Rational homology cobordisms of rational space forms (Q1118868)

This paper deals with a number of important problems in low-dimensional topology. The main theorem employs gauge theory to study rational cobordisms of rational homology 3-spheres. The results obtained generalize those of \textit{A. Casson} and \textit{C. McA. Gordon} [Algebr. Geom. Topol., Stanford/Calif. 1976, Proc. Symp. Pure Math. 32, Part 2, 39-53 (1978; Zbl 0394.57008)] and \textit{R. Fintushel} and \textit{R. Stern} [Topology 26, 385-393 (1987; Zbl 0627.57011)]. To be more precise, let W denote a smooth orientable 4-manifold having the same rational homology as a (perhaps multiply) punctured 4-sphere and whose boundary components are rational homology 3-spheres. The objective is to restrict \(\partial W\). To this end it is assumed that there is an epimorphism \(\phi\) : \(H_ 1(W)\to {\mathbb{Z}}_ d\) which satisfies: (1) if \(\tilde W\) denotes the d-fold cover of W associated to ker \(\phi\), then \(\partial \tilde W\) consists of rational homology 3-spheres, and (2) if \(\partial W=\coprod^{k}_{i=1}M_ i\), then each composition \(H_ 1(M_ i)\to H_ 1(W)\to^{\phi}{\mathbb{Z}}_ d\) is nontrivial. Each of the compositions in (2) above determines a Casson-Gordon invariant, \(\sigma (M_ i,\phi)\). The main theorem states that \(| \sum_{i}\sigma (M_ i,\phi)| \leq 3-k\) provided certain technical hypotheses are met.