Jet clustering with scikit-learn on lxplus.cern.ch

Instructions to setup the working area on lxplus:

export SCRAM_ARCH=slc6_amd64_gcc530
cmsrel CMSSW_8_0_26_patch2
cd CMSSW_8_0_26_patch2/src
cmsenv
git clone [email protected]:amadio/jet-clustering.git
cd jet-clustering
jet-constituents 10
for csv in data/\*.csv; do jet-clusters ${csv}; done

The code picks the first n events from the dataset

• BulkGravToZZToZhadZhad_narrow_M-1800_13TeV-madgraph

For each event, a csv data file is created with the list of jet constituents.

Data file format

Each data file contains the list of jet constituents with kinematic information necessary to seed the clusterization algorithm. The name of the data files follow the syntax event-%04d.csv. The momentum coordinates are normalized.

Plotting data and clusters

The jet-clusters script takes a data file as argument and generates a PNG file with the plots of real and fitted clusters.

Example using RecoJets

Example using GenJets