How to choose BSSNEig in slepc solver

[jasonhan0710]

Dear all,

I want to calculate the excitonic properties with BSE, however, the diagonalization solver consume too much memory. When I look into slepc solver, I am confused with the BSSNEig parameter. From the tutorial page, we know that

we will capture the lowest-energy BSSNEig states of the absorption spectrum. We set BSSNEig=4 since we know from the tutorial on how to analyse excitons that the first peak in the hBN spectrum is comprised of states 3 and 4

. It means we should do a diagonalization first to get the excitonic strength, which is impossible to implement. Can we get the excitonic strength by other method, for example, Haydock solver? If not, How to set BSSNEig in slepc solver?

Best,
Jason

[Daniele Varsano]

Dear Jason,
slepc it is an iterative method that provides you also the eigenvector. You need to indicate how many eigenvectors you want to calculate.
Of course, then the absorption will take into account also these eigenvalues.

The procedure I suggest you is to perform a Haydock calculation to have your spectrum and then use the slepc solver to have the eigenvector of the first excitons.

Best,
Daniele