Comparing results between Topas and Fullprof

dward

Dear All,

I am encountering an issue while fitting structural parameters to XRPD data (from simple oxides) using Topas and Fullprof. The initial parameters and the peak-shape function (pseudo-Voigt), are identical in both cases. However, the final results deviate significantly between two programs, exceeding the acceptable limits.

To address this, I decided to reduce the update step in each iteration. For example:
• a = 0.5035885 update = Val + 0.005 Change;
• b = Get(a);
• c @ 13.75243 update = Val + 0.005 Change;

Additionally, I employed LU decomposition instead of the default solver using the command:
use_LU

These modifications seem to improve the results, but I would like to confirm if this approach is valid and whether the commands I used are appropriate.

Best regards,
Dariusz

johnsoevans

Lots of people have compared TOPAS fits to both real and simulated data with other packages like gsas, fullprof, shelx and crystals. I'm not aware of any significant unexplained differences between them. If there are, then I'm sure the authors would want to sort them out.
In my experience TOPAS generally converges more effectively than other software so it's less likely to get stuck in a false minimum. Could this be your problem?
You should also remember that things like cell parameters will depend on the specific peak shape model you use if they are asymmetric. Quantities that depend on intensities (positions, occupancies, temperature factors) will depend on the specific corrections you use. These may differ between packages.
I'm not sure why you are using the "val + change" syntax in your example. Cell parameters are normally well defined by powder data so usually converge rapidly to the correct minimum with an appropriate model. If you're having to do this then something is probably wrong in your INP file.
Hope the above is some help.