On a certain numerical invariant of link types

Kunio Murasugi

Transactions of the American Mathematical Society · 1965 · 371 citations · 10 references

DOIFull text

Open access

Concepts

Abstract

KUNIO MURASUGI1. Introduction.Let / be an (oriented) tame link^) of multiplicity p in a 3-sphere S3, and let Lbe a diagram, i.e. the image under a regular projection of S3 into S2.To L we can associate an integral square matrix M, called the matrix of a link / ([13], [14], also see 3).The quadratic form f(xy,x2,---,xn) associated to the symmetric matrix M + M' induces some invariants of the original link type, where M' denotes the transposed matrix of M [19], [22].In this paper, we shall especially consider the signature a(f) of the quadratic form/(xx, ,x") associated to M + M'.The invariance of o(f) was proved by Trotter for p = 1 [22].For the general case, it is not so difficult to prove that it is also an invariant of the link type (Theorem 3.1).In 3, we shall define the matrix of a link for the convenience of the reader.In 5 we shall show that for any alternating link the signature is calculated immediately from the matrix.Then, from the properties of the matrix of a special alternating knot, it follows: Theorem 5.5.Any special alternating knot is not amphicheiral unless it is a trivial knot. 6-8 are concerned with slice knots or links, which were first defined by Fox and Milnor [4].In determining the signature of a link(2) of this kind, the nullity of the matrix plays an important role.The nullity is determined easily from the Alexander matrix [3, II] rather than the matrix of a link.This is the reason to establish some relations between them in 4.In 8, the concept of a slice knot is generalized to the case of links, and a slice link in various senses will be defined.Cf. [5].In particular, we obtain the following necessary condition for a knot to be a slice knot.Theorem 8.3.If a knot is a slice knot, then the signature is always zero.

References

10