Improve project CIF Exports

[briantoby]

Note comments below from 2/12/24 email "GSAS-II CIF requests"

Hi, Brian and Bob: A committee of the Commission on Powder Diffraction and IUCr staff in Chester have been working on making checkCIF and the publication process more powder diffraction friendly. Some progress has been made in eliminating irrelevant alerts, but a few things would be better fixed from your end.

1. Could GSAS-II add the weighting scheme the excluded regions H-atom treatment the number of restraints to the CIF it writes? Or do we make the author add these?
2. I gave them a CIF where the radiation was Mo. _diffrn_radiation_type was just K\a1,2. The Mo had to be added. Is this a glitch?
3. There is the matter of the maximum shift/esd. If the refinement has converged, could this value be set to the convergence criterion, or could you just do the one more cycle suggested? Are these things doable?

They have fixed the problem of senseless esds on bond distances and angles involving non-refined H (!). They are also working on recognizing wavelengths and residuals which are art of loops.

Reply 2/15/24

I have to say that writing code to create CIFs is one of my least favorite tasks, but I am certainly willing to work on improving the CIFs that GSAS-II creates. What would make this task a lot easier would be examples showing what GSAS-II produces and what would be better. I’ll include some comments and questions below.

On Feb 12, 2024, at 2:56 PM, kaduk@polycrystallography.com wrote:

1. Could GSAS-II add the weighting scheme

As far as I am aware, Rietveld is always weighted by experimental uncertainties (which depends on the data provider getting that right), except that we provide the possibility for changing the relative importance of different data sets. How would all this be indicated in CIF?

the excluded regions

This information is there, in that weights are set to zero for excluded points, but I could specify the actual regions. I’m not sure this was in pdCIF when I wrote the exporter or maybe I was just lazy.

H-atom treatment

What is being sought? How were the positions generated? Or how were they refined? The former is going to need to be put in by the user, but could go into the template for each phase. As for refinement, I don’t think we offer a riding mode, but people can create their own restraints and constraints so that could be messy.

the number of restraints

Is this number really worth anything? What about constraints? What about the relative contributions of restraints to the total chi squared? All of this can be produced.

2. I gave them a CIF where the radiation was Mo. _diffrn_radiation_type was just K\a1,2. The Mo had to be added. Is this a glitch?

I’m not sure if we store the anode or perhaps just the wavelengths. This will need some investigation and might be a no-can-do.

3. There is the matter of the maximum shift/esd. If the refinement has converged, could this value be set to the convergence criterion, or could you just do the one more cycle suggested?

I’m not sure if I understand this. Are you saying to have the refinement engine run for one more cycle after convergence is detected? Or run a cycle of refinement just before creating the CIF (I’d be reluctant to do that). What do you want reported for this?

I think most of this is doable. Maybe all.

Brian

cynarine_5.cif.gz cynarine_5.gpx.gz cynarine_H2O_R1.cif.gz vege083.gpx.gz vege083E_GSAS.cif.gz vege083E.cif.gz

[briantoby]

Outline of requests

• [ ] the weighting scheme

  _refine_ls_weighting_scheme sigma

  How to specify histogram weighting factors?

• [ ] the excluded regions

  _pd_proc_info_excluded_regions

  N.B. this is a text field that should have explanations. Should it be looped?

• [ ] H-atom treatment

  see https://www.iucr.org/__data/iucr/cifdic_html/1/cif_core.dic/Irefine_ls_hydrogen_treatment.html

  _pd_proc_info_excluded_regions

• [ ] the number of restraints

  This needs some discussion. There is a dictionary for individual restraints (https://www.iucr.org/resources/cif/dictionaries/browse/cif_core_restraints) but how to specify a number of restraints? See https://github.com/COMCIFS/Restraints_Dictionary/blob/main/cif_rstr.examples for examples. Use of weighting parameter needs thought. How to specify constraints?

• [ ] anode specification

  _diffrn_radiation_type 'Cu K\a~1~'

  I gave them a CIF where the radiation was Mo. _diffrn_radiation_type was just K\a1,2. The Mo had to be added.

  • [ ] can the shift/esd be computed without applying it?

  There is the matter of the maximum shift/esd. If the refinement has converged, could this value be set to the convergence criterion, or could you just do the one more cycle suggested?

[briantoby]

Note also IUCr Journals message about CIF problems from Nicola on 6/27/24