Issues with GW Band Convergence

[sunxl]

Dear all，

1、When calculating GW band structures, the tutorial recommends converging certain parameters using the variation of the direct band gap as the criterion. If my structure has an indirect band gap, I would like to perform convergence tests quickly. Can I assess convergence by calculating the direct band gap?

2、When performing convergence on GbndRnge, what is the acceptable error in eV to consider it converged?

Thank you in advance for your help.

Best，
sunxl

[Daniele Varsano]

Dear sunxl,

1) In general, you can check the convergence on the direct band gap. In any case, looking directly to the indirect gap is not an issue in the sense that does not add computational burden, as you can set the QPkrange as:

Code: Select all

%QPkrange                        # [GW] QP generalized Kpoint/Band indices
k1|k1|b1|b1|
k2|k2|b2|b2|
%

where k1,b1 and the MVB and k2,b2 are the mCB.

2) This really depends on the precision you need. In general, I do not think that looking for a convergence criteria smaller than few tenth of meV is a good idea.


Best,
Daniele