1 Definition and basic concepts

A self-adjoint operator is a linear operator that coincides with its adjoint relative to an inner product. The notion is most transparent in finite-dimensional linear algebra, but it becomes more delicate in infinite-dimensional analysis, where domain issues are essential. Self-adjointness is one of the main structural conditions ensuring that an operator has a stable spectral theory.

1.1 Inner product spaces and adjoints

An inner product space is a vector space equipped with a function that measures angles and lengths. For a linear operator on such a space, the adjoint is defined by shifting the operator from one side of the inner product to the other. When this is possible for all vectors in the relevant space, the adjoint captures how the operator interacts with the geometry of the space.

1.2 Formal definition of a self-adjoint operator

An operator is self-adjoint when it equals its adjoint. In finite dimensions, this means the matrix representing the operator is equal to its conjugate transpose. In infinite dimensions, the definition requires not only equality of formulas but also equality of domains, so that the operator and its adjoint are defined on the same set.

1.3 Symmetric versus self-adjoint operators

A symmetric operator satisfies the inner-product identity that characterizes an adjoint relation, but it need not have the same domain as its adjoint. Self-adjointness is therefore a stronger condition. This distinction is minor in finite-dimensional settings, yet it becomes central for unbounded operators on Hilbert spaces.

1.3.1 Difference in finite-dimensional spaces

In finite-dimensional spaces, every symmetric operator is automatically self-adjoint once the underlying field and inner product are fixed appropriately. There is no meaningful gap between the formal symmetry condition and full self-adjointness because all linear maps are defined on the entire space.

1.3.2 Difference in infinite-dimensional spaces

In infinite-dimensional spaces, an operator may satisfy the symmetry identity on its domain without being self-adjoint. This typically happens when the domain is too small or when the adjoint has a larger domain than the original operator. Such situations arise frequently for differential operators and other unbounded examples.

1.4 Examples of self-adjoint operators

Common examples include real diagonal matrices, Hermitian matrices, multiplication by a real-valued function on a function space, and many classical differential operators with suitable boundary conditions. In each case, self-adjointness reflects a compatibility between the operator and the inner product structure. These examples illustrate both algebraic and analytic forms of the same concept.

2 Finite-dimensional case

In finite-dimensional linear algebra, self-adjoint operators are among the best understood classes of transformations. Their theory is closely tied to matrix identities, orthogonality, and diagonalization. Many results that are subtle in infinite dimensions become straightforward here.

2.1 Hermitian matrices

A Hermitian matrix is a complex matrix equal to its conjugate transpose. Such matrices represent self-adjoint operators on complex inner product spaces. Their entries satisfy a pattern of conjugate symmetry across the diagonal.

2.2 Real symmetric matrices

A real symmetric matrix is one that equals its transpose. Over real inner product spaces, these matrices represent self-adjoint operators. They form the real counterpart of Hermitian matrices and share the same principal spectral properties.

2.3 Diagonalization

Self-adjoint matrices can be diagonalized by a change of orthonormal basis. This means the operator is reduced to a diagonal form whose entries are its eigenvalues. Diagonalization is one of the key reasons self-adjoint operators are so useful.

2.3.1 Orthonormal eigenbases

A self-adjoint operator on a finite-dimensional inner product space admits an orthonormal basis made of eigenvectors. This basis organizes the space into mutually perpendicular directions on which the operator acts by scalar multiplication. The result gives a clear geometric picture of the transformation.

2.3.2 Unitarly diagonalizable operators

A self-adjoint matrix is unitarily diagonalizable, meaning it can be written as a unitary change of basis followed by a diagonal matrix. The unitary factor preserves inner products and therefore preserves lengths and angles. This structure is a precise algebraic expression of orthogonal decomposition.

2.4 Spectral theorem in finite dimensions

The finite-dimensional spectral theorem states that a self-adjoint operator has real eigenvalues and an orthonormal eigenbasis. It also implies that the operator is completely determined by its spectrum and eigenvectors. This theorem serves as the model for more general spectral results in analysis.

3 Infinite-dimensional case

In infinite-dimensional analysis, self-adjoint operators are typically studied on Hilbert spaces. Unlike in finite dimensions, operators may be unbounded and defined only on a dense domain. These features make the theory more technical but also more powerful.

