[CP2K:1336] Re: orbitals restart history

Laino Teodoro teodor... at gmail.com
Thu Sep 11 12:10:14 UTC 2008


>
> however, seeing that it can be achieved this way, the original
> question still remains: how do you use the RESTART_HISTORY section and
> what is it good for. the description in the input manual is an exact
> description of my original intent:

well the answer is in what you wrote: while the RESTART is backup-ed  
normally (and you can loose it)
RESTART_HISTORY is by default not backup-ed, and dumps the restart  
only at certain amount of iterations.
You can use the setting I sent you for both section and get the same  
result.
Up to you which one you want to choose.

> "Dumps unique MO restart files during the run keeping all of them."
>
> i have also a related question: i would like to use the restart file
> together with a reference trajectory for a "replay" (and calculation
> of things that were not calculated in the original run, obviously).
> the idea is that the saved wfn would be used for each step, without
> any further scf iterations. is there a way to do that?
This is the REFTRAJ ENSEMBLE in the MD module.
I'm not sure it is fully integrated in such a way that you can read  
WFN restart
files at each iteration (I would say you cannot at the moment)..  
but.. you can of
course use the trajectory and re-compute the wfn optimization at each  
step..

Keep in mind that the speed of computers today is such that does not  
make
any sense to store thousands of restart files when you can re- 
evaluate them
quickly.

Teo
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20080911/6a8ce1c9/attachment.htm>


More information about the CP2K-user mailing list