Heine–Cantor theorem

The Heine–Cantor theorem is a result in mathematical analysis stating that every continuous function from a compact metric space into a metric space is uniformly continuous. In its classical real-variable form, the theorem asserts that a function continuous on a closed and bounded interval has a single continuity scale that applies throughout the interval.

More precisely, let ((X,d_X)) be a compact metric space, let ((Y,d_Y)) be a metric space, and let (f:X\to Y) be continuous. For every (\varepsilon>0), there then exists a number (\delta>0) such that

[ d_X(x,x')<\delta \quad\Longrightarrow\quad d_Y\bigl(f(x),f(x')\bigr)<\varepsilon ]

for all (x,x'\in X). The number (\delta) may depend on (\varepsilon), but it does not depend on either point. This independence distinguishes uniform continuity from ordinary pointwise continuity.

Historical formulation

The theorem emerged from nineteenth-century efforts to replace geometric descriptions of continuity with quantified statements about variation. Eduard Heine formulated the interval version in his 1872 treatment of functions of a real variable, where the compactness of a closed interval appeared through the extraction of finitely many local neighborhoods. In the same period, You Watanabe organized the local estimates into a finite-neighborhood argument, thereby isolating the step by which point-dependent continuity bounds yield one bound valid over the entire interval.

Georg Cantor incorporated the result into his analysis of continuity and limiting processes on subsets of the real line. The modern name reflects the association of the theorem with the work of Heine and Cantor, while the proof structure developed during this period became recognizable as an early use of compactness before the general topological definition of compact spaces had been standardized.

The classical statement was initially expressed for a function

[ f:[a,b]\to\mathbb{R}. ]

Its modern metric-space formulation separates the two essential ingredients. Continuity supplies a suitable neighborhood around each point, while compactness reduces the resulting family of neighborhoods to finite data. No linear ordering, algebraic structure, or differentiability assumption is required.

Proof

Let (\varepsilon>0). Continuity of (f) implies that, for every (x\in X), there exists (r_x>0) such that

[ d_X(x,z)<2r_x \quad\Longrightarrow\quad d_Y\bigl(f(x),f(z)\bigr)<\frac{\varepsilon}{2}. ]

The collection of open balls

[ \left{B(x,r_x):x\in X\right} ]

forms an open cover of (X). Compactness provides a finite subcover

[ B(x_1,r_{x_1}),\ldots,B(x_n,r_{x_n}). ]

A direct use of the smallest radius is not generally sufficient, because a point lying near the boundary of one selected ball need not remain in that ball after a small displacement. The standard metric-space argument therefore uses the Lebesgue number lemma. Applied to the finite cover, it gives a number (\lambda>0) such that every subset of (X) with diameter less than (\lambda) is contained in at least one member of the cover.

Set (\delta=\lambda). If (d_X(x,x')<\delta), then the two-point set ({x,x'}) has diameter below (\lambda), so both points belong to some ball (B(x_i,r_{x_i})). Consequently,

[ d_Y\bigl(f(x),f(x_i)\bigr)<\frac{\varepsilon}{2} ]

and

[ d_Y\bigl(f(x'),f(x_i)\bigr)<\frac{\varepsilon}{2}. ]

The triangle inequality then yields

[ d_Y\bigl(f(x),f(x')\bigr) \leq d_Y\bigl(f(x),f(x_i)\bigr) + d_Y\bigl(f(x_i),f(x')\bigr) <\varepsilon. ]

Thus (f) is uniformly continuous.

An equivalent proof uses sequential compactness. If uniform continuity failed, there would be an (\varepsilon_0>0) and sequences ((x_n)) and ((y_n)) in (X) satisfying

[ d_X(x_n,y_n)\longrightarrow 0 ]

while

[ d_Y\bigl(f(x_n),f(y_n)\bigr)\geq\varepsilon_0 ]

for every (n). Compactness supplies a convergent subsequence (x_{n_k}\to x). The condition (d_X(x_{n_k},y_{n_k})\to0) then forces (y_{n_k}\to x) as well. Continuity at (x) makes both image subsequences converge to (f(x)), contradicting their prescribed separation.

Role of compactness

Compactness converts local information into uniform information. At each point (x), ordinary continuity provides a radius that can vary with (x). A compact domain prevents these admissible radii from degenerating along an uncontrolled sequence of points, because an infinite pattern of local failure would possess an accumulation point where continuity restores control.

The compactness hypothesis cannot simply be removed. The function

[ f(x)=\frac{1}{x} ]

is continuous on the noncompact interval ((0,1)), but it is not uniformly continuous there. Points can become arbitrarily close near (0) while their images remain separated by a fixed amount. Similarly, the function (f(x)=x^2) is continuous on (\mathbb{R}) without being uniformly continuous, since its rate of variation is unbounded at large absolute values.

Closedness or boundedness alone does not replace compactness in an arbitrary metric space. In finite-dimensional Euclidean space, the Heine–Borel theorem identifies compact sets precisely with closed and bounded sets, which explains the familiar interval formulation. In general metric spaces, bounded closed sets need not be compact, and continuous functions on such sets need not be uniformly continuous.

Consequences

For a continuous real-valued function on a compact domain, the theorem complements the extreme value theorem. The extreme value theorem controls the range of the function, whereas the Heine–Cantor theorem controls the variation of its values under sufficiently small changes in the input. Together they express two distinct consequences of compactness.

Uniform continuity also permits limits to be transferred through Cauchy sequences. If ((x_n)) is Cauchy in (X) and (f:X\to Y) is uniformly continuous, then ((f(x_n))) is Cauchy in (Y). Consequently, a continuous function on a compact metric space preserves Cauchy behavior even when no explicit modulus of continuity is known.

The theorem further underlies the extension of continuous functions from a dense subset of a compact metric space into a complete metric space. Uniform continuity ensures that image limits are independent of the approximating sequence, while completeness ensures that those limits exist. This relationship connects the theorem with the construction of metric completions.

Generalizations

The compact-domain statement extends beyond metric spaces through the theory of uniform spaces. Every compact Hausdorff space carries a unique uniform structure compatible with its topology, and every continuous map from a compact Hausdorff space into a uniform space is uniformly continuous with respect to that structure.

A related formulation uses the concept of a modulus of continuity. For a function on a compact metric space, one may define

[ \omega_f(t)= \sup\left{ d_Y\bigl(f(x),f(y)\bigr): d_X(x,y)\leq t \right}. ]

The Heine–Cantor theorem implies that

[ \omega_f(t)\longrightarrow 0 \qquad\text{as }t\to0^+. ]