Notes about math, research, and more.
Let \((X_{n})_{n \in \mathbb{N}}\) be a sequence of finite graphs with corresponding standard graph metrics \((d_{n})_{n \in \mathbb{N}}\) and \((X, d)\) be the coarse disjoint union of \(\{ (X_{n}, d_{n})\}_{n \in \mathbb{N}}\). There exists a set of cut metrics \((d_{n,w})_{n \in \mathbb{N}}\) on \((X_{n})_{n \in \mathbb{N}}\) such that the identity function from \((X,d)\) to the coarse disjoint union \((X, d_{w})\) of \(\{(X_{n}, d_{n,w})\}_{n \in \mathbb{N}}\) is a coarse embedding if and only if there exists a coarse disjoint union, \((C,d_{C})\), of a sequence of hypercubes, \((C_{n})_{n \in \mathbb{N}}\), each equipped with the standard graph metric, \((d_{n,C})_{n \in \mathbb{N}}\), and a coarse embedding function \(f:(X, d) \to (C,d_{C})\).
Assume there exists a cut metric \(d_{w}\) on \(X\) such that \(id:(X,d) \to (X,d_{w})\) is a coarse embedding. Since each \((X_{n}, d_{n,w})\) isometrically embeds into a hypercube equipped with the standard graph metric, \((C_{n}, d_{n,C})\), \((X,d)\) coarsely embeds into the disjoint union of \(\{(C_{n}, d_{n, C})\}_{n \in \mathbb{N}}\)~\cite[Proposition~4.2.2]{deza:geometry}. Conversely, assume there exists a coarse disjoint union, \((C, d_{C})\), of a sequence of hypercubes, \((C_{n}, d_{n,C})_{n \in \mathbb{N}}\), equipped with the standard graph metric and a coarse embedding function \(f:(X, d) \to (C,d_{C})\) such that \(f(X_{n}) \subseteq C_{n}\) for all \(n \in \mathbb{N}\). Using the Hamming encoding of a hypercube we can create partitions of the vertices of each \(C_{n}\) such that for any \(x,y \in X_{n}\), \(d_{n,w}(x,y) = d_{n,C}(f(x),f(y))\). Fix an \(n \in \mathbb{N}\). \(C_{n}\) contains \(2^m\) vertices for some \(m \in \mathbb{N}\). Define partitions \(S^n_{1},\ldots,S^n_{m}\) of the vertices of \(X_{n}\) by \begin{equation*} S^n_{k} := \{ x : x \in X_{n}, h(f(x))_{k} = 1 \}. \end{equation*} The set of partitions \(S^n_{1},\ldots,S^n_{m}\) defines a cut metric, \(d_{n,w}\), on \(X_{n}\) by \begin{equation*} d_{n,w}(x,y) = |\{ k : |S^n_{k} \cap \{ x,y \}| = 1 \}|, ~\forall x,y \in X_{n} \end{equation*} which is equal to the number of bits which differ between \(h(f(x))\) and \(h(f(y))\). Hence, for all \(n \in \mathbb{N}\) and all \(x,y \in X_{n}\), \(d_{n,w}(x,y) = d_{n,C}(f(x), f(y))\). Let \(d_{w}:X \times X \to \mathbb{R}^{\geq 0}\) be defined such that \(d_{w}\restrict{X_{n}} = d_{n,w}\) and \(d_{w}(X_{n}, X_{m}) = d_{C}(f(X_{n}), f(X_{m}))\) for all \(m,n \in \mathbb{N}\). Then, for any \(x,y \in X\), \(d_{w}(x,y) = d_{C}(f(x),f(y))\) hence \(id:(X, d) \to (X, d_{w})\) is a coarse equivalence.