On \(\pi\)-hyperbolic knots and branched coverings (Q1125563)

The question has been around for some time whether a hyperbolic knot is determined by its 2-fold and 3-fold cyclic branched coverings (it occurs explicitly as problem 1.75 B in the Kirby list). In the present paper, this question is answered positively for \(\pi\)-hyperbolic or, more generally, Conway irreducible hyperbolic knots. Recall that a knot \(K\) in the 3-sphere is \(\pi\)-hyperbolic if its 2-fold branched covering is a hyperbolic manifold and the covering involution is an isometry; equivalently, the 3-orbifold \((S^3,K)\), with singular set \(K\) of branching index two, is hyperbolic. The \(\pi\)-hyperbolic knots form the generic class of hyperbolic knots: every \(\pi\)-hyperbolic knot is hyperbolic (in particular, its complement is atoroidal) and Conway-irreducible (there are no essential Conway spheres meeting the knot in exactly four points); conversely, for a Conway irreducible hyperbolic knot \(K\), the orbifold \((S^3,K)\) contains no essential spherical or Euclidean 2-suborbifolds so, by the orbifold geometrization theorem, it is geometric. The main result of the present paper states that, for any \(n\geq 3\), a Conway irreducible hyperbolic knot (and in particular any \(\pi\)-hyperbolic knot) is determined by its 2-fold and \(n\)-fold cyclic branched coverings (in the present paper, the problem remains open for arbitrary hyperbolic knots, but the author has recently constructed examples of different hyperbolic knots with the same 2-fold and \(n\)-fold cyclic branched coverings; the situation was known before for branching orders greater than two). It is also proved that a hyperbolic knot which is not determined by its \(m\)-fold and \(n\)-fold cyclic branched coverings, \(m>n>2\), has genus \((n-1)(m-1)/2\). The proofs are based on equivariant Seifert surfaces for knots, the positive solution of the Smith conjecture and algebraic considerations in the finite isometry group of a hyperbolic manifold occuring as a 2-fold branched covering.