Reliable inference of exoplanet light-curve parameters using deterministic and stochastic systematics models

摘要: Time-series photometry and spectroscopy of transiting exoplanets allow us to study their atmospheres. Unfortunately, the required precision extract atmospheric information surpasses design specifications most general purpose instrumentation, resulting in instrumental systematics light curves that are typically larger than target precision. Systematics must therefore be modelled, leaving inference curve parameters conditioned on subjective choice models model selection criteria. This paper aims test reliability commonly used As we primarily interested recovering rather favoured model, marginalisation over is introduced as a more robust alternative simple selection. can incorporate uncertainties into error budget well parameters. Its use demonstrated using series simulated transit curves. Stochastic models, specifically Gaussian processes, also discussed context found reliably recover for wide range functions. None tested criteria - including BIC routinely recovered correct model. means methods based may underestimate when extracting transmission eclipse spectra from real data, low-significance claims such techniques should treated with caution. In general, no modelling perfect; however, many helps mitigate poor selection, stochastic processes provide an even flexible approach systematics.