scientific article; zbMATH DE number 3857831
[For the entire collection see Zbl 0504.00017.] Let \(f=(f_ 1,...,f_ n)\), \(g=(g_ 1,...,g_ n)\) and \(L_{\ell}(s,z)=<s,\ell f(z)-g(z)>\) for \(\ell \geq 0\), where \(f_ i\) (and \(g_ i)\) is a convex (concave and positive, respectively) real- valued function on S, i.e. the (n-1) dimensional simplex such that \(z_ 1+...+z_ n=1\) and \(z_ i\geq 0\), \(i=1,...,n\). Hereinafter z and s denote some points of S, and \(S_ i=\{z:z_ i=0\}\), \(i=1,...,n\). Let the basic problem be \(V_ M(\ell)=\max_{s} \min_{z} L_{\ell}(s,z),\) where \(f_ i(z)\leq 0\) for \(z\in S_ i\), \(i=1,...,n\), and an index i exists such that \(f_ i(s)>0\) for each s. The basic problem possesses an equivalent dual pair of uniextremal separably- infinite programs being in perfect duality with attainment, and such a pair yields a saddle point for \(L_{\ell}(s,z)\). Moreover, \(V_ M\) has a unique zero.
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 Q3326322)