3.1 Hilbert spaces

A Hilbert space is a complete inner product space. Completeness allows limits of Cauchy sequences to remain inside the space, which is essential for operator theory and spectral analysis. Many spaces of functions used in analysis and physics are Hilbert spaces.

3.2 Bounded self-adjoint operators

A bounded self-adjoint operator is defined on all of a Hilbert space and satisfies the adjoint equality. Boundedness simplifies many arguments because the operator is continuous and has a well-behaved spectrum. Such operators retain the core spectral features of finite-dimensional self-adjoint matrices.

3.3 Unbounded self-adjoint operators

Unbounded self-adjoint operators are central in analysis because differential operators are often unbounded. Their definition requires careful attention to the set of vectors on which they act. Despite the added complexity, they still enjoy strong spectral properties when properly defined.

3.3.1 Domains of definition

The domain of an unbounded operator is the subspace on which it is actually defined. For self-adjointness, this domain must match the domain of the adjoint operator. Much of the theory concerns identifying the correct domain so that the operator becomes self-adjoint.

3.3.2 Closedness and extensions

A self-adjoint operator is always closed, meaning its graph is closed in the product space. Some symmetric operators are not self-adjoint until they are extended to a larger domain. Determining whether such an extension exists is a major part of operator theory.

3.4 Densely defined operators

To have an adjoint in the usual sense, an operator must be densely defined. This means its domain must be large enough to approximate every vector in the Hilbert space. Dense definition is a basic hypothesis for the standard theory of unbounded adjoints and self-adjointness.

4 Spectral properties

Self-adjoint operators have a rich spectral structure that generalizes the eigenvalue theory of matrices. The spectrum replaces the finite list of eigenvalues with a possibly continuous set of spectral values. This framework is one of the main reasons self-adjoint operators are so important.

4.1 Real spectrum

The spectrum of a self-adjoint operator lies on the real line. This fact distinguishes self-adjoint operators from general linear operators, whose spectra may be complex. Real spectrum is especially significant in applications where observables are expected to yield real measurements.

4.2 Eigenvalues and eigenvectors

When a self-adjoint operator has eigenvalues, those eigenvalues are real and eigenvectors for distinct eigenvalues are orthogonal. In infinite dimensions, however, not every point of the spectrum need be an eigenvalue. As a result, eigenvectors describe only part of the full spectral picture.

4.3 Spectral measures

Spectral measures decompose a self-adjoint operator into contributions from subsets of the real line. They allow the operator to be analyzed as an integral over its spectrum rather than as a finite sum. This measure-theoretic viewpoint is fundamental in modern spectral theory.

4.4 Functional calculus

Functional calculus lets one apply functions to a self-adjoint operator in a systematic way. It extends the idea of substituting a matrix into a polynomial or continuous function. The resulting operator reflects the same spectral information in transformed form.

4.4.1 Polynomial functional calculus

For polynomials, one can define p(T) by substituting the operator into the polynomial expression. This procedure is straightforward and works for all linear operators where composition is defined. For self-adjoint operators, it preserves many algebraic relations and is often the first step toward more advanced calculus.

4.4.2 Continuous functional calculus

Continuous functional calculus extends the polynomial case to continuous functions on the spectrum. It is especially powerful for bounded self-adjoint operators. Through it, one can define square roots, exponentials, and other derived operators.

4.5 Resolvent set and spectrum decomposition

The resolvent set consists of complex numbers for which the operator minus that scalar multiple of the identity has a bounded inverse. The complement is the spectrum. Spectral decomposition further distinguishes point, continuous, and residual components, although self-adjoint operators have especially restricted behavior in this regard.

5 Quadratic forms and positivity

Self-adjoint operators are closely linked to quadratic expressions built from inner products. These expressions reveal positivity and lower bounds, which are important in analysis and physics. Quadratic forms often provide a more flexible way to study operators than direct matrix or differential expressions.

5.1 Quadratic forms associated with operators

A quadratic form associated with an operator is typically given by evaluating the inner product of a vector with its image under the operator. Such forms summarize how the operator acts on vectors in terms of energy-like quantities. They are widely used when the operator itself is difficult to handle directly.

