Hacking exoplanet to fit two disparate transits without "islands" in the period PDF... about exoplanet

[dfm]

This is an interesting question!

I'd probably consider implementing it using a version of your last proposal, but analytically marginalizing over n. I'd probably fit for a parameter p0 then compute the full model for each value of period = p0 / n for n = .... Then you can just sum over the likelihoods (not the log likelihoods!) to compute the marginalized likelihood.

It might be better to move this conversation to email and we can try to work it out there.

[dfm]

Here's a demonstration of this method in practice: https://gist.github.com/dfm/122a184be253ead0466c424cf3932b5d