scientific article; zbMATH DE number 1329072
Let \(X\) be a locally convex Hausdorff space, \(f,g:X\to \mathbb{R}\) be (continuous) nonnegative positively homogeneous functions and \(c,d\in \mathbb{R}\) positive numbers. It is a simple result that the extremal problems \[ \begin{aligned} P(c):\quad &f(x)\to \max\quad\text{subject to }g(x)\leq c\\ & \text{with }\max\{f(x)\mid g(x)\leq c\}= d\end{aligned} \] and \[ \begin{aligned} Q(d):\quad & g(x)\to \min\quad\text{subject to }f(x)\geq d\\ & \text{with }\min\{g(x)\mid f(x)\geq d\}= c\end{aligned} \] have the same sets of solutions. In the paper it is shown that in case \(f\) is sublinear, the point \(x^0\) is a solution of \(P(c)\) and \(Q(d)\), respectively, if and only if the intersection \[ {1\over d} \underline\partial f(x^0)\cap {1\over c} \underline\partial g(x^0) \] contains a nonzero linear functional. In the same manner it is shown that in cases \(g\) is superlinear (on a certain closed convex cone) the point \(x^0\) is a solution of \(P(c)\) and \(Q(d)\), respectively, if and only if the intersection \[ {1\over d} \overline\partial f(x^0)\cap {1\over c} \overline\partial g(x^0) \] contains a nonzero linear functional. In both cases \(\underline\partial f\) and \(\overline\partial f\) are the classical subdifferential and superdifferential of \(f\) in sense of convex analysis. The proofs are based on some nice results regarding the duality between positively homogeneous functions and their different gauge functions and regarding some known properties of lower and upper support sets of positively homogeneous functions. At the end of the paper, the results are applied to the special problem of finding an element on the unit sphere which realizes the norm of a given linear operator. For the entire collection see [Zbl 0907.00033].
- Mathematical properties of optimization problems defined by positively homogeneous functions
- Some characterizations of the optimal solutions of a vector optimization problem
- Characterizations of optimal solution sets of convex infinite programs
- Homogeneous algorithms for multiextremal optimization
- Characterizations of the solution sets of convex programs and variational inequality problems
- Characterizations of solution sets of mathematical programs in terms of Lagrange multipliers
- Characterization of solution sets of convex programs
- scientific article; zbMATH DE number 58353
- Characterizations of linear suboptimality for mathematical programs with equilibrium constraints
- Characterizing optimality in mathematical programming models
- Duality of nonconvex optimization with positively homogeneous functions
- Saddle representations of positively homogeneous functions by linear functions
- scientific article; zbMATH DE number 1996060 (Why is no real title available?)
- Inverse problem in convex optimization with linear homogeneous constraints
- scientific article; zbMATH DE number 1421089 (Why is no real title available?)
- Best approximation by normal and conormal sets
- Mathematical properties of optimization problems defined by positively homogeneous functions
- On KKT points of homogeneous programs
- Linear programming related to homogeneous programming
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4257432)