2022MNRAS.517..257S

The properties of the Milky Way's nuclear stellar disc give crucial information on the epoch of bar formation. Mira variables are promising bright candidates to study the nuclear stellar disc, and through their period-age relation dissect its star formation history. We report on a sample of 1782 Mira variable candidates across the central $3\times 3\, \mathrm{deg}^2$ of the Galaxy using the multi-epoch infrared VISTA Variables in Via Lactea (VVV) survey. We describe the algorithms employed to select candidate variable stars and then model their light curves using periodogram and Gaussian process methods. By combining with WISE, 2MASS, and other archival photometry, we model the multiband light curves to refine the periods and inspect the amplitude variation between different photometric bands. The infrared brightness of the Mira variables means many are too bright and missed by VVV. However, our sample follows a well-defined selection function as expected from artificial star tests. The multiband photometry is modelled using stellar models with circumstellar dust that characterize the mass-loss rates. We demonstrate how ≳90 per cent of our sample is consistent with O-rich chemistry. Comparison to period-luminosity relations demonstrates that the bulk of the short period stars are situated at the Galactic Centre distance. Many of the longer period variables are very dusty, falling significantly under the O-rich Magellanic Cloud and solar neighbourhood period-luminosity relations and exhibit high mass-loss rates of $\sim 2.5\times 10^{-5}M_\odot \, \mathrm{yr}^{-1}$. The period distribution appears consistent with the nuclear stellar disc forming $\gtrsim 8\, \mathrm{Gyr}$ ago, although it is not possible to disentangle the relative contributions of the nuclear stellar disc and the contaminating bulge.