Testing predictions of semi-analytic models of galaxy evolution against observations help to understand the complex processes that shape galaxies. We compare predictions from the Garching and Durham models implemented on the Millennium Run with observations of galaxy-galaxy lensing (GGL) and galaxy-galaxy-galaxy lensing (G3L) for various galaxy samples with stellar masses in the range 0.5 < (M_* / 10^10 M_Sun) < 32 and photometric redshift range 0.2 < z < 0.6 in the Canada-France-Hawaii Telescope Lensing Survey (CFHTLenS). We find that the predicted GGL and G3L signals are in qualitative agreement with CFHTLenS data. Quantitatively, the models succeed in reproducing the observed signals in the highest stellar mass bin (16 < ( M_* / 10^10 M_Sun) < 32) but show different degrees of tension for the other stellar mass samples. The Durham models are strongly excluded at the 95% confidence level by the observations as they largely over-predict the amplitudes of the GGL and G3L signals, probably because they predict too many satellite galaxies in massive halos.

9 pages, 8 figures, submitted to A&A. Comments welcome