A linear integral transform that maps a function of time or space into a function of frequency, representing the original as a superposition of sinusoidal components.

Principle

Decompose signals into orthogonal frequency components so linear operations (convolution, differentiation) become algebraic manipulations in frequency coordinates.

Demonstration

The transform of f(x)=exp(-a x^2) (a>0) yields another Gaussian in frequency; convolution of two L1 functions corresponds to pointwise multiplication of their transforms.

Misapplication

Applying the transform to a non-tempered distribution without specifying distributional extension, or treating a transform that converges only in the distribution sense as an ordinary pointwise function.

Consequence

Provides spectral representations, simplifies linear differential equations, and yields Parseval-type identities that relate energy in original and frequency domains.

Reversal

The inverse transform reconstructs the original time/space function from its frequency representation; failing invertibility indicates loss of necessary domain assumptions.

Boundary

Defined classically for integrable functions or square-integrable functions and extended to tempered distributions; does not apply as a pointwise operation to arbitrary non-tempered generalized functions.

Semantic Tension

Often contrasted with the Laplace transform: both map to frequency-like variables but differ in domains of convergence and treatment of growth at infinity.

Synthesis

A canonical linear mapping between time/space and frequency descriptions that encodes a function as weighted sinusoidal components and turns convolution/differentiation into multiplication/algebraic operations.