What can a uniform reflection pipeline establish about supermassive black hole spin? An audit and a corrected analysis of 57 AGN with XMM and NuSTAR

Abstract

We audit an automated relativistic-reflection pipeline that fitted one XMM-Newton EPIC-pn and NuSTAR epoch for each of 57 AGN with relxillCp+xillverCp, and repeat the analysis with its failures corrected. The pipeline's apparent constraints were not controlled: Small Window pn backgrounds were never subtracted, the nearest-in-time pairing selected mismatched or nearly empty spectra, single-start fits often stopped short of the minimum, and most fits were inadequate or at a parameter limit. The corrected analysis uses in-window backgrounds, 41 sources with eligible pn--NuSTAR pairs, multi-start fits, bidirectional profiles, criteria fixed in advance and eight model variations. 24 of the 41 reference fits have $p\ge0.01$ and 18 pass all quality checks. Their 90% spin sets comprise 1 two-sided interval (NGC 4051, $0.46\le a_*\le0.93$), 6 lower bounds, 4 upper bounds, 4 disconnected sets and 3 unconstrained profiles. For 10 of the 18 the best fit lies at $i\ge75^\circ$, implausible for type-1 AGN, and most of those constraints disappear when $i\le60^\circ$ is imposed. Only 2 of the 15 constrained sets keep their type both under that limit and with pn restricted to 3--10 keV (an upper bound for Mrk 1040 and a weak lower bound for 1H 0419-577). Simulations show that, with a correct model, the 90% sets contain the true spin in 82--92% of cases without drifting edge-on, but at high inclination one low-spin simulation in six yields a false lower bound. We make no statement about the spin distribution or a spin--jet relation.