Geometrically, the inverse problem is to find the value of k25 such that the x-abscissa of the upper saddle node (SN2) is as close to zero as possible. As inequality constraints, we take 0.1 ≤ k25 ≤ 1.5. The result of the inverse analysis shows that the feedback strength should increase from its initial value of k25 = 0.9 to k25 = 1.099, resulting in a change of bifurcation diagram shown in Figure 12. Row 1 of Table 3 shows the number of optimization iterations as well as the number of functional evaluations (which in this case equals the number of one-parameter continuations) required to reach a tolerance of 10-3 on the function value.