Classification of links up to 0-solvability (Q2161359)

The knot concordance group is the set of concordance classes of oriented knots with the connected sum as the group operation. The knot concordance group can be generalized to \(m\)-component string links with the stacking operation. The concordance group \(\mathcal{C}^m\) of \(m\)-component string links admits the solvable filtration \(\mathcal{C}^m \supset \mathcal{F}_0 ^m \supset \mathcal{F}_{0.5}^m\supset \mathcal{F}_1^m \supset\mathcal{F}^m _{1.5} \supset \cdots\supset \{0\}\). As a convention the author denotes by \(\mathcal{F}^m_{-0.5}\) the concordance class of string links with vanishing pairwise linking numbers. This filtration measures how far a given link is from being slice, in the sense that the deeper a link is situated in the solvable filtration, the closer it is to being slice. The author studies \(m\)-component links with the linking numbers between any two components being zero. The set of such links is denoted by \(\mathcal{L}^m\). The author introduces the \(0\)-solve relation \(\sim_0\) on \(\mathcal{L}^m\) and shows that this relation is an equivalence relation. The main result of this paper provides various equivalent conditions regarding the \(0\)-solve equivalence. More precisely, two elements \(L=L_1\cup\cdots\cup L_m\) and \(K=K_1\cup\cdots\cup K_m\) of \(\mathcal{L}^m\) are 0-solve equivalent if and only if \(L\) and \(M\) are band-pass equivalent. In terms of various link invariants, this condition is also equivalent to \(\operatorname{Arf}(L_i)=\operatorname{Arf}(K_i)\), \(\overline{\mu}_L(ijk)=\overline{\mu}_K (ijk)\) and \(\overline{\mu}_L(iijj)=\overline{\mu}_K (iijj)\pmod{2}\) for \(i,j,k\in\{1,2,\cdots,m\}\), where \(\overline{\mu}(ijk)\) and \(\overline{\mu}(iijj)\) are Milnor's invariants. As a corollary, the author shows that the quotient group \(\mathcal{L}^m/{\sim_0}\) is isomorphic to \(\mathcal{F}^m _{-0.5}/\mathcal{F}^m_0\) which is also isomorphic to \(\mathbb{Z}_2 ^m \oplus \mathbb{Z}^{\binom{m}{3}}\oplus \mathbb{Z}_2 ^{\binom{m}{2}}\). In the proof, the author shows that if \(L_1\cup \cdots \cup L_m\) and \(K_1\cup\cdots \cup K_m\) are 0-solve equivalent, then \(L_{i_1}\cup \cdots \cup L_{i_n}\) and \(K_{i_1}\cup\cdots \cup K_{i_n}\) are also 0-solve equivalent for all choices of \(1\le i_1<\cdots <i_n \le m\). Thanks to this observation one only needs to handle \(2\)-component links or knots. Together with the Thom-Pontryagin construction, one can show that \(L\sim_0 K\) implies the equalities in Arf and Milnor invariants. To show that the conditions on link invariants imply the band-pass equivalence, the author uses an argument by \textit{K. Taniyama} and \textit{A. Yasuhara} [Topology Appl. 122, No. 3, 501--529 (2002; Zbl 1001.57011)]. Finally, by explicit handle calculus, the author shows that a band-pass move can be realized as 0-framed surgery on the Hopf link and using this, shows that band-pass equivalence implies 0-solve equivalence.