Description

This pull request (which is a stripped down version of #12304 that only includes minimal bug fixes and not deprecating support for units in scale or introduction of support for flux units) fixes the case where dimensionless (but not unscaled) units are passed to scale (#11547) and also changes the expected input_units based on the units passed to scale (which control the returned units). This second change is necessary in order to handle fitting BlackBody(..) + Linear1D(...) with respect to wavelength, as BlackBody was previously forcing the unitless array to Hz which then failed unit consistency checks with Linear1D.

Note that this does not address the second example in 11547 and that case remains ambiguous as scale is both providing a unitless scale factor and optionally responsible for determining the returned units. See #12304 for a possible solution to this by requiring scale to be dimensionless and handling the request for output units as a separate argument.

Revisiting the first case in #11547:

from astropy.modeling.models import BlackBody
from astropy import units as u
import numpy as np

T = 3000 * u.K
r = 1e14 * u.cm
DL = 100 * u.Mpc
scale = np.pi * (r / DL)**2

print(BlackBody(temperature=T, scale=scale).bolometric_flux)
print(BlackBody(temperature=T, scale=scale.to_value(u.dimensionless_unscaled)).bolometric_flux)

now returns the expected results (the passed scale with units of cm ** 2 / Mpc **2 are converted to dimensionless_unscaled before being applied):

4.823870774433646e-16 erg / (cm2 s)
4.823870774433646e-16 erg / (cm2 s)

The following case demonstrates the new treatment of input_units:

import astropy.units as u
from astropy.modeling import models
from astropy.modeling.fitting import LevMarLSQFitter
import numpy as np

SLAM = u.erg / (u.cm ** 2 * u.s * u.AA * u.sr)
SNU = u.erg / (u.cm ** 2 * u.s * u.Hz * u.sr)

fitter = LevMarLSQFitter()

x_lam = np.linspace(1000, 10000, 1000) * u.AA
x_nu = x_lam.to(u.Hz, u.spectral())

y_lam = np.random.random(x_lam.shape) * SLAM
y_nu = np.random.random(x_nu.shape) * SNU

b_nu = models.BlackBody(3000*u.K, scale=1*SNU)
l_nu = models.Linear1D(intercept=1*SNU, slope=0*SNU/u.Hz)
c_nu = b_nu + l_nu

print(b_nu.input_units, l_nu.input_units, c_nu.input_units)

fitter(c_nu, x_nu, y_nu, maxiter=1000)

which shows all models have expected Hz input units and the fitter runs successfully (as previously). However, the wavelength case:

b_lam = models.BlackBody(3000*u.K, scale=1*SLAM)
l_lam = models.Linear1D(intercept=1*SLAM, slope=0*SLAM/u.AA)
c_lam = b_lam + l_lam

print(b_lam.input_units, l_lam.input_units, c_lam.input_units)

fitter(c_lam, x_lam, y_lam, maxiter=1000)

used to have x input_units of Hz for the BlackBody model (but Angstrom for the other models), causing the following error during fitting: UnitConversionError: 'erg / (Angstrom2 cm2 s sr)' and 'erg / (Angstrom cm2 Hz s sr)' are not convertible, but now has expected input units for Angstrom for all models and fitting succeeds without errors.

Fixes #11547

Checklist for package maintainer(s)

This checklist is meant to remind the package maintainer(s) who will review this pull request of some common things to look for. This list is not exhaustive.

• Do the proposed changes actually accomplish desired goals?
• Do the proposed changes follow the Astropy coding guidelines?
• Are tests added/updated as required? If so, do they follow the Astropy testing guidelines?
• Are docs added/updated as required? If so, do they follow the Astropy documentation guidelines?
• Is rebase and/or squash necessary? If so, please provide the author with appropriate instructions. Also see "When to rebase and squash commits".
• Did the CI pass? If no, are the failures related? If you need to run daily and weekly cron jobs as part of the PR, please apply the Extra CI label.
• Is a change log needed? If yes, did the change log check pass? If no, add the no-changelog-entry-needed label. If this is a manual backport, use the skip-changelog-checks label unless special changelog handling is necessary.
• Is a milestone set? Milestone must be set but astropy-bot check might be missing; do not let the green checkmark fool you.
• At the time of adding the milestone, if the milestone set requires a backport to release branch(es), apply the appropriate backport-X.Y.x label(s) before merge.