On the Probability Distribution of Cosmological Microlensing Optical Depths

It is conventional to calculate the probability of microlensing for a cosmologically distant source based on the Press-Gunn approximation that the lensing objects are uniformly and randomly distributed in the intervening space with a constant comoving density. Here we investigate more realistic cosmological microlensing statistics by considering the strong spatial clustering of likely lensing objects with each other in galaxies and their association with the clumps of dark matter that make up the massive halos of galaxies. The distribution of microlensing optical depth (kappa) along randomly chosen sight lines is calculated as is the conditional distribution of kappa along sight lines near one which is strongly microlensed. Our overall result is that the Press-Gunn approximation is a useful order-of-magnitude approximation if the massive halos of galaxies are made of dark compact objects but that it fails badly and can be qualitatively misleading in the more likely case in which only the ordinary stellar populations of galaxies are the dominant source of cosmological microlensing events. In particular, we find that microlensing by stars is limited to of order 1 percent of high redshift sources at any one time. Furthermore, even though only a small fraction of high redshift sources are multiply-imaged (by galaxies), it is these sources that are most likely to be microlensed by stars. Consequently, microlensing by stars is usually observed at kappa's near 1 where the simple isolated point mass lens approximation is not appropriate. However, if CDM halos are composed of condensed objects, then more than 10 percent of high redshift sources are microlensed at any given time. The vast majority of these sources are not multiply-imaged, and have kappa's smaller than 0.01