MS 4.1 DMol3 cell optimization

Hi,

I've been trying to run a cell optimization of a hydrogen bonded system and moleucules using DMol3. The cell optimisation is a new feature of MS 4.1.

During the "optimisation" the cell doesn't seem to and the atom positions move to unphysical positions. The atoms in the hydrogen bonds get too close (less than an angstron) and some covalent bonds expand too much). This seems to be a bug in the optimisation routine. I have a feeling that, when cell optimisation is ON, it mistakenly adjusts the atoms positions instead of the cell parameters.

Nevertheless when I run a geometry optimisation (without cell optimisation) it gives very good results.

I wonder if anybody can help me. Is this problem known? What is the solution?

Luis