FM8.5.04 — Validation of transting planet candidates: a Bayesian view

Date & Time

Aug 5th at 11:45 AM until 12:00 PM

Track

Presentations 

Location

Rating ( votes)

Author(s): Rodrigo Fernando Díaz1, Jose Manuel Almenara3, Alexandre Santerne2

Institution(s): 1. Geneva Observatory, 2. Instituto de Astrofísica e Ciências do Espaço, 3. IPAG

Transiting candidate validation is essentially a Bayesian model comparison problem: different models, all explaining the observations comparably well, compete for the support of the available data. It has, however, two particularities that render it very complex and difficult to tackle: i) the relevant data sets are of diverse nature (transit light curves, broad band photometry, high angular resolution images, radial velocity observations, etc.), and ii) the models representing each hypothesis are highly non-linear and in some cases make the computation of the likelihood very time consuming.
Despite its clear Bayesian nature, the planet validation problem has received in the past mainly a frequentist treatment (BLENDER). Other techniques exist, but they employ unrealistic models that increase speed but only partially exploit the available datasets (ValFast).
The Planet Analysis and Small Transit Investigation Software (PASTIS) was developped keeping these issues and the characteristics of the problem in mind. It aims at computing the Bayesian evidence for a full set of false positive scenarios and the planet hypothesis, modelling in all cases the available data self-consistently, thus producing robust and rigorous Bayes factor for all models of interest. Its object-oriented architecture also permits constructing a vast set of false positive models easily.
I will review some key results of the planet validation technique, showing the limitations and dangers of some approaches and of the validation technique in general. I will also describe the PASTIS tool and present out results on CoRoT-22 b, Kepler-22 b, and other transiting candidates.