Dear experts,
I am performing optical-properties calculations within the RPA framework and would like to confirm whether my workflow is correct. My procedure is as follows: Step: 1 Ground-state DFT: I performed a standard DFT calculation and generated the WAVECAR.
Step 2: Generation of WAVEDER: In the second step, I used
ISTART = 1
NBANDS = two times of NBANDS what I obtained in step1
ALGO = Exact
LOPTICS = .TRUE.
CSHIFT = 0.1
This step successfully produced both WAVECAR and WAVEDER.
Step 3: RPA dielectric function
In the third step, I used
ISTART = 1
NBANDS = same as in step 2
ALGO = CHI
LRPA = .TRUE.
My main questions are:
1. Is this workflow correct for obtaining the dielectric function using RPA in VASP?
2. Parallelization issue:
In steps 1 and 2, parallelization works normally with NCORE = 24
However, in step 3, using ALGO = CHI, VASP stops with an error when NCORE is included in the INCAR. I also added NPAR= number of nodes in my INCAR, but still the same error is coming.
Should parallelization work for the CHI step?
Do I need to remove NCORE for this step?
3.Choice of NOMEGA:
What would be an appropriate value of NOMEGA for the third step? As my walltime is 24 Hrs.
I am attaching the error screenshot, obtained in 3rd step.
