Newton-X on Hyak

Create a NewtonX TDDFT Task

1. Prepare the input: Create a new folder and copy the optimized Gaussian structure in the format of Gaussian input (gjf/com) and Gaussian frequency output (log) to this folder.
2. Run ~/gaussian/newtonx.py file-name-of-opt-gjf/com file-name-of-freq-log. Then type the input following the instructions. If the question has a default answer, press <Enter> will use the default value as the answer.
3. Two ways to submit the job.
   1. Either enter the TDDFT_SPEC folder: cd TDDFT_SPEC, edit the partition, account, and/or running time in “newtonx.sh” file and submit it by bash newtonx.sh.
   2. Or enter the INITIAL_CONDITIONS folder under TDDFT_SPEC: cd TDDFT_SPEC/INITIAL_CONDITIONS. You can find several folders named with I1, I2, I3… Go the each subdirectory, edit the partition, account, and/or running time of “nx_submit.sh” file, and submit it using sbatch nx_submit.sh, respectively. Those I1, I2 … folders can be moved to anywhere to submit. Just make sure the directory in “nx_submit.sh” matches the real path.

Process Output

Go to the INITIAL_CONDITIONS folder that contains the I1, I2 and so on and run ~/gaussian/nxplot.py. This script will merge all completed initial conditions under those subfolders and compute absorption spectrum. These subfolders must be named with “I” plus number. The number must start with 1 and be consecutive. In the end, you will see all information in I_merged. Keep and save I_merged folder.

I_merged folders with different initial conditions can also be merged. Create a new directory and copy and rename those different I_merged folders as I1, I2 and so on to that directory. Then type ~/gaussian/nxplot.py.

Plot Spectrum

1. Load Python environment: conda activate. This is only needed to be do once every time you log in.
2. Copy the corresponding tddft csv file (*_uvvis.csv) and “cross-section.data” got from newtonx to the same folder
3. Plot the NewtonX spectrum with TDDFT spectrum: ~/gaussian/tdnx_plot.py file-name-tddft-uvvis.csv cross-section.data (number). If number present, all vertical transactions with wavelength larger than the number will be coated with aqua color.