End-to-end event reconstruction for precision physics at future colliders dataset

This dataset contains the raw version of the training and evaluation data used for the paper "End-to-end event reconstruction for precision physics at future colliders". Every file contains 100 events and includes generator particles, reconstructed tracks and calorimeter hits and their respective links in the EDM4HEP format. Evaluation files also contain reconstructed Pandora PF particles and their links. The dataset was generated using Key4HEP (2026-01-14) and version CLD_o2_v05 of the CLD detector concept for FCC-ee. For all datasets the detector simulation was done with Geant4 and the CLDConfig repository was used which has an implementation of PandoraPFA (our baseline).The included datasets are: cld_edm_Z_uds_root: pythia Zqq q=(uds) at 91 GeV sample used for training the clustering network (1M total) cld_edm_gun_root: particle gun with balanced PID classes used for training the particle properties networks (200k total) Contents This record includes the following files: cld_edm_Z_uds_root_part_*.tgz (1000 files included)~ 10k events cld_edm_gun_root: (1000 files included)~ 10k events cld_edm_Z_uds_eval_root: (1000 files included)~ 10k events dataset_full.txt: the full list of files, hosted on the EOS storage system at CERN HitPF_datageneration-main.zip: the code with the files for data generation which includes event generation (gun/pythia), ddsim for detector simulation, CLD reconstruction following CLDConfig repository and the ML dataset creation in parquet files. Dataset semantics Each file consists of 100 events. Each event contains structured branches including collections in the edm4hep format. The branches relevant to hit-based particle flow reconstruction include: MCParticles: the ground truth generator particles ECALBarrel, ECALEndcap, ECALOther, HCALBarrel, HCALEndcap, HCALOther, MUON: reconstructed hits in the various calorimeter subsystems SiTracks_Refitted: the reconstructed tracks PandoraClusters: the calorimeter hits, clustered by Pandora to calorimeter clusters CalohitMCTruthLink: the links between MC particles and reconstructed calorimeter hits SiTracksMCTruthLink: the links between MC particles and reconstructed tracks