Loop Space Splittings of Sphere Bundles over Highly Connected Poincaré Complexes

Wen Shen · 2026

Read Loop Space Splittings of Sphere Bundles over Highly Connected Poincaré Complexes by Wen Shen (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-00406.txt.

Introduction

Shen gives conditions under which the loop space of a sphere bundle total space breaks into a product of two loop spaces. The bundle sits over a highly connected Poincaré complex, with sphere fibres, and the split separates the base loop space from the fibre loop space. The result needs the fibre dimension parameter m above n, with n at least 2, and the base (n-1)-connected.

Catalog record

CREATORWen Shen
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE18691
LICENSEhttp://creativecommons.org/publicdomain/zero/1.0/
RIGHTSCC0
SOURCEarXiv

Source: https://arxiv.org/abs/2604.00406

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Loop Space Splittings of Sphere Bundles over Highly Connected Poincaré Complexes Wen Shen Abstract . Let m > n ≥ 2 m>n\geq 2 , and let N N be an ( n − 1 ) (n-1) -connected 2 ​ n 2n -Poincaré complex. In this paper, we establish sufficient conditions under which the loop space of the total space M M of the sphere bundle S m − 1 → M → N S^{m-1}\to M\to N (associated to a rank- m m real vector bundle over N N ) splits as a product of the loop spaces of N N and S m − 1 S^{m-1} . 2020 Mathematics Subject Classification Primary 55P15, 55P35 1. Introduction In the realm of algebraic and geometric topology, manifold classification stands as one of the most fundamental and far-reaching central problems. To classify manifolds, we first pick a suitable equivalence relation which describes a unique topological or smooth structure. The most widely studied among these relations are diffeomorphism, homeomorphism, and homotopy equivalence. Beyond these standard frameworks, there exists a more specialized and increasingly prominent classification paradigm: loop decomposition equivalence. Formally, two manifolds X and Y are deemed equivalent under this relation if their loop spaces satisfy \Omega X\simeq\Omega Y , where \Omega denotes the loop functor that maps a topological space X to the space of all based loops in X , and a continuous map f:X\to Y to the induced map on loops. The study of loop space equivalences not only reveals hidden homological and homotopical connections between manifolds but also provides a powerful lens for computing homotopy invariants that are otherwise intractable. A cornerstone of manifold classification lies in the study of highly connected manifolds, a class of spaces with vanishing homotopy groups in low dimensions. This line of inquiry has a long and illustrious history, with foundational contributions from C.T.C. Wall [ 12 , 14 ] . Up to loop decomposition, Beben and Theriault [ 4 ] proved that (n-1) -connected 2n -manifolds are equivalent if and only if their n -th Betti numbers are equal. Parallel to this work, the loop decomposition problem for the odd-dimensional counterpart, namely (n-1) -connected (2n+1) -manifolds, has been tackled by two independent teams: Beben and Wu [ 5 ] , and Huang and Theriault [ 9 ] . A key virtue of these loop decomposition theorems is their computational utility: they reduce the problem of computing the homotopy groups of highly connected manifolds to the computation of homotopy groups of spheres and Moore spaces. From a different perspective, Samik Basu and Somnath Basu [ 2 , 3 ] have presented related computations. Sphere bundles occupy a central role in topology, serving as a bridge between manifold theory, vector bundle theory, and homotopy theory. The homotopy classification of sphere bundles has a rich heritage dating back to the work of James and Whitehead [ 10 , 11 ] . In recent years, the study of loop decompositions has been extended to sphere bundles over 4 -manifolds [ 7 , 8 ] . A striking result from [ 8 ] asserts that the loop spaces of the sphere bundles of most real vector bundles over simply connected closed 4 -manifolds split into a product of simpler homotopy types. This