Yambo Community Forum

Dear developer,

I am trying to run a GW calculation on an isolated methanol molecule. I seem to be getting an extremely high (LUMO-HOMO) gap of more than 10 eV, against a PBE gap of ~5.5 eV. I suspect that my calculation might be underconverged with respect to some parameters but I'm not really sure what. I am attaching the input and the output of two separate calculations with different values of RandGvec , CUTRadius and NGsBlkXp. The QP gap does not seem to change much between the two calculations. Could you kindly take a look and let me know what you think?

Best regards,
Subhayan

Hi,

A quick update on this. I reran the calculation with a much higher number of (600) bands but unfortunately the gap doesn't seem to change much
Any suggestion will be extremely appreciated.

Best regards,
Subhayan

Dear Subhayan,
the value you get seems reasonable, please have a look at the GW100 paper (J. Chem. Theory Comput. 2015, 11, 5665−5687) where a gap of about 10.7 eV is reported for GW at Plasmon Pole level.

In general converging molecules is rather hard in terms of unoccupied bands and cutoff for the dielectric matrix.
Comment on your inputs:
* RandQpts this is the number of q vectors used for a Monte Carlo integration, set it to 1 million or more otherwise it could be inaccurate (large statistical error)
* CUTRadius you can keep it fix to half of your supercell
* GTermKind="BG" can help you in accelerating the convergencence wrt the bands in the GW summation.

Best,
Daniele

Hi Daniele,

Thanks a lot for the reference and the clarification

Best regards,
Subhayan