The coalescence history of massive black holes has been derived from cosmological simulations, in which the evolution of those objects and that of the host galaxies are followed in a consistent way. The present study indicates that supermassive black holes having masses greater than $\sim 10^{9} M_{\odot}$ underwent up to 500 merger events along their history. The derived coalescence rate per comoving volume and per mass interval permitted to obtain an estimate of the expected detection rate distribution of gravitational wave signals ("ring-down") along frequencies accessible by the planned interferometers either in space (LISA) or in the ground (Einstein). For LISA, in its original configuration, a total detection rate of about $15 yr^{-1}$ is predicted for events having a signal-to-noise ratio equal to 10, expected to occur mainly in the frequency range $4-9 mHz$. For the Einstein gravitational wave telescope, one event each 14 months down to one event each 4 years is expected with a signal-to-noise ratio of 5, occurring mainly in the frequency interval $10-20 Hz$. The detection of these gravitational signals and their distribution in frequency would be in the future an important tool able to discriminate among different scenarios explaining the origin of supermassive black holes.

The correlations between the mass of supermassive black holes and properties of their host galaxies are investigated through cosmological simulations. Black holes grow from seeds of 100 solar masses inserted into density peaks present in the redshift range 12-15. Seeds grow essentially by accreting matter from a nuclear disk and also by coalescences resulting from merger episodes. At z=0, our simulations reproduce the black hole mass function and the correlations of the black hole mass both with stellar velocity dispersion and host dark halo mass. Moreover, the evolution of the black hole mass density derived from the present simulations agrees with that derived from the bolometric luminosity function of quasars, indicating that the average accretion history of seeds is adequately reproduced . However, our simulations are unable to form black holes with masses above $10^9 M_{\odot}$ at $z\sim 6$, whose existence is inferred from the bright quasars detected by the Sloan survey in this redshift range.

Capture rates of compact objects were calculated by using a recent solution of the Fokker-Planck equation in energy-space, including two-body resonant effects. The fraction of compact objects (white dwarfs, neutron stars and stellar black holes) was estimated as a function of the luminosity of the galaxy from a new grid of evolutionary models. Stellar mass densities at the influence radius of central supermassive black holes were derived from brightness profiles obtained by Hubble Space Telescope observations. The present study indicates that the capture rates scale as $\propto M_{bh}^{-1.048}$, consequence of the fact that dwarf galaxies have denser central regions than luminous objects. If the mass distribution of supermassive black holes has a lower cutoff at $\sim 1.4\times 10^6$ M$_{\odot}$ (corresponding to the lowest observed supermassive black hole mass, located in M32), then 9 inspiral events are expected to be seen by LISA (7-8 corresponding to white dwarf captures and 1-2 to neutron star and stellar black hole captures) after one year of operation. However, if the mass distribution extends down to $\sim 2\times 10^5$ M$_{\odot}$, then the total number of expected events increases up to 579 (corresponding to $\sim$ 274 stellar black hole captures, $\sim$ 194 neutron star captures and $\sim$ 111 white dwarf captures).