/prepare-2LSS.cfg 
./initialize.py easyjet_example/config/master/initialize-2LSC.cfg
./analyze.py easyjet_example/config/master/analyze-2LSC.cfg

Running from an arbitrary location

Alternatively, each of the analysis scripts above can also be run from an arbitrary directory. The master config files still have to be specified relative to the share folder, although they can be discovered via tab completion:

e.g.

cd any/location/
./prepare.py flatNTuple/config/master/prepare-flatNTuple-Example.cfg
./initialize.py flatNTuple/config/master/initialize-flatNTuple-Example.cfg
./analyze.py flatNTuple/config/master/analyze-flatNTuple-Example.cfg
./visualize.py flatNTuple/config/master/visualize-flatNTuple-Example.cfg

On Every Login

Navigate to the working directory and

setupATLAS
lsetup git
cd build
asetup --restore
source setupAnalysis.sh