On the density of Henig efficient points in locally convex topological vector spaces (Q2349836)

Let \((X,\tau_0)\) be a Hausdorff locally convex space over \(\mathbb R\) with dual \(X^*\) and \(\tau\subset \tau_0\) another Hausdorff locally convex topology on \(X\), compatible with the duality \((X,X^*)\) and \(C\) a (not necessarily convex) cone in \(X\). A subset \(A\) of \(X\) is called \(\tau\)-asymptotically compact if \(U\cap\tau\)-cl\(([0,1]A)\) is relatively compact, for some \(\tau\)-neighborhood \(U\) of \(0\). The \(\tau\)-asymptotic cone of the set \(A\) is defined by \(A^\tau_\infty=\bigcap_{t>0}\tau\)-cl\(([0,t]A)\). A point \(x\in A\) is called an efficient point of \(A\) with respect to the cone \(C\) if \((x+C)\cap A\subset x-C\). If the cone \(C\) is pointed (i.e., \(C\cap(-C)=\{0\}\)), then \(x\in E(A,C)\) iff \((x+C)\cap A=\{x\}\). An efficient point \(x\in A\) is called a regular efficient point of \(A\) if it maximizes some strictly positive linear functional \(f\) on \(A\) (strictly positive means that \(f(x)>0\) for all \(x\in C\setminus\{0\}\)). The set of efficient points is denoted by \(E(A,C)\) and that of regular efficient points by \(\operatorname{Reg}E(A,C)\). \textit{M. I. Henig} [J. Optim. Theory Appl. 36, 387--407 (1982; Zbl 0452.90073); ibid. 36, 451--455 (1982; Zbl 0452.90072)] proposed a definition of efficient points with respect to a closed convex cone \(K\) such that \(C\setminus\{0\}\subset\) int\((K)\), the corresponding set of Henig efficient points being denoted by \(HE(A,C)\). In general, \(\operatorname{Reg}E(A,C)\subset HE(A,C)\), with equality when \(A\) is convex. A famous result of \textit{K. J. Arrow} et al. [Contrib. Theory of Games, II, Ann. Math. Stud. No. 28, 87--91 (1953; Zbl 0050.14203)] asserts that, if \(C\) is the non-negative orthant in \(\mathbb R^n\) and \(\emptyset \neq A\subset\mathbb R^n\) is convex compact, then the set of regular efficient point of \(A\) is dense in the set of efficient point of \(A\). Over the years, many extensions of this result were proved, among them a result of \textit{A. Göpfert} et al. [Optimization 53, No. 4, 369--376 (2004; Zbl 1055.90064)], on the denseness of the Henig efficient points in the set of efficient points, in the case of a normed space \(X\). The aim of this paper is to further extend this result to locally convex spaces: Under the hypotheses from the beginning of this review, suppose that \(\emptyset\neq A\subset X\) is \(\tau\)-closed and \(\tau\)-asymptotically compact and \(C\) is a closed convex cone with base \(B\). If there exists a \(\tau_0\)-closed convex cone \(K\neq X\) such that \(A^\tau_\infty\cap K=\{0\}\) and \(C\setminus \{0\}\subset \) int\((K)\), then \(HE(A,C)\) is \(\tau\)-dense in \(E(A,C)\) (Theorem 3.1).