Stiefel-Whitney classes and immersions of orientable and spin manifolds (Q2072118)

In [Ann. Math. (2) 122, 237--328 (1985; Zbl 0592.57022)], \textit{R. L. Cohen} resolved positively the famous immersion conjecture: every closed connected smooth manifold of dimension \(n>1\) can be immersed in \(\mathbb R^{2n-\alpha(n)}\), where \(\alpha(n)\) is the number of ones in the binary expansion of \(n\). This general result is optimal, since the appropriate product of real projective spaces (of total dimension \(n\)) cannot be immersed in \(\mathbb R^{2n-\alpha(n)-1}\). This is due to the fact that the dual Stiefel-Whitney class \(\overline w_{n-\alpha(n)}\) of that product is nonzero. The dimension of each projective space in the product in question is a power of two, and so the product is a non-orientable manifold. In the paper under review the authors consider the following problem: for a given \(n\), find the largest value \(c\) such that there is an \(n\)-dimensional \textit{orientable} manifold \(M\) with \(\overline w_c(M)\neq0\) (consequently, such a manifold \(M\) cannot be immersed in \(\mathbb R^{n+c-1}\)); and the same problem for Spin manifolds. In the case of orientable manifolds it is proved that \(c=n-\alpha(n)\) if \(n\equiv1\pmod4\), and \(c=n-\alpha(n)-1\) otherwise. In the case of Spin manifolds the authors show that all dual Stiefel-Whitney classes vanish if \(n<8\), and they calculate the exact value of \(c\) for \(8\leq n\leq23\) and \(32\leq n\leq33\). The proofs of these results rely on the following equivalences (also proved in the paper): \begin{itemize} \item there exists an \(n\)-dimensional orientable manifold \(M\) with \(\overline w_{n-k}(M)\neq0\) if and only if there exists \(\alpha\in H_n(K(\mathbb Z_2,k);\mathbb Z)\) such that \(\langle\chi Sq^{n-k}\iota_k,\rho(\alpha)\rangle\neq0\) (where \(\iota_k\in H^k(K(\mathbb Z_2,k);\mathbb Z_2)\) is the fundamental class, \(\chi\) is the canonical antiautomorphism of the mod \(2\) Steenrod algebra, and \(\rho:H_n(K(\mathbb Z_2,k);\mathbb Z)\rightarrow H_n(K(\mathbb Z_2,k);\mathbb Z_2)\) is the mod \(2\) reduction); \item there exists an \(n\)-dimensional Spin manifold \(M\) with \(\overline w_{n-k}(M)\neq0\) if and only if there exists \(\alpha\in ko_n(K(\mathbb Z_2,k))\) such that \(\langle\chi Sq^{n-k}\iota_k,h(\alpha)\rangle\neq0\) (where \(\iota_k\) and \(\chi\) are as above, \(ko_*(-)\) is connective \(KO\) homology, localized at \(2\), and \(h:ko_n(K(\mathbb Z_2,k))\rightarrow H_n(K(\mathbb Z_2,k);\mathbb Z_2)\) is the Hurewicz homomorphism). \end{itemize}