Mapping a sphere into Euclidean space (Q909290)

The following hypothesis of Knaster is well known: For any mapping \(f: S^ n\to {\mathbb R}^ m\) and any \(n-m+2\) points \(\{x_ i\}_{i=1}^{n- m+2}\) of \(S^ n\) there is \(g\in \mathrm{SO}(n+1)\) such that \(f(g(x_ i))=f(g(x_ j))\), \(1\leq i,j\leq n-m+2\). Recently Makeev proved that for \(m>2\) the Knaster hypothesis is not true. In the present paper it is shown that a counterexample of Knaster's hypothesis can be chosen to be a map defined by homogeneous polynomials. Precisely, the following theorem is proved: Let \(l(n-1)<m(C^ t_{t+l-1}+C^{t-1}_{t+l-2}-1)\). Then the set \(\{f\in H: f\) is not constant on any sphere \(S^{l-1}\subset \mathbb R^ n\}\) is open and dense in \(H\), where \(H\) is the space of all mappings from \({\mathbb R}^ n\) to \({\mathbb R}^ m\) given by polynomials with degree no more than \(t\).