search

desihub / specsim

Quick simulations of spectrograph response
2 stars 9 forks source link

validate fiberloss fractions against imaging data #97

Open moustakas opened 6 years ago

moustakas commented 6 years ago

Looking over #96 (and related code), I don't see where BGS-like targets are handled. Also, it looks like the fiberloss table is computed for just three "pure" astrophysical sources -- point source, EXP, and DEV.

We should do a more careful assessment of these assumptions -- and the resulting fiberloss fractions -- using the (DR7) imaging data.

Digging around some more, perhaps this issue instead belongs in desisim? https://github.com/desihub/desisim/blob/master/py/desisim/simexp.py#L470-L548

dkirkby commented 6 years ago

The specsim fiberloss machinery can handle a fairly general mixture of EXP + DEV, each with their own position angle, half-light radius and ellipticity. You can also mix in a point-like source for a AGN-like profile.

However, I believe desisim treats all ELGs as 0.45" round EXP and all LRGs as 1.0" round DEV, with no redshift evolution of the size, which could definitely be improved (or perhaps already has been?)

In the specsim notebooks, there are examples with more realistic size & shape distributions.