Consider whether exoplanet should use DensityDist rather than Potential about exoplanet

[dfm]

@bmorris3: There shouldn't be any issue with that syntax, but a simpler one might be:

pm.Normal("spitzer_depth_prior", mu=spitzer_depth, sigma=300, observed=0.0)

Which should work equivalently and I find somewhat more readable. (You can think of this "prior" as, instead, "data" from a previous experiment.)

[dfm]

I think this would be cool, but it would probably be better to separate y from mean + light_curve since that is actually the mean. I've been working on a better GP interface, but who knows how long it will take so I'd love a pull request to dfm/exoplanet-docs that changes the notation.

[mrtommyb]

Ok. I will try to work on it.

[dfm]

@mrtommyb: I implemented this on the flight to AAS. It's on the new_gp branch if you want to take a look: https://github.com/dfm/exoplanet/blob/new_gp/src/exoplanet/gp/celerite.py#L117

The syntax is now something like:

with pm.Model():
    kernel = xo.gp.terms....
    gp = xo.gp.GP(kernel, x, yerr**2, ...)
    gp.marginal("likelihood", observed=y)

[bmorris3]

Hi @dfm,

I'm working on a model which uses exoplanet to compute a transit+phase curve+eclipse model. I have a derived parameter (an eclipse depth) produced in my phase curve model which is wrapped in pm.Deterministic, and I would like to enforce a prior on this variable (eclipse depth), in order to be consistent with a value in the literature. I was thinking the easiest way to do this would be to run something like:

        pm.Potential('spitzer_depth_potential', pm.Normal.dist(0, 300).logp(spitzer_depth))

where spitzer_depth is the Deterministic, and its measured value is 0 ± 300 [ppm]. Will this work if I'm running gp.marginal later (like you eventually settled on) to create define the likelihood? Or is this a bad idea?

[bmorris3]

Thanks for this, that's definitely more readable!