Compactness and Bounded Linear Operators

Motivation: Finiteness in an Infinite-Dimensional World

In finite-dimensional spaces, a closed and bounded ball is compact. In infinite-dimensional spaces — it is not: an orthonormal sequence $e_n$ in $l^2$ does not have a convergent subsequence. Compact operators are an "intermediate class": they map infinite-dimensional balls into precompact sets. This is a key tool for the theory of Fredholm equations and integral equations.

Compactness in Metric Spaces

A compact set $K$: from any sequence in $K$ one can extract a subsequence converging to a point in $K$ (sequential compactness).

Hausdorff criterion: $K$ is compact $\iff$ $K$ is complete and totally bounded (for every $\epsilon > 0$ there exists a finite $\epsilon$-net: $K \subseteq \bigcup_{a\in A} B(a,\epsilon)$, $|A| < \infty$).

Heine–Borel theorem: In $\mathbb{R}^n$: $K$ is compact $\iff$ $K$ is closed and bounded. In infinite-dimensional spaces — false: the closed unit ball of $l^2$ is closed and bounded, but not compact.

Arzelà–Ascoli Theorem

Theorem: A family $F \subset C[a,b]$ is precompact (closure is compact) $\iff$ $F$:

Is uniformly bounded: $\sup_{f\in F} |f|_\infty < \infty$.
Is equicontinuous: $\forall \epsilon > 0;; \exists \delta > 0:$ $|x-y| < \delta \implies |f(x)-f(y)| < \epsilon$ for all $f \in F$.

Application: proof of Peano's theorem on existence of ODE solutions; compactness of integral-type operators.

Bounded Linear Operators

Linear operator $T: X\to Y$: $T(\alpha x+\beta y) = \alpha Tx+\beta Ty$. For a linear operator, the following are equivalent:

Continuity at zero.
Continuity everywhere.
Boundedness: $|T| = \sup_{|x| \leq 1} |Tx| < \infty$.

Space $L(X,Y)$: bounded linear $T: X\to Y$ with norm $|T|.$ When $Y$ is Banach: $L(X,Y)$ is Banach.

Compact operator $T: X\to Y$: the image of the unit ball is precompact. Examples: operators of finite rank; Hilbert–Schmidt operators (integral operators with $L^2$-kernel).

Numerical Example

Problem: Check precompactness of the family $F = {f_n}_{n\geq 1}$ in $C[0,1],$ where $f_n(x) = \sin(nx)/n$.

Step 1. Uniform boundedness: $|f_n(x)| = |\sin(nx)|/n \leq 1/n \leq 1$. Hence $\sup_n |f_n|_\infty \leq 1 < \infty ;\checkmark$.

Step 3. By the Arzelà–Ascoli theorem: $F$ is precompact in $C[0,1]$. In fact: $|f_n|_\infty = 1/n \to 0$, so $f_n \to 0$ uniformly $;\checkmark$.

Step 4. Comparison: $g_n(x) = \sin(nx)$ (without division by $n$). $|g_n(x) - g_n(y)| \leq n|x-y|$ — coefficient $n \to \infty$. Equicontinuity fails. Indeed, ${\sin(nx)}$ does not have a uniformly convergent subsequence in $C[0,1]$: $\sin(n_k x) \rightharpoonup 0$ weakly in $L^2$, but not in the $C[0,1]$ norm.

Step 5. Norm of the Volterra operator: $T: C[0,1]\rightarrow C[0,1],; Tf(x) = \int_0^x f(t)dt.$ $|T^n| \leq 1/n! \to 0.$ This is a compact operator (the image of a bounded set is an equicontinuous family by the inequality from Step 1 of the previous article).

Real-World Application

Signal processing: compression operators (JPEG, MP3) are compact operators. The Fredholm theorem (a consequence of compactness) guarantees that the problem of recovering a signal is either uniquely solvable or has a finite-dimensional space of exceptions.

Additional Aspects

In infinite-dimensional spaces, the unit ball is not compact — this is a central obstacle in analysis and the reason for the appearance of compact operators. A compact operator $T: X \rightarrow Y$ maps bounded sets into relatively compact ones. Spectral theory of compact operators (the Riesz–Schauder theorem) states that the spectrum is countable, the only possible accumulation point is $0$, and each nonzero eigenvalue has a finite-dimensional eigenspace. This is the basis of the Fredholm method for integral equations and of numerical methods for elliptic problems (FEM, Galerkin method): the resulting matrices are finite-dimensional approximations of compact operators, and convergence is guaranteed by general theory.

Connection with Other Areas of Mathematics

In the theory of differential equations, compactness of operators underlies existence theorems for solutions. The Leray–Schauder theorem and the Schauder fixed point theorem consider a compact operator in a Banach space and give existence of solutions for nonlinear boundary value problems for elliptic and parabolic equations. Linear elliptic problems reduce to Fredholm equations, where a linear closed operator splits into an isomorphism plus a compact operator, allowing one to apply the Fredholm–Riesz theorem.

In functional analysis and topology, compact operators are closely related to the concept of weak topology. The Eberlein–Šmulian theorem describes relative weak compactness in a Banach space via sequential compactness and applies to images of the unit ball under a compact operator. In locally convex spaces, compact sets are described by total boundedness in the system of seminorms, which connects operator theory with general topological constructions (Grothendieck's work on nuclear spaces).

In probability theory, compactness plays a role in limit theorems. The Prokhorov criterion formulates relative compactness of a family of distributions via uniform tightness of mass on compacts; in spaces of continuous trajectories (Skorokhod space) this leads to Kolmogorov–Chentsov conditions on modulus of continuity, formally very close to the Arzelà–Ascoli theorem. In numerical analysis, compactness ensures stability of approximations: Galerkin schemes for integral and elliptic problems use finite-dimensional subspaces, and convergence of solutions to the true one is guaranteed, in particular, by the compactness of Sobolev embeddings (the Rellich–Kondrachov theorem).

Historical Note and Development of the Idea

The first systematic studies of compact operators are linked to the works of Ernst Fredholm (Acta Mathematica, 1903), who studied integral equations with kernels in $L^2$ and introduced the determinant and resolvent. David Hilbert, developing this theory in the early 20th century, formulated the spectral properties of integral operators and laid the foundation of the notion of Hilbert space.

Work by Riesz (1918–1920) and Schauder in the 1930s generalized spectral theory to abstract Banach spaces; the Riesz–Schauder theorem became a central result. In the same years, Arzelà and Ascoli formulated the compactness criterion in spaces of continuous functions, which quickly became a standard tool in ODE theory (Carathéodory's book, 1935) and PDE (Sobolev's monograph, 1938).

In the second half of the 20th century, Grothendieck extended the concept of compactness to nuclear operators and tensor products of topological vector spaces (Mémoire de l’Académie des Sciences, 1955–1956). In probability theory, key results on compactness were systematized in the book by Billingsley–Convergence (Measure Theory, 1970s). In the 21st century, compact operators naturally arise in operator algebra theory, nonlinear functional analysis, and regularization methods in inverse problems (monograph by Engl and Hanke, 1996; further developments in works on total variation and sparse regularization).