Distribution (mathematics): generalized functions and their uses

Definition and basic properties

Formally, a distribution is a continuous linear functional on a space of test functions (typically the space of infinitely differentiable functions with compact support). This dual space is commonly denoted by D'. Classical functions that are locally integrable define distributions by integration against test functions, and such distributions are called regular. Measures and other generalized densities also embed naturally into this framework, so distributions unify many objects that fail to be pointwise-defined functions.

Common examples

Among the best-known distributions is the Dirac delta delta distribution, which models a unit point mass or point charge: it vanishes on every test function except that its action returns the value of the test function at a specified point. Other examples include the Heaviside step as a distributional primitive of the delta, derivatives of the delta (useful to represent dipoles), and principal value distributions that arise from limiting procedures. Tempered distributions form a subspace suited to Fourier analysis, allowing the transform to be defined for many singular objects (Fourier analysis).

Operations and limits

Distributions admit many operations that extend classical calculus, but some classical operations require care. Typical allowed operations include:

• Differentiation: every distribution can be differentiated any number of times, defined by how it acts on derivatives of test functions.
• Multiplication by smooth functions: a distribution can be multiplied by a smooth function, producing another distribution.
• Convolution: convolution with a compactly supported distribution or a suitable function is defined under compatibility conditions.

However, the product of two arbitrary distributions is not always defined; additional structure or regularization is needed in such cases.

History, motivation and applications

The theory arose from problems in physics where idealized sources—point masses, charge distributions, or impulses—cannot be represented by ordinary functions. Distributions provide a flexible language for boundary value problems and weak formulations, enabling existence and uniqueness results for partial differential equations (PDEs). In physics they underpin formal manipulations in quantum field theory and electrodynamics (QED), while engineers use related ideas in signal processing to handle impulses and filters (signal processing).

Importance and distinctions

Distributions are central to the concept of weak solutions in functional analysis and PDE theory and connect closely with Sobolev spaces and spectral methods. They should be seen as linear functionals rather than pointwise-defined values, which explains their ability to represent singular phenomena. For computational and theoretical work, tempered distributions and spaces of test functions with particular topologies are chosen to match the desired operations, for example to make the Fourier transform available and well-behaved.

For introductory material and historical notes see general expositions on the subject functions and generalized functions, and for applied perspectives consult resources on PDEs, Fourier methods, and physics (Fourier analysis) and (partial differential equations).