BSE Slepc issue

[elena.cannuccia]

Dear all,

I am studying a defective system. The size of the BSE matrix is 41472 (gamma point only)
I am trying to diagonalise the BSE with slepc, but I realise that to get closer to the structures at 9 eV that appear in the spectrum obtained with haydock, I need more and more states, more than 3000 states. (see attached file)
Does it make sense, or does it become meaningless beyond a certain limit?

Also, I was wondering if BSSEnTarget can be set to any energy value of the spectrum to explore the eigenstates around it or if it makes more sense to set it at the onset of the absorption spectrum?

Thanks,
Elena
Aix-Marseille University

[Daniele Varsano]

Dear Elena,

Does it make sense, or does it become meaningless beyond a certain limit?

It should make sense until it converges...

BSSEnTarget can be set to any energy value of the spectrum to explore the eigenstates around it

Yes, that's the meaning of the variable. You can find detailed info in this Master Thesis:

Best,
Daniele