launch_lammps_optimization_lj.py

from aiida.common.extendeddicts import AttributeDict
from aiida.engine import run_get_node
from aiida.orm import Code, Dict, StructureData
from aiida.plugins import CalculationFactory
import numpy as np

from aiida_lammps.data.potential import EmpiricalPotential

if __name__ == "__main__":

    from aiida import load_profile  # noqa: F401

    load_profile()

    codename = "lammps_optimize@stern"

    ############################
    #  Define input parameters #
    ############################

    cell = [
        [3.987594, 0.000000, 0.000000],
        [-1.993797, 3.453358, 0.000000],
        [0.000000, 0.000000, 6.538394],
    ]

    symbols = ["Ar"] * 2
    scaled_positions = [(0.33333, 0.66666, 0.25000), (0.66667, 0.33333, 0.75000)]

    structure = StructureData(cell=cell)
    positions = np.dot(scaled_positions, cell)

    for i, scaled_position in enumerate(scaled_positions):
        structure.append_atom(
            position=np.dot(scaled_position, cell).tolist(), symbols=symbols[i]
        )

    structure.store()

    # Example LJ parameters for Argon. These may not be accurate at all
    potential = {
        "pair_style": "lennard_jones",
        #                 epsilon,  sigma, cutoff
        "data": {
            "1  1": "0.01029   3.4    2.5",
            # '2  2':   '1.0      1.0    2.5',
            # '1  2':   '1.0      1.0    2.5'
        },
    }

    lammps_machine = {"num_machines": 1, "parallel_env": "mpi*", "tot_num_mpiprocs": 16}

    parameters_opt = {
        "units": "metal",
        "relax": {
            "type": "tri",  # iso/aniso/tri
            "pressure": 0.0,  # bars
            "vmax": 0.000001,  # Angstrom^3
        },
        "minimize": {
            "style": "cg",
            "energy_tolerance": 1.0e-25,  # eV
            "force_tolerance": 1.0e-25,  # eV angstrom
            "max_evaluations": 1000000,
            "max_iterations": 500000,
        },
    }

    LammpsOptimizeCalculation = CalculationFactory("lammps.optimize")
    inputs = LammpsOptimizeCalculation.get_builder()

    # Computer options
    options = AttributeDict()
    options.account = ""
    options.qos = ""
    options.resources = {
        "num_machines": 1,
        "num_mpiprocs_per_machine": 1,
        "parallel_env": "localmpi",
        "tot_num_mpiprocs": 1,
    }
    # options.queue_name = 'iqtc04.q'
    options.max_wallclock_seconds = 3600
    inputs.metadata.options = options

    # Setup code
    inputs.code = Code.get_from_string(codename)

    # setup nodes
    inputs.structure = structure
    inputs.potential = EmpiricalPotential(
        structure=structure, type="lennard_jones", data={"1  1": "0.01029   3.4    2.5"}
    )

    inputs.parameters = Dict(dict=parameters_opt)

    # run calculation
    result, node = run_get_node(LammpsOptimizeCalculation, **inputs)
    print("results:", result)
    print("node:", node)

    # submit to deamon
    # submit(LammpsOptimizeCalculation, **inputs)