Reverse polyblock approximation for optimization over the weakly efficient set and efficient set. (Q5942841)

Let \(X,C,f(x)\) be a polytope in the Euclidean space, a real matrix and a real concave function on \(X\) respectively. The paper is concerned with the problem: find \(\max\{f(x)\mid x\in X_{W_E}\}\) where \(X_{W_E}\) is a set of all efficient points properly defined on \(X\) using the matix \(C\). Problem was first considered for somewhat modified feasible set by \textit{J. Philip} in [Math. Program. 2, 207--229 (1972; Zbl 0288.90052)]. It is first proved that the problem is equivalent to the following one: find \(\max\{\theta (y)\mid y\in\mathbb{R}^p_+\), \(h(y)\leq 0\}\), where both \(\theta(y)\) and \(h(y)\) are decreasing functions on \(\mathbb{R}^p_+\). The author presents a new algorithm for the equivalent problem which is based on a representation of the feasible set by the intersection of a family of simple sets called reverse polyblocks. For given one several linearly constrained convex programs are solved at iteration \(k\) of the algorithm. The paper includes also results of some computational experiments with algorithm coded in Pascal for randomly generated small problems. Unfortunately, the numbers of iterations are not presented for resulting running times.