1 Definition and basic properties
A bounded operator is a linear transformation between normed vector spaces whose outputs do not grow faster than a fixed constant times the inputs. This simple condition makes bounded operators central in functional analysis, since it links linear algebra with topological notions such as continuity and convergence. In many settings, boundedness is the property that allows infinite-dimensional problems to behave in a controlled and analyzable way.
1.1 Linear operators on normed spaces
Let \(X\) and \(Y\) be normed vector spaces over the same field. A linear operator \(T : X \to Y\) satisfies \[ T(x_1+x_2)=T(x_1)+T(x_2), \qquad T(\alpha x)=\alpha T(x). \] The norm on \(X\) measures the size of an input, while the norm on \(Y\) measures the size of an output. The interaction between these norms determines whether the operator is bounded.
1.2 Boundedness condition
An operator \(T : X \to Y\) is bounded if there exists a constant \(C \ge 0\) such that \[
| \|Tx\| \le C\|x\| \quad \text{for all } x \in X. |
|---|
\] This inequality shows that the operator cannot magnify vectors without limit relative to their length. The smallest such constant is especially important and leads to the operator norm.
1.2.1 Operator norm
The operator norm of \(T\) is defined by \[
| \|T\| = \sup_{\|x\|\le 1} \|Tx\|. |
|---|
\]
| Equivalently, it is the least constant \(C\) for which \(\|Tx\|\le C\|x\|\) holds for every \(x\). When the norm is finite, the operator is bounded. |
|---|
1.2.2 Norm estimate and Lipschitz continuity
Boundedness implies the estimate \[
| \|Tx-Ty\| = \|T(x-y)\| \le \|T\|\|x-y\|. |
|---|
\]
| Thus bounded linear operators are Lipschitz continuous, with Lipschitz constant at most \(\|T\|\). This gives a direct quantitative measure of stability: small changes in input produce proportionally small changes in output. |
|---|
1.3 Equivalence with continuity
For linear maps between normed spaces, boundedness and continuity are equivalent. This is one of the most useful facts in the subject, since it allows analytic questions to be expressed in algebraic terms.
1.3.1 Continuity at the origin
| A linear operator is continuous at one point, especially at the origin, if and only if it is continuous everywhere. If \(T\) is continuous at \(0\), then near the origin it must send small vectors to small vectors, which forces a uniform bound of the form \(\|Tx\|\le C\|x\|\). |
|---|
1.3.2 Continuity on the whole space
Because linearity translates neighborhoods, continuity at the origin automatically implies continuity at every point. Therefore a linear operator is bounded exactly when it is continuous as a map between normed spaces. This equivalence is one of the foundations of functional analysis.
2 Examples of bounded operators
Bounded operators appear in many familiar settings, from finite-dimensional linear algebra to integral transforms and shift maps on sequence spaces. These examples show that the abstract definition captures a wide range of concrete phenomena.
2.1 Finite-dimensional linear maps
Every linear map between finite-dimensional normed spaces is bounded. In finite dimensions, all norms are equivalent, and linearity alone is enough to prevent uncontrolled growth. This is one reason matrix theory is a special case of operator theory.
2.2 Matrix operators
A matrix \(A\) defines a linear operator on \(\mathbb{R}^n\) or \(\mathbb{C}^n\) by multiplication: \[ x \mapsto Ax. \] Such operators are bounded because matrix multiplication is continuous and the space is finite-dimensional. Their operator norms depend on the chosen vector norm and can be computed or estimated from the matrix entries.
2.3 Projection and inclusion operators
Projection operators send a vector to its component in a subspace. If the projection is onto a closed subspace in a Hilbert space, it is bounded and often has norm \(1\). Inclusion maps between normed spaces may also be bounded when one norm dominates another, as happens in many function-space embeddings.
2.4 Integral operators
Integral operators act by \[ (Tf)(x)=\int K(x,y)f(y)\,dy, \] where \(K\) is a kernel. Such operators arise in differential equations, approximation theory, and harmonic analysis. Their boundedness depends on regularity and integrability properties of the kernel and on the underlying function spaces.
2.4.1 Kernels and boundedness criteria
A kernel may yield a bounded operator if it satisfies suitable size and smoothness bounds. Common criteria involve estimates from integral inequalities, such as those ensuring that the transform does not enlarge norms beyond a fixed constant. The precise conditions vary with the space, for example \(L^p\), \(L^2\), or spaces of continuous functions.
2.5 Shift and multiplication operators
On sequence spaces, a shift operator moves coordinates left or right, while on function spaces a multiplication operator acts by pointwise multiplication with a fixed bounded function. These operators are typically bounded when the underlying sequences or functions are controlled appropriately. They serve as standard examples in spectral theory and in the study of operator algebras.
3 Operator norm
The operator norm measures the maximal stretching effect of a bounded operator. It is both an analytic tool and an algebraic quantity, since it interacts well with addition, scalar multiplication, and composition.
