Closed iandroberts closed 4 months ago
Hi Ian,
Thanks for using the code, and sorry for the very slow reply. We had stopped using the calibration approach in MaNGA a long time ago, and the test coverage isn't good enough to catch a couple of bugs. Thanks for reporting this!
I've implemented a fix in #116 . Would you mind jumping on that branch (covar_approx
) and giving it a test run on your data? If it works, I'll merge the PR and release a new tagged version.
Thanks! Kyle
Thanks for the patch, I just tested it on one of my cubes and everything seems to be working properly. I think you should be good to merge.
Thanks, Ian! I'll get this merged and upload a new release to pip soon. Closing.
I am attempting to run (largely with success) the MaNGA DAP on a cube from the WEAVE Spectrograph Large IFU. However, I am running into an exception when I try to use a covariance calibration to calculate the SN for the Voronoi binning. Specifically, in my configuration file I am setting the following in order to make use of the 1 + alpha * log(Nbin) type calibration
This raised the following warning
The stacked S/N for my cube is comfortably above my Voronoi target S/N, so I didn't think that everything should be in a single bin (indeed I have single pixels that are above my Vornoi target S/N).
I dug a little deeper and the specific error that is being raised by the
voronoi_2d_binning
function is the followingMy best guess is that this is occurring because
index
is just a single pixel in this case and thereforesignal[index]
has nolen()
. Is this the intended behavior? If not and this is a bug, I would assume that for the case of a single-pixel 'bin',len(signal[index])
could be replaced by1.0
?This is all just my best-guess interpretation, I would love to get your insight.
Thanks! Ian