Infinite order corks via handle diagrams (Q2409810)

A cork in the context of 4-manifold theory is a smooth, contractible, compact 4-manifold \(W\) with boundary, together with a diffeomorphism \(f : \partial W \to \partial W\) that does not extend over \(W\). If one can embed \(W \hookrightarrow X\) in a compact 4-manifold \(X\) such that the manifolds \[ X_k := X \setminus W \cup_{f^k} W; \quad k \in \mathbb{Z} \] are pairwise non-diffeomorphic, then \((W,f)\) is an infinite order cork. The first example of infinite order corks were constructed in [Geom. Topol. 21, No. 4, 2475--2484 (2017; Zbl 1420.57058)] by the author. In this paper the proof is re-worked using handle diagrams, giving an alternative proof. The construction starts with the exterior \(E(r,s)\) of a ``double twist knot'' \(K(r,s)\), for suitable \(r,s\). Take \(E(r,s) \times I\) and add a 0-framed 2-handle to a meridian of \(K(r,s)\) in \(E(r,s) \times \{1\}\), to obtain a contractible compact 4-manifold \(W\). The diffeomorphism \(f : \partial W \to \partial W\) arises from torus twisting along an \(I\) thickening of \(\partial E(r,s) \times \{0\}\). A torus twist is a Dehn twist on \(S^1 \times I\) crossed with the identity on \(S^1\) to obtain a diffeomorphism of \(S^1 \times S^1 \times I\). Apply this to a thickening \(\partial E(r,s) \times I \times \{0\} \subset \partial W\) and extend to the rest of \(\partial W\) by the identity. This constructs the pair \((W,f)\) considered in the paper. The fact that this cork is infinite order is proven by embedding the cork in \(E(n) \#^N \overline{\mathbb{CP}^2}\) for suitable \(N\), where \(E(n)\) is the elliptic surface obtained from \(n\)-fold fibre sum on \(E(1) = \mathbb{CP}^2 \#^9 \overline{\mathbb{CP}^2}\) together with an elliptic fibration structure \(E(1) \to S^2\). It is shown through modifications on handle diagrams that cutting and pasting using \(f^k\) yields \(E(n)_{K(k)} \#^N \overline{\mathbb{CP}^2}\), where \(E(n)_{K(k)}\) is the result of Fintushel-Stern knot surgery on \(E(n)\) using \(K(k)\). These manifolds \(E(n)_{K(k)}\) are mutually non-diffeomorphic by Seiberg-Witten invariants. The proof is given in Sections 2 and 3. Section 4 studies when torus twisting produces diffeomorphisms of \(\partial W\) such that cutting and pasting, in any 4-manifold \(X\), does not change the diffeomorphism type of \(X\). The paper closes with a comparison to a recent paper of \textit{S. Akbulut} [in: Proceedings of the 23rd Gökova geometry-topology conference, Gökova, Turkey, May 30 -- June 3, 2016. Somerville, MA: International Press; Gökova: Gökova Geometry-Topology Conferences (GGT). 151--157 (2017; Zbl 1378.32007)]. Akbulut gave an alternative description of Gompf's construction using so-called \(\delta\)-moves. Section 5 shows that \(\delta\)-moves are in many situations, such as for irreducible 3-manifolds, equivalent to torus twists, so do not offer as much novelty as one might have hoped. The question of the well-definedness of \(\delta\)-moves is addressed. Curiously, a ``careful definition'' of \(\delta\)-moves is promised, but such a definition of a self-diffeomorphism of the 3-manifold \(M\) does not appear.