[CP2K-user] The problem of convergence of the wave function

[Fabian Ducry]

Is there a particular reason why you want to perform a calculations with a 
very low ELECTRONIC temperature?
In case you need a low "lattice temperature", you have to perform an 
optimization, GEO_OPT or CELL_OPT. This is different from 
ELECTRONIC_TEMPERATURE.

If you really want a low ELECTRONIC_TEMPERATURE:
This is notoriously difficult with diagonalization based methods. You can 
try to set SCF_GUESS RESTART and successively lower ELECTRONIC_TEMPERATURE 
until you reach the desired value.

A better option is to use OT:
    &QS
      EPS_DEFAULT 1.0E-12
    &END QS
    &SCF
      SCF_GUESS RESTART
      EPS_SCF 1.0E-6
      MAX_SCF 30
      &OT
        PRECONDITIONER FULL_SINGLE_INVERSE
        ENERGY_GAP 0.2
        MINIMIZER DIIS
      &END OT
      &OUTER_SCF
         EPS_SCF 1.0E-6
         MAX_SCF 10
      &END OUTER_SCF
    &END SCF

Fabian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20200507/e25b4bb1/attachment.htm>