Astropy updated constants

[pmaxted]

I have updated the the solar and planetary constants in constants/si.py to be consistent with 2015 IAU resolution B3. The constants are now "nominal values" or "conversion factors".

To convert the mass parameters like "GM_Sun" to actual masses like M_Sun, a value of G must be chosen. The user would expect that this would be the CODATA 2014 value mentioned in Resolution B3, so I have updated that value also. For consistency, I updated all the fundamental constants to CODATA 2014 values.

I have changed some of the variable names so that it is clear that they are polar/equatorial radii (e.g., R_earth -> Re_Earth or Rp_Earth). I have also changed some other variable names like R_jup to R_Jup. This will may require some users to change their code, but this should also mean that they can check that these new constants are really the numbers that they want.

The following files needed changing to update to the new constant names: units/astrophys.py, units/cds.py, coordinates/tests/accuracy/test_ecliptic.py, analytic_functions/tests/test_blackbody.py

I updated an example in units/core.py docstring to be consistent with the new output from u.m.find_equivalent_units()

The following tests required updated "hard-coded" test values to be consistent with CODATA 2014:

test_critical_density (test_cosmology.py)
9.31000324385361e-30 -> 9.30966845602090e-30
[2.70362491e-29, 5.53758986e-28] -> [2.70352772e-29, 5.53739080e-28]

(and delete the repetition of this last line)

test_doppler_energy_0 (test_equivalencies.py)
q1 = 0.000434286445543 * u.eV -> q1 = 0.000434286461222 *u.eV

test_brightness_temperature
tb = 7.05258885885 * u.K -> tb = 7.05259028913 * u.K

[astrofrog]

Thanks for the PR!

There are two other similar PRs: #4397 and #4109, and we need to make a decision on how to deal with versioning of constants.

[bsipocz]

A big 👎 for this name change from me.

• They need to be at least deprecated first before removal.
• doing such an API change just to change the case is totally unneccessary, and leaves inconsistency in with the rest of the code.
• I'm pretty sure most of us working with planets want to keep using R_earth and R_jup as is, without the extra complications of whether it is polar or equatorial.

[dhomeier]

Strongly second the concerns about the name change, except for the 3rd point:
the existing R_earth, R_jup already refer to the equatorial radii. One could therefore add the polar values and link the old name to the equatorial one:
R_earth = Re_earth etc.
Although it could also be argued for defining a volume-equal mean radius
R_earth = (Re_earth**2 * Rp_earth)**(1./3)
but that would again create inconsistencies with many people's current usage.

[bsipocz]

@astrofrog - another related issue that started the discussion: #3843

[mhvk]

I just created a new issue for the version control, which I think we indeed need to address first, as it is holding up so many PRs now. See #4958.

[pllim]

I see that the tag is 1.3 but given the ongoing discussions and Friday's feature freeze, it seems unrealistic.

[mhvk]

@pllim - agreed - I've removed it.

[pllim]

This does not appear to be the converging solution discussed in #4958, and no to name change.