Since we are (hopefully) on the home stretch for this project, I am looking at the different methods of extracting a 1D profile from our 2D convolved image, and how those methods effect our final output parameters.
As I showed to Ivana yesterday, we are finding that using certain methods appear to return satisfactory values (larger Sersic indices for the SDSS galaxies), but the actual fit of the data isn’t very accurate. We have also seen the opposite when taking a slice through the 2D image to retrieve a profile, where the fit appears very good, but the values seem far too small. We can see these effects by observing the fits directly, and comparing the catalogue and computed values.
One interesting idea that Ivana suggested yesterday was for me to fit the observed galaxies with the radial profile method, and observe their differences. Below are a few results of that test.
Here’s the blue profile is the profile we calculated using ellipse(), and the orange profile is calculated using a simple radial profile method I found.
It is evident from these figures that the radial profile returns a much different profile than Ellipse(), as expected. This makes sense given that even a slight ellipticity will cause the profile to “fall off” faster when calculating a radial profile.
If we want the best fits to our data, we will have to use Ellipse() to find them. Unfortunately this is incredibly time intensive, so I think moving it to Acenet will be a necessity. 100s of iterations will be required to fit each galaxy, so if we are ready to move to this final step, I will need to ensure things are as fast as possible! As well as attempt to fit a few myself just to get a ballpark estimate of the time required for a single galaxy. I have a few ideas that I would like to discuss, because I think we can use the information we already have (the tables for each galaxy calculated last summer) to our advantage!
HSC Research 