The \(\text{SU}(3)\) Casson invariant for integral homology \(3\)-spheres (Q1807879)

In 1985, Casson introduced an integer valued invariant for homology 3-spheres, which essentially counts the conjugacy classes of irreducible representations from the fundamental group of the manifold into \(\text{SU}(2)\). It is defined as the algebraic intersection number of the representation varieties of the two handlebodies of a given Heegaard splitting in the representation variety of the Heegaard surface. A perturbation is needed so that the two subvarieties are transversal at irreducible representations. Casson proved that the intersection number is independent of the perturbation and the Heegaard splitting used, thus proving the topological invariance. Various work have been done generalizing Casson's invariant. They generally fall into two different approaches: one is a finite dimensional approach using the symplectic structure on the representation varieties, and the other is an infinite dimensional approach using gauge theory. In this paper, the authors construct an \(\text{SU}(3)\) Casson type invariant for homology 3-spheres. The main difficulty is that one must first perturb so that the space of irreducible representations is cut out transversely, but the resulting counting will depend on the perturbation used. A similar problem was encountered by \textit{K. Walker} in generalizing the \(\text{SU}(2)\) Casson invariant to rational homology 3-spheres [An extension of Casson's invariant, Ann. Math. Stud. 126 (1992; Zbl 0752.57011)]. Walker introduced a correction term using the symplectic geometry and stratified structure of representation varieties associated to a Heegaard splitting of the 3-manifold. Such a correction term has been generalized by \textit{S. E. Cappell, R. Lee} and \textit{E. Y. Miller} using Maslov indices and anomaly cancellation to define the \(\text{SU}(n)\) Casson invariant [Bull. Am. Math. Soc., New Ser. 22, No. 2, 269-275 (1990; Zbl 0699.57009)]. The authors here use a gauge theory approach instead. Thus the representations (up to conjugacy) are viewed as flat connections (up to gauge equivalence). Such an approach, in the case of \(\text{SU}(2)\), has been used by \textit{C. H. Taubes} [J. Differ. Geom. 31, No.~2, 547-599 (1990; Zbl 0702.53017)], who showed that the \(\text{SU}(2)\) Casson invariant is \(-{1 \over 2}\) times the Euler characteristic of Floer's homology groups. No correction term is needed there. In order to deal with the perturbation dependence, the authors here carry out a detailed study of the moduli space of connections. They introduce a correction term involving spectral flow and Chern-Simons invariant, and then prove this lead to a topological invariant of homology 3-spheres. Several questions arise naturally from their work here. What kind of surgery relations does the invariant satisfy? Is their \(\text{SU}(3)\) Casson invariant the same as the one given by Cappell-Lee-Miller using the symplectic approach? More generally, what is the relationship between the gauge theory approach and the symplectic approach, if the \(\text{SU}(n)\) Casson invariants for arbitrary 3-manifolds can be defined in both ways? These and other related questions are remarked in the paper.