The topological slice genus of satellite knots (Q2163646)

Let \(P(K)\) denote the satellite knot with pattern \(P\) and companion \(K\). A very well known result in Knot Theory, by \textit{H. Schubert} [Acta Math. 90, 131--286 (1953; Zbl 0051.40403)], is: \[ g_3(P(K))= g_3(P)+ \vert \omega \vert g_3(K) \] where \(\omega\) is the winding number of \(P\) in the solid torus and \(g_3(K)\) is the \(3\)-genus of \(K\), i.e, the minimal genus of a Seifert surface for \(K\) in \(S^3\). In \(4\)-dimensional topology, instead of considering spanning surfaces in \(S^3\) one considers surfaces sitting in 4-dimensional space. In this setting, the topological 4-genus of a knot \(K\), \(g^{top}_4(K)\), is the minimal genus of any surface \(F\) properly embedded in \(B^4\) with \(\partial F=K\) via a locally flat embedding. The \(\mathbb{Z}\)-slice genus of \(K\), \(g_{\mathbb{Z}}(K)\), adds to the previous definition the condition \(\pi_1(B^4\setminus F)\cong \mathbb{Z}\). The main theorem in this article states: ``For any pattern \(P\) and knot \(K\), \(g_{\mathbb{Z}}(P(K)) \leq g_{\mathbb{Z}}(P(U)) + g_{\mathbb{Z}}(K)\)'', where \(U\) is the unknot. The authors translate this topological problem to an algebraic one by identifying \(g_{\mathbb{Z}}\) with the algebraic genus \(g_{alg}\). The algebraic genus is defined as follows. Let \(K \subset S^3\) be a link with \(r\) components: \(g_{alg}(K)=\min \Big\{\frac{1}{2}(m-r+1)\mid \) there is a Seifert surface \(F\) for \(L\) with an \(m\times m\) Seifert matrix of the form \(\left[\begin{smallmatrix} B& * \\ * & * \end{smallmatrix}\right]\), where \(B\) is a \(2n\times 2n\) matrix such that \(\det(tB-B^T)=t^n \Big\}\). In this article the following inequality is proven related with the algebraic genus of a link \(K\): \[ g_{alg}(P(K))\leq g_{alg}(P(U))+\min \{\vert \omega \vert , 1\}g_{alg}(K). \] The proof of this assertion is purely in terms of Seifert matrices. The main theorem follows since when \(K\) is a knot then \(g_{\mathbb{Z}}=g_{alg}\), by a result of [\textit{P. Feller} and \textit{L. Lewark}, ``Balanced algebraic unknotting, linking forms, and surfaces in three- and four-space'', Preprint, \url{arXiv:1905.08305}]. The following conjecture is stated: For any pattern \(P\) and a knot \(K\), \(g_4^{top}(P(K))\leq g_4^{top}(P(U))+g_4^{top}(K)\). The authors discuss why the known lower bounds on the topological \(4\)-genus, namely Tristram-Levine signatures and Casson-Gordon signatures, are not capable to disprove the above conjecture.