5.2 Positive and nonnegative self-adjoint operators

A self-adjoint operator is positive if its quadratic form is nonnegative for all vectors in its domain. Nonnegative operators are a slightly weaker but closely related class. These operators have especially useful spectral properties and often admit square roots.

5.3 Semibounded operators

An operator is semibounded if its quadratic form is bounded below by some real number. This condition generalizes positivity while still allowing useful estimates. Semibounded self-adjoint operators are common in variational methods and differential equations.

6 Extensions and classification

Not every symmetric operator is self-adjoint, so one important problem is deciding whether it can be extended to a self-adjoint operator. Classification results describe the possible extensions and the obstructions to them. These ideas are particularly important for differential operators with boundary conditions.

6.1 Self-adjoint extensions

A self-adjoint extension is a larger self-adjoint operator that agrees with a given symmetric operator on its original domain. The theory of extensions identifies when such enlargements exist and how many are possible. This is a central tool in the analysis of unbounded operators.

6.1.1 Von Neumann’s theory

Von Neumann’s theory gives a criterion for self-adjoint extensions in terms of deficiency spaces. It translates the extension problem into a comparison of certain subspaces associated with the adjoint operator. This framework is one of the standard methods in operator theory.

6.1.2 Deficiency indices

Deficiency indices measure the dimensions of the deficiency spaces. They determine whether self-adjoint extensions exist and, in favorable cases, how many there are. Equal deficiency indices are often the key condition for the existence of such extensions.

6.2 Essential self-adjointness

An operator is essentially self-adjoint if it has a unique self-adjoint extension, namely its closure. This property is valuable because it shows that a formally symmetric operator determines a canonical self-adjoint object. Many operators arising from physics are studied first in this way.

6.3 Closable operators

An operator is closable if it has a well-defined closure as an operator. Closability is weaker than self-adjointness but often serves as a starting point. If the closure is self-adjoint, the original operator can be viewed as a dense approximation to the full self-adjoint operator.

7 Applications

Self-adjoint operators appear throughout analysis and mathematical physics. Their real spectrum and orthogonal decomposition properties make them natural models for measurable quantities and stable dynamical systems. They also underpin many existence and uniqueness results for differential equations.

7.1 Quantum mechanics

In quantum mechanics, observable quantities are represented by self-adjoint operators. The spectral decomposition of such operators corresponds to possible measurement outcomes. This association is one of the foundational links between operator theory and physics.

7.2 Differential operators

Many differential operators become self-adjoint after suitable domain choices and boundary conditions are imposed. Examples include the Laplacian and certain second-order operators. Self-adjointness ensures that these operators have a robust spectral theory and well-controlled solutions.

7.3 Sturm-Liouville theory

Sturm-Liouville theory studies a class of second-order differential operators with boundary conditions that often yield self-adjointness. The eigenvalue problems arising there produce orthogonal families of functions. This theory is a classical source of special functions and expansion methods.

7.4 PDEs and boundary value problems

In partial differential equations, self-adjoint operators help formulate boundary value problems in a way that supports existence, uniqueness, and energy estimates. They also provide a framework for separation of variables and spectral expansion. Many linear evolution equations are most naturally expressed using self-adjoint generators.

Self-adjoint operators sit within a broader family of operator classes related by adjoints and symmetry. Understanding these neighboring notions clarifies what is special about the self-adjoint condition. Several of these concepts overlap in finite dimensions but differ more sharply in infinite-dimensional settings.

8.1 Normal operators

A normal operator commutes with its adjoint. Self-adjoint operators are normal, but not every normal operator is self-adjoint. Normality leads to a spectral theorem of its own, with self-adjointness as a special case.

8.2 Unitary operators

A unitary operator preserves the inner product and hence lengths and angles. While self-adjoint operators are associated with real spectra, unitary operators are associated with spectrum on the unit circle. Both classes are central in operator theory.

8.3 Skew-adjoint operators

A skew-adjoint operator is the negative of its adjoint. Such operators are closely related to self-adjoint ones by multiplication by the imaginary unit. They are often used in the study of continuous-time evolution and generator theory.

8.4 Symmetric bilinear forms

Symmetric bilinear forms are algebraic expressions that are unchanged when their arguments are swapped. They provide a finite-dimensional analogue of quadratic forms associated with self-adjoint operators. In analysis, they often serve as a bridge between form methods and operator methods.