[CP2K-user] [CP2K:11662] How to perfom zero temperature geometry optimization
Patrick Gono
patri... at gmail.com
Sun May 5 12:34:31 UTC 2019
Dear Mohammad,
Do you want to perform a geometry optimization (GEO_OPT), or a cell
optimization in which the cell parameters are varied as well?
Geometry optimization is by its very nature a zero Kelvin method - after
all, you are minimizing energy to find the ground state of the
computational cell geometry.
So is cell optimization by default, but you can choose the MD type, which
uses (finite temperature) molecular dynamics to compute the stress tensor.
The MD section with the appropriate temperature must be defined in order to
use this method. But I suspect this is not really what you are after.
Define the &GEO_OPT section and set up a GEO_OPT calculation to find the
optimized structure. Alternatively, if you want to relax the cell
parameters, go for CELL_OPT. The sections are quite similar, in both cases
I recommend the BFGS algorithm unless tou encounter instabilities in the
optimization process. You can find the appropriate tutorials on the
official cp2k website, or via Google.
Regarding the smearing: as you yourself mention your system has a gap, I
would opt for a different scf method. Instead of direct diagonalization
(with or without smearing) consider adopting an orbital transformation
method (defined in the & OT section). These provide improved performance,
although may not be as reliable as diagonalization in some cases. However,
the OT methods work well for virtually any system I study.
Regardless, you chose the fermi-dirac smearing method which takes one
parameter, a temperature. This is not the real temperature of the system,
it's only a parameter of the smearing method. Using 0 K basically means no
smearing (the FD distribution is a step function at zero temperature).
Consider commenting out the smearing section if you do not intend to use
smearing.
Yours sincerely,
Patrick Gono
On Sun, 5 May 2019, 08:20 Mohammad Shakiba, <mshakiba.... at gmail.com>
wrote:
> Hello everyone,
>
> What is the way to set the temperature during geometry optimization? Also,
> in the manual for CELL_OPT it says "*This section sets the environment
> for the optimization of the simulation cell. Two possible schemes are
> available: (1) Zero temperature optimization; (2) Finite temperature
> optimization." *but there is no section for setting the temperature (My
> system is a semiconductor).
>
> I also used Fermi-Dirac smearing function with the following inputs but
> that gives me an error after one scf step *"KS energy is an abnormal
> value (NaN/Inf).".*
>
>
> SCF WAVEFUNCTION OPTIMIZATION
>
> Step Update method Time Convergence Total energy
> Change
>
> ------------------------------------------------------------------------------
> 1 P_Mix/Diag. 0.40E+00 30.0 1.31109320 -882.6554194381
> -8.83E+02
>
>
> *******************************************************************************
> *
> ___ *
> * /
> \ *
> *
> [ABORT]
> *
> * \___/ KS energy is an abnormal value
> (NaN/Inf). *
> *
> | *
> *
> O/| *
> * /|
> | *
> * / \
> qs_ks_methods.F:866 *
>
> *******************************************************************************
>
>
>
> Can any one help me in this regard? Thanks in advance.
>
> The smearing input section which provides the electronic temperature:
>
> &SCF
> SCF_GUESS ATOMIC
> MAX_SCF 300
> EPS_SCF 1.0E-7
> ADDED_MOS 20
> &DIAGONALIZATION
> ALGORITHM STANDARD
> EPS_ADAPT 0.01
> &END DIAGONALIZATION
> &SMEAR ON
> METHOD FERMI_DIRAC
> ELECTRONIC_TEMPERATURE [K] 0
> &END
> &END SCF
>
> --
> 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 cp... at googlegroups.com.
> To post to this group, send email to cp... at googlegroups.com.
> Visit this group at https://groups.google.com/group/cp2k.
> For more options, visit https://groups.google.com/d/optout.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20190505/f24bfeb2/attachment.htm>
More information about the CP2K-user
mailing list