How narrow is the M87* ring? I. The choice of closure likelihood function

TL;DR

This paper investigates how the choice of closure likelihood function affects the inferred width of the M87* ring, proposing a new approximation that suggests a thicker ring consistent with theoretical models.

Contribution

It introduces a new closure likelihood approximation and demonstrates its impact on ring width estimates from EHT data, highlighting the importance of analysis methods.

Findings

The new likelihood approximation favors thicker rings.

Current methods may underestimate ring width.

Further analysis is needed to determine the best approximation.

Abstract

Event Horizon Telescope (EHT) observations of the core of the galaxy M87 suggest an observational appearance dominated by a ring of approximately 40$\mu$as in diameter. The thickness of the ring is less certain: imaging efforts constrained it to be less than half the diameter (consistent with an imaging resolution of 20$\mu$as), while visibility-domain modeling suggested a variety of fractional widths, including as low as $10\%$ on some days. The fractional width is very interesting as it has the potential to discriminate between different astrophysical scenarios for the source; in fact, the $10$--$20\%$ range is so narrow as to be in tension with theoretical expectations. In the first of a series of papers on the width of the observed ring, we reproduce a subset of EHT visibility-domain modeling results and we explore whether alternative data analysis methods might favor thicker rings. We point out that the closure phase (and closure amplitude) likelihood function is not independent of residual station gain amplitudes, even at high signal-to-noise, and explore two approximations of practical interest: one standard in the field (and employed by the EHT collaboration), and a new one that we propose. Analyzing the public data, we find that the new likelihood approximation prefers somewhat thicker rings, more in line with theoretical expectations. Further analysis is needed, however, to determine which approximation is better for the EHT data.