Symplectically knotted codimension-zero embeddings of domains in \(\mathbb{R}^4\) (Q2280534)

Consider inclusions of sympectic domains \(A\subset U\subset\mathbb R^4\). A natural question is whether a given symplectic embedding \(\phi\colon A\hookrightarrow U\) is equivalent to the inclusion. The embedding \(\phi\) is unknotted if there exists a symplectomorphism \(\Psi\colon U\to U\) such that \(\Psi(A)=\phi(A)\). It is knotted if it is not unknotted. The authors use constructions of symplectic embeddings \(B\hookrightarrow U\) such that their restrictions to subsets \(A \subset B \cap U\) lead to embeddings \(A\hookrightarrow U\) that are distinct from the inclusion. It is shown that certain toric domains \(X\) in \(\mathbb R^4=\mathbb C^2\) admit symplectic embeddings \(\phi\) into dilates of themselves \(U=\alpha X^\circ\) that are knotted. For example one can take a convex toric domain \(X\), such that \(X\) is a subset of a polydisk \(P(c, c)\) and \(X\) does not contain a ball \(B^4(c)\) of the symplectic capacity \(c\), for a certain \(c0\) (the space of symplectic embeddings from one 4-dimensional ball to another is always connected [\textit{D. McDuff}, Topology 30, No. 3, 409--421 (1991; Zbl 0731.53035)] that implies that that symplectic embeddings \(B^4(c) \hookrightarrow \alpha B^4(c)^\circ\) can never be knotted). Here, a toric domain \(X\subset \mathbb C^2\) is the inverse image \(\mu^{-1}(\Omega)\) of a domain \(\Omega\subset [0,\infty)^2\) under the moment mapping \(\mu(z_1,z_2)=(\pi\vert z_1\vert^2,\pi\vert z_2\vert^2)\). The toric domain is convex if \( \hat\Omega=\{(x_1,x_2)\in\mathbb R^2\,\vert\,(\vert x_1\vert,\vert x_2\vert)\in\Omega\} \) is a convex domain in \(\mathbb R^2\). In particular, the polydisk \(P(a_1,a_2)\) is a convex toric domain with \(\Omega=[0,a_1]\times [0,a_2]\). The fact that the embeddings are knotted is proven using filtered positive \(S^1\)-equivariant symplectic homology.