3.1 Definition and interpretation
| The operator norm is the size of the operator as a map between normed spaces. Geometrically, it describes how much the operator can enlarge vectors of unit length. If \(\|T\|=1\), then the map never expands a vector by more than a factor of one, though it may still preserve or reduce length. |
|---|
3.2 Basic inequalities
The operator norm satisfies \[
| \|Tx\| \le \|T\|\,\|x\|, |
|---|
\] and for two bounded operators \(S\) and \(T\), \[
| \|S+T\|\le \|S\|+\|T\|, \qquad \|\lambda T\|= | \lambda | \,\|T\|. |
|---|
\] For compositions, \[
| \|ST\|\le \|S\|\,\|T\|. |
|---|
\] These inequalities make the operator norm compatible with the algebraic structure of bounded operators.
3.3 Norm attainment
| In some cases, there exists a vector \(x\) with \(\|x\|=1\) such that \(\|Tx\|=\|T\|\). When this happens, the operator is said to attain its norm. Many operators do not attain their norm, especially in infinite-dimensional settings, but may be approximated by sequences that come arbitrarily close. |
|---|
3.3.1 Unit vectors and maximizing sequences
| A maximizing sequence is a sequence of unit vectors \((x_n)\) such that \(\|Tx_n\|\to \|T\|\). Such sequences are often used when exact maximizers do not exist. They are useful in variational arguments and in understanding compactness phenomena. |
|---|
3.4 Computation in special cases
In finite dimensions, the operator norm can often be computed from singular values or related matrix decompositions. For self-adjoint operators on Hilbert spaces, the norm is closely tied to the spectrum. In simple cases such as projections, isometries, or diagonal multiplication operators, the norm may be read off directly from the defining formula.
4 Algebraic structure
Bounded operators form an algebra under addition, scalar multiplication, and composition. This makes them a natural setting for combining analytic estimates with algebraic operations.
4.1 Vector space of bounded operators
The set of all bounded linear operators from \(X\) to \(Y\) is itself a vector space, usually denoted \(B(X,Y)\). When \(X=Y\), one writes \(B(X)\). The linear structure reflects the fact that sums and scalar multiples of bounded maps remain bounded.
4.2 Composition of bounded operators
If \(T : X \to Y\) and \(S : Y \to Z\) are bounded, then the composition \(S\circ T\) is bounded. The norm estimate \[
| \|S\circ T\|\le \|S\|\,\|T\| |
|---|
\] shows that composition is continuous with respect to the operator norm. This property is fundamental in iterative methods and dynamical systems.
4.3 Identity and zero operators
The identity operator leaves every vector unchanged and has operator norm \(1\) on any nonzero normed space. The zero operator sends every vector to \(0\) and has norm \(0\). These two operators serve as the basic reference points for the algebraic structure of \(B(X)\).
4.4 Algebra of operators
When \(X=Y\), bounded operators on \(X\) can be added and composed, giving an algebra. Many important questions concern how algebraic relations among operators influence their spectra, norms, and long-term behavior.
4.4.1 Banach algebra structure
If \(X\) is complete, then \(B(X)\) is complete in the operator norm and becomes a Banach algebra. Completeness means that limits of operator-norm Cauchy sequences remain bounded operators. This result underlies many powerful tools in analysis, including resolvent methods and perturbation theory.
5 Bounded operators on Banach spaces
When the underlying space is complete, bounded operators enjoy additional structural results. These theorems are among the most important in functional analysis because they guarantee that analytic behavior persists under limits and inversion.
5.1 Completeness of operator spaces
If \(X\) and \(Y\) are Banach spaces, then \(B(X,Y)\) is also a Banach space with the operator norm. This completeness makes it possible to use convergence arguments for sequences of operators, which is essential in approximation theory and the study of differential equations.
5.2 Closed graph and bounded inverse ideas
A linear operator between Banach spaces is bounded if its graph is closed and the domain is all of \(X\). Similarly, if a bounded operator is bijective between Banach spaces, then its inverse is automatically bounded. These principles often allow one to deduce continuity from purely algebraic or geometric hypotheses.
5.3 Open mapping theorem connections
The open mapping theorem states that a surjective bounded linear operator between Banach spaces maps open sets to open sets. This has many consequences, including the bounded inverse theorem and the closed graph theorem. Together, these results form a core part of the structural theory of Banach spaces.
6 Bounded operators on Hilbert spaces
Hilbert spaces provide additional geometric structure through inner products. In this setting, bounded operators can be studied using orthogonality, adjoints, and spectral decompositions, which often make the theory more explicit.
6.1 Adjoint operators
For a bounded operator \(T\) on a Hilbert space, there is a unique adjoint operator \(T^*\) satisfying \[ \langle Tx, y\rangle = \langle x, T^*y\rangle. \] The adjoint reflects how \(T\) interacts with the inner product and plays a central role in many classification results.
