A geometric computation of cohomotopy groups in co-degree one
Michael Jung, Thomas O. Rot
TL;DR
This work solves the geometric computation of the cohomotopy set $\pi^n(X)$ for closed $(n+1)$-manifolds with $n\ge3$ by translating it into a problem on normally framed 1-dimensional submanifolds via the Pontryagin–Thom construction. It distinguishes Type I and Type II manifolds to describe the exact extension data: Type I yields an isomorphism $\mathbb{F}_1(X)\to H_1(X; o_X)$, while Type II yields a nontrivial extension classified by $\mathrm{Ext}(H_1(X; o_X), \mathbb{Z}_2)$ that maps to $w_1^2+w_2$; the extension splits precisely when $X$ admits a $\mathrm{Pin}^-$-structure, linking splitting maps to Spin/Pin$^-$-structures. The paper further develops a refined obstruction to nonvanishing sections of oriented spin vector bundles through a degree invariant $\kappa(E)$ in Type IIa and shows how these invariants generalize Konstantis’ counting invariant in the orientable case. Overall, the results extend Kirby–Melvin–Teichner and Konstantis to non-orientable and non-spin settings, providing explicit geometric obstructions for vector-bundle sections and a complete cohomotopy description in co-degree one.
Abstract
Using geometric arguments, we compute the group of homotopy classes of maps from a closed $(n+1)$-dimensional manifold to the $n$-sphere for $n \geq 3$. Our work extends results from Kirby, Melvin and Teichner for closed oriented 4-manifolds and from Konstantis for closed $(n+1)$-dimensional spin manifolds, considering possibly non-orientable and non-spinnable manifolds. In the process, we introduce two types of manifolds that generalize the notion of odd and even 4-manifolds. Furthermore, for the case that $n \geq 4$, we discuss applications for rank $n$ spin vector bundles and obtain a refinement of the Euler class in the cohomotopy group that fully obstructs the existence of a non-vanishing section.
