Testing supermassive black hole quasi-periodic oscillation frequency scaling with latent-mass Bayesian inference
Annotatsiya
Context. High-frequency quasi-periodic oscillations (QPOs) in supermassive black-hole systems provide a potential probe of the frequency scale in the innermost accretion flow. The current supermassive black hole (SMBH) QPO catalogue is, however, small and heterogeneous, and it is not clear whether departures from the standard inverse-mass scaling can be identified robustly. Aims. We tested whether simple low-parameter corrections to the Schwarzschild epicyclic-resonance baseline improve the predictive description of the observed SMBH QPO frequencies. The goal was not to identify a unique physical mechanism but to quantify what the current data can support beyond the inverse-mass relation. Methods. We used a latent-mass errors-in-variables framework that explicitly propagated published black-hole mass uncertainties. We interpreted the reported QPO frequency as the upper branch of a common 3:2 epicyclic-resonance prescription. We used the same robust Student- t residual model for all tested frequency prescriptions and compared them using the widely applicable information criterion (WAIC) computed from the integrated predictive likelihood. Results. The current sample supports a heavy-tailed Student- t description of the observed log-frequency residuals conditional on latent mass as a stable predictive baseline, while the mass-measurement layer remains Gaussian. With this residual model, the robust inverse-mass baseline provides an adequate predictive description of the working sample. A free-slope control model provides only a marginal improvement. The wormhole or regular-geometry, magnetic-field, and dark-matter-shell benchmark prescriptions yield weak nominal predictive gains, with ΔWAIC ≃ −2.4 to −2.5, but none is decisively preferred. Pointwise diagnostics show that the gain is not uniform across the catalogue but is concentrated in a small number of high-leverage sources. Conclusions. The present SMBH QPO data support, at most, mild structured departures from the inverse-mass baseline. They do not strongly favour any specific geometric or environmental correction. Their main value is to define a conservative predictive benchmark: future SMBH QPO models should improve on the robust Student- t latent-mass baseline without relying on a small subset of sources. Within the benchmark models tested here, future QPO detections towards the high-mass end of the SMBH range offer the greatest discriminating leverage because the correction terms separate most clearly there.
Hali tarjima qilinmagan