[CP2K:7474] Re: Calculating forces in reftraj runs

[Ondrej Marsalek]

Hi Matt,

I tried that before, but unfortunately it turns off the whole
interaction calculation and just races through the trajectory without
doing anything electronic structure related. My guess is that this is
meant for postprocessing tools, i.e. only MSD at the moment. I should
have clarified that in the original email.

I do want electronic structure information after a converged SCF, I
just don't need atomic forces. Just like a sequence of ENERGY
calculations, but without the overhead of launching CP2K repeatedly
and taking advantage of wavefunction extrapolation. I imagine that
this is quite a common use case of the REFTRAJ mechanism (that forces
are not needed).

Ondrej

On Thu, Mar 3, 2016 at 2:20 AM, Matt W <MattWa... at gmail.com> wrote:
> Hi Ondrej,
>
> there is the an EVAL_ENERGY_FORCES keyword in the MD section, I guess it is
> meant to do what you are asking about. Maybe it only changes output or
> something at the moment?
>
> Matt
>
> On Friday, February 19, 2016 at 2:53:53 AM UTC, Ondřej Maršálek wrote:
>>
>> Dear CP2K developers,
>>
>> I am doing reftraj runs where I am interested in some electronic
>> properties, but not in the forces on atoms. It seem that these are
>> always calculated, though, with no setting in the input file - the one
>> that is available is energy+forces or nothing. I have tried disabling
>> forces in `force_env_methods.F` (just changing `calculate_forces`)
>> and, as an example, with a hybrid functional I got times per step go
>> down from ~135 s to ~115 s. While this is not a deal breaker by any
>> means, it would be nice to have a settings for reftraj, where you can
>> enable the calculation of the forces separately. My guess would be
>> that on line 1674 of `motion/integrator.F` the call to
>> `force_env_calc_energy_force` should get `calc_force` passed to it,
>> which would be read from a new setting in the REFTRAJ section.
>>
>> Thank you for considering this,
>> Ondrej
>
> --
> 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... 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.