Is it good practice to use OT when studying reactions (e.g. when using METADYNAMICS or NEB)

[Christopher O'Brien]

Does it make physical sense to use OT in conjunction with a technique meant 
to study reactions such as METADYNAMICS or NEB?

The reason I ask is that I am not entirely clear if it is possible to 
change MULTIPLICITY (a.k.a. orbital occupation) with OT as the 
diagonalization method. During a reaction, it seems that changing the spin 
state is a reasonable thing to expect. I am not looking at single 
molecules, but the interaction of a clay-like surface with an organic 
molecule. 

Judging from previous questions to the mailing list, it seems that OT 
cannot be used with smearing and, consequently, with added MOs.

OT is very fast and for my 990 atom system, it is the only way I can 
actually make progress in the calculations. Are there any long time CP2K 
users who can suggest a few techniques to either tune mixing parameters, 
which combinations of preconditioners and diagonalization algorithms work 
best? Could a tutorial be written similar to what is provided with VASP 
that has a nice flowchart of how to handle pathological convergence 
problems? I understand some of the algorithms are the same, but the 
implementation is key, and there is no substitute for experience in 
handling convergence problems. 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20150920/67d90d69/attachment.htm>