[Wien] way to confirm stable/transition structures

[Laurence Marks]

This is a slightly complicated question. What you sent is a transition
state, mathematically a saddle point. You are asking if within MINI
(and/or MSR1a) you can get trapped in one.

I can only answer for PORT & MSR1a. If you start a little away from a
saddle point, and there are no other issues such as problems with
touching RMTs or inconsistent forces, the program will converge to a
true local minimum. There are some standard traps that ensure that
this is always the case which are well known although often omitted in
simple textbooks. I am 99.8% certain that MSR1a will also not be
trapped in a saddle point although there is currently no analytic
proof of this.

A caveat. If you were to start at the exact saddle point (or very
close to it) the mini code may decide that you have already converged
and stop without checking. This is because the convergence criteria
are just that the forces are small, not that the Hessian is positive
definite (i.e. a local minimum).

Another caveat. If you have a soft mode distortion away from the
saddle point the code might also be fooled into thinking it has found
a minimum whereas in reality it has not. PORT is quite good in
avoiding this; MSR1a not quite so good for a number of reasons.

And yet a third one. It is possible to force a saddle point using
symmetry, for instance if you added a z-axis mirror.

If you are concerned whether you have got trapped in a saddle point
because you started there, move the atomic positions a bit and rerun.

N.B., a different question is how to find a saddle point/transition state.

On Fri, Oct 5, 2012 at 4:55 AM, tasaka at affinity-science.com
<tasaka at affinity-science.com> wrote:
> Hello,
>
> In MINI calculation, a relaxed structure is depend on initial structure
> in general. so, sometimes it might arrive to the transition structure.
> for example, please find the attached structure file.
> for that case, how do we confirm the obtained structure whether
> transition or stable structure, especially in wien2k-code?
>
> sorry if it's a silly question, but any information would be appreciated.
>
> With best regards,
> Tomo



-- 
Professor Laurence Marks
Department of Materials Science and Engineering
Northwestern University
www.numis.northwestern.edu 1-847-491-3996
"Research is to see what everybody else has seen, and to think what
nobody else has thought"
Albert Szent-Gyorgi