We propose to use the flux variability of lensed quasar images induced by gravitational microlensing to measure the transverse peculiar velocity of lens galaxies over a wide range of redshift. Microlensing variability is caused by the motions of the observer, the lens galaxy (including the motion of the stars within the galaxy), and the source; hence, its frequency is directly related to the galaxys transverse peculiar velocity. The idea is to count time-event rates (e.g., peak or caustic crossing rates) in the observed microlensing light curves of lensed quasars that can be compared with model predictions for different values of the transverse peculiar velocity. To compensate for the large time-scale of microlensing variability we propose to count and model the number of events in an ensemble of gravitational lenses. We develop the methodology to achieve this goal and apply it to an ensemble of 17 lensed quasar systems. In spite of the shortcomings of the available data, we have obtained tentative estimates of the peculiar velocity dispersion of lens galaxies at $zsim 0.5$, $sigma_{rm pec}(0.53pm0.18)simeq(638pm213)sqrt{langle m rangle/0.3 M_odot} , rm km, s^{-1}$. Scaling at zero redshift we derive, $sigma_{rm pec}(0)simeq(491pm164) sqrt{langle m rangle/0.3 M_odot} , rm km, s^{-1}$, consistent with peculiar motions of nearby galaxies and with recent $N$-body nonlinear reconstructions of the Local Universe based on $Lambda$CDM. We analyze the different sources of uncertainty of the method and find that for the present ensemble of 17 lensed systems the error is dominated by Poissonian noise, but that for larger ensembles the impact of the uncertainty on the average stellar mass may be significant.
Download