Adventures with planets and binaries in accretion discs

The primary idea behind the work in this thesis is that accretion discs interacting with astrophysical bodies, from planets to supermassive black holes (SMBHs), can strongly affect the dynamical behaviour of those bodies. While this idea is by no means new, observational and theoretical developments in recent years provide fresh motivation to consider this effect across a number of astrophysical contexts. Of particular relevance to the work here are three observational measurements which I attempt to reconcile with theory by invoking interactions with accretion discs. Firstly, observations of giant exoplanets show that they prefer to inhabit eccentric orbits, which is unexpected given the predictions of planet formation theory. Conversely, Kepler’s discovery of planets with low eccentricities around moderately eccentric binaries goes against theoretical expectation that their orbits should be eccentric. In galactic centres, binary supermassive black holes are not observed despite theoretical expectations that their evolution should drive them to ~ parsec separations and leave them there. In this thesis I use high-resolution smoothed particle hydrodynamics simulations to investigate each of these problems involving planet- and binary-disc interactions. I show that these interactions are unable to solve the problem of eccentric giant exoplanets, but that they can cause damping of circumbinary planetary eccentricity and so are able to explain Kepler’s circumbinary planets. I use this latter to place a limit on the surface density in which Kepler-16b in particular can have formed. I also show that a disc formed from a gas cloud moving prograde with respect to a SMBH binary will fragment to form stars sooner than a similar retrograde disc. Consequently, a retrograde disc is able to drive stronger binary evolution than is the prograde disc. Allowing that a large number of such encounters would be expected in the aftermath of the galaxy merger that formed the binary, this process may be able to solve the ‘last parsec problem.’