Page 1 of 1

BSE parameters

Posted: Tue Sep 16, 2025 6:32 pm
by xjxiao
Dear all,
I have performed the convergence tests for NGsBlkXp for the calculation of the dielectric function. Subsequently, I generated the input file for the Bethe-Salpeter Equation (BSE) calculation by pointing to the dielectric screening directory with the -J flag.
I would like to inquire if the default values provided for BSENGblk and BSENGexx in the generated input file can be considered converged and used directly.
This question arises from the observation that the default BSENGblk is set to the same value as my previously converged NGsBlkXp. I understand that it is generally required for BSENGblk to be less than or equal to NGsBlkXp, which seems to justify this default setting, but I wanted to confirm if separate convergence tests are still necessary for the BSE parameters.
Thanks!
Xiao

Re: BSE parameters

Posted: Wed Sep 17, 2025 2:59 pm
by Daniele Varsano
Dear Xiangjun,

in general, you need to converge the parameters. The two parameters BSENGexx and BSENGblk can be converged independently, ie. you can seek for the BSENGexx keeping a low value for BSENGblk. You can consider a large value for BSENGblk as the building of the X-kernel is not computationally intensive. And yes, BSENGblk is set to the maxim value determined by the NGsBlkXp present in the database. In most of the case this can be lowered (note, large NGsBlkXp value is computationally intensive).

Best,

Daniele