Sobolev mappings: Lipschitz density is not a bi-Lipschitz invariant of the target (Q2370118)

If \(n\in \mathbb N\), \(n\geq 2\), then there is a compact connected set \(X\subset \mathbb R^{n+2}\) and a bi-Lipschitz homeomorphism \(\Phi : \mathbb R^{n+2} \rightarrow \mathbb R^{n+2}\) such that, for any closed \(n\)-dimensional manifold \(M\), smooth mappings \(C^{\infty}(M,X)\) are dense in the Sobolev class \(W^{1,n} (M,X)\), but Lipschitz mappings Lip\((M,Y)\) are not dense in \(W^{1,n} (M,Y)\), where \(Y = \Phi(X)\). This is one of the main results of the paper. Note also that the class of Sobolev mappings \(W^{1,n}(M,X)\) is defined by \[ W^{1,n} (M,X) :=\{u\in W^{1,n} (M,\mathbb R^{n+2}) : u (x) \in X \;\text{a.e.}\} \] and is equipped with the metric of the Sobolev norm \(W^{1,n}\), i.e., \(\rho(u,v) := \| u-v\| _{1,n}\) for all \(u,v\in W^{1,n}(M,X)\).