The detection of gravitational waves has initiated a completely new way to study our Universe. With the future space-born gravitational wave experiments, such as the Laser Interferometer Space Antenna (LISA), low-frequency gravitational waves will allow to probe fundamental aspects of our understanding of the cosmos, from the nature of gravity to the rate of cosmological expansion, from early Universe conditions to the pace at which galaxies are assembled through hierarchical mergers. Mergers of intermediate mass and massive black holes, in the range 1000-10^7 Msun, will be the primary sources that LISA will target. Yet the processes by which such black holes form, pair into binaries, and ultimately sink at the center of colliding galaxies all the way to their final coalescence, are not well understood in a quantitive way. Using the results of supercomputer simulations I will discuss our current understanding of such processes and of their associated timescales. Merger timescales, in particular, are crucial in order to build credible forecasts for LISA event rates as well as to interpret the LISA datastream in the context of hierachical structure formation. I will show how different regimes of black hole binary evolution exist depending on the nature of their host galaxies. Intriguingly, for black holes in the range 10^4-10^5 Msun, which live in dwarf galaxies, the physical nature of dark matter also plays a role in the dynamics of black hole binaries, adding a novel unsuspected connection between LISA sources and fundamental physics.
Followed by wine and cheese in Pupin 1402.