Antipodal and fixed point theorems for sets in \(\Re^n\) bounded by a finite number of spheres (Q1291065)

The author proves an antipodal Borsuk-Lyusternik-Shnirel'man type theorem and a fixed point theorem. Let \(B^n_0\) be a closed \(n\)-ball in \(\mathbb R^n\), let \(B^n_j\), \(1\leq j\leq 2p+q,\) be pairwise disjoint closed \(n\)-balls all contained in the interior of \(B^n_0\), and let \(S^{n-1}_j\) be the boundary sphere of \(B^n_j\). Let \(D^n\) be the complement of the interior of the balls \(B^n_j\), \(1\leq j\leq 2p+q\), in \(B^n_0\). Two points of \(D^n\) are said to be antipodal if they lie on one of the spheres \(S^{n-1}_j\), \(0\leq j\leq 2p+q\), and if their midpoint is the center of \(S^{n-1}_j\). Theorem 1. Let \(A_i,A_{-i}\), \(1\leq i\leq m\), be closed subsets of \(D^n\) satisfying: \(\bigcup_{i=1}^m (A_i \cup A_{-i}) = D^n\). \(A_i \cap A_{-i} = \emptyset;\) for \(1 \leq i \leq m\). For any two antipodal points \(x\) and \(y\) on \(S^{n-1}_j\), there is an index \(i \in \{\pm 1, \dots ,\pm m\}\) such that \(x \in A_i\) and \(y \in A_{-i}\) if \( 0 \leq j \leq 2p\) and \(\{x,y\} \subset A_i\) if \( 2p+1 \leq j \leq 2p+q\). Then we have: \(m \geq n+1\). There exist \(n+1\) indices \(1 \leq k_1 < k_2 \dots < k_{n+1} \leq m\) such that at least one of the two intersections \(\bigcap _{i=1}^{n+1} A_{{(-1)^{i-1}}k_i}\) and \(\bigcap _{i=1}^{n+1} A_{{(-1)^{i}}k_i}\) is nonempty. For each \(j\), \(0 \leq j \leq 2p\), there exist \(n\) indices \(1 \leq h_1 < h_2 \dots < h_n \leq m\) such that \(S^{n-1}_j \cap \bigcap _{i=1}^{n+1} A_{{(-1)^{i-1}}h_i} \neq \emptyset\). When \(p=q=0\), this theorem reduces to a theorem proved by the author in [\textit{K. Fan}, Ann. Math., II. Ser. 56, 431-437 (1952; Zbl 0047.42004)]. Theorem 1 implies the following fixed point theorem: Theorem 2. Let \(D^n\) be as above and let \(g:D^n \to \mathbb R^n\) be a map \(g:x \mapsto (g_1(x), \dots ,g_n(x))\). Let \(N(x)=max_{1 \leq i \leq n}|g_i(x)-x_x|\). Suppose that for any two antipodal points \(x, y\) on \(S^{n-1}_j\) there is an index \(i, 1 \leq i \leq n,\) and \(\varepsilon =\pm 1\) such that \(g_i(x)-x_i= \varepsilon N(x)\) and \(g_i(y)-y_i=- \varepsilon N(y)\) if \(0 \leq j \leq 2p\); \(g_i(x)-x_i= \varepsilon N(x)\) and \(g_i(y)-y_i= \varepsilon N(y)\) if \(2p+1 \leq j \leq 2p+q\). Then \(g\) has a fixed point.