A note on nonconvex minimax theorem with separable homogeneous polynomials (Q637566)

The paper states and proves a minimax theorem for nonconvex bifunctions of a special type. The main result runs as follows. For each \(q\in \mathbb{N}\), let \(S_{q}\) be the set of polynomials of the form \(f(x_{1},x_{2},\ldots,x_{m})=\sum_{j=1}^{m}\alpha_{j}x_{j}^{q}+b\) where \(b\in \mathbb{R}\) and \(\alpha_{j}\in \mathbb{R}\), \(j=1,2\ldots,m\). Let \(\Delta\) be a compact box in \(\mathbb{R}^{m}\) and \(A\) a convex subset in \(\mathbb{R}^{n}\). If \(f:\mathbb{R}^{m}\times \mathbb{R}^{n}\rightarrow \mathbb{R}\cup \{+\infty \}\) is a bifunction such that \(f(x,\cdot)\) is convex for all \(x\in \Delta\) and \(f(\cdot,y)\in S_{q}\) for all \(y\in A\), then \[ \inf_{y\in A}\max_{x\in \Delta}f(x,y)=\max_{x\in \Delta}\inf_{y\in A}f(x,y). \] Some results follow as corollaries; among them, von Neumann's minimax theorem.