RE: design complexity of assemblers (was: Ramez Naam: redesigningchildren)

From: Ramez Naam (mez@apexnano.com)
Date: Tue Nov 26 2002 - 15:45:35 MST


From: Eugen Leitl [mailto:eugen@leitl.org]
> > current modeling methods that are able to handle 10^9 atoms.
>
> It's not at all current, the landmark paper of a billion
> atoms is quite a few years past (I think the figure has been
> several billion, but I can't produce the ref off hand, as the
> paper is somewhere in a box in the cellar).

I would love to see this reference. Since you mention GROMACS in your
post, let me use that as an illustration of where I consider the state
of the art to be. In Science 294 (2001), Bert L. de Groot and Helmut
Grubmuller present a 100 nano-second simulation of 10^5 water
molecules using the GROMACS code (which produces highly highly
approximate solutions). At the time it was published, this article
was hailed as one of the largest simulations ever done.

Even so, fully empirical forcefields are so error-prone, even when
highly tuned to the specific domain that they're being used for, that
one commonly hears in the field that only 10% of forcefield derived
solutions that are checked empirically turn out to be accurate.
Certainly you hear that in the field of protein + small drug
interaction modeling, where there are billions of dollars to be made
by producing highly accurate yet fast modeling methods.

You do make a good point (not quoted here) that a dry machine-phase
system will be easier to model due to lack of solvation effects, etc..
However, I question the assumption that assemblers will really be dry
machine-phase systems, or that they will conduct all of their work in
such environments. Remember, you have to be able to model not just
the assembler, but the assembler plus the environment it's working in.

> I don't see why one shouldn't be able to draft a sufficiently
> accurate all-purpose force field which can perform at least
> as well as current ones (say, GROMACS), it's just it's not
> relevant in practise. No one is modelling hybrid systems. At
> least yet.

Heh. First, GROMACS, while fast, is hardly accurate. Second, a
highly accurate, all-purpose force field as you're describing is one
of the holy grails of the field - everyone would love one, no one
knows how to build one.

At any rate, this is a stimulating conversation. I'm glad to be
having it.

cheers,
mez



This archive was generated by hypermail 2.1.5 : Wed Jan 15 2003 - 17:58:25 MST