Generate jet-images from Delphes output and train a LAGAN to mimic them

Code from my Part III research project, originally housed at pt3proj. All processes tested on Python 2.7.

Batch run Delphes

To batch produce events from Pythia and Delphes use run_delphes.py in the directory jetimage. Requires Delphes installation with internal Pythia8 https://cp3.irmp.ucl.ac.be/projects/delphes/wiki/WorkBook/Pythia8 Make sure the directory containing DelphesPythia8 command is in your PATH.

Requirements, tested on:

• ROOT v5.34
• Delphes v3.3.3 (Delphes installation directory must be in ROOT dynamic path, set in .rootrc).

run_delphes.py generates Pythia+Delphes events for wprime and qcd processes

Usage:
    run_delphes.py <process> <file_name> <n_events> [-s=SEED] [--boson-mass=BOSON_MASS] [--pt-hat-min=PTHMIN] [--pt-hat-max=PTHMAX]

Arguments:
    <process>    Process to generate events for
    <file_name>  Root file output name
    <n_events>   Number of events

Options:
    -s=SEED                   Random seed
    --boson-mass=BOSON_MASS   Specify boson mass GeV [default: 800]
    --pt-hat-min=PTHMIN       pthatmin GeV [default: 100]
    --pt-hat-max=PTHMAX       pthatmax GeV [default: 500]

Image processing

data_gen.py converts Delphes ROOT files to jet images and outputs hdf files

Usage:
    data_gen.py <in_file> [-w] [-o <out_file>] [--ptmin=<ptmin>] [--ptmax=<ptmax>]
    data_gen.py -h | --help

Arguments:
    <in_file>  Root file to extract from

Options:
    -h --help        Show this screen
    -o <out_file>    file to save to
    --ptmin=<ptmin>  Minimum pT of jets in GeV [default: 250]
    --ptmax=<ptmax>  Maximum pT of jets in GeV [default: 300]

Requirements, tested on:

• numpy v1.12.1
• scikit-image v0.12.3
• h5py v2.7.0
• tqdm

Training

See helper files in /proc for converting image hdf files in to a format suitable for training. Contents of /models contains training code, from adversarial-jets, see that repository for usage, or just run python models/train.py -h.

Addditional requirements, tested on:

• keras v1.2
• tensorflow v0.11