L2LFlows: Generating High-Fidelity 3D Calorimeter Images

DOI original : 10.5281/zenodo.8284809 | Plateforme : Zenodo

This upload contains the datasets used in arXiv:2302.11594. The file g4-showers_950k_10x10_train_val_test.pt contains the 760k training, 95k validation and 95k test showers as well as their incident energies. It should be loaded as follows: 

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 

import torch 

list_tensors = torch.load(args.file_path)

for (idx, tensor) in enumerate(list_tensors):

    [showers_train, showers_val, showers_test, inc_energies_train, inc_energies_val, inc_energies_test] = list_tensors

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 

The file g4-showers_665k_10x10_test.pt contains 665k additional showers that were used for the classifier scaling studies, in addition to the 95k test showers from the file g4-showers_950k_10x10_train_val_test.pt. It should be loaded as follows: 

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 

import torch

list_tensors = torch.load("g4-showers_950k_10x10_train_val_test.pt") 

[showers_geant, inc_energies_geant] = list_tensors

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 

A detailed description of how the datasets were simulated can be found in the paper.