Supporting data for manuscript: "Active Δ-learning with universal potentials for global structure optimization"

Datasets and Python scripts supporting the findings presented in the manuscript "Active $\Delta$-learning with universal potentials for global structure optimization". Contained are scripts to reproduce the results using AGOX v. 3.10.2 rss_sript.py: Random Structure Search AGOX script for $[\mathrm{Ag}_2\mathrm{S}]_X$ clusters. bh_script.py: Basin Hopping AGOX script for $[\mathrm{Ag}_2\mathrm{S}]_X$ clusters. gofee_script.py: GOFEE AGOX script for $[\mathrm{Ag}_2\mathrm{S}]_X$ clusters. rex_script.py: Replica Exchange X AGOX script for $[\mathrm{Ag}_2\mathrm{S}]_X$ clusters. rss_surface_script.py: Random structure search AGOX script for $(\sqrt{7} \times \sqrt{7})$ Ag(111) sulfur reconstructed surface. rex_surface_script.py: Replica Exchange X AGOX script for $(\sqrt{17} \times \sqrt{17})$ Ag(100) sulfur reconstructed surface. And datasets: figure_2_ags_clusters/Ag<2*size>S<size>_<umlip>.xyz: Global minima configurations of Ag<2*size>S<size> clusters with different <umlip> for size = [4, 6, 8, 10, 12] and umlip = [chg, mace, mace_mpa]. figure_3_ags_clusters/ag16s8_<umlip>.xyz: Minima configurations for the $[\mathrm{Ag}_2\mathrm{S}]_8$ clusters. figure_6_ag111_sq7xsq7/dft.xyz: DFT Global minimum configuration for the $(\sqrt{7} \times \sqrt{7})$ Ag(111) sulfur reconstructed surface. figure_7_ag100_sq17xsq17/<mulip>.xyz: DFT, CHGNet, MACE-MP0 and MACE-MPA global minima configurations for the $(\sqrt{17} \times \sqrt{17})$ Ag(100) sulfur reconstructed surface. figure_8_pretraining_data/training.xyz: Training data to pretrain $\Delta$-model for $(\sqrt{17} \times \sqrt{17})$ Ag(100) sulfur reconstructed surface. ag17x17.xyz: $(\sqrt{17} \times \sqrt{17})$ Ag(111) surface slab. ag7x7.xyz: $(\sqrt{7} \times \sqrt{7})$ Ag(100) surface slab.