Metric Spaces

Job: quantitative topology — completeness, compactness criteria, uniform structure, and Baire category as used in analysis.

Core

Definition. A metric \(d:X\times X\to[0,\infty)\) satisfies \(d(x,y)=0\Leftrightarrow x=y\), symmetry, triangle inequality. Topology: open balls \(B(x,r)\) form a basis. Isometry preserves \(d\); uniformly continuous: \(\forall\varepsilon\exists\delta\) independent of basepoint.

Concept Definition / fact
Cauchy sequence \(d(x_n,x_m)\to 0\) as \(n,m\to\infty\)
Complete every Cauchy sequence converges in \(X\)
Completion \(\widehat{X}\) unique complete space with \(X\) dense isometrically embedded
Totally bounded \(\forall\varepsilon\): finite \(\varepsilon\)-net
Compact metric complete + totally bounded \(\Leftrightarrow\) sequentially compact
Heine–Borel (\(\mathbb{R}^n\)) closed bounded \(\Leftrightarrow\) compact
Lipschitz / Hölder \(d(fx,fy)\le K\,d(x,y)^\alpha\)
Equivalent metrics same topology; strongly equivalent: \(c\,d\le d'\le C\,d\)

Completeness

Closed subset of complete \(\Rightarrow\) complete. Product of countably many complete metric spaces is completely metrizable (e.g. \(\mathbb{R}^{\mathbb{N}}\)).

Theorem (Baire category). \(X\) complete metric (or locally compact Hausdorff). Countable intersection of dense open sets is dense. Equiv.: \(X\) is not a countable union of nowhere dense sets.

Consequences: uniform boundedness (BANACH–Steinhaus), open mapping, existence of continuous nowhere differentiable functions (comeager in \(C[0,1]\)), residual sets in dynamical systems.

Compactness in metric spaces

Theorem. For metric \(X\), TFAE: (i) compact; (ii) sequentially compact; (iii) complete and totally bounded; (iv) every infinite set has an accumulation point.

Lebesgue number: for open cover of compact metric, \(\exists\delta\) s.t. any set of diam \(<\delta\) lies in some cover element — used in refining covers for \(\pi_1\) and homology.

Continuous functions

\(C(X,Y)\) with sup metric when \(Y\) bounded metric; if \(X\) compact, \(C(X,\mathbb{R})\) Banach under \(\|\cdot\|_\infty\).

Theorem (Arzelà–Ascoli). \(X\) compact metric, \(\mathcal{F}\subset C(X,\mathbb{R})\): relatively compact iff pointwise bounded and equicontinuous.

Theorem (Stone–Weierstrass). Subalgebra of \(C(X,\mathbb{R})\) (\(X\) compact Hausdorff) separating points, containing constants, dense in \(\|\cdot\|_\infty\). Complex version: closed under conjugation.

Proper metric spaces

Closed balls compact \(\Leftrightarrow\) proper (as a map to a point in coarse geometry language: inverse images of compacts are compact). Riemannian manifolds with geodesic distance: Hopf–Rinow.

See it

\(C[0,1]\) vs \(L^p\). \(C[0,1]\) complete in \(\|\cdot\|_\infty\), not in \(\|\cdot\|_1\); completion of \(C_c\) in \(L^p\) norm is \(L^p\). Metric completeness is the analytic content of Banach space hypotheses.

p-adic metric. \(d_p(x,y)=p^{-v_p(x-y)}\) on \(\mathbb{Q}\); completion \(\mathbb{Q}_p\). Ultrametric inequality \(d(x,z)\le\max(d(x,y),d(y,z))\) — balls are clopen; totally disconnected locally compact field.

Check

Why is \((0,1)\) not complete in the standard metric, and what is its completion?

Cauchy sequence \(x_n=1/n\) has no limit in \((0,1)\). Completion is \([0,1]\) (or isometrically, the abstract completion is homeomorphic to \([0,1]\)).

Show that infinite-dimensional normed spaces have noncompact closed unit ball (Riesz lemma sketch).

Riesz: for closed proper subspace \(Y\), \(\exists x\), \(\|x\|=1\), \(d(x,Y)\ge 1-\varepsilon\). Iterate to get sequence with \(d(x_n,x_m)\ge 1/2\); no convergent subsequence.

Pitfalls

  • Topologically equivalent metrics need not share Cauchy sequences (e.g. \(d\) and \(\arctan d\) on \(\mathbb{R}\): same topology, different completeness — \(\mathbb{R}\) complete in \(d\), bounded metric not complete? Actually \((\mathbb{R},\arctan|x-y|)\) is not complete). Completeness is metric, not topological.
  • Total boundedness \(\neq\) boundedness in infinite dimensions.
  • Uniform continuity on noncompact sets fails (\(x\mapsto x^2\) on \(\mathbb{R}\)).
  • Length spaces / geodesic structure is extra geometric data beyond the metric topology.

Refs

  • Rudin, Principles of Mathematical Analysis
  • Munkres, Topology
  • Burago–Burago–Ivanov, A Course in Metric Geometry (length spaces)

Next

Real analysis · Foundations · Functional analysis · Topology overview