Multivariate signatures of iterated torus links (Q2234384)

Given a component of a link, a cabling operation creates a new link by taking a solid torus neighbourhood of the component and putting a torus knot into that solid torus. Iterated torus links can be created by repeated cabling operations beginning from the unknot. In the case that the link is a knot, the Levine-Tristram signatures (which are defined in terms of a Seifert matrix) can be fully calculated using two existing results: Hirzebruch's formula gives the signatures for torus knots, and Litherland's formula describes the result of cabling operations. For multicomponent iterated torus links there is a multivariate generalization of the Levine-Tristram signatures. Work of \textit{A. Degtyarev} et al. [Int. Math. Res. Not. 2017, No. 8, 2249--2283 (2017; Zbl 1405.57011)] gives an analogue of Litherland's formula for this context. This article provides the equivalent of Hirzebruch's formula by calculating the multivariate signatures for suitable generalizations of torus links. This is done by reducing the task of finding the multivariate signatures of any Seifert link to calculating (univariate) Levine-Tristram signatures, which are in turn given by work of \textit{W. D. Neumann} [Adv. Stud. Pure Math. 8, 349--361 (1987; Zbl 0652.32011)]. The proof works by viewing a (suitably oriented) Seifert link as built up by splicing (a more general operation than cabling). A splice diagram is a decorated tree (or forest) that encodes a sequence of splices. The key observation is that the same Seifert link can be broken down in a number of different ways.