[Wien] run_lapw -it0

Bertoni Giovanni giovanni at cemes.fr
Wed Jan 7 13:51:47 CET 2004


dear users,
I'm testing convergence on a big system.
I know that lapw1 -it makes the diagonalization using an iterative mode
(reducing calculation time)
usually I do run_lapw -it that means that the first iteration uses a
full diagonalization, while the others the iterative.
at the end I do a full diagonalization. Is this correct for the
convergence?
I have a question: I'm testing with -it0 (iterative mode in the first
iteration also). Why does the first iteration take 3 times the others ?
Is it a full diagonalitazion and not an iterative diagonalization?

giovanni
-----



More information about the Wien mailing list