Uniform spaces add a global language of closeness to topology without requiring numerical distances. They are the right setting for uniform continuity, Cauchy filters, and completion.
Entourages
An entourage records which pairs of points are close at a chosen scale.
Definition. A uniformity on \(X\) is a filter \(\mathcal U\) of subsets of \(X\times X\), called entourages, such that every \(U\in\mathcal U\) contains the diagonal, \(U\in\mathcal U\) implies \(U^{-1}\in\mathcal U\), and for every \(U\in\mathcal U\) there is \(V\in\mathcal U\) with \(V\circ V\subseteq U\).
Example. A metric \(d\) defines a uniformity by entourages \(U_\varepsilon=\{(x,y):d(x,y)<\varepsilon\}\). Topological groups have natural uniformities from neighborhoods of the identity.
Proposition. Every uniformity induces a topology by declaring \(U[x]=\{y:(x,y)\in U\}\) to be a neighborhood basis at \(x\).
Proof. The diagonal condition gives \(x\in U[x]\). Finite intersection follows from the filter property. If \(U\) is an entourage, choose symmetric \(V\) with \(V\circ V\subseteq U\). Then for \(y\in V[x]\), one has \(V[y]\subseteq U[x]\), giving the neighborhood axiom.
Uniform continuity
Uniform continuity controls all points with one entourage condition, not point-by-point neighborhoods.
Definition. A map \(f:X\to Y\) between uniform spaces is uniformly continuous if for every entourage \(V\) of \(Y\), there is an entourage \(U\) of \(X\) such that \((x,x')\in U\) implies \((f(x),f(x'))\in V\).
Proposition. Uniformly continuous maps are continuous for the induced topologies.
Proof. Let \(f(x)\in O\) with \(O\) open in \(Y\). Choose an entourage \(V\) such that \(V[f(x)]\subseteq O\). Uniform continuity gives \(U\) with \((f\times f)(U)\subseteq V\). Then \(U[x]\) is a neighborhood of \(x\) mapped into \(O\).
Example. The function \(x\mapsto x^2\) on \(\mathbb R\) is continuous but not uniformly continuous for the usual metric uniformity.
Cauchy filters and completeness
Uniform spaces support Cauchy notions even when there is no metric.
Definition. A filter \(\mathcal F\) on a uniform space \(X\) is Cauchy if for every entourage \(U\), there exists \(F\in\mathcal F\) such that \(F\times F\subseteq U\). A uniform space is complete if every Cauchy filter converges.
Proposition. Every convergent filter is Cauchy.
Proof. If \(\mathcal F\to x\) and \(U\) is an entourage, choose symmetric \(V\) with \(V\circ V\subseteq U\). Since \(V[x]\in\mathcal F\), take \(F=V[x]\). For \(y,z\in F\), \((y,x)\in V\) and \((x,z)\in V\), hence \((y,z)\in V\circ V\subseteq U\).
Completion
Completion adds limits for all Cauchy filters while preserving uniform structure.
Theorem. Every separated uniform space admits a completion: a complete separated uniform space \(\widehat X\) and a uniformly continuous dense embedding \(X\hookrightarrow\widehat X\) with the expected universal property for uniformly continuous maps into complete separated spaces.
Proof. One constructs \(\widehat X\) from minimal Cauchy filters, identifying filters that are eventually close in every entourage. Entourages lift by requiring representative filters to contain sets that are pairwise close. The original point \(x\) embeds as its neighborhood filter. Density follows because every Cauchy filter is approximated by its members, and completeness follows by taking limits of Cauchy filters of minimal Cauchy filters. The universal property is obtained by sending a Cauchy filter to the limit of its image filter.
Remark. Metric completion is the special case where Cauchy filters can be represented by Cauchy sequences in first-countable settings.