ParameterOutOfBounds: crash for some stellar parameter values

[iancrossfield]

When attempting a PandExo (v3) transit simulation and using the following 'star' input:

{'type': 'phoenix', 'starpath': 'filename_star', 'w_unit': 'Angs,um,nm,Hz,sec,cm', 'f_unit': 'FLAM or W/m2/um or erg/s/cm2/Hz or Jy', 'mag': 7.009, 'hmag': 'H Band Magnitude', 'jmag': 'J Band Magnitude', 'ref_wave': 2.2, 'temp': 4780.0, 'metal': 0.31, 'logg': 4.66, 'radius': 0.68, 'r_unit': 'R_sun'}
The code crashes with
ParameterOutOfBounds: Parameter '[4800.0, 0.5, 5.0, 'phoenixp05/phoenixp05_4800.fits[g50]']' has no valid data.

This seems to be because the file 'g50' field of file grp/hst/cdbs/grid/phoenix/phoenixp05/phoenixp05_4800.fits contains only zeros!

Not really a problem with PandExo per se -- just that slightly more rigorous error-catching could be useful to handle this (and any similar) odd edge-cases.

[iancrossfield]

For what it's worth, I got around this by hacking engine/jwst.py to read in the star_spec as:

try:
   star_spec = create.outTrans(pandexo_input['star'])
except:
   pandexo_input['star']['metal'] = min(0.3, pandexo_input['star']['metal'])
   star_spec = create.outTrans(pandexo_input['star'])

[natashabatalha]

Ugh.. These Phoenix files are full of issues.. Will implement a check for this moving to pandexo 4.

[MartianColonist]

I just always set the metallicity to 0.0 for PandExo runs using the PHOENIX models. This is because the solar metallicity models have the highest spectral resolution and interpolation reliability with pysynphot, while some of the non-solar grid locations have issues... They actually replaced the solar metallicity PHOENIX models to fix some of these issues shortly before pysynphot was discontinued.