[CP2K-user] [CP2K:16895] Re: What is the precision on forces emitted by CP2K's energy forces job?
Matt D
ucecmtd at gmail.com
Wed Apr 27 08:50:02 UTC 2022
Hi Alex,
The quality of the forces you calculate will depend on the accuracy to
which the SCF problem is solved. The tolerance for the latter is given by
EPS_SCF - a typical value for EPS_SCF during a geometry optimisation would
be around 1e-7 Ha. Another important parameter for the SCF tolerance is
EPS_DEFAULT, this should be decreased from the default value of 1e-10 to
1e-14 to complement the value of 1e-7 for EPS_SCF. The error in the forces
with respect to these parameters should be sqrt(EPS_SCF) < error on force;
and sqrt(EPS_DEFAULT) < EPS_SCF.
The default units in CP2K are always atomic units, so the force is in
Hartree/Bohr. If you want to change the units you put in the input, this is
possible but I have forgotten the exact syntax. I think it is something
like:
KEYWORD [unit] 0.025
and the "unit" is replaced by one of the units in this list:
https://manual.cp2k.org/trunk/units.html
Best wishes,
Matt
On Monday, 25 April 2022 at 21:56:44 UTC+1 alexvd... at gmail.com wrote:
> Hi,
>
> I'm curious what the precision on forces in a cp2k energy_forces job. In
> particular, how does it depend on various input variables, such as EPS_SCF?
>
> Also, in the forces section, the printout says the units are in atomic
> units. Is that also the default unit for all force dependent convergence
> criteria, such as in a NEB job?
>
> Thanks,
> Alex
>
--
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+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/0dfdc809-93cf-41af-976f-68e7bf7a56b9n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20220427/e7bbe725/attachment.htm>
More information about the CP2K-user
mailing list