Chapter 7 — Metric and Normed Spaces
One idea — a distance satisfying the triangle inequality — and the whole of analysis follows: open sets, completeness, compactness and connectedness in $\mathbb{R}^n$, in sequence spaces and in spaces of functions.
How far apart are two functions? For $f(x)=x$ and $g(x)=x^2$ on $[0,1]$ one natural answer is the largest vertical gap, $\max|f-g|=\tfrac14$; another is the area between the graphs, $\int_0^1|f-g|=\tfrac16$. Both are perfectly good "distances", and both obey the rules we expect of distance: zero only for equal functions, symmetric, and never shortened by a detour. At the beginning of the twentieth century Maurice Fréchet, Felix Hausdorff and Stefan Banach realised that once these rules are fixed, the theorems of real analysis — on convergence, continuity, compactness and the intermediate value property — can be proved once and for all, for points of $\mathbb{R}^n$, for infinite sequences, for continuous functions and for anything else that carries a distance.
This chapter builds that general theory, and it is not abstraction for its own sake. The choice of distance decides what "converges" means: the functions $x^n$ approach $0$ if we measure by area, but not if we measure by the largest gap. Completeness — the absence of holes — is what makes iterative methods work, from Picard's existence theorem for differential equations to the ranking algorithms of web search. Compactness explains why a continuous function on $[a,b]$ has a maximum and why the same is false on the closed unit ball of an infinite-dimensional space. And connectedness is the reason the intermediate value theorem holds, and the simplest tool for proving that two spaces are genuinely different.
- decide whether a given formula is a metric, and compute distances in $\mathbb{R}^n$, $\ell^p$ and $C[a,b]$ using the Hölder and Minkowski inequalities;
- find interiors, closures, boundaries and limit points, and decide whether a set is open, closed, dense or nowhere dense — in $\mathbb{R}^n$, in a discrete space and in function spaces;
- prove or disprove completeness, compare metrics, apply the Banach contraction principle to equations and integral equations, and use the Baire category theorem;
- use the equivalence of compactness, sequential compactness and "complete + totally bounded", apply Heine–Borel in $\mathbb{R}^n$, and produce closed bounded sets that are not compact;
- recognise connected and path-connected sets, prove that the connected subsets of $\mathbb{R}$ are the intervals, and use connectedness to show two spaces are not homeomorphic;
- work with norms, Banach spaces and bounded linear operators, show that all norms on $\mathbb{R}^n$ are equivalent, and explain why the unit ball of an infinite-dimensional normed space is not compact.
How this is examined
Metric spaces are among the most heavily examined topics of Unit 1, because they let an examiner test definitions rather than computations. Part B typically asks whether a given function of two variables is a metric, what the closure or boundary of a set is in an unusual metric (the discrete metric is a favourite), whether a specific map is a contraction and for which parameter values, which of four sets is compact or connected, or what the norm of a simple linear operator is. Part C, where several options may be correct, prefers statements to be judged: "every closed and bounded subset is compact", "a continuous bijection is a homeomorphism", "the space is complete under this metric", "this subset of $\ell^2$ or of $C[0,1]$ is compact", "these two spaces are homeomorphic", "this set is a $G_\delta$". Almost every wrong option is one of the counterexamples collected in the ⚠️ boxes of the four sections — the discrete metric, the unit ball of $\ell^2$, $x^n$ in $C[0,1]$, the topologist's sine curve, $\mathbb{Q}$ and the Cantor set. GATE MA asks the same material in single-answer and multiple-select form.
Sections in this chapter
Metric Spaces and Completeness
Metrics on $\mathbb{R}^n$, $\ell^p$ and $C[a,b]$; open and closed sets, closure, boundary, density; equivalent metrics; completeness and completion; the Banach contraction principle; the Baire category theorem.
Compactness in Metric Spaces
Open covers, sequential compactness, total boundedness and their equivalence; Lebesgue numbers; Heine–Borel in $\mathbb{R}^n$; non-compact closed balls and the Hilbert cube; continuous functions on compact spaces.
Connectedness
Connected sets, the intervals of $\mathbb{R}$, continuous images, path-connectedness, the topologist's sine curve, components, totally disconnected sets, and connectedness as an obstruction to homeomorphism.
Normed Linear Spaces and Spaces of Continuous Functions
Norms and Banach spaces, $\ell^p$ and $C[a,b]$, equivalence of norms in finite dimensions, bounded operators and the operator norm, Riesz's lemma, Arzelà–Ascoli and Stone–Weierstrass.
This chapter generalises results you have already met on the real line, and it is much easier if they are fresh. Review: suprema and completeness of $\mathbb{R}$ (§1.2) and countable sets (§1.1); convergent and Cauchy sequences and contractive sequences (§2.1); subsequences and the Bolzano–Weierstrass theorem (§2.2); open and closed sets, compactness and the Heine–Borel theorem in $\mathbb{R}$ (§3.1); uniform continuity (§3.2); uniform convergence and the sup norm (§5.1) and the Weierstrass approximation theorem (§5.2). Linear-algebra language (vector spaces, linear maps, bases — Chapter 8) is used in §7.4, but only at the level of a first course in matrices. Throughout, the habit that pays off most is testing every statement on four examples: $\mathbb{R}$, a discrete space, $\ell^2$ and $C[0,1]$.