Nonhomeomorphic closed aspherical four-manifolds

The explanation

Start with a cycle of 3 points, a triangle. Try to colour each point red or blue so that neighbours get different colours. Red, blue, then the third point touches both. It fails. A cycle of 2 points works at once, red then blue. Every odd cycle fails in the same way. This tiny parity fact sits at the heart of a new topology result.

First, the setting. A manifold is a space that looks like ordinary space near each point. A torus, the surface of a doughnut, is a two-dimensional example. Its universal cover is the flat plane, which can shrink to a point. A manifold with this property is called aspherical. For such a manifold, the fundamental group, its group of loops, decides its homotopy type. Homotopy type means its shape up to continuous deformation.

The Borel conjecture is usually stated as follows. Every homotopy equivalence between closed aspherical manifolds should be homotopic to a homeomorphismSource 1. A homeomorphism is a perfect, reversible matching of points. A weaker form asks only that homotopy-equivalent closed aspherical manifolds be homeomorphic. The question already appears in Borel's letter to Serre of 2 May 1953Source 2. It holds in dimensions one and two, and for closed orientable three-manifolds. Hsiang and Wall proved it for homotopy tori in dimensions five and upSource 3. Farrell and Jones did so for closed negatively curved manifolds thereSource 4. Dimension four is the hard case. Bartels and Lück's four-dimensional result assumes the fundamental group is good in Freedman's senseSource 5. Khan proves rigidity only up to a weaker relation called s-cobordismSource 6. Separately, Davis, Hayden, Huang, Ruberman and Sunukjian found smooth examples that are homeomorphic but not diffeomorphicSource 7. That is a different question.

The paper claims to disprove the weaker form in dimension fourSource 8. It builds two closed aspherical four-manifolds M and N. They are homotopy equivalent but not homeomorphic. Their common fundamental group is word-hyperbolic, a strong negative-curvature property of groupsSource 9. It also claims a homotopy equivalence of M to itself that no homeomorphism can matchSource 10.

Here is where parity enters. The paper starts from a four-dimensional building block and looks at its cyclic covers, unwrapped copies of degree d. The boundary of the double cover can be filled by a manifold. For odd d of at least 3, it claims no filling existsSource 11. A filling would give a colouring of a cycle of d points. Neighbours could never share a colour. So the colours would have to alternate around the cycle, which is impossible when the cycle is oddSource 12.

The obstacle is to rule out every homeomorphism, not just one homotopy class. The paper glues the blocks in two ways, giving M and N. It proves the group of outer automorphisms, the symmetries of the fundamental group, is finiteSource 13. Two subgroups realized on M and N then differ. They contain different numbers of elements with finite-order liftsSource 14. A homeomorphism would force them to match. The proof also uses two companion constructions from other papersSource 15.

How sure is this? No Lean check covers this paper. No specialist review has been published. The argument depends on two companion papers, which also lack published review. Treat the counterexample as a claim.

What it doesn't show. This does not affect the Borel conjecture in dimensions five and up, where it is proved for large classes of groups.