Hi François.
Sorry for the delay in answering you. I took that time to make some
simulations in here. ;)
Basically, choosen to simulate cyclohesane instead of my derivatives
molecules. It's (not) "funny", but, the densities I've got for ciclohexane
area almost perfect, while the intermolecular energies deviate from -19%
from the experimental one. And this EXACT the trend I've noticed for my
derivatives of cyclohexane.
I'm reporeting it in here in case someones has any different information in
this subject. Will try decalins now and will report later.
Btw: can those error arise from something simple like polarization?
Thanks a lot for all help!
Sincerally yours,
Jones
On 10/1/07, FyD <fyd.q4md-forcefieldtools.org> wrote:
>
> Quoting Jones de Andrade <johannesrs.gmail.com>:
>
> > Can I make one question? Well, unfortunatelly I could not yet get access
> to
> > the reference you sent me (will have to get in touch with the author for
> > this specific one). But, from what I could get from its abstract, it was
> not
> > fairly easy to reproduce the eectrostatic potencial around a simple
> molecule
> > like cyclohexane with only simple atomic point charges.
>
> You get charge values that do not reproduce the MEP computated by QM...
>
> > I wonder what would lead to such a difficult on simple alcanes for that.
> > Specially considering the "trick" used of placing extra punctual charges
> in
> > midle of the hidrogens. I would not in a first instance expect such
> problems
> > on such simple molecules. Could you please give me any indication of
> what
> > sort of molecule "characteristic" or "property" would lead to such
> problems?
> > I still wandering what would be a good reason for that, even considering
> > some kind of "polarizability effect" that first came to my mind, but I'm
> > absolutelly unsure on this.
>
> We are working on adding extra-points in R.E.D. We might do something
> wrong, but, in these alkane cases, and from all our tests the
> improvements given by those extra-points are not that impressive, so
> far...
>
> Please, consider molecule H-H, the two hydrogens must be equivalent,
> the total charge of H2 is zero. This means to me that the charge of
> each hydrogen has to be zero... Is it real ?
>
> I do not know really what would be the impact of using such charge
> values for alkanes in MD simulation. May be someone in the AMBER
> mailing list has an idea ? From what I read, many examples of FF
> parameters are developped using those charges for alkanes, and it
> looks like that no problem is reported. The best example is the
> CT-CT-CT-CT dihedral parameter. Here, either no answer is known, or
> this is not a problem...
>
> My feeling is that "being aware of this problem" is already a good
> starting point ;-) (to be able to relativize the results obtained in
> MD in case of a reviewer would discuss about this point).
>
> Sorry, I did not help a lot. I do not have the answer.
> regards, Francois
>
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber.scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
>
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Wed Oct 03 2007 - 06:07:46 PDT
