[Paper Review] On the high-dimensional geography problem

Robert Burklund, Andrew Senger|arXiv (Cornell University)|Jul 10, 2020

Geometric and Algebraic Topology4 citations

TL;DR

This paper completes the classification of smooth, closed, oriented, (n−1)-connected 2n-manifolds for all n ≠ 63 by resolving the high-dimensional geography problem via n-space invariants. It establishes realizability conditions involving signature, Kervaire invariant, and characteristic classes, and provides a counterexample to conjectures of Galatius–Randal-Williams and Bowden–Crowley–Stipsicz in dimension 23, linked to the Witten genus and E∞-ring spectra refinement by Ando–Hopkins–Rezk.

ABSTRACT

In 1962, Wall showed that smooth, closed, oriented, $(n-1)$-connected $2n$-manifolds of dimension at least $6$ are classified up to connected sum with an exotic sphere by an algebraic refinement of the intersection form which he called an $n$-space. In this paper, we complete the determination of which $n$-spaces are realizable by smooth, closed, oriented, $(n-1)$-connected $2n$-manifolds for all $n eq 63$. In dimension $126$ the Kervaire invariant one problem remains open. Along the way, we completely resolve conjectures of Galatius-Randal-Williams and Bowden-Crowley-Stipsicz, showing that they are true outside of the exceptional dimension $23$, where we provide a counterexample. This counterexample is related to the Witten genus and its refinement to a map of $\mathbb{E}_\infty$-ring spectra by Ando-Hopkins-Rezk. By previous work of many authors, including Wall, Schultz, Stolz and Hill-Hopkins-Ravenel, as well as recent joint work of Hahn with the authors, these questions have been resolved for all but finitely many dimensions, and the contribution of this paper is to fill in these gaps.

Motivation & Objective

• To complete the classification of smooth, closed, oriented, (n−1)-connected 2n-manifolds for all n ≠ 63, resolving the high-dimensional geography problem.
• To determine which n-spaces—algebraic invariants combining homology, intersection form, and normal data—are realizable by such manifolds.
• To resolve long-standing conjectures of Galatius–Randal-Williams and Bowden–Crowley–Stipsicz, identifying a counterexample in dimension 23.
• To clarify the role of the Kervaire invariant one problem in dimension 126 and its impact on realizability.
• To apply results to Stein fillability of contact structures and mapping class group computations in high-dimensional topology.

Proposed method

• Use Wall’s n-space framework to algebraically classify (n−1)-connected 2n-manifolds via homology, intersection form, and normal bundle data.
• Apply homotopy-theoretic tools, including the cokernel of the J-homomorphism and stable homotopy groups, to analyze boundaries of manifolds.
• Leverage the Witten genus and its refinement to E∞-ring spectra to detect exotic structures in dimension 23.
• Employ obstruction-theoretic arguments and computations in stable homotopy theory to verify realizability conditions for n-spaces.
• Use the Kervaire–Milnor exact sequence to relate homotopy spheres to parallelizable manifolds and determine which bound (n−1)-connected 2n-manifolds.
• Combine results from Hill–Hopkins–Ravenel, Stolz, and Hahn–Burklund–Senger to close remaining gaps in the geography problem.

Experimental results

Research questions

• RQ1 Which n-spaces are realizable by smooth, closed, oriented, (n−1)-connected 2n-manifolds for all n ≠ 63?
• RQ2 Are the conjectures of Galatius–Randal-Williams and Bowden–Crowley–Stipsicz true in all dimensions, or are there exceptions?
• RQ3 What is the role of the Witten genus and its E∞-refinement in detecting exotic structures in dimension 23?
• RQ4 How does the Kervaire invariant one problem in dimension 126 affect the classification of 126-manifolds?
• RQ5 Which homotopy spheres admit Stein fillable contact structures, and how does this relate to bounding parallelizable manifolds?

Key findings

• An n-space is realizable if and only if it satisfies signature, Kervaire invariant, and characteristic class conditions depending on n mod 4, with explicit exceptions for n = 4, 8, 9, 12, 63.
• For n = 63, realizability depends on the existence of a Kervaire invariant one manifold in dimension 126, which remains open.
• A counterexample to the conjectures of Galatius–Randal-Williams and Bowden–Crowley–Stipsicz exists in dimension 23, where a homotopy sphere with [Σ] = η³κ admits a Stein fillable contact structure but does not bound a parallelizable manifold.
• The Witten genus and its E∞-refinement by Ando–Hopkins–Rezk explain the exceptional behavior in dimension 23.
• The abelianization of the mapping class group Γⁿ_g is smaller than expected when n = 11 (g ≥ 3), due to the 23-dimensional counterexample, with H₁(Γ¹¹_g) ≅ (Z/4Z ⊕ Z/2Z) ⊕ Z/4Z.
• The Stein fillability of homotopy spheres in odd dimensions is fully characterized: a (2q+1)-sphere admits a Stein fillable contact structure iff [Σ] = 0 in coker(J)₂q₊₁, except when q = 11, where the image is {0, η³κ}.