Studying gravitational waves with pulsars : results from the Parkes pulsar timing array

This thesis addresses the problem of gravitational-wave (GW) detection using radio timing observations of pulsars. We study GW signals in real and simulated pulsar timing observations, and describe the astrophysical implications for cases in which no GW signal is detected. We simulate timing observations from a range of hypothetical pulsar timing array projects. The pulse arrival times are then perturbed by an individual source of GWs. One of the simulated data sets comprises an array of 20 pulsars timed with a root-mean-square residual of 10 ns over 10 years. If there is no detectable GW signal in this data set, then the merger rate of supermassive black-hole binaries (SMBHBs) with a chirp mass of 109 solar masses is less than one merger every 105 years up to a redshift of z = 2. This constraint rules out estimates of the SMBHB merger rate based on hierarchical galaxy formation with standard assumptions on the merger parameters. Applying a similar analysis to recently published observations from the Parkes Pulsar Timing Array (PPTA) yields a constraint on the merger rate of SMBHBS of less than one merger every five years for SMBHBs with chirp mass ~ 1010 solar masses up to a redshift of z = 2. The results also indicate that it is unlikely that an individual GW source could be detected with existing data sets. We consider the signal caused by an isotropic stochastic gravitational-wave background (GWB), and show that, with a few more years of observations, either the GWB will be detected or the parameter space of most current galaxy evolution models will be significantly constrained. An analysis of the cross-correlation between the timing residuals of different pulsars in the PPTA shows that there is no detectable GWB signal in the current data. The GWB detection statistic is dominated by only a few pulsars in the current PPTA data. There are good prospects for detection of the GWB using radio timing of pulsars in the next decade. We conclude that the effect of instabilities in realisations of the terrestrial timescale and inaccuracies in the solar system ephemeris must be removed from the timing residuals in order to detect a GW signal in pulsar timing observations in the future. The forthcoming International Pulsar Timing Array project will also significantly increase the probability of detection of GW signals using pulsar timing