Pretzel links, mutation, and the slice-ribbon conjecture (Q2053651)

Consider the four-stranded pretzel link \( P (p, q, r, s) \) (where \( p, q, r, s\) denotes the number of half twists) which only depends on \( (p, q, r, s) \) up to cyclic permutation and overall reversal. \( P (p, q, r, s) \) has two components if and only if none or exactly two parameters are even. This paper is concerned with \( P (p, q, r, s) \) with two components. The following classification is complete. (1) At least three of \( p, q, r, s\) are distinct: not slice (2) \( P (p, -p, q, -q) \): ribbon link (3) \( P (p, p, -p, -p) \): ribbon link (4) \( P (\pm 1, p, \mp 1, -p) \): ribbon link (5) \( P (0, p, 0, -p) \): not slice (6) \( P (p, q, -p, -q) \), \( 1<p<q \): not slice The content of this paper is to prove the nontrivial cases (1) and (6). Only when \( p, q, r, s\) are all odd it is difficult to judge sliceness. The authors set that the slice-ribbon conjecture holds for such links as their main theorem. The main theoretic result is Theorem 2.4, as a slice obstruction. Section 2: Case (1) is proved by considering whether the 2-fold branched cover bounds a manifold with only nontrivial Betti number \( b_1 = 1\). Section 3: Case (6) is proved by construction of a manifold bounded by the 3-fold branched cover using Kirby diagrams, and then calculation shows that a morphism of integral lattices asserted in Theorem 2.4 cannot exist. This paper uses definitions and many results from [\textit{P. Aceto}, Algebr. Geom. Topol. 20, No. 3, 1073--1126 (2020; Zbl 1446.57010)] without explanation, thus it is not self-contained. The authors use three paragraphs in the introduction to emphasize that \( P (p, -p, q, -q) \) and \( P (p, q, -p, -q) \) are positive mutants, and it is difficult to find a pair of positive mutant links such that one is slice while the other one is not. However, as they explain in the initial paragraph of Section 3, it is easy to see that \( P (3, -3, 2, -2) \) is slice while \( P (3, 2, -3, -2) \) is not slice.