take the Lyman-limit into account

[londumas]

The Lyman-limit, lambdaRF=911.8 A, reveals itself for quasars at z>3.
However, the Lyman-break can be at rest-frame wavelength a bit lower than that.
This feature could be very useful to help measure the redshifts for these quasars, but for the moment, as far as I know, the Lyman-break is not in the desi simulation.
We know that the catastrophic failure rate starts to be high for z>3.5, maybe this would help.
We could simply set the flux at zero for lambdaRF<=911.8, but we need simulations to help us.
@sbailey, do we have simulations with quasars at z in [3,7], with the Lyman-break?

[moustakas]

We had this discussion early on when developing the desisim templates, and decided to not include IGM absorption, but I agree that we should, both in our data challenge spectra and in our spectral/redshift fitting.

[mifumagalli]

LLSs with column density log NHI > 17 cm2 can be incorporated in a probabilistic way following a similar syntax to what is currently done for BALs. Basically, for each quasar, we can compute the probability of finding a LLS along the line of sight using the measured number density per unit redshift. We can then draw a redshift and column density from observed distributions and construct a LLS with Lyman series and Lyman limit, which gets injected in the mock. This will automatically incorporate DLAs as well. One can then expand the model by allowing for more than one LLS in the same LOS (and it is indeed worth thinking about allowing at least a second system, which would be injected iterating on the above procedure, but with knowledge of the previous system). I have a bit of code that does some of this already, and it can be adapted for the desisim quasar template.

[mifumagalli]

Somewhat related to this, currently the quasar template seems to assume a mean attenuation at <912A in the quasar rest-frame due to the mean free path. It should be easy (1 line of code) to generalise that a bit more to allow for a distribution with a given scatter around the mean value, to reflect the actual diversity in real quasar spectra.