1 General concepts

1.1 Definition and purpose

In mathematical analysis, decomposition is the act of splitting an object into simpler parts that are easier to examine or use. The object may be a function, measure, operator, space, or domain. The aim is often to isolate structure, reduce complexity, or make calculations more tractable.

Decomposition is valuable because many analytic problems become manageable once they are expressed in smaller pieces. These parts may interact in a controlled way, allowing one to study each component separately and then combine the results.

1.2 Decomposition versus representation

A decomposition is related to a representation, but the two notions are not identical. A representation describes an object in some chosen form, while a decomposition emphasizes breaking the object into constituent pieces. For example, writing a function as a series is a representation; viewing that series as a sum of simpler building blocks is a decomposition.

In practice, the distinction is sometimes subtle. Many analytic constructions serve both purposes at once, since a useful representation often reveals a natural decomposition.

1.3 Existence and uniqueness

Not every object admits a useful decomposition in a given setting. Whether such a breakdown exists depends on the structure being used, such as linearity, orthogonality, or measurability. Even when a decomposition exists, it may not be unique.

Uniqueness is often tied to additional conditions. A direct sum decomposition may be unique once the subspaces are fixed, while a series expansion may depend on the chosen basis. Canonical decompositions are especially valued because they are determined by the object itself rather than by arbitrary choices.

1.4 Canonical and non-canonical forms

Canonical decompositions arise from intrinsic properties and usually have a distinguished status. Examples include spectral decompositions of operators or the separation of a function into positive and negative parts. These forms are often stable and conceptually informative.

Non-canonical decompositions depend more heavily on choices, such as a basis, partition, or coordinate system. They can still be extremely useful, especially for computation or approximation, even when they are not unique.

2 Decomposition in function analysis

2.1 Series expansions

A central theme in function analysis is expressing a function as a sum of simpler functions. These expansions may be finite or infinite, depending on the context and on the function space under consideration.

Series expansions often serve as both analytical tools and computational devices. They can reveal smoothness, periodicity, localization, or frequency content.

2.1.1 Fourier series

Fourier series decompose periodic functions into sums of sines and cosines, or equivalently into complex exponentials. This form is especially effective for analyzing oscillatory behavior and frequency content.

The coefficients in a Fourier series measure how much of each basic wave is present. Under suitable hypotheses, the series converges to the original function in an appropriate sense.

2.1.2 Orthogonal expansions

Orthogonal expansions generalize Fourier series by using any orthogonal family of functions. Common examples include expansions in Legendre polynomials, Hermite functions, or orthonormal bases in Hilbert spaces.

Orthogonality simplifies coefficient extraction and often leads to efficient error estimates. Such expansions are widely used in approximation theory and in the study of function spaces.

2.1.3 Wavelet decompositions

Wavelet decompositions break a function into components localized in both scale and position. Unlike purely global expansions, wavelets can capture sharp changes and local features effectively.

This makes them especially useful for signals with nonuniform behavior. They also support multiresolution analysis, where coarse and fine details are separated systematically.

2.2 Decomposition into even and odd parts

Any suitably defined function on a symmetric domain can often be written as the sum of an even part and an odd part. The even part is symmetric under sign change of the variable, while the odd part changes sign.

This decomposition is elementary but useful. It can simplify integration, differential equations, and symmetry arguments by separating contributions with distinct parity.

2.3 Decomposition into positive and negative parts

A real-valued function can be split into its positive and negative parts. The positive part records where the function is above zero, and the negative part records where it is below zero, typically with a sign convention that keeps both pieces nonnegative.

This decomposition is important in integration theory and in measure-related arguments. It allows one to handle signed quantities through nonnegative components.

2.4 Decomposition of functions by domain

Functions can also be decomposed by restricting them to subdomains or by expressing them piecewise. This is common when a function behaves differently on different intervals or regions.

Such domain-based decompositions are useful in local analysis, especially when singularities, discontinuities, or boundary effects are present. They also underpin piecewise definitions in numerical methods.

