Quadratic forms between Euclidean spheres (Q1330991)

The story begins with identities of the type \[ (x_ 1^ 2+\cdots +x_ r^ 2) (y_ 1^ 2+\cdots +y_ s^ 2)= z_ 1^ 2+\cdots +z_ n^ 2,\tag \(*\) \] \(z_ k\in \mathbb{R}(x_ i, y_ j)\) (\(\mathbb{R}\) real numbers), and the famous Hurwitz-Radon theorem (1922). \((*)\) leads to normed bilinear maps \(f: \mathbb{R}^ r\times \mathbb{R}^ s\to \mathbb{R}^ n\) which in turn lead (say by the Hopf construction) to the map \(F: S^{r+s-1}\to S^ n\) of Euclidean spheres defined by \(F(x,y)= (\| x\|^ 2- \| y\|^ 2, 2f(x,y))\), \(x\in\mathbb{R}^ r\), \(y\in \mathbb{R}^ s\), \(\| x\|^ 2+ \| y\|^ 2=1\). Given \(m\) (respectively \(n\)), the author determines the least \(n= q(m)\) (respectively the greatest \(m= p(n)\)) for which there is a non-constant homogeneous quadratic map \(S^ m\to S^ n\) of Euclidean spheres (theorems 4 and 5). Useful alternative expressions for \(q(m)\) and \(p(n)\) are determined in \S5.