bug in SFH basis use
changhoonhahn opened this issue · comments
There seems to be some inconsistencies in how the SFH bases are applied. Check the SFH calculations in the methods get_SFR
and _fsps_model
.
Also check that _emulator
and _fsps_model
generate the same SED for the same parameter values
Once I went back to v0.3.0 of python-fsps
, the emulator and fsps model are consistent. Hence, this issue seems to have been a red herring.
@kgb0255 Could you confirm this by running DESI_speculator_comparison.ipynb and checking that you reproduce the same results?
@changhoonhahn I don't see the exactly same spectra as yours at the end of the notebook (maybe because I ran the whole notebook multiple times), but it seems that the plotted spectra have good agreement. Attached the result below.
@kgb0255 Great. You have a different spectrum since you have a different random seed. What important is the agreement between the emulator and fsps.