Cause of peak-time clustering among Subaru HSC M31 microlensing candidates

Determine why the peak times of the purported microlensing event candidates reported by Sugiyama et al. (2026) in the Subaru Hyper Suprime-Cam high-cadence Andromeda Galaxy (M31) dataset are clustered within short intervals during the nights 2014-11-24 and 2017-09-20, and ascertain whether this clustering is an artifact introduced by the event selection cuts used in that analysis or arises from other systematic effects.

Background

Given the cadence and sampling of the Subaru HSC observations, peak times of genuine microlensing events should be uniformly distributed over time. However, in the dataset analyzed by Sugiyama et al. (2026), all five candidates on 2014-11-24 peaked within 17 minutes, and six of seven candidates on 2017-09-20 peaked within 32 minutes.

The reanalysis in this study finds that the candidates are variable stars (mostly RR Lyrae), offering a natural explanation for their spatial distribution but not for the unusually tight temporal clustering. The authors hypothesize that event selection cuts in the original pipeline may have produced this clustering but do not establish the mechanism.