Data as Code Artifacts

These files are anonym versions of the working examples from the paper.

MD Simulation

The input file molecular_dynamics.py contains the adapted script to create the results from the paper. In addition to the package1, our pre-existing package for the study of machine learned interatomic potentials is used in combination with the new node to showcase the easy interoperability between packages using DaC (package4).

flowchart TD
        node1["Lammps"]
        node2["Packmol"]
        node3["RDF"]
        node4["SmilesToAtoms"]
        node5["SmilesToAtoms_1"]
        node1-->node3
        node2-->node1
        node4-->node2
        node5-->node2

Random Forest Classifier

The adapted example from https://github.com/iterative/example-get-started can be found in the directory dvc-example/. The source code for the individual nodes and the workflow file are seperated. The workflow can be found in the main.py.

flowchart TD
        node1["Evaluate"]
        node2["Featurize"]
        node3["Prepare"]
        node4["Train"]
        node5["data/data.xml.dvc"]
        node2-->node1
        node2-->node4
        node3-->node2
        node4-->node1
        node5-->node3

Structural Database Repository

This is an exemplary input script for creating a geometry optimized BMIM-BF4 system, which is directly available from the GitHub repository. It uses package4 from above which is built using package1.

import package4

project = package4.Project(automatic_node_names=True)

with project.group(name="BMIM_BF4"):
    cation = package4.configuration_generation.SmilesToAtoms("CCCCN1C=C[N+](=C1)C")
    anion = package4.configuration_generation.SmilesToAtoms("[B-](F)(F)(F)F")
    
    single_structure = package4.configuration_generation.Packmol(
        data=[cation.atoms, anion.atoms],
        count=[1, 1],
        density=1210,
    )

    structure = package4.configuration_generation.Packmol(
        data=[single_structure.atoms],
        count=[10],
        density=1210,
    )
    
    cp2k = package4.calculators.CP2KSinglePoint(
        data=structure,
        cp2k_files=["GTH_BASIS_SETS", "GTH_POTENTIALS", "dftd3.dat"],
        cp2k_shell=cp2k_shell,
    )

    geopt = package4.calculators.ASEGeoOpt(
        model=cp2k,
        data=structure.atoms,
        optimizer="BFGS",
        run_kwargs={"fmax": 0.1},
        name="x10",
    )

project.build()

pythonfz / dataascodeartifacts Goto Github PK