Closed jdtsmith closed 1 year ago
Did a quick test at a couple of wavelengths, for each nirspec high-resolution segment. All the returned resolution values are reasonable now, when compared to the plots on the webpage that I linked in the issue.
Thanks. I realized with the convert
method I could fit with auto-domain then convert to domain-less, so I could remove that complexity. Mind taking another quick peek?
I also checked the medium resolution just to be sure. It all seems correct now.
NISPEC polynomial fits to resolution had implicit 'domain' set from numpy.polynomial.Polynomial. The domain was not saved, so evaluating the polynomial led to differet results. All segments were re-fit with default ([-1,1]) domain, except for prism, which did not converge without wavelength rescaling due to unusual shape. Direct support of domain was added.
Please check against published resolution results. Closes #235.