astropy: astropy units for blackbody_lambda does not do proper si conversion

If I run the following code, it correctly converts the unit “bar” to SI:

from astropy import units as u

test = 1 * u.bar
print(test)
print(test.si)
# 1.0 bar
# 100000.0 N / m2

But for blackbody_lambda, it fails to convert the result to correct si unit:

from astropy import units as u
from astropy.modeling.blackbody import blackbody_lambda

testB = blackbody_lambda(1 * u.um, 100 * u.K)
print(testB)
print(testB.si)
print(testB.to('N/s/m^2/sr'))
# 3.8960173143757327e-56 erg / (Angstrom cm2 s sr)
# 3.896017314375732e-54 bar / (rad2 s)
# 3.896017314375732e-49 N / (m2 s sr)

That is, astropy does not convert bar to N/m^2 (which gives factor of 10^5 difference).

I tested with some other complicated combinations, e.g., test = 1 * u.cm * u.bar / u.sr * u.erg, but the SI conversion worked just fine. Maybe the failure is only for blackbody_lambda…?

About this issue

Original URL
State: closed
Created 5 years ago
Comments: 15 (14 by maintainers)

Most upvoted comments

I don’t think this is a bug in blackbody per se. The SI conversion maps J / m3 to Pa (and bar is 1e5 * Pa). When we’re talking about flux, it makes more sense to have testB.si return something like J / (m3 rad2 s), no? For now, you can force this manually:

>>> testB.to(u.J / (u.m ** 3 * u.s * u.rad ** 2))
<Quantity 3.89601731e-49 J / (m3 rad2 s)>

p.s. Astronomer calculating flux in SI units? That’s crazy talk!

pllim on Feb 6, 2019

Very much so. Though bar/rad2 just doesn’t make sense! If someone has the time, I’d do the swap of power and pressure in si.py and see if that helps…

mhvk on Mar 11, 2019

Hi, I am triaging and noticed that some level of closure might have been reached… Should this issue be closed for now and a new issue be opened should a similar problem arise?

I second @pllim’s observation: This does not look like a bug to me, but rather some kind of rigidity in the default units adopted in the .si units conversion mechanism. Maybe in the future some sort of flexibility could be introduced to the overarching scheme of unit conversion to suit the user’s needs better? But that would be a separate issue.

kakirastern on Mar 9, 2019

@MSeifert04 Oh, sorry. I am using version 3.1.1:

import astropy
print(astropy.__version__)
# 3.1.1

ysBach on Feb 6, 2019