Comparison of spectral models for disc truncation in the hard state of GX 339-4

We probe models of disc truncation in the hard spectral state of an outburst of the well-known X-ray transient GX 339--4. We test a large number of different models of disc reflection and its relativistic broadening, using two independent sets of codes, and apply it to a \textit{Rossi X-ray Timing Explorer} spectrum in the rising part of the hard state of the 2010/11 outburst. We find our results to be significantly model-dependent. While all of the models tested show best-fits consistent with truncation, some models allow the disc to extend close to the innermost stable circular orbit (ISCO) and some require substantial disc truncation. The different models yield a wide range in best-fit values for the disc inclination. Our statistically best model has a physical thermal Comptonization primary continuum, requires the disc to be truncated at a radius larger than or equal to about two ISCO radii for the maximum dimensionless spin of 0.998, and predicts a disc inclination in agreement with that of the binary. Our preferred models have moderate Fe abundance, $\gtrsim$2 times solar. We have also tested the effect of increasing the density of the reflecting medium. We find it leads to an increase in the truncation radius, but also to an increase in the Fe abundance, opposite to a previous finding.