1919 ## Permute the components of z back to components of x
-> 1920 numpy.put(x, ipvt, wa)
1921 return(r, x, sdiag)
I've copied nmpfit.py to nmpfit_mos.py and edited so this bug no longer happens. However, this means that MOSFIRE will need to use a different version of nmpfit for now until stsci fixes theirs.
I've fixed this in my branch 'install_update'. This is the repackaging branch. I had to do it in this branch because it was the only way for me to test it in my python environment. However, you can apply the fix to the main branch by just copying over three files from my branch:
I've tested it and the interactive wavelength fitting seems to run. I haven't gone through the entire interactive fitting process though. The bypass=True mode does run completely.
There is a problem when using nmpfit from numpy 1.9 when the routine goes through the interactive wavelength fit. The error is as follows:
This can be traced to nmpfit.py:
I've copied nmpfit.py to nmpfit_mos.py and edited so this bug no longer happens. However, this means that MOSFIRE will need to use a different version of nmpfit for now until stsci fixes theirs.
I've fixed this in my branch 'install_update'. This is the repackaging branch. I had to do it in this branch because it was the only way for me to test it in my python environment. However, you can apply the fix to the main branch by just copying over three files from my branch:
Fit.py nmpfit_mos.py numerixenv.py
into the MOSFIRE directory
From: https://github.com/Keck-DataReductionPipelines/MosfireDRP/tree/install_update/MOSFIRE
I've tested it and the interactive wavelength fitting seems to run. I haven't gone through the entire interactive fitting process though. The bypass=True mode does run completely.