GbndRnge < BndsRnXp convergence test

[xjxiao]

Dear all,
Hello,
I have a question regarding the validity of a convergence study I performed.
In my test procedure, I kept GbndRnge constant at 120 while varying two parameters: BndsRnXp across the set {120, 240, 360, 480} and NGsBlkXp across {5, 10, 15, 20} Ry. Based on this, I found the calculation to be converged with BndsRnXp = 360 and NGsBlkXp = 15 Ry.
However, it has recently come to my attention that the condition GbndRnge > BndsRnXp should be satisfied. This raises concerns about my results, since my fixed GbndRnge (120) was less than the apparent converged value of BndsRnXp (360).
Therefore, could you clarify if a convergence test is considered reliable when conducted under conditions where GbndRnge < BndsRnXp?

[Daniele Varsano]

Dear Xiangjun,

you can raise GbndRnge to verify the calculation is properly converged. Please note, if you have not already used, you can use the terminator technique (GTermKind= "BG") explained in F. Bruneval and X. Gonze, Physical Review B 78, 085125 (2008 ). This will allow to accelerate the convergence with respect to the bands in the self-energy summation.

Best,

Daniele