[CP2K:4962] Outputting cell information and fractional coordinates

Matthias Krack matthia... at psi.ch
Fri Jan 31 14:59:27 UTC 2014

There is now an alternative dump format called DCD_ALIGNED_CELL which dumps 
the coordinates based on a cell alignment as described which should allow a 
reconstruction of the scaled coordinates only from the DCD dump data (to 
single precision accuracy).


On Thursday, January 30, 2014 10:13:29 AM UTC+1, FX wrote:
> you are right, the information about the alignment of the cell with 
> respect to the Cartesian axes is lost. Thus it is currently not possible to 
> retrieve the scaled coordinates for an atomic configuration in an unique 
> manner from the dcd file information involving non-orthorhombic cells. This 
> is however something which could be fixed. CP2K adopts a common convention 
> for the cell alignment when an explicit cell vector definition is not 
> available, namely, the cell vector *a* is aligned with the x axis and the 
> cell vectors *a* and *b* lie in the xy plane. I think this convention 
> could also be applied in this case and the Cartesian coordinates could be 
> dumped only after the cell has been mapped to this representation using 
> scaled coordinates.
> Either that or allowing to dump into a format with support for non-aligned 
> unit cell (like xtc: http://manual.gromacs.org/online/xtc.html). 
> Otherwise, people currently using DCD files might get annoyed that the unit 
> cell is constrained along an axis, which makes it not very suitable for 
> visualization (it looks like it’s jumping all around).
> In my particular case, since my simulations have already run, I’m writing 
> a short program to reconstruct fractional coordinates from the .cell file 
> and the .xyz trajectory.
> FX
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140131/0f37e5fe/attachment.htm>

More information about the CP2K-user mailing list