Skip to content

README.md

Latest commit

 

History

History
109 lines (81 loc) · 4.04 KB

README.md

File metadata and controls

109 lines (81 loc) · 4.04 KB

materials.

"materials." is a srcipty tool that generates input files for Quantum Espresso. it also has scripts to post-process the results of the calculations (scf, nscf, dos) and extract useful information from them.

running.

setting up potentials and environment variables.

  1. download pseudopotentials from SSSP precision library which contains all pseudopotential files and place them in some directory (e.g. ./pseudos/).
  2. set the PSEUDO_DIR variable in .env file to the path of this directory.
  3. run python ./src/rename_pseudos.py to rename the pseudopotential files for other scripts to recognize.
  4. copy .env.example to .env and set the PROJECT_NAME, leave the rest to defaults.

generating input files.

you have to build your structure in main.py's main() function. you'll have to two more functions to generate the input files. here's an example of how to do that:

def main():
    # get project directory
    project_dir = setup_output_project(PROJECT_NAME)

    # build your structure here
    structure = get_structure_from_id("mp-149") # Silicon ID from Materials Project

    # generate the input files
    generate_espresso_input(
        structure,
        out_dir=project_dir,
        prefix="si",
        like="insulator", # or "metal"; this decides what occupation scheme to use i.e. smearing (mv*) or fixed occupations;
        xc="pbe", # exchange-correlation functional; this sets the `input_dft` variable in the input files
        kpts=(7, 7, 7),
    )

then run,

python ./src/main.py

although phonon files are generated for functionals other than pbe, they are not supported by Quantum ESPRESSO as of build 7.4.1 (which i used).

running calculations.

after generating the input files, you can run the calculations using

# run relaxation calculation
pw.x < ./<OUTPUT_DIR>/<PROJECT_NAME>/relax.in > ./<OUTPUT_DIR>/<PROJECT_NAME>/relax.out

# copy the relaxed structure to scf.in and nscf.in (bottom)

# run scf calculation
pw.x < ./<OUTPUT_DIR>/<PROJECT_NAME>/scf.in > ./<OUTPUT_DIR>/<PROJECT_NAME>/scf.out

# run nscf calculation
pw.x < ./<OUTPUT_DIR>/<PROJECT_NAME>/nscf.in > ./<OUTPUT_DIR>/<PROJECT_NAME>/nscf.out

# run dos calculation
dos.x < ./<OUTPUT_DIR>/<PROJECT_NAME>/dos.in > ./<OUTPUT_DIR>/<PROJECT_NAME>/dos.out

# run phonon calculation
ph.x < ./<OUTPUT_DIR>/<PROJECT_NAME>/ph.in > ./<OUTPUT_DIR>/<PROJECT_NAME>/ph.out

you can check calculation progress by running tail -f ./<OUTPUT_DIR>/<PROJECT_NAME>/<calculation_type>.out in a separate terminal, where <calculation_type> is one of scf, nscf, dos or ph.

post-processing.

after running the calculations, you can post-process the results using

python ./src/process.py

this will generate ./<OUTPUT_DIR>/<PROJECT_NAME>/processed.json file which contains useful informaton extracted from scf and nscf calculations.

tips.

  • you can use mpirun -np <num_procs> pw.x < ./<OUTPUT_DIR>/<PROJECT_NAME>/scf.in > ./<OUTPUT_DIR>/<PROJECT_NAME>/scf.out to run the calculations in parallel on multiple processors.
  • get your pseudopotentials from here.
  • although phonon calculations are always generated, they are not supported by Quantum ESPRESSO as of build 7.4.1 (which i used) for exchange-functionals other than pbe.
  • certain exchange-functionals (eg. SCAN) require norm-conserving pseudopotentials to be used. i used sg15_oncv_upf_2020-02-06.tar.gz SG15 ONCV potentials in UPF format from SG15 ONCV potentials for this purpose. you are still required to rename them just like the normal psuedos and update your .env file if working with such functionals.

license.

   ╱|、
 (˚ˎ 。7
  |、˜〵
 じしˍ,)ノ

the repository and everything within is licensed under the GNU General Public License v3.0. Refer to COPYING.md for the full license text.
⚡