6.2 Self-adjoint and normal operators
An operator is self-adjoint if \(T=T^*\), and normal if \(TT^*=T^*T\). Self-adjoint operators model observable quantities in many contexts and have real spectral behavior. Normal operators include self-adjoint, unitary, and many multiplication operators, and they admit especially tractable spectral analysis.
6.3 Unitary and isometric operators
A unitary operator preserves inner products and hence norms. More generally, an isometry preserves distances and lengths, though it need not be surjective. These operators represent rigid symmetries of Hilbert spaces and preserve important geometric and spectral features.
6.4 Orthogonal projections
Orthogonal projections map each vector to its nearest point in a closed subspace. They are self-adjoint idempotent operators satisfying \(P^2=P\) and \(P^*=P\). Their norm is \(1\) unless the projection is zero, and they are essential in decomposition theory and approximation.
7 Spectral theory
Spectral theory studies how operators behave through the values that make them fail to be invertible. For bounded operators, the spectrum generalizes eigenvalues and provides a powerful framework for understanding structure and dynamics.
7.1 Spectrum and resolvent
The spectrum of a bounded operator \(T\) consists of complex numbers \(\lambda\) for which \(T-\lambda I\) is not invertible. The resolvent set is its complement, and the resolvent operator \((T-\lambda I)^{-1}\) is defined there. This framework encodes both algebraic and analytic information about \(T\).
7.2 Spectral radius
The spectral radius is the largest modulus of points in the spectrum. For bounded operators, it can be estimated by powers of the operator and is always no larger than the operator norm. In many cases, it captures long-term growth under iteration.
7.3 Eigenvalues and eigenvectors
An eigenvalue \(\lambda\) is a scalar for which there exists a nonzero vector \(x\) with \(Tx=\lambda x\). Such vectors are eigenvectors. While not every point of the spectrum is an eigenvalue, eigenvalues remain especially important because they provide direct information about invariant directions and modes of action.
7.4 Spectral properties of compact operators
Compact operators have particularly rich spectral behavior. Their nonzero spectrum consists only of eigenvalues of finite multiplicity, with possible accumulation only at zero. This structure makes them resemble finite-dimensional operators in many respects, even though they act on infinite-dimensional spaces.
8 Special classes of bounded operators
Certain families of bounded operators play a privileged role because they have additional structure or especially useful approximation properties. These classes appear repeatedly in analysis and its applications.
8.1 Compact operators
A compact operator maps bounded sets to relatively compact sets. Such operators often arise as integral operators or as limits of finite-dimensional approximations. They are central to eigenvalue problems and to the study of perturbations.
8.2 Finite-rank operators
Finite-rank operators have finite-dimensional range. They are the simplest nontrivial bounded operators and often serve as building blocks for approximations of more complicated maps. Many compact operators can be approximated by finite-rank operators in suitable settings.
8.3 Isometries
Isometries preserve norms and distances. They may be injective without being surjective, especially in infinite-dimensional spaces. Their rigidity makes them useful in geometry, harmonic analysis, and semigroup theory.
8.4 Contractions
A contraction is a bounded operator with norm at most \(1\). Contractions are central in fixed-point arguments, iteration schemes, and stability analysis. Their norm control often leads to convergence results for repeated application.
8.5 Fredholm operators
Fredholm operators are bounded operators with finite-dimensional kernel and cokernel and closed range. They occupy an important middle ground between invertible operators and compact perturbations of invertible ones. Their index is stable under many perturbations and is a key invariant in analysis.
9 Applications in applied mathematics
Bounded operators provide a unifying language for many applied problems where linearity and stability are essential. Their norm estimates are often what make existence, convergence, and error control possible.
9.1 Stability of dynamical systems
In discrete and continuous dynamical systems, bounded operators describe linear evolution and perturbations. Norm bounds help determine whether solutions remain controlled over time or grow too rapidly. Contractions in particular are associated with stable iterative behavior.
9.2 Solving linear operator equations
Many equations can be written as \(Tx=f\), where \(T\) is a bounded operator. Solvability depends on invertibility, spectrum, and perturbation size. In practical settings, one often studies approximate solutions and the robustness of the equation under changes in data.
9.3 Numerical analysis and discretization
Numerical methods often replace infinite-dimensional operators with matrices or finite-rank approximations. Boundedness helps ensure that discretization does not introduce uncontrolled errors. Operator norms provide a natural way to measure approximation quality and algorithmic stability.
9.4 Signal processing and Fourier methods
In signal processing, bounded operators model filtering, modulation, translation, and reconstruction. Fourier methods often convert complicated operators into multiplication or diagonal action in frequency space. This viewpoint simplifies analysis and clarifies how signals are transformed.
9.5 Quantum mechanics and control theory
In quantum mechanics, bounded operators on Hilbert spaces represent many observables and physical transformations in an idealized mathematical framework. In control theory, bounded operators describe linear input-output systems and help assess responsiveness and stability. In both areas, operator norms and spectra are key tools for understanding system behavior.