Some additional words about the "problematic step of
extracting the intensities of individual reflections from
the powder data".
I am not an expert in Monte Carlo or Genetic Algorithm
methods that try to fit the raw pattern with a model (or piece
of model) plus rotation and so on. Nevertheless, when I try
to figure out what is really done, it seems to me that the
"problematic step of extracting the intensities of individual
reflections from the powder data" cannot be avoided. Let me
explain why.
In order to compare the model-calculated pattern to the a raw
pattern, you have to know at best : the profile shapes, widths, cell
parameters, zeropoint, etc. And what is the best way to know
these values nowadays ? IMHO the Pawley and Le Bail (pub) methods !
So tell me if I abuse myself in believing that, anyway, before to
apply Monte Carlo or Genetic Algorithm (...) you have to
extract at least your profile and cell parameters at best by doing
this "problematic step of extracting the intensities of individual
reflections from the powder data", but discarding the extracted
intensities :-). And if I am not wrong (of course this step is never
explained in those publications...), how can you resist to a fast
direct or Patterson application to these by-product extracted
intensities ?
Finally the important point is to determine the structure whatever
the method. All of us are then using the Rietveld method for the
final refinement. Nothing has been found to replace this final
step, otherwise this mailing list would have no reason to exist.
Comments ?
Armel