Knot and braid invariants from contact homology. II (Q2388876)

This paper gives an alternative and equivalent interpretation of the 0-dimensional homology group invariants introduced in Part I of this paper [Geom. Topol. 9, 247--297 (2005; Zbl 1111.57011)]. The 0th contact homology group for a knot introduced there is shown in this paper to be equivalent (roughly speaking) to a ring of homotopy classes of paths beginning and ending on the boundary of a tubular neighborhood of the knot. This structure (the ``cord ring'') is in turn shown (in the appendix, which is joint work with Siddhartha Gadgil) to be determined by the double coset space of the group of the knot with respect to its peripheral subgroup. The notions are seen to be more general in that they may be extended to knots in an arbitrary 3-manifold as well as to imbeddings of \(S^2\) in \(S^4\) (the latter in the appendix where it is shown that many spun knots are distinguishable via their cord rings from the unknotted \(S^2\) in \(S^4\) ). Several methods are demonstrated to calculate the cord ring from a knot diagram. The cord rings are presented explicitly for 2-bridge knots. By using the cord ring, the author obtains a lower bound on the number of minimal chords of a knot in terms of the number of invariant factors of the 1-dimensional homology of the double cover of \(S^3\) branched over the knot.