Hi Ignacio,
I believe the problem you describe is a function of the fact that the PME
terms are not pairwise decomposable and thus it is not possible to remove
the contribution between fixed atoms when using Belly. The general advice is
that one should NOT be using belly with PME/Ewald calculations. Instead it
is better to use ntr=1 harmonic restraints.
Others may be able to provide more in depth comments.
All the best
Ross
> -----Original Message-----
> From: Ignacio J. General [mailto:ijgeneral.gmail.com]
> Sent: Thursday, July 28, 2011 12:49 PM
> To: AMBER Mailing List
> Subject: [AMBER] Energy output when using ibelly
>
> Dear Amber users:
>
> After experimenting a little with Belly using Ewald, I found something
> I didn't expect about the MD output energies. Could anybody please
> check that what I understand is correct?
>
> In the standard output there are, among several other quantities, one
> called EELEC and another called 1-4EEL. The 1st one gives the total
> Ewald electrostatic energy of the system, with the exception of the
> 1-2, 1-3 and 1-4 interactions (or are those terms also included?). The
> 2nd one gives a 1-4 electrostatic energy, calculated outside Ewald, I
> imagine in the same way that the others (BOND, ANGLE, etc.) are
> calculated. So that if you put together EELEC + 1-4EEL, you get the
> total electrostatic energy of the whole system (of course, except 1-2
> and 1-3 interactions).
>
> So far, so good. But I found out that if you use belly, then the
> 1-4EEL (along with BOND, ANGLE, etc.) ignore the energies between
> fixed atoms. On the other hand, the Ewald calculation, and thus EELEC,
> is unaffected by belly, so that energy does include even the fixed
> atoms. So EELEC and 1-4EEL are incompatible! Consequently, if we do
> EELEC + 1-4EEL, we don't get the total electrostatic energy of the
> system.
>
> This could not be a problem, depending on what you are calculating. In
> particular, if you do charge TI, DV/DL is obtained from (EELEC +
> 1-4EEL), so you may have big problems here...
>
> If anybody has some comment on this I would really appreciate it.
>
> Ignacio
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jul 28 2011 - 13:30:03 PDT
