Geometric formulas for Smale invariants of codimension two immersions (Q1869159)

It is well-known that the correspondence which associates to an immersion \(f\) of an \(n\)-sphere \(S^n\) into \((n+k)\)-space \(\mathbb{R}^{n+k}\) its Smale invariant \(\Omega (f)\) induces an isomorphism between the set (group) \(\text{ Imm}(S^n,\mathbb{R}^{n+k})\) of regular homotopy classes of immersions of \(n\)-sphere into \((n+k)\)-space and the \(k\)-th homotopy group \(\pi_k(V_{n+k,k})\) of the Stiefel manifold of \(k\)-frames in \((n+k)\)-space. In this article, the authors mainly study the Smale invariant \(\Omega (f)\in \mathbb{Z}\) of an immersion \(f\) of a \((4k-1)\)-sphere into \((4k+1)\)-space and show three formulas (Theorem 1.1), expressing the Smale invariant \(\Omega (f)\) by geometric characteristics of some generic maps \(g\) from oriented \(4k\)-dimensional manifolds \(M^{4k}\) having the \((4k-1)\)-sphere as its boundary to \((6k-1)\)-space or the upper half space of \(6k\)-space. For example, the first formula (1.1) is given as follows: If \(g\) is a generic map from an oriented \(4k\)-dimensional manifold \(M\) with boundary the \((4k-1)\)-sphere to \((6k-1)\)-space such that \(g\) is an immersion on some neighbourhood of the boundary and its restriction to the boundary is regularly homotopic to \(f\) in \((6k-1)\)-space, then the equation \(a_k(2k-1)!\Omega (f)=-{\bar P}_k[{\hat M}]+\# \Sigma^{1,1}(g)\) holds, where \({\hat M}\) is a closed manifold obtained by adding a disk to \(M\) along its boundary, the term \({\bar P}_k[{\hat M}]\) is the \(k\)-th normal Pontryagin number of \(\hat M\), the term \(\# \Sigma^{1,1}(g)\) denotes the algebraic number of cusps of \(g\), and \(a_k\) is equal to 1 or 2 according to \(k\) being even or odd. As a corollary to theorem 1.1, the authors show that the natural map \(\text{ Imm}(S^{4k-1},\mathbb{R}^{4k+1})\to \text{ Imm}(S^{4k-1},\mathbb{R}^{6k-1})\) is injective. This means that the following theorem of \textit{M. A. Kervaire} [Comment Math. Helv. 34, 127-139(1959; Zbl 0108.18002)] is the best-possible for \(n=4k-1\): If \(2q>3n+1\), then every embedding of \(n\)-sphere into \(q\)-space is regular homotopic to a standard embedding. Using the techniques adopted in the proof of Theorem 1.1, the authors give a formula for the linking number of the \(2k\)-fold self-intersection set of a generic immersion \(f\) of 2-connected closed oriented \((4k-1)\)-dimensional manifold into \((4k+1)\)-space. In the proof of this theorem, the equation on the last line of p.192 should be read as \(<{\bar P}_1^k,[N]>=(2k+1)\# D_{2k+1}(h)\).