\(\text{Preparing } \textit{ab-initio} \text{ data}\)

Displacements using CFOUR

To create displaced geometries in .xyz format, we will follow these steps:

Geometry Optimization and Frequency Calculation:
- Use CFOUR to optimize the molecular geometry and compute harmonic frequencies.
- Extract the results from the output file generated by the harmonic frequency calculation.
Generating Displacements:
- Utilize the cfour_displacements.py script located in the Tools directory to process the output and generate the displaced geometries in .xyz format.

Configuring and Running the `cfour_displacements.py` Script

To generate displaced geometries using the cfour_displacements.py script, configure the following variables within the script:

############## Variables to Change ###############

inputfile = ""  # Output file from CFOUR frequency calculation
natoms = 0      # Number of atoms in the molecule
is_linear =   # Indicate whether the molecule is linear (True or false)

min_displacement = 0.0  # Minimum displacement in dimensionless units
max_displacement = 0.0  # Maximum displacement in dimensionless units
step_displacement = 0.0  # Step size for displacement in dimensionless units

outputfile = ""  # Output directory for the displaced geometries
##################################################

After configuring the variables, execute the script using the following command:

python3 cfour_displacements.py

Displaced Geometries

Upon executing the cfour_displacements.py script, a directory will be created containing all displaced geometries in .xyz format. The naming convention for these files is as follows:

geo_v{normal_mode}_{displacement}.xyz

``normal_mode``: Indicates the vibrational normal mode.
``displacement``: Specifies the displacement value in dimensionless units, including the sign for negative displacements.

For example, geo_v1_-01.xyz for a negative displacement and geo_v1_01.xyz for a positive displacement.

Preparing ADCC Input Files

Running ADCC Calculations

To process these geometry files and run ADCC calculations, you can use the provided Python script run_adc.py. This script takes a single geometry file as input, performs an SCF calculation using pyscf, computes excited states using adcc, and generates a molden file.

run_adc.py

import numpy as np
import sys
import adcc
import pyscf
from pyscf.tools import molden

# Create the molden filename
def change_extension(filename):
    if filename.endswith('.xyz'):
        return filename[:-4] + '.molden'
    return filename

def main():
    # Run SCF in pyscf
    mol = pyscf.gto.M(
        atom=sys.argv[1],  # Take the geometry from the displaced folder
        basis="cc-pvdz"
    )

    scfres = pyscf.scf.RHF(mol)
    scfres.conv_tol = 1e-13
    scfres.kernel()

    adcc.set_n_threads(8)
    hf = adcc.ReferenceState(scfres)
    print(adcc.LazyMp(hf).energy(level=3))  # Print MP3 energy; for better precision, CCSD(T) could be used
    state = adcc.cvs_adc3(scfres, n_singlets=10, core_orbitals=4)
    print(state.describe())  # Print the data obtained
    print(state.describe_amplitudes())  # Print the transitions

    new_filename = change_extension(sys.argv[1])
    # Creating the .molden file
    with open(new_filename, 'w') as f1:
        molden.header(mol, f1)
        molden.orbital_coeff(mol, f1, scfres.mo_coeff, ene=scfres.mo_energy, occ=scfres.mo_occ)

if __name__ == "__main__":
    main()

Submitting Jobs to the Cluster

To run the run_adc.py script on a cluster, you can use the following sbatch script. Replace the placeholders (X) with appropriate values for your system.

run_sbatch_adcc.sh

#!/bin/bash
#SBATCH --partition=X
#SBATCH --nodes=X
#SBATCH --ntasks-per-node=X
#SBATCH --time=XX:XX:XX
#SBATCH --mem=Xgb

new_filename="${1%.xyz}.out"  # This will create an output file
python3 run_adc.py ${1} > $new_filename

Running Multiple Calculations

To efficiently run calculations for all displaced geometries, you can use the following bash script. This script iterates over all normal modes and displacements, submitting a separate job for each geometry file. Ensure that the filename formatting matches the output of cfour_displacements.py.

run_everything.sh

#!/bin/bash
nmodes=x
min_disp=x
max_disp=x
step=x
for nmode in $(seq 1 1 $nmodes); do
    for ((disp=$min_disp; disp<=$max_disp; disp+=$step)); do
        if [ $disp -lt 0 ]; then
            inpgeo=$(printf "geo_v%d_%03d.xyz" $nmode $disp)
        else
            inpgeo=$(printf "geo_v%d_%02d.xyz" $nmode $disp)
        fi
        sbatch run_sbatch_adcc.sh $inpgeo
    done
done

Note: Adjust the values of nmodes, min_disp, max_disp, and step to match your specific calculation parameters. Also, verify that the filename formatting in the script (e.g., geo_v%d_%03d.xyz for negative displacements and geo_v%d_%02d.xyz for positive displacements) matches the naming convention used by cfour_displacements.py. For example, this script assumes filenames like geo_v1_-01.xyz and geo_v1_01.xyz.

Extracting Data from ADC Output Files

. Configure the Extraction Script:

Open Tools/extract_adcc.py in a text editor.
Modify the configuration parameters as needed
Running it by executing bash python3 extract_adcc.py
This will create the data files gs_v{mode}.dat and ce{state}_v{mode}.dat, each file consisting of two columns, the displacement and the energy.