Unsolvable problems about higher-dimensional knots and related groups (Q993658)

This paper gives a clear and detailed account of the constraints on algorithmic solutions to questions of classification in higher-dimensional knot theory. In particular, it highlights the interest in the classical case (in which there is at least an algorithm for recognizing the unknot), and the case of knotted 2-spheres in \(S^4\) (which the authors expect shall be as in higher dimensions). The issues considered may be novel for many geometric topologists, but they reflect the increasing influence of computing (i.e., algorithmic thinking) on the practice of mathematics. Although the results were foreshadowed by earlier work of Nabutovsky and Weinberger on presentations of the trivial \(n\)-knot, they are substantially new. The algebraic constructions rely in the main on delicate applications of the familiar notion of HNN extension. The main result considers the classes of groups \(\mathcal{K}_0\), \(\mathcal{K}_1\), \(\mathcal{K}_2\), \(\mathcal{K}_3\) and \(\mathcal{G}\), corresponding to the infinite cyclic group, \(n\)-knot groups for \(n=1,2,3\), and all finitely presentable groups. These classes are nested and distinct. It is shown that if \(\mathcal{A}\) contains all 3-knot groups and \(\mathcal{B}\) is another of these classes properly contained in \(\mathcal{A}\) then there is no algorithm to decide, given a finite presentation of a group \(G\) in \(\mathcal{A}\), whether \(G\) is in \(\mathcal{B}\). Other issues considered include the impossibility of deciding whether the second homology of a finitely presented group is trivial, various incomputability results for Whitehead torsion and surgery obstruction groups, and the undecidability of the knotting problem for \(n\)-spheres in \(S^{n+2}\).