The geometric Hopf invariant and surgery theory

Abstract

The first author's geometric Hopf invariant of a stable map $F:Σ^{\infty}X \to Σ^{\infty}Y$ is a stable ${\mathbb Z}_2$-equivariant map $h(F):Σ^{\infty}X \to Σ^{\infty}(Y \wedge Y)$ constructed by an explicit difference construction applied to $(F \wedge F)Δ_X - Δ_Y F$. The stable ${\mathbb Z}_2$-equivariant homotopy class of $h(F)$ is the primary obstruction to desuspending $F$ up to homotopy. The explicit nature of the construction allows for a $π$-equivariant version of $h(F)$ in the case of a $π$-equivariant $F$, with $π$ a discrete group. In earlier joint work we applied the $π_1(N)$-equivariant geometric Hopf invariant of the Umkehr map $F:Σ^{\infty}N^+ \to Σ^{\infty}T(ν_f)$ of an immersion $f:M \to N$ to capture the double points of $f$ in ${\mathbb Z}_2$-equivariant homotopy theory. In this manuscript we use the $π$-equivariant geometric Hopf invariant $h(F)$ to unify all the previous homotopy theoretic treatments of double points. Furthermore, $h(F)$ is combined with the second author's algebraic surgery theory of chain complexes with Poincaré duality to provide the homotopy theoretic foundations for non-simply-connected geometric surgery. For an $n$-dimensional normal map $(f,b):M \to X$ the $π_1(X)$-equivariant geometric Hopf invariant $h(F)$ of the Umkehr map $F:Σ^{\infty}X^+ \toΣ^{\infty}M^+$ is shown to induce the $π_1(X)$-equivariant quadratic structure $ψ_F$ on the chain complex kernel $C$ of $(f,b)$. Previously $ψ_F$ had only been constructed using the chain complex analogue of the functional Steenrod squares. The Wall surgery obstruction $σ_*(f,b)=(C,ψ_F) \in L_n({\mathbb Z}[π_1(X)])$ is the cobordism class of the corresponding $n$-dimensional quadratic Poincaré complex $(C,ψ_F)$, as in the original theory.