Successful _pΔMSE fits with different stochastically varied SU start vectors. The variation of the scaled SU start vectors followed the algorithm described for Fig. 9 A with the difference that the data points were generated de novo for each SU vector. Successful_pΔMSE fits were selected as described in Fig. 9 B by using a threshold (red dashed lines) at 10⁻¹⁰ for the corresponding ΔMSE_x values. Again, for all four models, the sums of all obtained successful_pΔMSE fits out of 19,000 fits are indicated.