ASE calculator

This example shows how to use the trained potential in Python as an ASE calculator. Suppose you finished 0.0_train, and we then write the following run.py.

import ase.io
from ase.visualize.plot import plot_atoms

from motep.calculator import MTP
from motep.io.mlip.mtp import read_mtp

We first load the atomic configuration to compute.

atoms = ase.io.read("../0.0_train/ase.xyz")
ax = plot_atoms(atoms)

We next load the trained potential.

potential = read_mtp("final.mtp")  # copied from ../0.0_train
potential.species = [6, 1]

We then assign the potential to atoms and trigger the calculation.

atoms.calc = MTP(potential)
atoms.get_potential_energy()

np.float64(-199.52118216266535)

We finally store the energy and the forces in a new file.

atoms.write("final.xyz")

Run (local)

python run.py

Run (slurm)

You can also submit the above job to a computational node using SLURM.

The batch script script_001_1h.sh is in the slurm directory.

#!/usr/bin/env bash
#SBATCH -J script
#SBATCH --time 1:00:00
#SBATCH --export=HOME

source $HOME/.bashrc

# https://www.gnu.org/software/bash/manual/bash.html#Special-Parameters
# ($@) Expands to the positional parameters, starting from one.
$@

For later convenience, we put the script in close to the home directory.

mkdir -p ~/slurm
cp ../slurm/script_001_1h.sh ~/slurm

We can submit the above python script as

sbatch ~/slurm/script_001_1h.sh python run.py