The Seiberg-Witten invariants and 4-manifolds with essential tori (Q1426826)

One of the reasons the Seiberg-Witten invariant is so useful in 4-dimensional topology is that there are reasonable ways to compute the Seiberg-Witten invariants for a wide variety of 4-manifolds. The main theorem in this paper has been quoted for several years and has found many applications. Most of these applications stem from an elegant consequence of this gluing theorem given in [\textit{R. Fintushel} and \textit{R.~Stern}, Invent. Math.~139, 363--400 (1998; Zbl 0914.57015)] where the Seiberg-Witten invariants of knot and link surgery manifolds are expressed using Alexander polynomials. This paper defines Seiberg-Witten invariants for 4-manifolds with cylindrical ends modeled on a 3-torus cross a ray. This invariant associates an integer to every pair consisting of a \(\hbox{Spin}_c\) structure and relative cohomology class so that the first Chern class of the \(\hbox{Spin}_c\) structure is the image of the relative cohomology class in the absolute cohomology. The most important cases occur when the induced map from the second cohomology of the ends is non-trivial. The main theorem expresses the Seiberg-Witten invariant of 4-manifolds obtained by gluing two ends of one together. The author does a good job explaining the geometric set-up. Other gluing results for Seiberg-Witten invariants have appeared in the literature. It is typical of these papers to state that some technical point is a simple modification of a technical argument from Donaldson theory. This paper avoids this particular dodge. The proof is mostly self-contained and everything is explained in the framework of Seiberg-Witten theory. At a couple of points the author does quote results from [\textit{J.~W.~Morgan}, \textit{Z.~Szabó} and \textit{C.~H.~ Taubes}, J. Differ. Geom. 44, 706-788 (1996; Zbl 0974.53063)]. This paper provides a slightly over-due proof of an important result in Seiberg-Witten theory and this proof was worth the wait.