Hello again,
I've been making some headway on this constraint virial problem, and I
now know that my error involves the computation of constraint forces for
SETTLE'd water molecules.
Does anyone know where I can find formula that tell you the forces to
apply along each of the three constrained bonds to reproduce the
accelerations that the SETTLE routine applies to the water? The problem
can be phrased fairly simply: SETTLE is considered to have applied forces
along each of the constrained bonds to return the molecule to its original
shape after a Verlet-integration (half) step deformed the water molecule
slightly. But, at the end of the SETTLE calculation, all we have are the
new velocities, from which we derive the net accelerations and thus the
net forces on each atom. Ww must account for these force vectors with a
unique set of forces acting along each of the bonds. Three constrained
bonds, three atoms and one net force on each atom resulting from the
SETTLE calculation.
But, when I try to write down these equations and then solve for the
magnitude of the force along each bond, I run into all sorts of trouble
and I can't figure out why. It would seem that the system is
over-determined; I can write nine different equations relating each of the
three components of the three net atomic forces computed by SETTLE to two
other terms involving the forces applied along each of the bonds ending at
each of the three atoms. I can pick three of the equations and solve them
for the magnitude of the force along each bond, but for some resaon I get
different results depending on which three equations I pick--they don't
even seem to be self-consist when I look closely at the equations.
But, at the same time, it seems that there are three unknowns (the
magnitudes of the forces along each bond) and three knowns (the net force
on each atom applied by SETTLE) so there should be a fairly concise
solution to these equations...
Thanks!
Dave
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun Feb 18 2007 - 06:07:13 PST
