Evolution of binary supermassive black holes and the final-parsec problem

TL;DR

This paper reviews the evolution of binary supermassive black holes, focusing on stellar-dynamical mechanisms that may resolve the final-parsec problem, and introduces a novel Monte Carlo simulation method to study these effects.

Contribution

It presents a new Monte Carlo approach to simulate both collisional and collisionless evolution of non-spherical stellar systems, applied to the binary black hole problem.

Findings

In triaxial galaxies, the final-parsec problem is largely resolved.

In axisymmetric galaxies, the problem persists in collisionless regimes but not in N-body simulations.

Simulation results depend on the number of particles, affecting the dominance of collisional effects.

Abstract

I review the evolution of binary supermassive black holes and focus on the stellar-dynamical mechanisms that may help to overcome the final-parsec problem - the possible stalling of the binary at a separation much larger than is required for an efficient gravitational wave emission. Recent N-body simulations have suggested that a departure from spherical symmetry in the nucleus of the galaxy may keep the rate of interaction of stars with the binary at a high enough level so that the binary continues to shrink rather rapidly. However, a major problem of all these simulations is that they do not probe the regime where collisionless effects are dominant - in other words, the number of particles in the simulation is still not sufficient to reach the asymptotic behaviour of the system. I present a novel Monte Carlo method for simulating both collisional and collisionless evolution of non-spherical stellar systems, and apply it for the problem of binary supermassive black hole evolution. I show that in triaxial galaxies the final-parsec problem is largely non-existent, while in the axisymmetric case it seems to still exist in the limit of purely collisionless regime relevant for real galaxies, but disappears in the N-body simulations where the feasible values of N are still too low to get rid of collisional effects.