Isotopies of 3-manifolds (Q1924687)

It is not easy, in a short review, to present the variety of results contained in this nicely written paper. It is on the line of better understanding of the space \(\text{Diff} (M)\) for three-dimensional manifolds. The considered object is \(\pi_1(\text{Diff} (M))\) for some 3-manifolds. Such a research has been already started by the second named author in [Trans. Am. Math. Soc. 264, 151-163 (1981; Zbl 0471.55004)] where he proved that if \(M\) is a connected sum of at least three closed aspherical 3-manifolds, then \(\pi_1(\text{Diff}(M))\) is not finitely generated. Here this is refined and pushed forward. The main result here reads as follows: Let \(M\) be a compact orientable 3-manifold which has no 2-sphere boundary components. (a) Suppose \(M\) is prime or is the connected sum of two irreducible summands, and that for each irreducible summand \(P\) of \(M\), \(\pi_1(\text{Diff} (P))\) is finitely generated. Then \(\pi_1 (\text{Diff} (M))\) is finitely generated. (b) Suppose \(M\) has at least three nonsimply connected prime summands, or is the connected sum of \(S^2 \times S^1\) with a prime summand with infinite fundamental group. Assume that for any irreducible summand with finite fundamental group, the fundamental group is isomorphic to a subgroup of SO(4) which acts freely on \(S^3\). Then \(\pi_1(\text{Diff} (M))\) has infinite rank as an abelian group. One may notice from the stated theorem that the main objective is to show whether \(\pi_1 (\text{Diff} (M)) \) is finitely or infinitely generated, but many intermediary results are proved and used which are hidden behind the main statement. Most part of the text occupies the proof of part (b) of the stated main theorem. The authors craftily use the configuration space constructed by [\textit{H. Hendriks} and \textit{F. Laudenbach}, Topology 23, 423-443 (1984; Zbl 0579.57009)] in order to find these isotopies. The isotopies are constructed on submanifolds of \(M\) which are homotopy equivalent to a wedge of two 2-spheres and a finite number of 1-spheres, while the integral linear independence is proven by employing obstruction-theoretic methods. The last section treats manifolds with 2-sphere boundary components where it is proved: Let \(N\) be a 3-manifold which is compact and orientable. The \(\pi_1(\text{Diff} (N\# D^3)\) is finitely generated if and only if one of the following three statements holds: (i) \(\pi_1(N)\) is finite and \(\pi_1 (\text{Diff} (N))\) is finitely generated; (ii) \({\mathcal P}(N)\) is prime and \(\pi_1(\text{Diff} (N))\) is finitely generated; (iii) \(N\simeq\mathbb{R}\mathbb{P}^3\#\mathbb{R}\mathbb{P}^3\) and \(\pi_1(\text{Diff} (N))\) is finitely generated, where \(\widehat N\) denotes the manifold obtained by filling in all 2-sphere boundary components of \(N\) with 3-balls, and \({\mathcal P}(N)\) denotes the Poincaré associate of \(N\). The case when \(M=S^2 \times S^1 \# P\) where \(\pi_1 (P)\) is finite is also treated. The obtained partial result in this case is: If \(P\) is a closed 3-manifold with finite fundamental group of order \(n\), then \(\pi_1 (\text{Diff} (S^2 \times S^1 \#P))\) contains a free abelian subgroup of rank \(n-1\).