The emission line is in fact on the red side of the blue camera, but the code returns the red camera, and so it misses the line completely.
This bug probably affects a pretty significant number of objects, but those objects can be identified by looking for very large differences between MODELAMP and AMP.
@jdbuhler has found that the code used to estimate the (observed, not model) amplitude of certain emission lines can be grossly incorrect.
One example is
HBETA_AMP
onsv1/bright/5060/39632929894174164
-- https://fastspecfit.desi.lbl.gov/target/sv1-bright-5060-39632929894174164The
main
branch of the code gives ~1e-5, whereas the correct answer is ~2.3 (slightly smaller than the model amplitude, ~2.6, as expected).The bug is here, where I compute which camera the line falls in-- https://github.com/desihub/fastspecfit/blob/main/py/fastspecfit/emlines.py#L1029
The emission line is in fact on the red side of the blue camera, but the code returns the red camera, and so it misses the line completely.
This bug probably affects a pretty significant number of objects, but those objects can be identified by looking for very large differences between
MODELAMP
andAMP
.