Dear Doroteya,
there is no systematic way in GSAS to make sequential Rietveld refinements
on a large amount of data files, but it is possible to write a sort of
script in a separate language, and couple it with the GSAS scripting to
achieve your goal. I routinely run through tens or hundreds of files using
this method. Here are the steps that are needed (at least the way I do it):
1) Refine a 'starting' file, e.g., the lowest temperature file, say
sam_001.gss.
2) Copy that file to a new file, sam_002.gss.
3) Edit the new file to change the input file names to the new data for the
next temperature.
4) Run POWPREF and GENLES on the new file.
5)Check for convergence. If not converged, run them again etc. If a
divergence occurs, stop and issue a warning. Clearly, the process will fail
through structural phase transitions.
6) When convergence is achieved, write out parameters with PUBTBL, DISAGL,
BIJCALC etc.
7) Repeat step 2 with the next file, and so on.
8) At the end of the process, one needs to extract all the parameters and
put them in tables. This is relatively easy for the .tbl files, since the
variable parameters are all followed by a bracket containing the errors, so
one can extract all of those and worry about what they are later. For
distances and angles and magnetic moments, one has to search through the
.lst files using some keywords (e.g., CU_O for Cu-O distances).
9) After running through tens or hundreds of files, one typically discovers
that he would like to change something in the structural model. Therefore,
it is important to be able to edit the .exp files and re-run them in a
sequential way.
The way this is implemented depends very much on the operating system. For
example, points 2), 4) and 6) reqire a system call, e.g. of the form
'POWPREF sam_002.gss'. Point 9) can similarly be achieved (in DOS) through
a system call of the type 'EXPEDT sam_002.gss < macro.mac', where macro.mac
is a GSAS macro file that implements the changes. I believe there is a
similar syntax for other operating systems. Points 3),5) and 8) require
physically editing files. This can easily be done in some operating system
shells (e.g. UNIX), less so in others. Alternatively, one can use scipting
languages (e.g. PEARL) or higher-level languages (e.g. FORTRAN).
I wrote a couple of working packages of this type, one in UNIX and the other
for Windows, using IDL as a scripting/GUI language, but I have got nothing
polished enough to be distributed. It takes something like 3 or 4 days to a
modestly competent programmer (like myself) to write the whole thing, and it
is well worth the effort, because you recuperate the time immediately on the
first big project.
I very much wish that somebody (Bob?, Brian?) will take the time to do this
once and for all. I suspect that Bob will see the light as soon as HIPPO
will start producing data.
I hope this helps.
Paolo
-----Original Message-----
From: Doroteya (details omitted)
Sent: Tuesday, August 21, 2001 10:34 AM
To: [EMAIL PROTECTED]
Subject:
Dear Mr. Radaelli,
I am a PhD student in the ... and I am working with GSAS to make Rietveld
refinement on powder diffraction data.
May I ask you, please, for an advice:
I need to make Rietveld analysis with big amount of data files and I would
like to ask you if you know is it possible to make GSAS refinement for many
files at the same time or is there some procedure to do analysis
automatically
for some consequence of files.
Thank you very much in advance,
Sincerely yours,
Doroteya
