Enhanced spectral index expression

[sjperkins]

@LauraRichter Remember to provide link to the paper here.

[LauraRichter]

Here is a paper which uses this parameterization:

http://adsabs.harvard.edu/abs/2013ApJS..204...19P

^^ See Section 8, Equation 10, and Table 10 for the coefficient values for the set of calibrators they fit in that paper.

In katpoint there are an additional two optional parameters, and a reference for that in the code:

https://github.com/ska-sa/katpoint/blob/master/katpoint/flux.py

[o-smirnov]

Is this still not available? Bumping up...

[o-smirnov]

Another bump. Need this badly.

See MeqTrees implementation: https://github.com/ska-sa/meqtrees-cattery/commit/1961bf7e4f98832bde197e6f77feaf0071d430b6#diff-775eeeadde559783f5086161e301aab7

[sjperkins]

This still isn't feasible given the architectural inflexibility in the current master branch. I see two solutions:

1. Use the ddfacet branch which removes the single spectral index parameter, but provides the ability to set per-channel stokes values (in which your desired and fairly arbitrary spectral model should be integrated). It looks like this for a single spectral index parameter.
2. Wait for the dask development version.

[o-smirnov]

Option 1 works for me, but can we merge the ddfacet branch into master? Otherwise how's CubiCal's pip install going to play with a branch version of montblanc?

[sjperkins]

Option 1 works for me, but can we merge the ddfacet branch into master?

We could, since it seems as if everyone is now essentially using that version (DDFacet, Cubical, @chrisfinlay).

Otherwise how's CubiCal's pip install going to play with a branch version of montblanc?

Alternatively, this can be achieved through the use of dependency_links in setup.py

[sjperkins]

Alternatively, this can be achieved through the use of dependency_links in setup.py

In fact, this method is deprecated. PEP 508 is now available in the latest pip