Arc statistics with realistic cluster models

Arc statistics is known to be a powerful cosmological tool. Numerical lensing simulations show that orders of magnitude differences in the number of {\em giant} arcs on the whole sky are expected in different cosmological models. In this paper, we discuss the analytic and numerical methods in arc statistics and show that analytic models fail to reproduce the efficiency for strong lensing of more realistic numerical cluster models. Then, we discuss two recent extensions of the lensing simulations, i.e. the effects of cD galaxies in the lensing clusters and the impact of cluster mergers on arc statistics. We show that cD galaxies can increase the lensing cross sections for long and thin arcs by perhaps up to $\sim50%$, while major mergers can change the cluster efficiency for producing such arcs by up to one order of magnitude.