Thank you sooooo much! I am learning a lot here!

Lily

On Tue, May 17, 2011 at 2:46 PM, Emilio Artacho <[email protected]> wrote:

>
> On 17 May 2011, at 19:41, lily zheng wrote:
>
> Just got one general question about structure optimization.   Usually we
> need provide initial(wild) guess about coordinates of atoms in one
> systems.
> could the structure optimization  procedure make those wildly-guessed
> coordinates reasonable?
>
> Thanks
>
> Lily
>
>
>
> That would be fantastic. Unfortunately not. Or at least not necessarily.
>
> The relaxation techniques built in in siesta as such (conjugate gradients,
> Broyden variants) are local relaxations, as opposed to global ones,
> meaning that siesta will find the atomic positions given by
> a local minimum of the energy, in particular, and most probably,
> the one your initial position is in the basin of.
>
> If your global energy minimum is in a different basin of the
> energy landscape (or valley or whatever way you prefer to call it)
> it won't be found by those algorithms: it would need to cross energy
> barriers (and the right ones) in order to get there, and there is no
> easy way to do that. This happens for instance if your initial structure
> is making wrong bonds among its atoms (you'll never get to the
> graphite structure if starting from diamond just by relaxing).
>
> A traditional way of trying to address that problem is simulated annealing,
> but it is extremely costly.
>
> More modern approaches:
> - ex nihilo structure prediction: see recent papers by
> Chris Pickard and Richard Needs
> - genetic algorithms: recent re-discovery by Artem Oganov and co.
>
> Otherwise you have to guess a good initial structure, with the right
> bonds, coordinations, connectivity ...
> The credibility of your answer depends on how good your guess was.
>
> See J. Maddox, Nature 335 201 
> (1988)<http://www.nature.com/nature/journal/v335/n6187/pdf/335201a0.pdf>
> .
>
> Emilio
>
> --
> Emilio Artacho
>
> Department of Earth Sciences, University of Cambridge
> Downing Street, Cambridge CB2 3EQ, UK
> Tel. (+44/0) 1223 333480, Fax  (+44/0) 1223 333450
> [email protected], http://www.esc.cam.ac.uk/~emilio
>
>
>
>

Responder a