Lectures on the topology of 3-manifolds. An introduction to the Casson invariant (Q5906422)

This is an excellent introduction to the Rokhlin and Casson invariants for homology 3-spheres (closed 3-manifolds with the homology of the 3-sphere), and in particular also to the necessary background material from the theory of 3- and 4-manifolds such as Heegaard splittings, Dehn surgery on links, the Kirby calculus for framed links, Seifert fiber spaces, some knot theory including Seifert surfaces and matrices, Alexander polynomials and Arf invariants, and 4-manifolds and their intersection forms, so the book may serve also as a reasonable short and efficient introduction to some important parts of low-dimensional topology. It grew out of a course for second year graduate students and concentrates 19 lectures on less than 200 pages, including also a glossary on back-ground material from algebraic topology, a collection of exercises, open problems and comments on recent developments as Floer-homology (of which the Casson invariant occurs as the Euler-characteristic) and Donaldson polynomials, and also an extensive bibliography. So each of these chapters is rather short, and I think that the author has succeeded in presenting the main ideas and constructions involved in a lucid and intuitive, but nevertheless rigorous way, avoiding technicalities and excessive notation, presenting instead many examples and nice pictures (a few of them taken from the book ``Algorithmic and Computer Methods for Three-Manifolds'' by Fomenko and Matveev to which the author gives credit as one of his two major sources, the other being the book of Akbulut and McCarthy on the Casson invariant). Throughout the book there are comments on further developments, open and recently solved problems and references to the relevant literature. Rokhlin defined his invariant in a seminal paper from 1952. Rokhlin's theorem states that, if the intersection form (a non-degenerate, integer-valued symmetric bilinear form on the second homology which determines such a manifold up to homotopy) of a closed smooth simply-connected 4-manifold is even (the associated quadratic form has only even values) then its signature is divisible by 16 (the signature of any even form is divisible by 8). Now any closed homology 3-sphere \(\Sigma\) occurs as the unique boundary component of a simply-connected 4-manifold \(W\) with even intersection form, and the Rokhlin invariant of \(\Sigma\) is defined as the signature of \(W\) divided by 8, modulo 2 (so it has values in \(\mathbb{Z}/2\mathbb{Z}\); by Rokhlin's theorem, and because the signature is additive when glueing two 4-manifolds along a common boundary \(\Sigma\), it is well-defined). On the other hand, the Casson invariant of a closed homology 3-sphere was defined around 1985 and has values in \(\mathbb{Z}\). It is obtained by counting, in a suitable and delicate way with signs determined by orientations, the number of irreducible representations of its fundamental group in the 3-dimensional Lie group SU(2) of unitary \((2\times 2)\)-matrices of determinant one. As both invariants satisfy the same type of surgery formula (involving the second derivative of the Alexander polynomial of a knot, evaluated at one), the mod 2 reduction of the Casson invariant gives the Rokhlin invariant. In particular, by the definition of the Casson invariant, both invariants are zero for homotopy 3-spheres (trivial fundamental group). This was Casson's solution of an old problem about the Rokhlin invariant which thus cannot be used to distinguish a homotopy 3-sphere, that is a possible counterexample to the Poincaré conjecture, from the 3-sphere itself. As explained in the book, together with Freedman's result that every even form is the intersection form of a unique closed topological 4-manifold (e.g. the 4-sphere for the trivial form), this implies the existence of closed 4-manifolds without simplicial triangulations (note also that, by Rokhlin's theorem, a closed 4-manifold cannot be smooth unless its signature is divisible by 16, so Freedman's theorem implies immediately the existence of closed 4-manifolds without smooth structures). The corresponding question is still open in dimensions greater than four and also related to the Casson invariant; in fact, every closed topological manifold of dimension \(\geq 5\) is simplicially triangulable if and only if there exists a homology 3-sphere with Casson invariant one such that the connected sum of two copies of it is homology-cobordant to zero (but by Kirby and Siebenmann there exist topological manifolds without a piecewise linear structure or combinatorial triangulation in all dimensions \(\geq 5)\). The first nine chapters of the book present the above-mentioned basic material on 3- and 4-manifolds and on knot theory. In chapter 10, following a geometric approach of Freedman and Kirby, a more general version of Rokhlin's theorem is proved (and applications to the Thom conjecture on the genus of surfaces representing homology classes in \(\mathbb{C} P^2\) arc discussed). In chapter 11, the Rokhlin invariant is introduced, computed for Seifert fibered homology spheres, its surgery formula is derived and the homology cobordism group of homology 3-spheres is discussed. Chapter 12 gives an axiomatic version, the main properties and the unicity of the Casson invariant, chapters 13-15 deal with SU(2)-representation spaces, and finally in chapters 16 and 17 the Casson invariant is constructed using Heegaard splittings (following the book of Akbulut and McCarthy, but avoiding some technical details). In the last chapter the Casson invariant is computed for Seifert fibered homology 3-spheres. To conclude, the author has succeeded in presenting a lot of material in a clear and efficient way, and the book is interesting and stimulating to read.