From: Jim Phillips (jim_at_ks.uiuc.edu)
Date: Wed Dec 03 2003 - 16:24:45 CST
Zhu seems to be right on this one (and he's actually used it). I've just
modified the manual entry for fixedAtomsForces to reflect this.
I general, mixing fixed atoms with constant pressure is asking for
trouble. A few fixed atoms is probably OK, and is handled like any other
external force as either tension or compression based on the input atomic
coordinates. If you fix a globular protein to equilibrate the solvent
around it you're OK. If you fix the membrane around your channel, you
will probably see some strange behavior.
On Tue, 2 Dec 2003, Fangqiang Zhu wrote:
> I think in the current version of NAMD, "fixedAtomsForces on" is not required
> for constant pressure simulations, at least this is not enforced by the program.
> my opinion, those forces shouldn't be included in the pressure calculation at
> However, one should be aware that when a large fraction of the atoms are fixed,
> the pressure calculation will be somewhat inaccurate. But, in my opinion, this
> problem can't be solved at all by including forces between fixed atoms.
> things can even be worse if those forces are included. In practice, at early
> equilibrium stage, I often need to fix some atoms and run constant pressure
> simulations, with forces on fixed atoms ignored, and everything seems fine. So
> I disagree with the instruction in the NAMD manual regarding the
> "fixedAtomsForces" option.
> Again, this is my personal opinion, and Jim may have different views...
> Jan Saam wrote:
> > Hi all,
> > When you are trying to simulate a system using constant pressure then NAMD
> > requires that all forces, including the ones between fixed atoms are
> > calculated ("fixedAtomsForces on").
> > I assume this is somehow due to the the pressure calculation scheme that uses
> > the atomic virial and kinetic energy.
> > But couldn't one just take omly the moving atoms into account?
> > I remember that "fixedAtomsForces on" was not required in former versions of
> > NAMD. How did it work then?
> > Does anybody know about this?
> > Thanks,
> > Jan
> > ++++++++++++++++++++++++++++++++++++++
> > Jan Saam
> > Institut für Biochemie
> > Monbijoustr. 2
> > 10117 Berlin
> > Germany
> > 030-450 528 449
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:37:12 CST