Title: Microlensing Predictions: Impact of Galactic Disc Dynamical Models

Abstract: Galactic model plays an important role in the microlensing field, not only for analyses of individual events but also for statistics of the ensemble of events. However, the Galactic models used in the field varies, and some are unrealistically simplified. Here we tested three Galactic disc dynamic models, the first is a simple standard model that was widely used in this field, whereas the other two consider the radial dependence of the velocity dispersion, and in the last model, the asymmetric drift. We found that for a typical lens mass $M_{\rm L}=0.5M_{\odot}$, the two new dynamical models predict $\sim16\%$ or $\sim5\%$ less long-timescale events (e.g., microlensing timescale $t_{\rm E}>300$ days) and $\sim 5\%$ and $\sim 3.5\%$ more short-timescale events ($t_{\rm E}<3$ days) than the standard model. Moreover, the microlensing event rate as a function of Einstein radius $\theta_{\rm E}$ or microlensing parallax $\pi_{\rm E}$ also shows some model dependence (a few percent). The two new models also have an impact on the total microlensing event rate. This result will also to some degree affect the Bayesian analysis of individual events, but overall, the impact is small. However, we still recommend that modelers should be more careful when choosing the Galactic model, especially in statistical works involving Bayesian analyses of a large number of events. Additionally, we find the asymptotic power-law behaviors in both $\theta_{\rm E}$ and $\pi_{\rm E}$ distributions, and we provide a simple model to understand them.