Alternating links and definite surfaces (Q2403209)

In knot theory, alternating knots and links form one of the most important classes. Such a knot or link is defined as one admitting an alternating diagram, whose crossings alternate over and under along each component. It was a long-standing problem to find an intrinsic characterization of alternating knots and links, which is independent from diagrams. The present paper under review finally gives an answer to this problem. For an alternating diagram of a non-split alternating link in the \(3\)-sphere, there are two checkerboard surfaces, which may be non-orientable. For such a surface, there is a natural pairing on the first homology group introduced by \textit{C. McA. Gordon} and \textit{R. A. Litherland} [Invent. Math. 47, 53--69 (1978; Zbl 0391.57004)]. It is easy to see that the pairings for checkerboard surfaces are positive definite and negative definite, respectively. The answer to the question above is that the existence of such two surfaces implies conversely that the link is alternating. In fact, the ambient manifold is generalized to a \(\mathbb{Z}_2\)-homology sphere. Here is a concise statement of the characterization. Let \(L\) be a link in a \(\mathbb{Z}_2\)-homology sphere with irreducible complement. If it bounds both a positive definite surface and a negative definite surface, then \(L\) is a non-split alternating link in the \(3\)-sphere. Moreover, \(L\) has an alternating diagram whose checkerboard surfaces are isotopic relative boundary to the two given definite surfaces. As a corollary, this characterization gives a new geometric proof of Tait's conjecture, which claims that any two connected reduced alternating diagrams of a link have the same crossing number and writhe. This has been already solved by using Jones polynomials or Kauffman polynomials. Another corollary is that a Seifert surface for a special alternating link has minimal genus if and only if it is obtained by applying Seifert's algorithm to a special alternating diagram. There is an appendix by A. Juhász and M. Lackenby, which gives an algorithmic detection of prime alternating knots. Independently, \textit{J. Howie} [Geom. Topol. 21, No. 4, 2353--2371 (2017; Zbl 1375.57011)] has given a characterization of alternating knots in the \(3\)-sphere in terms of a pair of spanning surfaces. It is equivalent to the condition above.