Closed ameliesaintonge closed 1 year ago
This "feature" is due to an intentional fitting prior in FastSpecFit which removes SPS models which are older than the age of the universe at the redshift of each galaxy (+a pad of 500 Myr). Although physically motivated, this choice clearly introduces an unwanted redshift-dependence.
I've rerun fastphot
using the https://github.com/desihub/fastspecfit/tree/no-age-prior branch where this prior is removed and things look much better:
And here's an updated comparison with an independent stellar mass catalog generated for the same set of BGS targets:
Fixed in #132. Some additional analysis figures can be reviewed in this notebook-- https://github.com/moustakas/fastspecfit-projects/blob/main/mstarchallenge/mstarchallenge-bug.ipynb
I just noticed a couple of "features" in the stellar mass-redshift distribution, based on the fastspec output for Iron. There is a clear break at z~0.27, and narrow plateau at z>0.45, both of which almost certainly are not physical:
This figure is made using the Z and LOGMSTAR columns of this file: '/global/cfs/cdirs/desi/spectro/fastspecfit/iron/v1.0/catalogs/fastspec-iron.fits'