We review how the recent increase in X-ray and radio data from black hole and neutron star binaries can be merged together with theoretical advances to give a coherent picture of the physics of the accretion flow in strong gravity. Both long term X-ray light curves, X-ray spectra, the rapid X-ray variability and the radio jet behaviour are consistent with a model where a standard outer accretion disc is truncated at low luminosities, being replaced by a hot, inner flow which also acts as the launching site of the jet. Decreasing the disc truncation radius leads to softer spectra, as well as higher frequencies (including quasi periodic oscillations, QPOs) in the power spectra, and a faster jet. The collapse of the hot flow when the disc reaches the last stable orbit triggers the dramatic decrease in radio flux, as well as giving a qualitative (and often quantitative) explanation for the major hard-soft spectral transition seen in black holes. The neutron stars are also consistent with the same models, but with an additional component due to their surface, giving implicit evidence for the event horizon in black holes. We review claims of observational data which conflict with this picture, but show that these can also be consistent with the truncated disc model. We also review suggested alternative models for the accretion flow which do not involve a truncated disc. The most successful of these converge on a similar geometry, where there is a transition at some radius larger than the last stable orbit between a standard disc and an inner, jet dominated region, with the X-ray source associated with a mildly relativistic outflow, beamed away from the disc. However, the observed uniformity of properties between black holes at different inclinations suggests that even weak beaming of the X-ray emission may be constrained by the data. After collapse of the hot inner flow, the spectrum in black hole systems can be dominated by the disc emission. Its behaviour is consistent with the existence of a last stable orbit, and such data can be used to estimate the black hole spin. By contrast, these systems can also show very different spectra at these high luminosities, in which the disc spectrum (and probably structure) is strongly distorted by Comptonization. The structure of the accretion flow becomes increasingly uncertain as the luminosity approaches (and exceeds) the Eddington luminosity, though there is growing evidence that winds may play an important role. We stress that these high Eddington fraction flows are key to understanding many disparate and currently very active fields such as ULX, Narrow Line Seyfert 1's, and the growth of the first black holes in the Early Universe.
ASJC Scopus subject areas