OT for Aluminum Oxide?

[MD Simulation]

Does anyone know if the OT method could be used for aluminum oxide (Al2O3)?  It seemed a bit slow when I tried it compared to using SMEAR.

Here's my two SCF solving routines, the OT method is slow, ~9s per SCF iteration, while the SMEAR method is ~3s per SCF iteration.  Thanks for the help.

#### OT

        &SCF

            SCF_GUESS ATOMIC

            EPS_SCF 1.0E-8

            MAX_SCF 200

            &OT ON

                MINIMIZER DIIS

                PRECONDITIONER FULL_SINGLE_INVERSE

            &END OT

            &OUTER_SCF

                MAX_SCF 10

                EPS_SCF 1.0E-8

            &END OUTER_SCF

        &END SCF

####

#### Fast SMEAR

        &SCF

            SCF_GUESS ATOMIC

            EPS_SCF 1.0E-8

            MAX_SCF 200

            ADDED_MOS 10

            &DIAGONALIZATION ON

                ALGORITHM STANDARD

            &END DIAGONALIZATION

            &MIXING  .TRUE.

                METHOD BROYDEN_MIXING

                ALPHA 0.4

                NBROYDEN 8

            &END MIXING

            &SMEAR ON

                METHOD FERMI_DIRAC

                ELECTRONIC_TEMPERATURE [K] 300.0

            &END SMEAR

        &END SCF

####

[Nicholas Winner]

OT should certainly be fast for a large band gap system like Al2O3. You said that the speed of the scf iterations are slower, but is the full scf convergence also slower? Things to consider are your preconditioner and your outer loops I think. Full_single_inverse will be less effective than Full_all in most situations. Generally use Full_all unless your system is very large. Maybe also try reducing your max inner SCF cycles and increasing your outer loop number. It could be that your initial state is not so good, and so resetting often using more frequent outer loops will help. 

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/579141ea-3662-4d4f-b54e-f160834b37c8%40googlegroups.com.

[Fabian Ducry]

Hi,

I can confirm that OT works well with alumina. Settings that I typically use (in MD) are

        &SCF

            SCF_GUESS RESTART

            EPS_SCF 1.0E-6

            MAX_SCF 30

            &OT

                MINIMIZER DIIS

                ALGORITHM IRAC

                ENERGY_GAP 0.5

                PRECONDITIONER FULL_SINGLE_INVERSE

            &END OT

            &OUTER_SCF

                MAX_SCF 20

                EPS_SCF 1.0E-6

            &END OUTER_SCF

        &END SCF

IRAC is much faster than CG, but apparently not always reliable. I haven't noticed any problems so far though.

If you have more than a few dozen atoms, I recommend FULL_SINGLE_INVERSE over FULL_ALL with a relatiely large ENERGY_GAP (~0.5).

OT is not necessarily faster than diagonalization, it just scales better with number of atoms. Without knowing more about your systems it's difficult to tell what's happening.

Best,

Fabian