astropy.modeling.models has no Planck function

[ibusko]

That seems to me a pretty basic model for spectra fitting. Should we formally post a request in astropy? Or develop our own?

[pllim]

There is one. I added it to astropy. It just is not in modeling because that function is highly depended on units and modeling cannot handle Quantity yet. See http://docs.astropy.org/en/stable/analytic_functions/index.html#using-astropy-analytic-functions

I hacked together a Blackbody model for pysynphot a while back but that project is on hold.

[ibusko]

So, should we add it to specviz's list? Or the units dependency may create new issues that we won't be able to deal before the release?

[pllim]

Look at https://github.com/spacetelescope/pysynphot_DONOTUSE/blob/master/synphot/models.py#L30 and tell me what you think...

[ibusko]

The blackbody_lambda and blackbody_nu models are in fact functions, as suggested by the example. Is that right? If so, in order to be usable in specviz, they should instead subclass Fittable1DModel In that case, they can be used as the core of a model class that we will have to develop ourselves.

[nmearl]

A basic custom one should be straightforward

from astropy.modeling.models import custom_model
from astropy.units import Quantity
from astropy.analytic_functions import blackbody_lambda

@custom_model
def blackbody(wavelength, temperature):
    wavelength = Quantity(wavelength, "Angstrom")
    temperature = Quantity(temperature, "K")
    return blackbody_lambda(wavelength, temperature).value

[pllim]

If you can make it work, go for it. The only thing that stopped me was lack of unit support in modeling machinery. And as I understood from astropy/astropy#4615, the fitting itself would not support units internally. So, whatever you put together now will break when that PR is merged.

[ibusko]

Implemented Blackbody model. Hence, this ticket can be closed.

[cheekujodhpur]

@ibusko Where is this model implemented? I do not see it in the latest code.

What am I doing wrong?