A ribbon obstruction and derivatives of knots (Q2089813)

A knot in \(S^3\) is smoothly slice if it bounds a smoothly embedded disk in the 4-ball \(B^4\) and is doubly slice if it is realized as the intersection of a trivial 2-knot in \(S^4\) and the equatorial \(S^3\subset S^4\). A knot is algebraically slice (algebraically doubly slice) if a Seifert form of the knot is metabolic (hyperbolic, respectively). Being algebraically (doubly) slice is weaker than being (doubly) slice. To quantify how close a given knot is to being slice, we use the solvable filtration of the knot concordance group. Similarly, the doubly solvable filtration \(\{\mathcal{F}_{m,n}\}_{m,n\in \frac{1}{2}{\mathbb{Z}_{\ge 0}}}\) measures how close a knot is to being doubly slice. This paper provides examples suggesting that being algebraically slice or algebraically doubly slice can be very far from being slice or being doubly slice. More precisely, the authors find a ribbon, algebraically doubly slice knot that is not in \(\mathcal{F}_{1,1}\), an algebraically doubly slice knot that is not in \(\mathcal{F}_{0.5,1}\), and an algebraically slice knot that is not homotopy ribbon (1)-solvable. The last example has the interesting property that any derivative of any Seifert surface has non-vanishing triple linking number. The main tool is the obstruction \(\psi(K,P)\) sitting inside \(H_3(BG;\mathbb{Z})\), where \(BG\) is the classifying space of a group \(G\) depending on the knot \(K\) and a choice of a Lagrangian \(P\) for the rational Blanchfield pairing. This obstruction vanishes when the knot is \(\mathbb{Z}[\mathbb{Z}]\)-homology ribbon (1)-solvable. In particular, for a knot \(K\) in \(\mathcal{F}_{0.5,1}\), we have that \(\psi(K,P)=0\) for some Lagrangian \(P\); for a knot \(K\) in \(\mathcal{F}_{1,1}\), there are two Lagrangians \(P_1\) and \(P_2\) such that \(\psi(K,P_1)=\psi(K,P_2)=0\). As explicit examples, the authors investigate genus 3 algebraically doubly slice knots \(K_1\), and \(K_2\). By relating the obstruction \(\psi\) with the triple linking numbers of a derivative of a genus 3 knot, the authors can show that \(\psi(K_1,P)\ne 0\) for all but exactly one Lagrangian \(P\) and \(\psi(K_2, P)\ne 0\) for any Lagrangian \(P\), and consequently, \(K_1\notin \mathcal{F}_{1,1}\) and \(K_2\notin\mathcal{F}_{0.5,1}\).