Boundary-reducing surgeries and bridge number (Q2283338)

Suppose \(K\) is a tame knot in a three-manifold \(M\). A cosmetic surgery is a nontrivial Dehn surgery on \(K\) that yields \(M\) again. The paper under review treats cosmetic surgeries on knots in handlebodies. It was conjectured that, in analogy with knots in a solid torus (the genus-one handlebody), if a knot in a handlebody admits a cosmetic surgery, then that knot must be 1-bridge -- that is, admit an isotopy to a knot whose intersection with the interior of the handlebody is a single arc. The second named author has constructed counterexamples to this conjecture for handlebodies of genus 2. The paper under review strengthens this result in two ways. First, the authors show that the previous counterexamples include knots of arbitrarily large bridge number. Second, the authors construct counterexamples (of arbitrarily large bridge number) in handlebodies of arbitrary genus. Cosmetic surgeries on handlebodies are special cases of boundary-reducible surgeries. The paper under review also treats these surgeries, constructing, for all handlebodies of finite genus \(g\), all positive integers \(N\), and all relatively prime pairs of naturals \((p,q)\), an infinite family of hyperbolic knots in the given handlebody with the same bridge number at least \(N\), and each admitting a surgery to a Seifert-fibered space over a disc with two exceptional fibers and with \(g-1\) one-handles attached to its boundary. The authors also raise two interesting questions about knots in handlebodies, and provide evidence for a positive resolution to one of them. The first question is whether or not an irreducible, boundary-irreducible, atoroidal, anannular knot with more than one cosmetic surgery in a handlebody must be one-bridge. The second question depends upon the surgery dual knot, which is the core of the Dehn surgery's attached solid torus. If this is a cosmetic surgery, this is again a knot in a handlebody admitting a cosmetic surgery. The authors provide upper bounds on the bridge numbers of the surgery duals of enough of the knots constructed before, to provide an infinite family of knots in handlebodies with cosmetic surgeries and arbitrarily large bridge number, but whose surgery duals do not have arbitrarily large bridge number. So they pose the second question, of whether or not there is a family of irreducible, boundary reducible, atoroidal, and anannular knots in a handlebody having cosmetic surgeries and arbitrarily large bridge number, but whose surgery duals also have arbitrarily large bridge number.