Maximizing a linear fractional function on a Pareto efficient frontier (Q700704)

Let \(F_i(z)\) \((i=1,2,3)\) be nonconstant linear fractional functions defined on a polytope \(P\subset\mathbb{R}^n\). The authors study the problem: Maximize \(F_3(z)\) s.t. \(z\in E\subset \mathbb{R}^n\) where \(E\) the set of all Pareto efficient points. A point \(z^0\in P\) is said to be Pareto efficient w.r.t. \(F_1,F_2,P\) if for all \(z\in P\), \(F_1(z)\geq F_1(z^0)\) and \(F_2(z)\geq F_2 (z^0)\) implies \(F_1(z)=F_1(z^0)\) and \(F_2(z)=F_z(z^0)\). The authors describe a pivoting-type algorithm that generates a sequence of points in \(E\) that contains an optimal solution. They have performed the sensitivity analysis on the convex simplex method to generate the two-dimensional efficient frontier \(F(E)\) where \(F(z)= (F_1(z),F_2(z)) \subset\mathbb{R}^2\). It is claimed that the algorithm takes care of degeneracy and the usual nondegeneracy assumption is not required, e.g. \textit{H. Konno} et al. [J. Global Optim. 1, 65-81 (1991; Zbl 0746.90056)]. The computational procedure is illustrated with the help of an example.