[CPMD-list] How to do unit cell optimization?

[HW Sheng]

Hi, Mark:

If you try to optimize the cell, there are two keywords that you might use.
(1) STEEPEST DESCENT CELL (equivalently, TSDC). (2) PARRINELLO-RAHMAN.
If you use TSDC alone, the cell can be optimized isotropically, that is, the
shape of the cell doesn't change, but the volume changes. For example:
&CPMD
   OPTIMIZE GEOMETRY
   RESTART WAVEFUNCTION LATEST
   TSDC
   SPLINE RANGE
   4.0
&END
...

If you use PARRINELLO-RAHMAN together with TSDC, you are using a variable
cell method, that is the shape will also be optimized. For example:
& CPMD
    OPTIMIZE GEOMETRY
    PARRINELLO-RAHMAN
   RESTART LATEST WAVEFUNCTION
   TSDC
   ..
&END

(Note, in this way, you will not be able to keep the unit cell orthorhombic
as it was before optimization. In order to keep the unit cell orthorhombic,
the off-diagonal elements of the stress tensor have to be eliminated during
optimization.)

To my understanding, the keyword  ISOTROPIC CELL is not invoked in geometry
minimization. It's only useful in MD. Also, I am not aware whether or not
the internal symmetry can be preserved during geometry minimization. Most
likely, no.

Best regards.

Howard

>
> Do you mean that I can either optimize the cell with no constraints or use
the
> ISOTROPIC CELL keyword to constrain everything relative to each other but
that
> I cannot, for instance, constrain a system to remain orthorhombic but to
allow
> x, y, and z to optimize independant of each other?  Also, is it possible
to
> preserve internal symmetry while optimizing the unit cell?
>
> Thank you again,
>
> Mark Kosmowski
>
> Chemistry Department
> Syracuse University
> mkosmows at syr.edu
>
> _______________________________________________
> CPMD-list mailing list
> CPMD-list at cpmd.org
> http://www.cpmd.org/mailman/listinfo/cpmd-list