On the intersection forms of spin four-manifolds with boundary (Q2509866): Difference between revisions

A smooth spin four-manifold may have a number of copies of the negative definite \(E_8\) lattice in its intersection lattice, however it must have a sufficient number of hyperbolic pairs. Conjecturally, there must be at least three hyperbolic pairs for every two \(E_8\)s. This is known as the \(11/8\) conjecture because of the resulting inequality relating the rank and signature of the intersection form. The current state of the art is a version of the inequality with \(10/8\) proved by Furuta shortly after the introduction of the Seiberg-Witten equations. Furuta's proof followed from the \(\text{Pin}(2)\)-equivariant homotopy theory of the Seiberg-Witten moduli space. Recently, Manolescu resolved the triangulation conjecture via similar analysis. In this paper, Manolescu proves similar bounds for spin cobordisms, introduces a new invariant for homology spheres and describes a condition that may be useful for an inductive attack on the \(11/8\) conjecture. Technically, he uses a finite dimensional approximation to the Seiberg-Witten equations by considering the restriction to a space spanned by eigenfunctions with bounded eigenvalue. After the introduction, he recalls a number of facts about the equivariant \(K\)-theory of \(\text{Pin}(2)\) spaces. He then restricts to a special type of \(\text{Pin}(2)\) space that appears in the context of the Seiberg-Witten equations. In particular he defines a few invariants for such spaces. After recalling the finite dimensional approximation of the Seiberg-Witten equations, he defines his new \(3\)- manifold invariants. One drawback of these new invariants is that they are currently difficult to compute. In the fifth section the invariants are computed for several infinite families of Seifert Fiber spaces. These computations are based on the explicit description of the Seiberg-Witten moduli space obtained in [\textit{T. Mrowka} et al., Commun. Anal. Geom. 5, No. 4, 685--791 (1997; Zbl 0933.57030)] by reducing to the vortex equations on \(2\)-dimensional orbifolds. It is shown that the invariants and restrictions on the intersection forms of manifolds bounding a given three manifold, so obtained are better in some cases than older invariants defined via orbifods and filling. The paper concludes with a comparision of invariants arising from \(K\)-theory and homological invariants.