The study of the frequency or eigencontent of a function, operator, matrix, or time series by expressing it in terms of sinusoidal modes, eigenfunctions, or eigenvalues and analyzing the relative strength and structure of those components.

Principle

A linear or linearized object can be decomposed into modes (Fourier basis, eigenfunctions, singular vectors) whose amplitudes and phases summarize persistent oscillatory or modal behavior; transforms and spectral measures convert convolutional or operator structure into multiplicative or diagonal form for analysis.

Demonstration

Compute the discrete Fourier transform of a regularly sampled time series to reveal dominant periodic components; compute the eigenvalues and eigenvectors of a symmetric operator to identify slow/fast modes for model reduction.

Misapplication

Applying a global Fourier spectrum to a strongly nonstationary signal without time–frequency adaptation, leading to misleading stationary peaks; interpreting aliasing-affected FFT peaks as true physical frequencies when sampling conditions are violated.

Consequence

Provides modal decompositions used for filtering, denoising, system identification, and reduced-order modeling; enables stability criteria via spectral radius or growth rates from eigenvalues.

Reversal

A pure time-domain or pathwise description that emphasizes transient events, arrival times, or non-oscillatory structure rather than modal content; for some problems, transient impulse responses or local time features convey more relevant information than global spectra.

Boundary

Applies most directly to linear operators, stationary stochastic processes, and sufficiently sampled deterministic signals; limitations include finite-data effects, windowing leakage, nonlinearity-induced harmonics, and situations where modes are non-orthogonal or ill-conditioned.

Semantic Tension

‘Spectrum’ can mean frequency content (Fourier power spectrum) or operator eigenvalues; power spectra, periodograms, and singular spectra are related but answer different questions about energy, variance, or operator action.

Synthesis

Spectral analysis is the set of transforms and decompositions that convert structure in time, space, or operator form into modal amplitudes and phases or eigenvalues, enabling identification, reduction, and filtering of persistent oscillatory or modal behavior while requiring attention to stationarity, sampling, and linearity assumptions.