[sdpd] Re: I am new here!

["Dr. Jaap Vente" <jaap... @jade.cieamer.conacyt.mx>]

To continue the discussion. 

First of all I did not intend replace direct methods with Gulp, nor did I
propose to throw the good old diffractometer in the bin. I have no
intention of that at all! Diffraction data are to be collected. I only
proposed to combine the techniques, and in doing so to increase the number
of observations. This only to solve the structure, not to determine and
refine. Of course we are not able to model any random system as yet, but
that is not a reason not to pursue it. Indeed there are very few
structures done this way, but also that is not a reason not to think of it
as a promising way forward. I know only of Li3RuO4 by Bush and Battle,
that was really helped by using Gulp. But do not underestimate the
potential power of it. At this moment it is possible to predict rather
precisely the cell parameters if you know the structure. Handy if you
expect isomorphism. 



> Direct method always need diffraction data. So few structures were
> really predicted and so many were truely determined by direct methods
> that I am not sure if adding prediction possibilities to the direct methods
> would change anything to the result ?


The result would not change I guess, but the likelyhood of automatically
finding the 'best' model would become greater. And it may be faster I do
not know that.

> My bibliography about truely predicted structures is probably false now
> since I vaguely remember of only one not disputable successfull case.
> We are certainly far from being able to predict which solid phases
> would occur in the system ThF4/ZnF2/GaF3/LiF for various syntheses
> conditions, and what would be their structures, without doing any 
> experiment. This would be true prediction (hope you agree with that
> definition).


Jep Agreed, and agreed that we are a long way away from that. Failure to
model complicated systems is not a reason to look into it though. How long
ago was XRD only finger printing?


> 
> So, I would recommend some diffraction data, if you have a sample. 
> Then I would recommend applying direct methods or
> Patterson, or random starting models by ESPOIR, or molecule location,
> in that order, and finally, if all that fail, then prediction, or prayer, maybe.
> 

Or combine things!


> What is the estimated complexity of your problem (how many
> independent atoms to find) ?
> 

Well the problem is it contains oxygen, aluminium and magnesium, and I
only have x-ray data as yet. And the cell is large!


Best

Jaap




Jaap Vente
Cinvestav-IPN Unidad Merida
Departamento de Fisica Aplicada
Carretera Ant. a Progreso km 6
Apartado Postal #73 Cordemex
Merida, Yucatan, 97310
Mexico
Fax: (..) 52 - 99 - 812917
Tel: (..) 52 - 99 - 812960 ext 246/233
e-mail: jaap... @jade.cieamer.conacyt.mx