breakthrough not only generalizes earlier loop decomposition results for manifolds but also raises compelling questions about the extent to which such splittings hold for sphere bundles over more general spaces. Inspired by this work on sphere bundles over 4 -manifolds, we turn attention to a natural and far-reaching generalization: sphere bundles associated to real vector bundles over highly connected manifolds. Now we present the main result: Theorem 1.1 . Let m>n\geq 2 . Let N be an (n-1) -connected 2n -Poincaré complex. Let S^{m-1}\to M\to N be the sphere bundle of a rank- m real vector bundle over N . Then \Omega M\simeq\Omega N\times\Omega S^{m-1} if n,m , and d=\mathrm{rank}(H^{n}(N;\mathbb{Z})) satisfy one of the following conditions (1) n\geq 3 , d\geq 2 . (2) n=2 , (m,d)\neq(3,0) and (4,0) . (3) m>2n , d=0 . For d=0 , the complex N as in Theorem 1.1 is homotopy equivalent to S^{2n} . For d=1 , Adams’ result [ 1 ] implies that n must be 2,4 or 8 . Specifically, for n=2 , N\simeq\mathbb{CP}^{2} ; however, for n=4 or 8 , N is not necessarily homotopy equivalent to \mathbb{HP}^{2} or \mathbb{OP}^{2} respectively. This constitutes a key obstruction to the study of the cases (n,d)=(4,1) and (n,d)=(8,1) . For d\geq 2 , \Omega N has already been established in [ 4 , Theorem 1.4] . Notably, for n\neq 2,4,8 , Adams’ result [ 1 ] and Poincaré duality imply d\neq 1 . As an immediate consequence of Theorem 1.1 , we obtain an isomorphism of homotopy groups: \pi_{\ast}(M)\cong\pi_{\ast}(S^{m-1})\oplus\pi_{\ast}(N) Among the results in Theorem 1.1 , Item (2) coincides with the main theorem of [ 8 ] . For Item (3), we follow the idea of [ 8 ] to complete the proof. Notably, Item (1) is proven from a novel and unified perspective, which distinguishes it from the existing arguments in the literature. Moreover, the argument for Item (1) also recovers the cases for n=2 and d\geq 2 as in [ 8 ] . The structure of this paper is organized as follows: In Section 2 , we first investigate vector bundles over highly connected complexes and their classifying maps, presenting key observations that lay the foundation for proving Theorem 1.1 . Subsequently, in Section 3 , we provide the detailed proofs of Items (1) and (3) of Theorem 1.1 . 2. Vector bundles over highly connected complexes First, we recall a classic result in homotopy theory that is fundamental to our subsequent arguments: Lemma 2.1 . For any connected space X and any integer k\geq 2 , there exists a (k-1) -connected space X\langle k\rangle with a fibration p:X\langle k\rangle\to X that induces isomorphisms on \pi_{i} for i\geq k . Proof. This lemma is a direct consequence of the Whitehead tower construction for the space X . ∎ Let N be an (n-1) -connected 2n -Ponincaré complex satisfying H^{n}(N;\mathbb{Z})=\mathbb{Z}^{d} . By the minimal cell structure [ 13 , Proposition 4.1] , N has the following homotopy type N=(\vee_{i=1}^{d}S^{n})\cup_{\alpha}e^{2n} Let \bar{N}=\vee_{i=1}^{d}S^{n} denote the 2n-1 -skeleton of N . Recall that a rank- m real vector bundle over N admits a classifying map f:N\to\mathrm{BO}_{m} where \mathrm{BO}_{m} is the classifying space of the orthogonal group \mathrm{O}_{m} . Lemma 2.2 . Any map f:N\to\mathrm{BO}_{m} admits a lift g:N\to\mathrm{BO}_{m}\langle n\rangle along the fibration p:\mathrm{BO}_{m}\langle n\rangle\to\mathrm{BO}_{m} . Proof. Let F be the homotopy fiber of p:\mathrm{BO}_{m}\langle n\rangle\to\mathrm{BO}_{m} . Since p induces isomorphisms on \pi_{i} for i\geq n , we have \pi_{i}(F)=0 for i\geq n . Note the cell structure of N . We can use the obstruction theory [ 6 , p.418] to prove the lemma. ∎ Note that the space \mathrm{BO}_{m}\langle n\rangle is (n-1) -connected. By the Hurewicz theorem, we have \pi_{n}(\mathrm{BO}_{m}\langle n\rangle)\cong H_{n}(\mathrm{BO}_{m}\langle n\rangle) . Hence, the homotopy class of any map \iota:S^{n}\to\mathrm{BO}_{m}\langle n\rangle is uniquely determined by its induced homomorphism on integral homology. We now generalize this observation to maps defined on wedge sums of n -spheres, which is critical for

Read the complete text