LeBail fit divergence when profile fitting 2 crystal phases

[Charlaiz]

Hello,

I've encountered issues when trying to perform a LeBail profile fit on a powder pattern with multiple phases. This problem does not occur when working with single phase patterns. I have fairly good lattice constants from both phases (it is a mix from two pure phases I deliberately made to try multi phase refinement), so the starting parameters should be ok.
For these particular phases, Fox profile fitting works very well when dealing with phase-pure samples; however in the mixed powder pattern I am unable to get a decent profile fit - it diverges from the start. I have verified in fullprof that this pattern that a good profile fit can be easily obtained for this pattern.
I add the input file (as .txt) as an attachment.
Thanks in advance!
gis-ana.txt

[vincefn]

Thanks for the example. Indeed it is not converging correctly. By deactivating some of the most sensitive group of parameters (asymmetric profiles, displacement. - background seems OK in this case) there is no divergence but something is wrong e.g. with the Le Bail fit.

Strange, with a slightly different procedure it works with multiple phases in python using pyobjcryst - cf https://pyobjcryst.readthedocs.io/en/latest/examples/Quantitative-phase-analysis.html