[CP2K-user] unwrapped ions during CELL_OPT

Travis polla... at gmail.com
Fri Oct 25 14:44:52 UTC 2019


Your input structure is very compressed. It should expand and your final 
structure looks much more reasonable. The amount it is expanding by, I 
would actually suggest redoing the CELL_OPT from the final structure as 
your effective cutoff decreases when the cell volume increases. Coordinates 
are unwrapped intentionally in all calculations. You can impose PBC to wrap 
the coordinates using an external program. I recommend Atomic Simulation 

Grab last image from foo-pos-1.xyz:

Add cell parameters from command line args:

Generated structure is in POSCAR format, viewable with VESTA. Adapt to your 

Delete this bit, it's specific to my setup.
from require import check_ase


On Friday, October 25, 2019 at 5:46:21 AM UTC-3, Pierre-André Cazade wrote:
> Dear CP2K users,
> I am running a CELL_OPT on a 280 atoms unit cell. It was poorly resolved 
> experimentally, yet I am surprised that some atoms wander out of the unit 
> cell and are not wrapped back in. It is quite unexpected. According to the 
> following topic: https://groups.google.com/forum/#!topic/cp2k/rqOO2W04OPc , 
> CP2K does not wrap atoms, at least in MD mode. This is very strange. Is it 
> intentional? Is it a bug? And does it affect in any kind of way 
> calculations?
> I am bothered by the optimized structure which shows a considerable 
> increase of the "a" parameter. Again, I compared with VASP and the final 
> structure with the latter is much more reasonable. I do not expect 
> identical optimized structure but they should not drift so much away from 
> each other. Could this lack of wrapping explain the divergence or it comes 
> from inadequate SCF parameters?
> I attached the input file and the final restart file.
> Regards,
> Pierre
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20191025/b8568d0e/attachment.htm>

More information about the CP2K-user mailing list