Generating disjoint incompressible surfaces (Q624394)

In a compact, orientable \(3\)-manifold \(M\), how many connected components can a collection of disjoint, incompressible, non-parallel, non-boundary-parallel surfaces contain? It is known that such a collection cannot contain infinitely many surfaces [\textit{H. N. Howards}, Topology Appl. 99, No.~1, 117--122 (1999; Zbl 0937.57018)]. The main result in the paper under review is that there is a collection containing an arbitrary large number of surfaces if \(M\) has at least one boundary component of genus greater than or equal to two. This contradicts Theorem III.24 in [\textit{W. Jaco}, Lectures on three-manifold topology. Regional Conference Series in Mathematics, 43. Providence, R.I.: American Mathematical Society (AMS) (1980; Zbl 0433.57001)], which has been proven correctly for a collection of annuli in the book. The first counter example of this theorem was given in [\textit{W. Sherman}, On the Kneser-Haken finiteness theorem: A sharpness result, PhD thesis, University of California, Los Angeles (1992)]. The number of surfaces in such a collection is bounded if the boundary \(\partial M\) consists of spheres and tori by \textit{H. Kneser}'s theory of normal surfaces [Jahresbericht D. M. V. 38, 248--260 (1929; JFM 55.0311.03)], Jaco's theorem and \textit{B. Freedman} and \textit{M. H. Freedman}'s result [Topology 37, No.~1, 133--147 (1998; Zbl 0896.57012)]. The surfaces realizing the main result must have larger and larger negative Euler Characteristic because the number of surfaces in a collection is bounded if all the surfaces are required to have Euler Characteristic at least \(n\) by Freedman and Freedman's result. An affirmative answer to Question III.16 in Jaco's book is also given, i.e., it is shown that any \(3\)-manifold with a boundary component of genus \(2\) or higher contains a separating incompressible surface of arbitrary high genus. This implies that the free group on two generators may be split into a free product with amalgamation over two arbitrarily large free groups. The collection of \(n\) surfaces realizing the main result is constructed from \(n\) parallel copies of a non-separating surface representing a non-trivial element of Ker(\(H_1(\partial M) \rightarrow H_1(M)\)) by iterated tunneling operations. The tunneling operations must be performed very carefully because M. Freedman's conjecture on tunneling operations expected to yield an incompressible surface (Conjecture 3.1 in the paper under review) remains unsolved.