Parametric set-wise injective maps (Q2411479)

Spaces in this paper are assumed to be metric, and map means continuous function. Before stating results, we recite some definitions that are given beginning on page 338. One says that a space \(M\) has the \(m\)-\(\mathrm{DD}^{\{n,k\}}\)-property if any maps \(f:I^m\times I^n\to M\) and \(g:I^m\times I^k\to M\) (\(I=[0,1]\)) can be respectively approximated by maps \(f':I^m\times I^n\to M\) and \(g':I^m\times I^k\to M\) such that \(f'(\{z\}\times I^n)\cap g'(\{z\}\times I^k)=\emptyset\) for all \(x\in I^m\). A map \(g:X\to M\) is called set-wise injective in dimension \(k\) if for any closed subsets \(A\), \(B\subset X\) with \(\mathrm{dim}(A\setminus B)\geq k\), \(g(A)\neq g(B)\). One notes that set-wise injective in dimension \(0\) is equivalent to the injectivity of \(g\). Two additional definitions are worth giving. A map \(f:X\to Y\) is said to be \(\sigma\)-perfect if \(X\) is the union of a countable collection of closed sets \(B\) such that each restriction \(f|B:B\to f(B)\) is a perfect map. Also, \(f\) is called an \(n\)-dimensional map if \(\mathrm{dim}(f^{-1}(y))\leq n\) for all \(y\in Y\). When dealing with the space \(C(X,M)\) of maps of \(X\) to \(M\), the source limitation topology will be in play. This is defined on the first page of the article. Here are the main results. Theorem 1.1. Let \(f:X\to Y\) be a \(\sigma\)-perfect surjective \(n\)-dimensional map between metric spaces such that \(\mathrm{dim}Y\leq m\), and let \(M\) be a complete separable metric \(\mathrm{LC}^{2m+n-1}\)-space with the \(m\)-\(\mathrm{DD}^{\{n,k\}}\)-property where \(k\leq n\). Then the function space \(C(X,M)\) contains a dense \(\mathrm{G}_\delta\)-set of maps \(g\) such that all restrictions \(g|f^{-1}(y)\), \(y\in Y\), are set-wise injective in dimension \(n-k\). Corollary 1.2. Let \(X\), \(Y\), and \(f\) be as in Theorem 1.1, and let \(P\subset Q\subset X\) be \(\mathrm{F}_\sigma\)-subsets of \(X\) such that \(\mathrm{dim}(P\cap f^{-1}(y))\leq p\) and \(\mathrm{dim}(Q\cap f^{-1}(y))\leq q\) for every \(y\in Y\), where \(0\leq p\leq q\leq n\). Then for every complete separable metric \(\mathrm{LC}^{2m+n-1}\)-space \(M\) with the \(m\)-\(\mathrm{DD}^{\{q,p\}}\)-property, the function space \(C(X,M)\) contains a dense \(\mathrm{G}_\delta\)-set of maps \(g\) satisfying the following condition: \(g^{-1}(g(z))\cap Q\cap f^{-1}(y)=\{z\}\) for all \(z\in P\cap f^{-1}(y)\) and all \(y\in Y\).