Note
Previous documentation version, by Kaylan Burleigh and John Moustakas is to be found on the former Obiwan website 1.
User documentation
Developer documentation
Introduction¶
Obiwan is a Monte Carlo method for adding fake galaxies to Legacy Survey 2 images, and re-processing the modified images with legacypipe. The legacypipe documentation is available here: legacypipe docs 3.
What for?¶
Targets for spectroscopic follow-up are selected among the sources detected by legacypipe. The target density includes cosmological clustering signal (to be measured) but is also impacted by so-called “imaging systematics”, due to the telescope, the opacity of the atmosphere, extinction and dust of the Milky Way, bias and variance of legacypipe. These systematics can be (partly) removed by regressing the target density against photometric templates (linear model, or neural nets 4), but these methods can only remove dependence of the target density on known systematics.
Obiwan rather forward models the source detection and target selection processes, by injecting fake galaxies into raw images, running legacypipe and applying the target selection colour cuts.
Obiwan has been applied on eBOSS ELGs 5.
A picture is worth a 1000 words
Why the name Obiwan?¶
Just as Obi-Wan Kenobi was the only hope in Star Wars: Episode IV - A New Hope 6; Obiwan is one of the only hopes for removing (most of) photometric systematics in the sample of galaxies selected from the imaging data.
Quick start-up¶
For a quick start-up on NERSC, see Example on NERSC.
Acknowledgements¶
See the offical acknowledgements for the Legacy Survey.