CP2K Problems with Restarting the Localization of Wannier-Centers

[hut... at chem.uzh.ch]

Can somebody comment on this issue or does somebody have some advice? Juerg Hutter

i am a student from Erlangen and i currently do my master thesis. Therefore a have to work with the software CP2K and the HPC-Team of our university gave me the advice to contact you with my problem. 

I have reached the maximum jobtime at the HPC-Clusters of the University with my simulations and i tried to use the checkpoint-restart-possibilities of CP2K. My calculations are of MD-type with the input of an external trajectory-file (REFTRAJ-mode) and the target is to calculate the wannier-centers to obtain in the end Raman-spektra with the software TRAVIS.

I have started several tests with small molecular systems to simulate. I have a single-run CP2K simulation of 100 timesteps and one Serine-molecule to compare it with the results of the checkpoint-restart-tries. Restarting the input of the external trajectory works as desired. BUT there are always inconsistencies with restarting the method of the localization of the wannier-centers. The coordinates of the atoms are the same as in the single-run CP2K simulation and the coordinates of the wannier-centers differ in an inacceptable way.
I hope to get help from you and your team and not to take up too much of your time!
Best regards,

Tobias Lechner