The bounding problem for infra-solvmanifolds (Q260595): Difference between revisions

In the late 70's many compact manifolds were tested in cobordism theory whether they are cobordant to zero or not. Among them, it was a striking result that flat manifolds are boundaries proved by Hamrick and Royster. In the early 1980s, M. Gromov introduced the notion of almost flat manifold \(M\) which is a convergence of Riemannian metrics to the flat metric over \(M\). Naturally the next target of the bounding problem became these almost flat manifolds. On the other hand, it is known that almost flat manifold are diffeomorphic to infranilmanifolds. Let \(G\) be a simply connected Lie group and \(\text{ Aut}(G)\) be the automorphism group. Let \(K\) be a maximal compact subgroup of \(\text{ Aut}(G)\). A semidirect product \(E(G)=G\rtimes K\) acts on \(G\) by \(\displaystyle (g,\alpha)x=g\alpha (x)\) \((\forall\, x\in G,\forall\, \alpha\in K)\). Then \(\text{ E}(G)\) is said to be the \textit{euclidean group} of \(G\) due to J. Wolf. Any discrete cocompact (respectively torsionfree) subgroup \(\Gamma\leq \text{ E}(G)\) acts properly discontinuously (respectively freely) on \(G\) which is called a (respectively torsionfree) crystallographic group. The quotient \(\Gamma\backslash G=\Gamma\backslash \text{ E}(G)/K\) is said to be an infra-homogeneous orbifold (respectively manifold). In particular when \(G\) is a vector space (respectively simply connected nilpotent group), \(\Gamma\backslash G\) is said to be a flat manifold (respectively an almost flat manifold, equivalently infranilmanifold as above). So the bounding problem has been shifted to infra-homogeneous manifolds along this direction. Since \(\Gamma\backslash G\) is a locally homogeneous manifold, it is natural to ask whether homogeneous manifolds are boundaries. It is worthwhile to mention that solvmanifolds (compact homogeneous manifold which is a quotient of a simply connected solvable Lie group by a discrete subgroup) are boundaries proved by Auslander and others. It is known that a flat (respectively almost flat) manifold \(\Gamma\backslash G\) is finitely covered by a torus (respectively nilmanifold), thus \(\Gamma\) is a finite group extension of a free abelian group (respectively finitely generated nilpotent group) which is called a Bieberbach group (respectively almost Bieberbach group.) However, this property does not hold in general for infrasolvmanifolds. It is interesting to know when an infrasolvamnifold is a boundary. Let \(\Gamma\leq \text{ E}(G)\) be a discrete subgroup. The image of \(\Gamma\) by the canonical projection from \(\text{ E}(G)\) to \(K\) is called the holonomy group of \(\Gamma\). So the holonomy is finite for Bieberbach groups or almost Bieberbach groups and the holonomy for infrasolvmanifolds is not necessarily finite (but simply the closure is compact). In [\textit{F. T. Farrell} and \textit{S. Zdravkovska}, Mich. Math. J. 30, 199--208 (1983; Zbl 0543.53037)] it was shown that an infranilmanifold is a boundary when the holonomy group is cyclic of order \(m\) or \(2m\) (\(m\) odd). K. B. Lee and F. Raymond studied similar problems from the viewpoint of group actions in the 1980's. In fact an infranilmanifold \(M\) admits an injective \(S^1\)-action provided \(\text{ rank}\, Z(\pi_1(M))\geq 1\). In case it acts freely, then \(M\) is a boundary of \(W=D^2\times M/S^1\). Similarly if it admits a free involution \(T\), \(M\) bounds \(W=D^1\times M/T\) originally due to R. Lee. It is difficult to find such an action in general. In the early 2000s, it was a question whether in connection with \textit{hyperbolic geometry} a Seifert manifold (respectively Heisenberg infranilmanifold) can be a boundary of real hyperbolic manifold (respectively a complex hyperbolic manifold). Most of the cases were negatively answered. Recently in 2015 there was a progress that J. Davis and F. Fang showed that every infranilmanifold \(\Gamma\backslash N\) with cyclic or generalized quaternion holonomy is a boundary. (S. Upadhyay has also proved this assuming that \(N\) is \(2\)-step nilpotent in 2000.) The author has generalized this result to the case when the holonomy is a dihedral group or product of cyclic groups. Moreover the author has proved a certain infrasolvmanifold bounds using the algebraic criterion. An infrasolvmanifold is of type (E) if \(\exp:\mathfrak g\rightarrow G\) is onto and the center \(Z(G)\) is nontrivial. Any infrasolvmanifold of type (E) with certain holonomy conditions is a boundary (Compare Theorem 2.10.) It is not clear whether there is a plentiful of examples of infrasolvmanifolds of Theorem 2.10.