Xref: utzoo sci.physics:3315 sci.research:376
Path: utzoo!mnetor!uunet!husc6!bbn!gatech!gtss!chas
From: chas@gtss.UUCP (Charles Cleveland)
Newsgroups: sci.physics,sci.research
Subject: Re: Atomic simulation software.
Message-ID: <249@gtss.UUCP>
Date: 2 May 88 19:55:56 GMT
References: <14718@oddjob.UChicago.EDU>
Reply-To: chas@gtss.UUCP (Charles Cleveland)
Organization: Georgia Tech School of Physics
Lines: 42
Keywords: simulations atomic software properties

In article <14718@oddjob.UChicago.EDU> harris@oddjob.UChicago.EDU (Jonathan Harris) writes:
  ... Excerpts from something I wrote ...
)>But totally apart
)>from quantum mechanical questions, we can't even take an arbitrary
)>simple potential (of the order of Lennard-Jonesium, for example) and
)>predict what crystal structure it prefers.  Even classically.  At zero
)
)This situation is much brighter. I cannot remember who has done the most
)recent work on the crystallization of Leonard Jonesian, maybe Keith
)Gubbins at Cornell. Molecular dynamics can be used to simulate the
)nucleation of crystals from the melt or vapor from the crystal. 

Yes, I've done it.

Molecular dynamics, which I more or less do for a living, cannot certify
that the structure it finds is the preferred structure in nature.  It
can only suggest its structure as a possible structure, perhaps even
with high confidence in its veracity.  It cannot exclude the possibility
that a lower energy structure could occur.  Many things can make it
impossible for a MD simulation to find the true structure (not the least
of which are the effects of periodic boundary conditions in 'bulk'
systems, and size effects).  But even if it is possible for the simulation
to find the lowest (free) energy structure, it isn't possible for the
researcher to be certain that it has done so.

And now that we have Lennard-Jonsium worked out (presuming that we do),
suppose we make a tiny change in it.  With MD, we have no recourse except
to start our sequence of simulations all over again, doing everything
in our power to be sure that we don't bias the system into reproducing
the same structure as the similar L-J potential, without introducing
artifacts that would make it difficult or impossible to obtain the
L-J structure if it were in fact preferred.  I was very careful in my wording,
for this very reason.  As your quoted above, I said that we can't take an
*arbitrary* simple potential (Lennard-Jones mentioned only to gauge
the level of complexity of the potential) and predict what crystal
structure it prefers.  This is a very fundamental and, to the best of
my knowledge, outstanding problem.
-- 
-Life would be so much easier if we could just look at the source code.-

Charles Cleveland    Georgia Tech School of Physics    Atlanta, GA 30332
UUCP: ...!gatech!gtss!chas         INTERNET:  chas@ss.physics.gatech.edu