Hi everybody,
after extensive discussion on this list, since I have partially filled
states in the gap of GaAs surface, I'm trying to use MP smearing to
improve convergence vs. k-grid cutoff.
But what I observe that for any non-zero value of smearing (I used up to
0.7eV (which increases entropy quite noticeably) with 50A k-grid cutoff
(while even for metals it is enough about 35A cutoff)) I cannot obtain
convergence in SCF steps even with mixing value of 0.001!
It is not the question of the amount of SCF steps (I use about 200
steps), since at some point the dDmax and E_tot stop changing with dDmax
values 0.0005-0.0010
So I've switched back to Fermi-Dirac smearing and it works quite
predictably, I use up to 600K (0.05eV) smearing which ensures small
value of entropy while gives good convergence vs. k-grid and number of
SCF steps.
Actually even 0K and quite small k-grid cutoff (12A) gives good
convergence of E_tot and SCF as well, although E_Fermi is not converged
well.
What's interesting that E_tot with MP smearing differs quite
significantly from the converged value with F-D smearing.
Moreover I've tried to use OccupationMPOrder 2 and again, E_tot differs
significantly from both Order 1 MP and FD smearing.
I just want to understand why MP smearing doesn't work well in my case?
Has anybody actually used MP smearing in SIESTA?