In 2009, the discovery of two planets orbiting the evolved binary star system HW Virginis (HW Vir) was announced, based on systematic variations in the timing of eclipses between the two stars. The planets invoked in that work were significantly more massive than Jupiter, and moved on orbits that were mutually crossing – an architecture which suggests that mutual encounters and strong gravitational interactions are almost guaranteed.
In this work, we perform a highly detailed analysis of the proposed HW Vir planetary system. First, we consider the dynamical stability of the system as proposed in the discovery work. Through a mapping process involving 91 125 individual simulations, we find that the system is so unstable that the planets proposed simply cannot exist, due to mean lifetimes of less than a thousand years across the whole parameter space.
We then present a detailed re-analysis of the observational data on HW Vir, deriving a new orbital solution that provides a very good fit to the observational data. Our new analysis yields a system with planets more widely spaced, and of lower mass, than that proposed in the discovery work, and yields a significantly greater (and more realistic) estimate of the uncertainty in the orbit of the outermost body. Despite this, a detailed dynamical analysis of this new solution similarly reveals that it also requires the planets to move on orbits that are simply not dynamically feasible.
Our results imply that some mechanism other than the influence of planetary companions must be the principal cause of the observed eclipse timing variations for HW Vir. If the system does host exoplanets, they must move on orbits differing greatly from those previously proposed. Our results illustrate the critical importance of performing dynamical analyses as a part of the discovery process for multiple-planet exoplanetary systems.