GW calc error - the frequency range is not sufficient

[pnikacevic]

Dear VASP community,

I have started doing GW calculations relatively recently, and I am having a problem that my calculations often crash with the following message:

ERROR in DETERMINE_SLOT_INTER_WEIGHT: the frequency range is not sufficient

I attach the necessary input/output files. I have not attached WAVECAR and WAVEDER which I obtained from a previous DFT calculation (for some reason it crashes even earlier when I tried to do it all in one calculation, which should've been possible with VASP 6)

Thank you in advance
Pavle

[martin.schlipf]

The number of bands has been changed from the values supplied in the INCAR file. This is a result of running the parallel version. The orbitals not found in the WAVECAR file will be initialized with random numbers, which is usually adequate. For correlated calculations, however, you should redo the groundstate calculation. I found NBANDS = 96. Now, NBANDS = 240.

Did you consider this warning message? You should use the same NBANDS for the preceding run, otherwise you get random orbitals which will not work for GW. Note that you could do an NSCF run just to increase the number of orbitals, though.

[pnikacevic]

Thanks for your reply! After changing this, the calculation did work, however, when I try to do the GW calc from scratch (without the preceding DFT), it crashes with the same error as before. I attach the relevant files. I would appreciate any further help!

[martin.schlipf]

I'm not sure what you mean without preceding DFT run. Where do you get your WAVECAR file from?

Here is our guide how to conduct GW calculations. Is this the procedure that you do?

[pnikacevic]

Yes, that's the guide I've been following. Calculations consist of two steps: 1) DFT calculation, and 2) GW step.

But in the link, it also says the following:
"As of VASP.6 all GW approximations can be selected directly via the ALGO tag (omitting NBANDS) without a preceding DFT calculation."
This method didn't work for me (I attached the relevant files in the previous message), both when omitting and not omitting the NBANDS tag

[martin.schlipf]

Okay, I could not reproduce the issue that you mention. If I do not have NBANDS in my input, everything works as expected (i.e. a DFT calculation is performed first). However, I still would not recommend this approach, because NELM is used for both the DFT as well as the GW run. For G0W0 that is fine (only one GW step), but if you want some kind of self-consistency (like EVGW0) this is not a good way, because the two methods need very different amount of steps to converge.