>So here's some more fat in the fire! Major figures though Alan, Armel
>and Bob are, I suspect their comments stems from a lack of personal
>experience in the use of the correlated integrated intensity method
>of refinement. We experimentalists who use the technique find that it
>works very well indeed - our latest structure solution success from
>the ESRF shows the two methods - Rietveld and correlated integrated
>intensity methods producing the same results.
I confirm a total incompetency about the correlated integrated
intensity method of refinement. What is it exactly ?
>Your problem about equipartioning applies to the data not to the
>model. Again, I'll go through that off-line with you if you want.
What you do with the data seems important to me. Equipartitioning
was an example of what one can do : apparently this is not your
choice. So what is your choice ?
>Again, I'll go into the details off-line as I guess that they are
>beyond the interests of most of the mailing list. Suffice to say
>that, in practice, the two methods give the same results.
Going into the details off-line is not fair. You should realize
that if the two methods give the same results, then it means that
you have mathematically formulated the same problem in two
equivalent forms... If this is the case, the more simple form will
be retained in the future (meaning that if your process qives the
same result as the Rieveld method but in more complex way, it will
not be kept). Now if the two approaches do not give exactly the
same result, then one is better than the other. If correlated
integrated intensity method is the best, then we will have to
close the Rietveld mailing list and to open the David mailing
list :-). That's why it is unfair to continue the discussion
off line.
Armel
Le Bail - Residence Cristal,
2, rue de Gasperi, 72100 Le Mans
http://www.cybercable.tm.fr/~cristal/