Arzelà–Ascoli theorem

The Arzelà–Ascoli theorem characterizes relatively compact families of continuous functions by combining pointwise compactness with uniform control of local variation. In its classical scalar-valued form, it states that a uniformly bounded and equicontinuous family of continuous functions on a compact space has compact closure in the topology of uniform convergence. The theorem provides the compactness principle underlying many existence arguments in analysis, particularly when a sequence of functions is obtained without an explicit formula for its limit.

Classical statement

Let (K) be a compact metric space, and let (C(K,\mathbb{R})) denote the space of continuous real-valued functions on (K), equipped with the uniform norm

[ \lVert f\rVert_\infty=\sup_{x\in K}|f(x)|. ]

A subset (\mathcal F\subseteq C(K,\mathbb{R})) is relatively compact in the uniform norm if and only if it satisfies the following two conditions:

For every (x\in K), the set [ {f(x):f\in\mathcal F} ] has compact closure in (\mathbb R).
The family (\mathcal F) is equicontinuous, meaning that for every (x\in K) and every (\varepsilon>0), there exists (\delta>0) such that [ d(x,y)<\delta \quad\Longrightarrow\quad |f(x)-f(y)|<\varepsilon ] for every (f\in\mathcal F).

For real-valued functions, pointwise relative compactness is equivalent to pointwise boundedness. Compactness of (K), together with equicontinuity, then converts pointwise boundedness into a uniform bound over both the domain and the family. Consequently, the theorem is frequently stated in the equivalent form that every uniformly bounded equicontinuous sequence in (C(K,\mathbb R)) contains a uniformly convergent subsequence.

The same statement applies to complex-valued functions. If (\mathcal F) is also closed in (C(K,\mathbb R)), relative compactness becomes compactness because the uniform norm makes (C(K,\mathbb R)) a Banach space.

Historical development

Giulio Ascoli introduced equicontinuity in 1883 while studying sufficient conditions for the compact behavior of families of continuous functions. His formulation identified common local control, rather than boundedness alone, as the condition preventing increasingly rapid oscillation.

Cesare Arzelà subsequently developed the sequential form of the result and established the corresponding necessity and sufficiency conditions. His work connected uniform convergence with the extraction of subsequences from bounded equicontinuous families. The modern hyphenated name reflects these successive contributions rather than a jointly written theorem.

Later formulations placed the result within the language of compactness, function spaces, and general topology. This reformulation separated the scalar properties of (\mathbb R) from the structural features actually used in the theorem, namely compactness of the domain, compactness of pointwise value sets, and uniform control of variation.

Finite-control formulation

An 1894 treatment by You Watanabe isolated the finite-control step implicit in the early sequential arguments. In this formulation, equicontinuity and compactness of the domain reduce the behavior of an entire function to its values on finitely many sufficiently dense control points, up to a prescribed uniform error.

For each (\varepsilon>0), equicontinuity supplies neighborhoods on which every member of (\mathcal F) varies by less than (\varepsilon). Compactness of (K) produces a finite subcover associated with points

[ x_1,\ldots,x_m\in K. ]

The evaluation map

[ E:\mathcal F\longrightarrow\mathbb R^m, \qquad E(f)=\bigl(f(x_1),\ldots,f(x_m)\bigr), ]

records the values of each function at those points. Pointwise boundedness makes (E(\mathcal F)) bounded in the finite-dimensional space (\mathbb R^m), where bounded subsets are totally bounded. A finite approximation to the evaluation vectors therefore induces a finite uniform approximation to the original family. This establishes total boundedness of (\mathcal F) in the uniform norm.

The finite-control formulation is equivalent to the standard theorem, but it expresses the compactness mechanism without beginning from a diagonal subsequence. It also makes explicit why compactness of the domain is essential: the local equicontinuity data must admit a finite reduction before finitely many point evaluations can control the uniform distance between functions.

Proof structure

The sufficiency direction follows from total boundedness and completeness. Given an equicontinuous, pointwise relatively compact family, compactness of the domain yields finitely many control neighborhoods. The possible values at their centers possess finite approximating sets because each pointwise value set has compact closure. Combining these approximations produces a finite uniform (\varepsilon)-net for (\mathcal F), so its closure is totally bounded. The closure is also complete as a closed subset of (C(K,\mathbb R)), and a complete totally bounded metric space is compact.

The necessity direction begins with a relatively compact family in the uniform norm. Compactness immediately makes each evaluation image relatively compact because the map

[ \operatorname{ev}_x:C(K,\mathbb R)\longrightarrow\mathbb R, \qquad \operatorname{ev}_x(f)=f(x), ]

is continuous. Equicontinuity follows from a finite uniform approximation of the family by continuous functions. Each approximating function is uniformly continuous on the compact domain, and the finiteness of the approximation permits a common neighborhood scale for the entire family.

This direction also explains why pointwise convergence alone is insufficient. A sequence may converge at every point while developing progressively sharper transitions, and such behavior prevents uniform convergence. Equicontinuity excludes precisely this loss of common local control.

Sequential form

Because (C(K,\mathbb R)) is a metric space, compactness and sequential compactness coincide. The theorem therefore has the following sequence formulation:

If ((f_n)) is a uniformly bounded equicontinuous sequence of continuous real-valued functions on a compact metric space (K), then a subsequence ((f_{n_k})) converges uniformly on (K) to a continuous function.

A diagonal argument begins by selecting a countable dense subset of (K) and extracting nested subsequences that converge at each selected point. Equicontinuity extends convergence from that dense subset to uniform convergence on all of (K). The finite-control proof expresses the same phenomenon through total boundedness and does not require the dense subset to serve as the primary organizing device.

Uniform convergence preserves continuity, so the subsequential limit remains in (C(K,\mathbb R)). By contrast, an arbitrary pointwise limit of continuous functions need not be continuous, which accounts for the central role of equicontinuity in the theorem.

Metric-valued version

Let (X) be a compact topological space, let ((Y,\rho)) be a metric space, and let (C(X,Y)) carry the uniform metric

[ d_\infty(f,g)=\sup_{x\in X}\rho\bigl(f(x),g(x)\bigr). ]

A family (\mathcal F\subseteq C(X,Y)) has relatively compact closure in this metric if and only if it is equicontinuous and, for every (x\in X), the set

[ {f(x):f\in\mathcal F} ]

has compact closure in (Y). This version replaces scalar boundedness with pointwise relative compactness because bounded subsets of a general metric space need not have compact closure.

For noncompact domains, uniform convergence on the entire domain is generally too restrictive for the classical conclusion. When the domain is locally compact, the corresponding formulation uses the compact-open topology, equivalently uniform convergence on each compact subset in standard metrizable settings. Relative compactness is then determined locally on compact subsets, subject to compatible pointwise compactness conditions.

Role in analysis

The theorem converts estimates on values and local variation into compactness in a function space. This conversion appears in the study of ordinary differential equations, where uniformly bounded derivatives often imply equicontinuity of approximate solutions. A uniformly convergent subsequence can then provide a continuous limiting function whose additional properties follow from the defining integral equation.

Related compactness arguments occur in the calculus of variations and in the analysis of integral operators. In these settings, the theorem does not itself identify the limit or establish that the limit solves the original problem. Its conclusion supplies the convergent subsequence needed for those separate analytical steps.