NameError: name 'atomic_nums' is not defined

[Vahid999]

I am going through the steps to generate 2x2x2 supercells in order to train the NN for GaAs.
When I run "python perturb_supercell.py", I get

NameError: name 'atomic_nums' is not defined

The utils.py file returns atomic_numbers not atomic_nums. Could this be the cause of the error?

Thanks,
Vahid

[QuantumLab-ZY]

Hello @Vahid999,

Thank you for your question! I’d like to clarify that the perturb_supercell.py script is not intended for generating training structures with random perturbations. Its primary purpose is to create perturbed configurations specifically for calculating the gradient matrix (grad_mat) by perturbing atomic coordinates within the central unit cell of a supercell.

If your goal is to generate training data with randomly perturbed structures, you can use a custom script like the example below. This script creates perturbations on a larger supercell while saving the output in the desired format:

from pymatgen.core import Structure
import glob
import natsort

# Specify the input file (POSCAR or CIF file)
f_vasp = glob.glob("/public/home/zhongyang/NiBrI_monolayer.cif")  # Update the file path as needed
f_vasp = natsort.natsorted(f_vasp)
idx = 1

# Loop over the input files and generate perturbed structures
for i, poscar in enumerate(f_vasp):
    for j in range(200):  # Number of perturbations per input structure
        crystal = Structure.from_file(poscar)
        crystal.make_supercell((2, 2, 2))  # Create a 2x2x2 supercell
        crystal.perturb(0.03)  # Apply random perturbation with a magnitude of 0.03 Å
        output_path = f'/public/home/zhongyang/perturb/NiBrI_{idx}.vasp'
        crystal.to(filename=output_path, fmt='poscar')  # Save the perturbed structure
        idx += 1

Best regards,
Yang Zhong

[Vahid999]

Dear Yang Zhong, Thank you for this clarification and the script. I am very interested in the potential of this code and look forward to being able to use it in my research. Best, Vahid On Dec 15, 2024, at 11:06 PM, Yang Zhong ***@***.***> wrote: CAUTION: The Sender of this email is not from within Dalhousie. Hello @Vahid999<https://github.com/Vahid999>, Thank you for your question! I’d like to clarify that the perturb_supercell.py script is not intended for generating training structures with random perturbations. Its primary purpose is to create perturbed configurations specifically for calculating the gradient matrix (grad_mat) by perturbing atomic coordinates within the central unit cell of a supercell. If your goal is to generate training data with randomly perturbed structures, you can use a custom script like the example below. This script creates perturbations on a larger supercell while saving the output in the desired format: from pymatgen.core import Structure import glob import natsort # Specify the input file (POSCAR or CIF file) f_vasp = glob.glob("/public/home/zhongyang/NiBrI_monolayer.cif") # Update the file path as needed f_vasp = natsort.natsorted(f_vasp) idx = 1 # Loop over the input files and generate perturbed structures for i, poscar in enumerate(f_vasp): for j in range(200): # Number of perturbations per input structure crystal = Structure.from_file(poscar) crystal.make_supercell((2, 2, 2)) # Create a 2x2x2 supercell crystal.perturb(0.03) # Apply random perturbation with a magnitude of 0.03 Å output_path = f'/public/home/zhongyang/perturb/NiBrI_{idx}.vasp' crystal.to(filename=output_path, fmt='poscar') # Save the perturbed structure idx += 1 Best regards, Yang Zhong — Reply to this email directly, view it on GitHub<#9 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ARHYUBBRXKYXY35Z6CYZGD32FY7SRAVCNFSM6AAAAABS2BVY32VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKNBUGQ2DKOBWGI>. You are receiving this because you were mentioned.Message ID: ***@***.***>