Add sky spectrum residuals to target spectra in quickbrick about desisim HOT 3 OPEN

Thanks for posting this, @julienguy. Just planting a seed: I've had great success implementing (with Vivienne's help) a second-pass of sky-subtraction using PCA to mitigate these kinds of issues (see http://adsabs.harvard.edu/abs/2010arXiv1010.2500W) in AGES (another fiber redshift survey using the MMT/Hectospec instrument). Obviously here we're trying to introduce these effects for better realism, but we should also keep this in mind for when we start taking data.

from desisim.

Should the random sky leakage fraction be the same in each camera for a given fiber?

from desisim.

In reality, it won't be random, following mostly a radial pattern so that it will be a similar radial variation for each "petal" of the focal plane and hence similar for each camera.

Images showing this effect (in ideal conditions, and it won't be the case)

• light on screen
  
• portion of screen seen by a central fiber
  
• portion of screen seen by a fiber at the edge of fov (1.6 deg)
  
• non uniformity pattern (relative to sky)
  

from desisim.