Comparing Approximate Bayesian Computation with the Poisson-Likelihood Method for Exoplanet Occurrence Rates

TL;DR

This study compares approximate Bayesian computation and Poisson-likelihood methods for estimating exoplanet occurrence rates using Kepler data, finding consistent results that validate both approaches.

Contribution

First direct comparison of ABC and Poisson-likelihood methods for exoplanet occurrence rate estimation using the same dataset.

Findings

Occurrence rate F0 ≈ 0.596 with ABC

Corrected occurrence rate F0 ≈ 0.421 after reliability correction

Results agree within 1σ with previous Poisson-likelihood estimates

Abstract

We present Kepler exoplanet occurrence rates inferred with approximate Bayesian computation (ABC). By using the same planet catalogue, stellar sample, and characterization of completeness and reliability as Bryson et al. (2020), we are able to provide the first direct comparison of results from ABC to those derived with the popular Poisson-likelihood method. For planets with orbital periods between 50 and 400 days and radii between 0.75 and 2.5 $R_{\oplus}$, we find an integrated occurrence rate $F_{0} = 0.596_{-0.099}^{+0.092}$ planets per GK dwarf star. After correcting for reliability against astrophysical false positives and false alarms, we find $F_{0} = 0.421_{-0.072}^{+0.086}$. Our findings agree within 1$\sigma$ of Bryson et al. (2020), indicating that the results are robust and not method-dependent.