Force field for fully flexible water

nadler rod... at gmx.ch
Wed Oct 8 09:04:43 UTC 2008


Well, I have to say that I am VERY glad that you and all the people
around you are still up and healthy! :D

I tried a time step of 0.05 fs and it works... But it is strange
because in the paper they reported a time step of 0.75 fs (which I
tested and it doesn't work as well) and the parameters I used where
from Table 1, "Parameteres for water in ambient condition in this
work". Therefore the numbers should be ok, according to the paper. So
I changed the units to kjmol*rad^-2 and it worked too!! The
fluctuations in the conserved quantity are small.

Thanks Axel and Teo for guiding me to a working input :)

Roger



On 7 oct, 23:38, Laino Teodoro <teodor... at gmail.com> wrote:
> Dear Roger,
>
> do you have an idea of what 353.0  kjmol*deg^-2  mean?
> if you are at the equilibrium position, and you compress or enlarge  
> the H-O-H angle of just one small irrelevant degree
> the energy of your system would increase of ~ 85 kcal. This means  
> that thermal fluctuations of few degrees normally connected with the
> bending H-O-H would be enough to break chemical bonds.
>
> That's why I'm not wondered to hear from you that your water is  
> exploding!.. I'm not sure.. I had no time to test your input.. but the
> result should be roughly like the famous Tsar bomb (if you're  
> curious:http://www.youtube.com/watch?v=FfoQsZa8F1c)
>
> Since I drunk few minutes ago a fresh glass of water (kindly offered  
> from the Zurich water supply network), I can witness
> that it was not like the plasma you call water! ;-)
>
> So.. I would think a bit more on your units!
>
> of course as Axel suggested, with a potential like this one.. maybe  
> an integration time step of 0.05-0.1 fs is enough.. not for
> getting a proper conserved quantity.. but at least to tame the  
> integrators...
>
> Teo
>
> >       &BEND
> >         ATOMS H O H
> >         KIND HARMONIC
> >         K [kjmol*deg^-2] 353.0
> >         THETA0 [deg] 109.4712
> >       &END BEND


More information about the CP2K-user mailing list