Gaussian on Hyak

  1. Prepare Gaussian 16 input file
    • The Route section (# line) of the input specifies the calculation type.
      • sp or no keyword: single point energy;
      • opt: geometry optimization;
      • freq: frequency and thermochemical analysis;
      • td:  excited state calculation.
    • If the input is in XYZ format,  edit the route or other keywords in “~/gaussian/xyz2gjf.py” and run
      • Either ~/gaussian/xyz2gjf.py input.xyz if the original input name is “input.xyz”;
      • Or for i in *.xyz; do ~/gaussian/xyz2gjf.py $i; done to convert all files ending with “.xyz” to Gaussian input format.
    • If the input is already in Gaussian input format, the only thing that need to be changed might be the route card. For example, to change “opt b3lyp/6-31g(d,p)” on the 4th line of all gjf files to “td(Nstates=20) um062x/6-31+g(d,p)”, type for i in *.gjf; do sed -i "4s#opt b3lyp/6-31g(d,p)#td(Nstates=20) um062x/6-31+g(d,p)#g" $i; done. You can check if it is changed correctly by typing head *gjf.
  2. Generate slurm file using  ~/gaussian/gaussian-klone.py input-file-name partition number-of-node time-in-hour.
    • For example, to create a task to run “input.gjf” on stf partition for 2h 30min on 1 node: type ~/gaussian/gaussian-klone.py input.gjf stf 1 2.5. Then “input.sh” will be created.
    • To create multiple tasks for all files ending with “.gjf”: for i in *.gjf; do ~/gaussian/gaussian-klone.py $i stf 1 2.5; done.
  3. Submit the task: sbatch file-name-of-sh
    • To submit multiple sh files: for i in *.sh; do sbatch $i; done.

© 2024 Tureček Research Group — Powered by WordPress

Theme by Anders NorenUp ↑