X-Message-Number: 26004
From: 
Date: Tue, 12 Apr 2005 03:49:22 EDT
Subject: Uploading (2.V.0) Q.M. and M.D.

Uploading (2.V.0) Q.M. and M.D.
 
Beyond Boolean, differential and stochastic models, there are tow more 

possibilities : Quantum and molecular dynamics ones. The most detailed model is 
the 
quantum Schrodinger equation. Taken head on with brute force computation, it 
can solve only the hydrogen and helium atoms and the molecular hydrogen H2. 

Everything beyond needs some approximation. Computation recipes have been cooked
so that any atom can be modeled and molecules with up to 200 elements are 

worked out. This bring the model to the 2 nm and some femtosecond scale. May be

someday, quantum computers will extend this domain in the macroscopic realm, but
for now, there is few interest in this level for brain modeling.
 
These quantum models can be found under the name of : Hartree-Fock (HF), 

Density Functional Theory (DFT) or Configuration Interaction (CI). See for 
example 
Schaefer (1984) (1). More phenomenological models, using some experimental 
inputs, have been built too. They can be found under names such: MINDO ( 

Modified Intermediate Neglect of Differential Overlap), or Extended Huckel model
(EH) 
or AM1...(2). Such models are interesting to see for example a large 

molecular complex, may be a ion channel. From what is seen here, more simpler 
models 

can then be evolved producing nearly the same output of interest at far reduced
computational cost. This is not a model class for uploading, only a way to 
check that a simpler system works properly. This is a research tool, not 
something being implemented in an artificial neuron.
 
At an even broader range, quantum mechanics produces only  a point source for 
a force field at each atom. The work is done exclusively at the force field 
level. This is the so-called Molecular Dynamics (MD) scale. Objects up to the 
viral or small cell organ can be computed here. One interesting use of MD is 
the simulation of nonrandom diffusion. Molecular diffusion rests on the Fick's 
law built on the Brownian motion. This is the archetype of the random walk and 
of all random processes. When Ca++ ions enter a set of ion channels, they 

create a local and brief pulse of Ca++ concentration and this one diffuse 
rapidly, 
too rapidly in fact to act on some other channel such the so-called 

metabotrophic ones. What is really going on ? It seems we face here a case of 
nonrandom 
diffusion. Water molecule are polarized and at short distance they organize 
temporarily as a crystal-like structure. Near a membrane cell where many big 
molecules have an electric charge, the water may be extensively organized. 
Anything entering this domain will not diffuse at random, here will be some 
anisotropy of the medium. This may be very important for loading Ca++ on some 
molecules.
 
Beyond long range electrostatic force, the MD scheme takes into accout the 
short range van der Waals field. There are three equations used for simulating 
it: The Lennard-Jones, the Buckingham exponential and the Morse function. One 
of the best MD simulation technique is the Cell Multipole Method (CMM) (3).
 

 
(1) Schaefer, H. F. (1984) in: Quantum Chemistry: The Development of Ab 

Initio Methods in Molecular Electronics Structure Theory. Clarendon Press, 
Oxford.
(2) Pople, J. A., and Beveridge, D. L., (1970). In: Approximate Molecular 
Orbital Theory. McGraw-Hill, New-York.
(3) Figueirido F., Levy R.M., Zhou R.H.,and Berne B.J. (1997). Large scale 

simulation of macromolecules in solution: Combining the periodic fast multipole
method with multiple time step integrator. J. Chem. Phys. vol.107: p.7002.

Yvan Bozzonetti.


 Content-Type: text/html; charset="US-ASCII"

[ AUTOMATICALLY SKIPPING HTML ENCODING! ] 

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=26004