3 Decomposition in linear algebra and functional analysis

3.1 Vector space decompositions

Vector spaces often admit decompositions into smaller subspaces whose algebraic relationship is well understood. These decompositions clarify the structure of linear problems and make coordinates easier to manage.

They are especially useful when a space can be split according to invariant properties, geometric constraints, or orthogonality relations.

3.1.1 Direct sums

A direct sum decomposition expresses a space as a sum of subspaces with trivial overlap. Each vector can then be written uniquely as a sum of components from the specified subspaces.

Direct sums are among the most basic and important forms of decomposition in linear algebra. They provide a clean framework for separating independent degrees of freedom.

3.1.2 Orthogonal decompositions

In inner product spaces, orthogonal decompositions split a vector into mutually perpendicular components. This structure is central in Hilbert space theory and in least-squares methods.

Orthogonality often yields explicit formulas for projections and norms. It also makes it possible to analyze geometry through simple Pythagorean relationships.

3.1.3 Complemented subspaces

A subspace is complemented if there exists another subspace such that the whole space is their direct sum. Complemented subspaces are significant because they permit projection operators and a controlled splitting of the ambient space.

In infinite-dimensional settings, complemented subspaces need not exist for every subspace. This makes the concept important in functional analysis, where geometry and topology interact closely.

3.2 Operator decompositions

Operators can often be decomposed into simpler constituents that reflect algebraic or analytic properties. Such decompositions are central to understanding linear transformations on finite- and infinite-dimensional spaces.

They may separate magnitude from rotation, isolate eigenstructure, or reduce an operator to standard blocks.

3.2.1 Spectral decomposition

Spectral decomposition expresses an operator in terms of its eigenvalues and associated eigenspaces, or more generally through spectral measures. It is one of the most powerful tools in linear and functional analysis.

This decomposition clarifies how an operator acts on different modes. In self-adjoint contexts, it underlies much of modern operator theory.

3.2.2 Polar decomposition

Polar decomposition factors an operator into a product of a partial isometry and a positive operator. This is analogous to writing a complex number in terms of magnitude and phase.

The decomposition separates directional behavior from size. It is useful in matrix analysis, operator theory, and geometry.

3.2.3 Jordan-type decompositions

Jordan-type decompositions refine the structure of linear operators by organizing them into blocks associated with eigenvalues and generalized eigenvectors. They are particularly important when an operator is not diagonalizable.

Such decompositions reveal nilpotent structure and the extent to which an operator fails to have a basis of eigenvectors. They are widely used in finite-dimensional linear algebra.

3.3 Decomposition of Hilbert and Banach spaces

Hilbert and Banach spaces may be decomposed into subspaces or into orthogonal and complemented components, depending on the available structure. In Hilbert spaces, inner products make orthogonal splitting especially natural.

In Banach spaces, decompositions are more delicate because orthogonality is absent in general. As a result, questions of complementability and bounded projections become central.

4.1 Decomposition of measures

Measures can often be decomposed into parts with different analytic behavior. This is fundamental in modern measure theory and in the study of integration against signed or complex measures.

Such decompositions help separate absolute continuity from singularity, or positive from negative contributions.

4.1.1 Lebesgue decomposition

The Lebesgue decomposition theorem splits one measure into an absolutely continuous part and a singular part relative to another measure. This distinction describes how a measure relates to a reference measure.

The theorem is a cornerstone of measure theory. It is frequently used when comparing integrals, densities, and singular phenomena.

4.1.2 Hahn decomposition

Hahn decomposition applies to signed measures and partitions the space into a positive set and a negative set. On the positive set, the measure assigns nonnegative values to measurable subsets, and on the negative set it assigns nonpositive values.

This result is a basic step in understanding signed measures. It also leads naturally to the Jordan decomposition.

4.1.3 Radon–Nikodym decomposition

The Radon–Nikodym theorem provides a derivative-like description of an absolutely continuous measure with respect to another measure. In decomposition form, it expresses a measure through a density relative to a reference measure, together with any singular remainder when combined with Lebesgue decomposition.

