Accompanying dataset for the paper: "AutoPot: Automated and massively parallelized construction of Machine-Learning Potentials"

Contributions Max Hodapp did contribute to the convertion of existing MLIP trainings into a Motoko workflow. This led to the Autopot repository. Guillaume Anciaux did the code architecture conception of Motoko workflow manager, built upon Blackdynamite, a parametric study helper. Funding sources Max Hodapp gratefully acknowledges the financial support by the Austrian Federal Ministry for Labour and Economy and the National Foundation for Research, Technology and Development and the Christian Doppler Research Association. Data structure and information The directory data contains the resulting MLIP presented in the paper for MoTa and W cases, as well as the full execution result for the unary example which is part of the autopot package. In details: data : workflow and results export supporting the paper MoTa : MLIP produced by the workflow for Molybdenum-Tantalum alloy, as presented in paper MoTa_elastic_constants.pdf : elastic constants report MoTa_lattice_constants.pdf : lattice constants report MoTa_lev16.mtp : configuration file for moment tensor potential MoTa_stacking_fault_energies.pdf : stacking fault report check_errors.ipynb : cheking validity of produced potential and produces figures W : MLIP produced by the workflow for Tungsten, as presented in paper W_dd_plot.pdf : differential displacement plot W_energy_curve.pdf : minimum energy path for screw dislocation W_lev16.mtp : configuration file for moment tensor potential check_errors.ipynb : checking validity of produced potential and produces figures unary : Full workflow execution of the unary autopot example check_errors.ipynb : checking validity of produced potential and produces figures create_configurations.ipynb : Creates the initial configuration set init_training_cfgs.xyz : initial configurations used as initial training set lammps.log : lammps log produced while checking errors (check_errors.ipynb execution) md_cfgs.xyz : configurations for intitial md_select task mlip.ini : produced by check_errors.ipynb workflow : MLIP workflow full description and execution files motoko.yaml : workflow metadata description orchestrator.py : python script providing the orchestration wf.zip : workflow database saved as a zip eam : sp_calc Task, using eam classical MD runs BD-sp_calc-runs.zip : all run produced files bd.yaml : Metadata description for the task bd.zip : task database saved as a zip doIt.py : Actual execution script launch.sh : Bash setup for environment (mlp, potential paths, ...) md_select : MD selection Task BD-md_select-runs.zip : all run produced bd.yaml : Metadata description for the task bd.zip : task database saved as a zip doIt.py : Actual execution script launch.sh : Bash setup for environment (mlp, potential paths, ...) select : select Task, based on configuration grade BD-select-runs.zip : all run produced bd.yaml : Metadata description for the task bd.zip : task database saved as a zip doIt.py : Actual execution script launch.sh : Bash setup for environment (mlp, potential paths, ...) train : train Task, based on mlip-2 BD-train-runs.zip : all run produced bd.yaml : Metadata description for the task bd.zip : task database saved as a zip doIt.py : Actual execution script launch.sh : Bash setup for environment (mlp, potential paths, ...) install-deps.sh : Script installing all necessary dependencies to execute the workflow (tested on debian) pkg.txt : list of debian packages to install as dependencies reproduce.sh : Clone autopot, install dependencies and launch the unary example requirements.txt : Python package dependencies ro-crate-metadata.json : Metadata extracted from the dataset