"Internal periodicity" constraints for CELL_OPT with multiple unit cells

[Konstantin Tokarev]

Hello CP2K community,

As far as I understand, usage of  non-screened hybrid functionals (e.g. B3LYP) requires setting of HF CUTOFF_RADIUS to value not larger than half of smallest unit cell dimension. So, if single unit cell is to small to provided sufficiently large cutoff radius, it seems to be a good idea to use multiple unit cells so that CUTOFF_RADIUS could be enlarged.

However, when optimizing such a "multicell" I see that positions atom coordinates inside "subcells" can differ from each other substantially, leading to lack of internal periodicity. While this is not surprising, deviations in my particular case are quite substantial and much larger than what could been expected from error values in CIF file.

Is there any way to enforce internal periodicity when using multiple unit cells?