load structure information using diffpy.structure package

[dragonyanglong]

Hi @pavoljuhas, I am currently working on PDFgui2 project, replacing PDFFIT engine with your CMI engine. So in the future, one important functionality is that people can use magnetic PDF in PDFgui2 to analyze data from neutron PDF. I may sometimes have questions about the codes and functionalities, like in PDFgui and CMI, during my developing. Can I ask you questions for them in the future? I really appreciate your help and time!

Currently, I am working on calculation (no refinement yet) in the pdfgui2 branch on my own forked repo. I meet a problem about loading structure information for CMI engine in the GUI. Could you help give any clues when you have time?

In calculation.py , I would like to load structure information for CMI PDFcalculator. How can I do similar thing like this part of code for CMI PDFcalculator load structure for PDFFIT engine ? The original self.owner.strucs is in such format: [[Ni 0.000000 0.000000 0.000000 1.0000, Ni 0.000000 0.500000 0.500000 1.0000, Ni 0.500000 0.000000 0.500000 1.0000, Ni 0.500000 0.500000 0.000000 1.0000]]

I tried to use diffpy.Structure.loadStructure function to read the structure information self.owner.strucs, but failed. Any solutions?

Or perhaps I even don't need to use diffpy.Structure.loadStructure for CMI PDFcalculator (although that's the only method I use when using CMI for refinements)? I could just keep such format and add structure information into CMI PDFcalculator load structure for PDFFIT engine ?

Thank you very much, Pavol!

[pavoljuhas]

Can I ask you questions for them in the future? I really appreciate your help and time!

Sure, I'd be glad to help.

How can I do similar thing like this part of code for CMI PDFcalculator load structure for PDFFIT engine?

You can directly pass FitStructure instances from self.owner.strucs to PDFCalculator call. No need to write any file. There are however several caveats:

1. unlike pdffit2, PDFCalculator has no utility for constraint equations, therefore you need to make sure any constraints have been applied to the structure before passing it to PDFCalculator. This can be done by calling FitStructure.applyParameters method; this is effectively the same as the |Apply parameters| button in the Parameters panel.

2. PDFCalculator works with one structure. To model multi-phase PDF, you need to sum up partial PDFs in the Calculation.

3. PDFCalculator currently applies some metadata from FitStructure.pdffit dictionary to the calculation, but I am going to remove that functionality. To make sure your code will not break, I recommend to disable metadata using the nometa function and set any calculator attributes as needed yourself. Example:

   
   from matplotlib.pyplot import subplots, show
   from diffpy.srreal.pdfcalculator import PDFCalculator
   from diffpy.srreal.structureadapter import nometa
   from diffpy.pdfgui.tui import LoadProject

load example FitStructure from PDFgui project file

prj = LoadProject('src/diffpy/pdfgui/tests/testdata/ni.ddp') nickel = prj.getPhases()[0]

calculation with implicit metadata

nickel.pdffit['delta2'] = 4 nickel.pdffit['spdiameter'] = 20 pc1 = PDFCalculator() r, g1 = pc1(nickel)

calculation with no metadata

pc2 = PDFCalculator() r, g2 = pc2(nometa(nickel))

calculation with explicit calculator setup

pc3 = PDFCalculator() pc3.delta2 = 4 pc3.addEnvelope('sphericalshape') pc3.spdiameter = 20 r, g3 = pc3(nometa(nickel))

fig, ax = subplots() ax.plot(r, g1, label='implicit metadata') ax.plot(r, g2, label='no metadata') ax.plot(r, g3, label='explicit metadata') ax.legend()

show()