Free 2-tangles from the point of view of c-Heegaard splittings (Q1928413)

A genus \(g\) Heegaard splitting of \(S^3\) is said to be a \((g,0)\)-bridge splitting of \(K\) if one of the handlebodies \(H\) entirely contains \(K\) and the exterior of \(K\) in \(H\) is a compression body; it is said to be a \((g,b)\)-bridge splitting with \(b>0\) if both handlebodies intersect \(K\) in a trivial set of properly embedded arcs. Note that we can obtain a \((g+b,0)\)-bridge splitting from a \((g,b)\)-bridge decomposition by drilling neighbourhoods of the arcs out of one of the handlebodies and adding them to the other. The Heegaard genus \(hg(K)\) is that of the exterior of \(K\). The meridional destabilization number \(md(K)\) was defined in [\textit{T. Saito}, Algebr. Geom. Topol. 11, No. 2, 1205--1242 (2011; Zbl 1221.57011)] to be the maximal number \(m\) such that \((S^3, K)\) admits an \((hg(K)-m,m)\)-bridge splitting. \(K \in {\mathcal K}_g^m\) means \(hg(K)=g\) and \(md(K)=m\). K. Morimoto studied composite knots with \((3,0)\)- or \((2,1)\)-bridge splittings. See, for example, [\textit{K. Morimoto}, Topology Appl. 64, 165--176 (1995; Zbl 0827.57002), ibid. 146-147, 149--158 (2005; Zbl 1086.57006) and J. Knot Theory Ramifications 10, No. 6, 823--840 (2001; Zbl 1002.57017)]. A \(2\)-tangle \((B, J)\) is said to be free if the exterior of the \(2\) arcs \(J\) in the \(3\)-ball \(B\) is a handlebody. The paper under review introduces the notion of a \(3\)-bridgelike tangle, which is a \(2\)-tangle \((B, J)\) such that there is a collar neighbourhood \(\partial B \times [0,1]\) intersecting \(J\) in four vertical arcs and a trivial arc and the complement is a \(3\)-tangle with a trivial set of three arcs. It is shown that, if a knot \(K\) has a \(2\)-tangle decomposition with one tangle being a \(3\)-bridgelike tangle and the other one a free tangle, then \(K\) admits a \((2,1)\)-bridge splitting. Moreover, if the free tangle has an unknotting arc component, then \(K\) admits a \((1,2)\)-bridge splitting. The proof uses the inverse operation of the untelescoping introduced and formulated in [\textit{A. Casson} and \textit{C. McA. Gordon}, Topology Appl. 27, 275--283 (1987; Zbl 0632.57010)], [\textit{M. Scharlemann} and \textit{A. Thompson}, Contemp. Math. 164, 231--238 (1994; Zbl 0818.57013)] and [\textit{J. Schultens}, Proc. Lond. Math. Soc., III. Ser. 67, No. 2, 425--448 (1993; Zbl 0789.57012)]. The above theorem together with a result in [\textit{C. McA. Gordon} and \textit{A. W. Reid}, J. Knot Theory Ramifications 4, No. 3, 389--409 (1995; Zbl 0841.57012)] implies that, if a knot \(K\) has an essential free \(2\)-tangle decomposition with one of the tangles being a \(3\)-bridgelike tangle, then \(hg(K)=3\) and \(md(K) \geq 1\). Moreover, if the other tangle has an unknotted arc component, then \(md(K) \geq 2\). As applications, the knot \(8_{16}\) is in \({\mathcal K}_3^3\), and Morimoto's knot \(K_M\) is in \({\mathcal K}_3^2\). It is also shown that, if a knot \(K'\) in \(S^3\) has a \((g,2)\)-bridge splitting and a knot \(K\) in \(S^3\) has an essential free \(2\)-tangle decomposition with one (resp. both) of the tangles having an unknotted arc component, then the composite knot \(K \sharp K'\) admits a \((g+3, 0)\)-bridge splitting (resp. a \((g+2, 1)\)-bridge splitting). As a corollary, if a knot \(K\) has an essential free \(2\)-tangle decomposition with one of the tangles being \(3\)-bridgelike and both having an unknotted arc component, then \(K \sharp K\) admits a \((3,1)\)-bridge splitting. This implies \(hg(K_M \sharp K_M)=4\) and \(md(K_M \sharp K_M) \geq 1\).