This theorem connects measure decomposition with differentiation in an abstract setting. It is essential in probability, integration theory, and functional analysis.

4.2 Decomposition of functions into measurable components

Functions may be broken into measurable pieces through truncation, level sets, or positive-negative separation. These techniques are useful when one wants to reduce an argument to bounded, simple, or otherwise manageable functions.

Measurable decomposition is especially important in integration theory, where approximation by simple functions often provides a route to defining or studying the integral.

4.3 Decomposition of domains and integrals

Integration over a complicated region is often simplified by dividing the domain into smaller parts. The integral over the full region is then expressed as a sum of integrals over subregions.

This approach is common in the use of partitions, change of variables, and local estimates. It is also the basis for many convergence and comparison arguments.

5 Decomposition in approximation and numerical analysis

5.1 Piecewise decomposition

Approximation often relies on dividing a problem into pieces on which simpler models are accurate. Piecewise methods are especially effective for functions with varying smoothness or localized features.

They allow one to tailor the approximation to the geometry of the data or the behavior of the target function.

5.1.1 Partitioning intervals

Partitioning an interval means subdividing it into smaller subintervals for separate treatment. This is a basic strategy in numerical integration, interpolation, and approximation.

Finer partitions generally improve local accuracy, though they may increase computational cost. The choice of partition often reflects the function’s variation.

5.1.2 Piecewise polynomial approximation

Piecewise polynomial approximation replaces a function by different polynomials on different subintervals or regions. Splines are a common example of this approach.

Such approximations balance flexibility and efficiency. They are widely used because they can fit complex shapes while remaining computationally manageable.

5.2 Decomposition for numerical integration

Numerical integration frequently uses decomposition to break an integral into simpler pieces. Each piece can then be estimated using quadrature rules adapted to the local behavior of the integrand.

This strategy improves accuracy when a function has sharp variations, oscillations, or localized irregularities. It is also a standard way to handle integrals over composite domains.

5.3 Error analysis via decomposition

Error estimates often become clearer when the total error is decomposed into separate contributions. For example, one may isolate truncation error, discretization error, and rounding error.

By analyzing each component individually, one can identify the dominant source of inaccuracy. This makes decomposition a practical tool in numerical analysis as well as a theoretical one.

5.4 Multiscale methods

Multiscale methods decompose a problem into levels of resolution. Coarse components capture broad trends, while finer components record detail.

These methods are especially effective in situations where structure appears at different scales. Wavelet-based algorithms and hierarchical solvers are prominent examples.

6 Applications and examples

6.1 Solving differential equations

Decomposition is widely used in differential equations, where a complicated solution is expressed in terms of simpler modes, basis functions, or local pieces. Separation of variables is a classical example, as are eigenfunction expansions.

The method can transform a difficult equation into a family of simpler subproblems. This is particularly useful for boundary value problems and linear partial differential equations.

6.2 Signal analysis and reconstruction

In signal analysis, decomposition separates a signal into frequency, scale, or component-wise contributions. Fourier and wavelet methods are common tools for this purpose.

Reconstruction combines the components to recover the original signal or an approximation to it. Decomposition also helps identify noise, trends, and localized events.

6.3 Simplifying proof strategies

Analytic proofs often become easier when the object under study is decomposed into manageable parts. A function may be split into bounded and unbounded portions, or a measure into absolutely continuous and singular pieces.

This technique can reduce a general statement to several simpler claims. It is a standard method in both rigorous proofs and heuristic reasoning.

6.4 Worked analytical examples

A simple example is the decomposition of a real-valued function \(f\) into even and odd parts: \[ f(x)=\frac{f(x)+f(-x)}{2}+\frac{f(x)-f(-x)}{2}. \] The first term is even, and the second is odd.

Another example is the decomposition of a signed function \(f\) into positive and negative parts: \[ f=f^+-f^-, \] where \(f^+\) and \(f^-\) are nonnegative and do not overlap. Such formulas illustrate how decomposition converts one object into components with clearer properties.