Grouped sequential fits now possible; fixes significant TOF error with Dij terms: v5.9.0

[briantoby]

Up to now, one could perform a combined fit, fitting phase(s) to multiple histograms, or a sequential fit, where one fits phase(s) to a series of histograms one at a time. However, one might have a parametric series of measurements with multiple datasets at each setting. Examples might be with multiple x-ray wavelengths each collected at a series of temperatures, or where neutron and x-ray data have been collected with some common temperatures, or simply TOF data collected with multiple banks with a series of compositions or temperatures.

In v5.9.0 (#5844) a new mode in GSAS-II allows for grouped sequential fits. Grouping is done by designating what histograms are in each group (in Controls) using the histogram names, and then a combined fit is performed on the histograms in each group. This is a fairly major reworking of the sequential refinement code that was started about a year ago. The former sequential fitting mode has been tested and works the same, but new bugs are always possible; please let us know if anything breaks.

This new version does fix one very significant bug (thanks to Angus & Chenguang for reporting it). For TOF fits, the Dij terms (the hydrostatic strain tensor) were being applied twice. It appears that this bug goes back to June, 2025. Fortunately, use of the Dij terms with TOF data is not used all that often.

If anyone has a good set of data for a grouped sequential fit tutorial, please get in touch. Also, again we have managed to do something that is not yet supported in pdCIF, so that is going to need some thought.

[briantoby]

An e-mail to the GSAS-II mailing list with more information on the bug is here: https://mail-archive.com/gsas-ii@aps.anl.gov/msg00396.html