A general purpose Gaussian Approximation Potential (GAP) [1,2] for iron. The training database has been computed at the PBE level of theory [3] using the VASP code [4-7]. Fitting of the potentials was done using QUIP/GAP [1,2,8]. The potential uses 2-body (distance_2b), 3-body (angle_3b) and SOAP (soap_turbo) [9,10] descriptors, as implemented in the TurboGAP code [11]. All files necessary to use the potential QUIP/GAP (compiled with the TurboGAP libraries) are in QUIP_files.tar.gz, all files to use with TurboGAP in TurboGAP_files.tar.gz. The database used to train the potential is in training_database.tar.gz. More details can be found in this publication: Searching for iron nanoparticles with a general-purpose Gaussian approximation potential Richard Jana, Miguel A. Caro https://doi.org/10.1103/PhysRevB.107.245421 The authors are grateful to the Academy of Finland for financial support under projects #321713 (R. J. & M.A. C.) and #330488 (M.A. C.), and CSC -- IT Center for Science as well as Aalto University's Science-IT Project for computational resources.

2023