Don't bother trying to understand how all of it works. It just does.
Be sure to install Python3, Tensorflow 2.X, h5py, scikit-learn and the usual packages. If something's missing, you will know anyways.
The paths are stored as key-value pairs in the form of Python dictionaries present in the following scripts (only necessary ones to get it all running are mentioned below):
-
pofah/path_constants/sample_dict.py
- base directory where all samples are stored:
base_dir_events
- base directory where all samples are stored:
-
pofah/path_constants/sample_dict_file_parts_input.py
- same as above:
base_dir
- directories where individual samples are stored:
sample_dir
qcdSigTrain
: Data on which the VAE is to be trained (likewise for test data)qcdSigMCOrig
: The original MC signals before sampling is carried out to replace it with QCD jets having similar kinematics.
- same as above:
-
pofah/path_constants/sample_dict_file_parts_reco.py
- writing the reconstructed files to disk:
base_dir
(these files are moderately large so watch your disk quota)
- writing the reconstructed files to disk:
-
pofah/path_constants/experiment_dict.py
- saving trained model:
model_dir
- rest should be self-explanatory.
- saving trained model:
Once you are done setting the paths, move to vande/
. Don't look at the other folders for now.