birth cloud dust attenuation applied beyond dust_tesc

[changhoonhahn]

I ran into the following issue regarding the birth cloud dust attenuation:

I'm running StellarPopulation with dust_tesc=7.0, which I think means that only stellar populations with age < 1e7 should be affected by birth cloud dust attenuation. However, I find that when birth cloud dust attenuation is applied to stellar populations with age 1.1e7.

Below is an illustration. I generate spectra for a stellar population with tage=1.1e7. In one of the spectra, I set dust1 = 0.9; in the other I set dust1=0. Since tage > dust_tesc, I didn't expect there to be a difference between the two spectra but the figure below shows a significant difference.

I find that for dust_tesc=7.0, birth cloud dust attenuation is applied up to tage=1.12e7yr

[bd-j]

Thanks for the detailed code. The units of dust_tesc are log(years) while the units of tage are Gyr. So the population you are obtaining from get_spectrum(tage=1.1e-7) is extremely young, try tage=1.1e-2 and see if the problem persists.

The docstring for get_mags and get_spectrum should be made explicit about the units of tage, I'll add that to updates I'm working on at the moment.

[changhoonhahn]

Oops, that was a typo when I was trying to reproduce the bug. Here's the same issue with tage=1.1e-2:

[bd-j]

This is because for requested ages between the native isochrone age gridpoints the output spectrum is interpolated from two SSPs. In your case one of the bracketing SSPs is at log(t)=7.0 and the second is at log(t)=7.05. The the first gets attenuated by dust1 and the second does not. This is actually an FSPS issue, can you reopen there?