CP2K Problems with Restarting the Localization of Wannier-Centers

[M. Brehm]

Hi,

there is no "canonical" ordering of the Wannier centers in the output, so 
the order will change (as it also changes from step to step within one CP2k 
MD run). TRAVIS does not have problems with this, it works fine, as TRAVIS 
does not rely on any order, but assigns the Wannier centers to adjacent 
molecules in each step.

Is the problem only related to changing order of the centers, or are the 
coordinates really differing, i.e., are there Wannier centers after the 
restart at some position where there was no Wannier center before the 
restart? This might also happen if the electronic structure is delocalized 
(e.g., in aromatic systems). This might impose problems for the calculation 
of IR spectra...

Best regards,
Martin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20150310/d731169e/attachment.htm>