TST: Adjust Lorentz1D FWHM result for devdeps

ejeschke / ginga

The Ginga astronomical FITS file viewer

BSD 3-Clause "New" or "Revised" License

122 stars 77 forks source link

TST: Adjust Lorentz1D FWHM result for devdeps #1106

Closed pllim closed 3 months ago

pllim commented 3 months ago

Adjust Lorentz1D FWHM result to be compatible with astropy v7.0 probably due to astropy/astropy#16794 bug fix upstream. Without this patch, the devdeps job fails. Example failure log: https://github.com/ejeschke/ginga/actions/runs/10448173088/job/28928224243

Would be nice if @larrybradley can confirm that such a change downstream is actually expected from the Lorentz1D fit_deriv fix. 🙏

larrybradley commented 3 months ago

@pllim Yes, having incorrect derivatives can result in the fit not fully converging to the correct values. One way to test this is to set estimate_jacobian=True when calling the fitter object. That option will not use the model fit_deriv, and then you can compare results with and without estimate_jacobian (that's how the astropy model fit_deriv tests are performed)

pllim commented 3 months ago

Thanks, all!