Distinguished waves and slopes in genus two

Abstract

If R is a nonseparating simple closed curve on the boundary of a genus two handlebody H and H[R] has incompressible boundary, then there exists a unique arc omega in bdry(H), meeting R only in its endpoints, such that, omega is isotopic in bdry(H), keeping its endpoints on R, to a nontrivial wave based at R in each Heegaard diagram of R on bdry(H) which has no cut-vertex. Then surgery on R along its "distinguished-wave" omega yields a pair of simple closed curves, say m_1 and m_2, in bdry(H), each representing an "omega-determined-slope" m on bdry(H[R]), that depends only on R and H. A few consequences: 1) Only Dehn filling of H[R] at slope m can yield S^3, (S^1 X S^2) # L(p,q), or S^1 X S^2. So H[R] embeds in at most one of S^3, (S^1 X S^2) # L(p,q), or S^1 X S^2. And, if such an embedding exists, it is unique. 2) Theta curves arising from unknotting tunnels of tunnel-number-one knots in S^3, S^1 X S^2, or (S^1 X S^2) # L(p,q), have canonical component knots. 3) One can recognize (1,1) tunnels of (1,1) knots in S^3 or S^1 X S^2. 4) Algorithms for recognizing genus two Heegaard diagrams of S^3, S^1 X S^2, or (S^1 X S^2) # L(p,q) that use waves can be streamlined. 5) Efficient procedures for computing the depth of an unknotting tunnel of a knot in S^3, S^1 X S^2, or (S^1 X S^2) # L(p,q) exist. Finally, if H[R_1] is homeomorphic to H[R_2], but (H,R_1) and (H,R_2) are not homeomorphic, then the omega-determined slopes on bdry(H[R_1]) and bdry(H[R_2]) may differ. However, computation suggests that, if 'R' is a set of simple closed curves on bdry(H) such that R_1 in 'R' and R_2 in 'R' means H[R_1] is homeomorphic to H[R_2], then at most two distinct slopes appear as omega-determined slopes for curves in 'R', and that, if such distinct omega-determined slopes exist, they are never more than distance one apart.