Refinement fails due to TypeError: must be real number, not str for goniometer angles with value

[HamidReza-Azimaee]

When attempting to refine a powder pattern in GSAS-II (latest version, Windows 11), the refinement fails with the following traceback:

TypeError: must be real number, not str
...
pFile.write(' Goniometer omega = %.2f, chi = %.2f, phi = %.2f\n' %
(Sample['Omega'], Sample['Chi'], Sample['Phi']))

Upon inspection, the values for Omega, Chi, and Phi in the Sample Parameters tab were shown as 0. — which appear to be interpreted as strings rather than real numbers.

Workaround that resolved the issue:
Manually changing the values to a very small real number (1e-09) allowed refinement to proceed without error. Attempts to enter 0.0 directly reverted back to 0., and the error persisted.

Steps to reproduce:

Import a PWDR dataset and CIF.

Attempt refinement without modifying sample orientation values (which appear as 0.).

Refinement fails with TypeError.

Suggested Fix:
Ensure that default or user-entered values like 0. are parsed and stored as floats (0.0) instead of strings.

Let me know if you want to include system details or contact info.

[briantoby]

I suspect that this problem is related to the method that the instrument/sample parameters were set, since it is not happening all the time. I wonder if it is coming from to the instrument parameter file?

Could you supply a set of input files that you are using that generate this problem. If you are using a default set rather than reading from an instrument parameter file, I'd like to know which one you are using.

[HamidReza-Azimaee]

Hi Dr. Toby,

Thank you for your reply. I use the default instrument parameter "Defaults for CuKa lab data". Also I import CIF file downloaded from COD for phase data, and a Bruker RAW file for powder data, all of which attached here.

Input files.zip