Non-stationary variability in accreting compact objects

Accreting compact objects show variations in source flux over a broad range of time-scales and in all wavebands. The light curves typically show a lognormal distribution of flux and a linear relation between flux and rms. It has been demonstrated that an exponential transform of an underlying (and unobserved) Gaussian stochastic process provides a very good description of the light curves with these observed properties. Recently, a non-stationary power spectrum was observed on fast time-scales (similar to days) in the active galaxy, IRAS 13224-3809, as well as a non-lognormal flux distribution and non-linear rms-flux relation. Here, we investigate the effects of piecewise non-stationary power spectra on the resultant flux distribution and rms-flux relation. We demonstrate that the simple 'exponentiation' model successfully reproduces the observed quantities, even when the light curves are non-stationary. We also demonstrate how non-lognormal flux distributions and rms-flux relations inconsistent with a linear model can be erroneously produced from poorly sampled PSDs. This is of particular importance for AGN surveys where very long baselines are required to sample the PSD down to low enough frequencies.