Application of 'CLEAN' in the power spectral analysis of non-uniformly sampled pulsar timing data

Spectral analysis of the residual pulse-arrival times of pulsars is a useful tool in understanding the nature of the underlying processes that may be responsible for the timing noise observed from pulsars. Power spectra of pulsar timing residuals may be described by one or a combination of power-laws. As these spectra are expected to be very steep, it is important to ensure a high dynamic range in the estimation of the spectrum. This is difficult in practice since one is, in general, dealing with timing measurements made at unevenly placed epochs. In this paper, we present a technique based on 'CLEAN' to obtain high dynamic range spectra from unevenly sampled data. We compare the performance of this technique with other techniques including some that were used earlier for estimation of power spectra of pulsar timing residuals.