CaloArt -- Generated samples for CaloChallenge datasets 2 and 3

Generated samples used for the evaluation of CaloArt, a refined DiT-style conditional flow-matching transformer for direct full-shower calorimeter simulation. This repository contains the generated calorimeter showers used for the evaluation of the final CaloArt models on CaloChallenge dataset 2 and dataset 3. The samples are provided in HDF5 format and correspond to the configurations reported in the CaloArt paper. The repository also includes two zip files containing the high-level-feature histogram outputs used for the evaluation on the two datasets. Included files: - samples_ds2.hdf5: generated CaloArt samples for CaloChallenge dataset 2.- samples_ds3.hdf5: generated CaloArt samples for CaloChallenge dataset 3.- CaloChallenge_ds2_hlfs.zip: high-level-feature histogram outputs for CaloChallenge dataset 2.- CaloChallenge_ds3_hlfs.zip: high-level-feature histogram outputs for CaloChallenge dataset 3. The high-level-feature histograms include the standard CaloChallenge observables used in the paper, including: - ratio between total deposited energy and incident energy;- layer energy deposition;- layer-wise center of energy;- layer-wise width of the energy distribution;- layer-wise sparsity;- voxel energy deposition;- weighted shower depth in radial slices;- weighted shower depth in angular slices.