Velcoities printed are different

[Rad]

Hi Teo,

Thanks for the reply. I would like keep the atoms constrained, that is
part of the simulation requirement. So do avoid scaling can I give the
initial temperature as 0.
I am testing the possibility - will let you know.


Rad

On May 17, 6:55 am, Teodoro Laino <teodor... at gmail.com> wrote:
> Rad,
>
> >    &CONSTRAINT
> >      CONSTRAINT_INIT  T
> >      &FIXED_ATOMS
> >        RANGE               1         2
> >      &END FIXED_ATOMS
> >    &END CONSTRAINT
> > If you compare the initial set of velocities with the ones in the
> > input, you can see that they are different. Is it a bug in cp2k or the
> > velocities are converted to a different unit or somehow the velocity
> > of the center of mass is subtracted?
>
> Normally velocities (when read through the input file) are just accepted
> as they are (no postprocessing, i.e. rescaling or vcom removal).
>
> But in your case you're explicitely asking for
>
> CONSTRAINT_INIT
>
> what this keyword does is:
> (1) apply constraints at the 0 step of the MD (so that you don't see
> any jump between the
> step nr.0 and the nr. 1 in the energy)
> (2) after application of constraints rescale consistently the initial
> energy of the system..
>
> Now in your case you don't have any constraint that requires
> specifically the application
> of initial constraints, because fixed atoms have just to stay fixed
> (the story is different if you
> want to keep bonds constrained or anything else).
> But since you asked for :
>
> CONSTRAINT_INIT
>
> this means that once the velocities of the fixed atoms are nullified
> the velocities of the all particles
> are scaled in such a way to get the requested temperature.
> If you dont' want that, please remove the keyword:
>
> CONSTRAINT_INIT
>
> No bugs at the moment  ;-)..
> Cheers
> Teo
>
>
>
> > Could you please look into this and let me know where the issue might
> > be. I got into this issue on my large system and so I tried it on a
> > small test system and verify the consistency of the behaviour.
>
> > Please let me know if you need any additional information on the test.
>
> > Thanks
> > Rad