SE SCP-NDDO MD of Bulk Water

[p.ba... at googlemail.com]

Dear cp2k community,

I am trying to perform a SE MD simulation with 64 water-molecules with PBC 
using the SCP-NDDO method, as described in Murdachaew, Mundy, Schenter, 
Laino and Hutter, J. Phys. Chem. A 115, 6048 (2011). Unfortunately, I have 
not been able to reproduce the reported results yet. 

I know that the usage of PBC in SE simulations needs a lot of insight into 
the algorithms. Hence I tried to optimize the parameters in the SE and 
POISSON/EWALD sections, but after a couple of MD steps the calculation 
ceases to convergence. 

Therefore I tried using PERIODIC EWALD_GKS in the SE-section instead of 
PERIODIC EWALD. According to the manual and some testings this method 
disables the usage of multipoles (the manual rather says that PERIODIC 
EWALD enables multipoles) and is independent from the definitions inside 
the COULOMB and EXCHANGE methods. 
However, the revised version works well till 600 steps, then the kinetic 
energy of the system starts rising and a few H-O bonds seem to break. 

What is the difference between EWALD and EWALD_GKS?
Is an outer SCF loop of TYPE SCP necessary? 

Many thanks in advance!
Bassirian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140721/7344d017/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: EWALD.rar
Type: application/rar
Size: 40381 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140721/7344d017/attachment.rar>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: GKS.rar
Type: application/rar
Size: 4987665 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140721/7344d017/attachment-0001.rar>