Electromagnetic + Hadronic Sampling Calorimeter Shower Images

We include multiple hdf5 files:

The files 'piplus_1.hdf5' and 'piplus_2.hdf5' each contains 100,000 calorimeter showers originating from incoming charged pions with incident energies uniformly distributed in the range [1,100] GeV. The file 'piplus_1.hdf5' was used to train our models. The second file 'piplus_2.hdf5' was used in evaluation of generation performance.

The file 'piplus_log.hdf5' contains 100,000 calorimeter showers originating from incoming charged pions with incident energies log-uniformly distributed in the range [1,100] GeV.

The files with names 'piplus_{E_true}.hdf5' each contain 100,000 calorimeter showers originating from incoming charged pions with fixed incident energy at E_true = {10, 20, 30, 40, 50, 60, 70, 80, 90} GeV. These files were used for evaluation of calibration performance.

Each file has the following structure:

energy    Dataset {100000, 1}

layer_0   Dataset {100000, 3, 96}

layer_1   Dataset {100000, 12, 12}

layer_2   Dataset {100000, 12, 6}

layer_3   Dataset {100000, 3, 96}

layer_4  Dataset {100000, 12, 12}

layer_5   Dataset {100000, 12, 6}

overflow  Dataset {100000, 6}

The sampling calorimeter we built is segmented longitudinally into six layers with different depths and granularities. In units of mm, the six layers have the following (eta, phi, z) dimensions:Layer 0: (5, 160, 90) | Layer 1: (40, 40, 347) | Layer 2: (80, 40, 43) | Layer 3: (20.83, 666.67, 375) | Layer 4: (166.67, 166.67, 667) | Layer 5: (333.33, 166.67, 958)