cp2k - ab initio simulations - reproduction of the denisty for condensed phases

[Anna Wójcik]

Hi :)
I'm a new user of cp2k program. Till now I was using only QM, QM/MM and 
sometimes MD methods (MD only to model enzyme structure before QM or 
QM/MM), but I need to use cp2k for requirement parametrization of some 
molecules. I did some test modelling on condensed phase including 30 
molecules of triacetin and also for 30 water molecules as a reference, but 
so far I'm receiving overestimated density for triacetin and for water. 
Ab-initio MD was lasting about 2 ps using NPT ensemble. The model was 
equilibrated using classical MD in GROMACS in OPLS force field, afterwards 
I chose two geometries which have density equal to experimental one and box 
size I took exactly for these geometries from classical MD, than I did 3000 
steps in NVT ensemble of ab initio MD using cp2k and because temperature 
was OK, and all energies including ConsQty reached plateau I performed 
ab-initio MD in NPT ensemble, but in all models I reached overestimated 
density and the error equals about 0.1 g/cm3 in all models tested for 
triacetin and for water. My question is: Do You have any suggestion what is 
the reason?? Maybe the reason is an algorithm or maybe settings or maybe 
simulations were to short or models two small?? Or maybe for calculations 
in NPT ensemble I should took bigger box than in classical MD?
I would be very grateful for any suggestions
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20160314/bbf8aca5/attachment.htm>