Gaia16aye was a binary microlensing event detected by Gaia in 2016 as one of the first transients of that type reported by Gaia Science Alerts. The world-wide follow-up campaign resulted in enormous number of photometric data collected, which revealed one of the most complicated and demanding microlensing event ever observed. Multiple second-order effects detectable in the microlensing lightcurve, including Keplerian rotation of binary lens components, provided almost complete information about the system. Thanks to the large Einstein radius \(\theta_{\mathrm E}= 3.04\pm0.24\) mas and high brightness of the event, it was predicted to have an astrometric microlensing signal measurable by the Gaia mission.
In this work we have analyzed the astrometric time-series for this event provided by the Gaia space mission. The 1-D projected positions of the light centroid could not be explained by the simple 5-parameter astrometric model (\(\alpha_0\), \(\delta_0\), \(\varpi\), \(\mu_{\alpha}\), \(\mu_{\delta}\)). Including the astrometric microlensing shift calculated based on the photometric model heavily improved the fit and allowed to measure the value of \(\theta_{\mathrm E}=3.06\pm0.06\) mas, which is consistent with the value derived from the photometric model. This is third detection of astrometric microlensing and the first for an event with unresolved source and lens. This also shows that measuring astrometric microlensing effect is possible with the 1-D Gaia astrometric time-series, and confirms that this is a viable method for measuring masses of lenses.