`sigmalos` errors with potential as argument

[cgarling]

🐛 Bug

galpy.df.jeans.sigmalos errors when provided with a potential from galpy.potential while galpy.df.jeans.sigmar does not.

Reproducible example

import astropy.units as u
from galpy.potential import PlummerPotential
from galpy.df.jeans import sigmar, sigmalos
prof = PlummerPotential(1693*u.solMass, b=100*u.pc)
print(sigmar(prof, 1.0*u.pc, beta=0.0)) # 0.1030855301393923 km/s
print(sigmalos(prof, 1.0*u.pc, beta=0.0)) # errors
print(sigmalos(prof, 1.0, beta=0.0)) # also errors without unit

Expected behavior

Docstrings indicate that galpy.df.jeans.sigmalos should be able to take the same arguments as sigmar but it errors instead.

System Details

Ran on the galpy REPL at https://www.galpy.org/repl
Version info is

Emscripten-3.1.14-wasm32-32bit
Python 3.10.2 (main, Aug 29 2022, 18:09:32) [Clang 15.0.0 (https://github.com/llvm/llvm-project 7effcbda49ba32991b8955821b8f
numpy 1.22.4
scipy 1.8.1
matplotlib 3.5.2
galpy 1.10.1.dev0
astropy 5.1

Additional context

Stacktrace seems to point to some unit-related code but sigmalos also errors when R is a plain number.

[jobovy]

Thanks for reporting this! Yes, there's a minor bug in the jeans.sigmalos code where it isn't consistently using internal units. For now, you can do

prof.turn_physical_off()
print(sigmalos(prof, 1.0*u.pc, beta=0.0,use_physical=True,ro=8.,vo=220.))

and you'll get output in km/s as you would if the code worked properly. For some reason this doesn't seem to work on the REPL, but I think it should work when you run the code locally. I think this doesn't work in the REPL because of an issue with the scipy version that is used in the browser (which has known issues with numerical integrals that are irrelevant to the regular Python version of scipy).