> What I would like try to do is integrate the hamiltonian of some molecule
>using a symplectic numerical integrator
A typical molecular mechanics Hamiltonian has many terms, usually for 1, 2, 3,
and 4 bodies. There are any number of books which describe them, for instance
McCammon and Harvey, "Dynamics of proteins and nucleic acids"
___ 2 ___ 2
V = (1/2) \ ( K (b-b ) ) + (1/2) \ (K (theta - theta ) ) +
/__ b 0 /__ theta 0
bonds angles
___
(1/2) \ ( K (1+cos[n phi - delta)]) +
/__ phi
dihedrals
___ [ A C q1 q2 ]
\ [ --- - --- + ----- ]
/__ [ r^12 r^6 D r ]
nonbonded
pairs
> What I would like try to do is integrate the hamiltonian of some molecule
> using a symplectic numerical integrator running on several loosely coupled
> processors.
You might want to take a look at our program 'namd'. Information about it is
on the web at http://www.ks.uiuc.edu/Research/namd . The program is designed
for workstation clusters, which is about as loosly coupled as you can get. It
is also designed to be used by others, so there is complete source code
documentation (the "programmer's maual") as well as commented C++ source code
which is freely available and for free from our ftp site (ftp.ks.uiuc.edu in
/pub/mdscope/namd). Plus, we plan to experiment with different simpletic
integrators - though only a couple Verlet methods are there now.
Also, there is a section in the the documentation describing (in glourious
LaTeX equations) the different terms of the Hamiltonian, and differentiating
them to get the force terms as well. BTW, the documentation is in Postscript.
Andrew Dalke
dalke at ks.uiuc.edu
