Azimuthally averaged radial convergence profiles, for the different lens models applied to fit the sample of six mock lenses (Fig. 7). Upper part of each panel: true profiles are shown in dotted lines; power-law, cored power-law and composite models are shown in blue continuous, green dashed and red dot-dashed lines, respectively. The spectroscopic (square) aperture used for computing velocity dispersions is indicated as a vertical dotted line, and the true Einstein radius location is indicated as a vertical dashed line. The gray area encloses lensed quasar image positions. For each model, the inferred H₀ values are indicated (in km s⁻¹ Mpc⁻¹), and must be compared to the input value . Lower part of each panel: relative error computed as (truth − model)/truth. The pixel size in the simulated images is .