???: discrete cellular particle assembly modeling

[Eugene Leitl <Eugene.Leitl (+ at +) lrz.uni-muenchen.de>]

 Dear CCLers,
 in about 3 months I am to embark on a PFP (protein folding) project by a
 novel modelling paradigm. I'd like to amalgamate the discrete-particle CA
 (cellular automata) modelling paradigm (pioneered by B. Hasslacher et al.,
 recently expanded to many other systems) and the featureless/smooth field,
 resulting in a new hybrid method. I'd like to represent both the state of
 the system, and the Hamiltonian as (large) data blobs, possibly GA-tuning
 the Hamiltonian (lots of look-up tables) using empirical data (Brookhaven
 PDB). Keeping code small and simple, moving complexity into the data
 structure instead. What I want to do is finding (whether manually, or by
 GA) the right discrete Hamiltonian, so that the right system evolution
 emerges naturally in the course of successive iteration. (I hope I do make
 sense). Due to the modeling paradigm's embarassing parallelism, resulting
 code should show roughly linear speedup in 3-d node grid supercomputer
 architectures. (I am thinking particularly of recent SuperDSP clusters, as
 SHARC, and TI's new SuperDSP product line).
 I'd like to learn whether anybody is pursuing similiar goals. An url/
 reference, perchance?
 And finally, an embarrassingly trivial question: since I'd like to start
 with an argon box, could somebody provide me with an equation for the
 exponentally corrected Lennard-Jones potential? Is there a simple way to
 parametrize the three-body interactions?
 XXL thanks in advance.
 Regards,
 Eugene Leitl.