PDC

[flo]

I agree its a bit confusing in CP2K with the non periodic calulations.
You have the positions of the atoms wich are periodic or not. In
addition, the charge distribution you get from a FFT of the plane wave
basis is periodic as well.
To make the positions non periodic it is simple, just don't apply
periodic boundary conditions.
To make the charge distribution non periodic, it's a bit more
involved, since this is intrisically periodic. Therefore, you have to
chose an algorithm which is able to do so.

The MT scheme requires therefore a simulation cell about twice as big
as yous charge distribution, for you molecule in X about 40 Angstrom.
The Wavelet solver requires less vacuum, but the molecule has to be
centered, and during MD it might hit the boundary.

Don't know how big your cell was, but using MT the first time one
tends to amke it to small. Maybe you can check.

About the FULL all problem. I'm a bit surprised, usually this
converges almost independent of the energy gap, only the number of
iteration steps differs.

Cheers
Flo