SCF calculation

Introduction

In a simple self-consistent field (SCF) calculation, we get the (1) energy and (2) forces of the system. There are three (3) files that you need.

File	Description
SLURM script	Submits the job to the cluster with required resources and environment
Python script	Controls the workflow
Structure file	Contains the atomic positions and lattice vectors of your system

In this example, we will perform a SCF calculation on a simple HB sheet system. The tutorial files associated in this example can by found here: link to files

SLURM script

The SLURM script is responsible for submitting the job to the cluster with the required resources and environment.

Script content

Warning

Please note that you need to adapt the environment variables and paths according to your specific setup.

#!/bin/bash
#SBATCH --partition=i8cpu
#SBATCH --nodes=2
#SBATCH --ntasks-per-node=128
#SBATCH --time=00:30:00
#SBATCH --job-name=calc
#SBATCH --array=0

# -- ENVIRONMENT --------------------------------------------------------------
source ~/.bashrc
module purge
module load oneapi_compiler/2023.0.0 oneapi_mkl/2023.0.0 oneapi_mpi/2023.0.0
module list

export FI_PROVIDER=psm3
ulimit -s unlimited

# Activate QE environment ------------------------------------------------------
# export PATH='/home/k0107/k010725/apps/qe/QE/src/qe-7.2/bin':$PATH
export PATH=$HOME/apps/qe/QE/src/qe-7.2/bin:$PATH
export ESPRESSO_PSEUDO=$HOME/apps/qe/QE/pseudo/gbrv/pbe
export ESPRESSO_TMPDIR=TMPDIR
export ESPRESSO_PROFILE='srun pw.x -nk 2'

# Activate Python environment --------------------------------
mamba activate qe_tutorial
echo "CONDA ENV: " $CONDA_PREFIX

# -- EXECUTION ----------------------------------------------------------------
echo "========= Job started  at `date` =========="

python run_scf.py

echo "========= Job finished at `date` =========="

Explanation

Here is an explanation of the important parameters and environment settings used in the script:

• #SBATCH --partition=i8cpu : Specifies the partition on the cluster where the job will run. In this case, it is set to i8cpu.

• #SBATCH --nodes=2 : Specifies the number of nodes requested for the job. In this case, it is set to 2.

• #SBATCH --ntasks-per-node=128 : Specifies the number of tasks per node. In this case, it is set to 128.

• #SBATCH --time=00:30:00 : Specifies the maximum time for the job to run. In this case, it is set to 30 minutes.

• #SBATCH --job-name=calc : Specifies the name of the job. In this case, it is set to calc.

• #SBATCH --array=0 : Specifies the job array index. In this case, it is set to 0.

The following environment settings are also important:

• source ~/.bashrc : Sources the .bashrc file to load any necessary environment variables.

• module purge : Removes all loaded modules from the environment.

• module load oneapi_compiler/2023.0.0 oneapi_mkl/2023.0.0 oneapi_mpi/2023.0.0 : Loads the specified modules for the Intel oneAPI compiler, MKL, and MPI.

• export FI_PROVIDER=psm3 : Sets the fabric interface provider to psm3.

• ulimit -s unlimited : Sets the stack size limit to unlimited.

The following Quantum Espresso (QE) environment settings are important for the job:

• export PATH=$HOME/apps/qe/QE/src/qe-7.2/bin:$PATH : Adds the QE binary directory to the PATH environment variable.

• export ESPRESSO_PSEUDO=$HOME/apps/qe/QE/pseudo/gbrv/pbe : Sets the path to the QE pseudopotential files.

• export ESPRESSO_TMPDIR=TMPDIR : Sets the temporary directory for QE.

• export ESPRESSO_PROFILE='srun pw.x -nk 2' : Sets the QE profile for parallel execution.

Finally, the script activates the Python environment and executes the run_scf.py script.

Python workflow

The Python script controls the workflow of the SCF calculation. It sets up the input parameters, runs the calculation, and analyzes the results.

Script content

import os
import numpy as np
from ase.calculators.espresso import Espresso, EspressoProfile
from ase.io import read

# 1. INPUT STRUCTURE FILE
atm = read("pristine_vdw.vasp") # Read the input structure file

# 2. INPUT PARAMETERS
# 2.a - KPOINTS GRID
kpts = (4, 4, 1)

# 2.b - PSEUDOPOTENTIALS
pseudopotentials = { 
    "H": "h_pbe_v1.4.uspp.F.UPF",
    "N": "n_pbe_v1.2.uspp.F.UPF",
    "B": "b_pbe_v1.4.uspp.F.UPF",
}

# 2.c - PW PARAMETERS
input_data = {
    "control": {
        "tprnfor": True,  # -- Print forces
    },
    "system": {
        "ecutwfc": 60, 
        "ecutrho": 480,
        "occupations": "smearing",
        "smearing": "gaussian",
        "degauss": 0.01,
        "input_dft": "vdw-df2-b86r", # -- Use the rev-vdW-DF2 functional
    },
    "electrons": {
        "electron_maxstep": 100,
        "conv_thr": 1.0e-09,
        "mixing_beta": 0.7,
        "mixing_mode": "plain",
        "diagonalization": "rmm-davidson",
    },
}

# 3. CALCULATOR
calc = Espresso(
    kpts=kpts,
    pseudopotentials=pseudopotentials,
    input_data=input_data,
    directory="DFT",
    profile=EspressoProfile(argv=os.environ["ESPRESSO_PROFILE"].split()),
)
atm.set_calculator(calc)

# 4. RUN CALCULATION

energy = atm.get_potential_energy()  # -- Run the calculation (will return energy)
forces = atm.get_forces()  # -- get the forces

# 5. SAVE OUTPUT
print("Energy: ", energy, "eV")
print("Forces (min,max): ", np.linalg.norm(forces, axis=1).min(), np.linalg.norm(forces, axis=1).max())

Explanation

Step	Description
1	Read the input structure file named “pristine_vdw.vasp” using the read function from ase.io
2	Set up input parameters for the SCF calculation, including k-points grid, pseudopotentials, and various parameters for the electronic structure calculation
3	Create an instance of the Espresso calculator from ase.calculators.espresso and attach it to the atomic structure
4	Run the SCF calculation by calling the get_potential_energy method on the atomic structure
	Calculate the forces on the atoms using the get_forces method
5	Print the energy and forces of the system

Structure file

The structure file contains the atomic positions and lattice vectors of your system. In this example, the structure file is named “pristine_vdw.vasp” and is in the VASP POSCAR format.