Harmonic Analysis · aiwiki.page

Harmonic analysis is a branch of mathematical analysis that studies functions and operators by decomposing them into elementary oscillatory components and examining how those components interact. Its classical foundations are Fourier series and the Fourier transform. Modern harmonic analysis also investigates singular integrals, maximal functions, frequency localization, and decompositions associated with symmetry groups. Its central questions concern reconstruction, convergence, regularity, and quantitative bounds for operators. (math.ucla.edu)

Fourier series and frequency decomposition

For an integrable function $f$ of period $2\pi$ , its Fourier coefficients are

\widehat f(n)=\frac{1}{2\pi}\int_{-\pi}^{\pi} f(x)e^{-inx}\,dx,\qquad n\in\mathbb Z.

The associated Fourier series is formally

f(x)\sim\sum_{n\in\mathbb Z}\widehat f(n)e^{inx}.

Each exponential represents a frequency, while its coefficient records the component’s amplitude and phase. The symbol $\sim$ does not assert convergence: determining whether and in what sense the series reconstructs $f$ is part of the theory. (arxiv.org)

In the Hilbert space $L^2([-\pi,\pi])$ , equipped with the normalized inner product

\langle f,g\rangle=\frac{1}{2\pi}\int_{-\pi}^{\pi} f(x)\overline{g(x)}\,dx,

the functions $e^{inx}$ form an orthonormal basis. Fourier partial sums converge in the $L^2$ norm, and Parseval’s identity states

\frac{1}{2\pi}\int_{-\pi}^{\pi}|f(x)|^2\,dx =\sum_{n\in\mathbb Z}|\widehat f(n)|^2.

Thus frequency decomposition preserves the squared norm. Convergence in this norm is distinct from pointwise convergence and uniform convergence, which require separate analysis. (arxiv.org)

The Fourier transform

For nonperiodic functions on Euclidean space $\mathbb R^d$ , frequencies vary continuously. One common normalization defines

\widehat f(\xi)=\int_{\mathbb R^d} f(x)e^{-2\pi i x\cdot\xi}\,dx.

For sufficiently regular, rapidly decreasing functions, inversion gives

f(x)=\int_{\mathbb R^d} \widehat f(\xi)e^{2\pi i x\cdot\xi}\,d\xi.

The Plancherel theorem extends the transform to a unitary operator on $L^2(\mathbb R^d)$ , with

\|\widehat f\|_2=\|f\|_2.

For general $L^2$ functions, this extension need not be represented by an absolutely convergent defining integral. (arxiv.org)

The transform converts convolution into multiplication:

\widehat{f*g}=\widehat f\,\widehat g.

Under suitable regularity and decay assumptions, it also converts a partial derivative into multiplication by frequency:

\widehat{\partial_j f}(\xi)=2\pi i\xi_j\widehat f(\xi).

These identities connect integral operators, differentiation, and algebraic operations in frequency space. (arxiv.org)

Operator estimates and real-variable methods

A major part of harmonic analysis concerns inequalities of the form

\|Tf\|_{L^q}\leq C\|f\|_{L^p},

where $T$ is an operator and $C$ is independent of $f$ . Such estimates describe whether a transformation preserves or changes integrability. The framework relies on $L^p$ spaces, measure theory, and functional analysis. (math.ucla.edu)

Two fundamental classes of operators are:

Maximal functions, which take suprema of local averages of $|f|$ . They control exceptional sets and support results about differentiation and convergence.
Singular integral operators, whose kernels are not ordinarily integrable at a singularity. Their behavior depends on cancellation as well as kernel size and regularity. Calderón–Zygmund theory provides a systematic framework for establishing their boundedness. (math.ucla.edu)

A prototype is the Hilbert transform,

Hf(x)=\frac{1}{\pi}\operatorname{p.v.} \int_{\mathbb R}\frac{f(y)}{x-y}\,dy.

The principal value excludes a symmetric neighborhood of the singularity before taking a limit. The transform is bounded on $L^p(\mathbb R)$ for $1<p<\infty$ ; endpoint behavior requires different estimates. (math.ucla.edu)

Localization, scales, and oscillation

Fourier components extend across space, so many problems require decompositions that also track location and scale. Littlewood–Paley theory divides frequency space into bands, often at geometrically increasing scales, and relates the resulting pieces to function-space norms. This helps quantify regularity and analyze interactions between low and high frequencies, including in Sobolev spaces. (math.ucla.edu)

Wavelets provide another form of multiscale decomposition, using translated and dilated functions to combine spatial and frequency information. Oscillatory-integral methods instead exploit cancellation in expressions with rapidly varying phases. Both approaches extend harmonic analysis beyond finding explicit Fourier coefficients. (math.ucla.edu)

Harmonic analysis on groups

Abstract harmonic analysis extends frequency decomposition to groups. This connects the subject with group theory and representation theory: elementary components are described by representations of the underlying symmetry group. Classical Fourier series correspond to the circle group, whose irreducible representations are the one-dimensional functions $e^{in\theta}$ . (math.columbia.edu)

For compact Lie groups, the Peter–Weyl theorem generalizes Fourier-series decomposition. With respect to normalized invariant Haar measure, suitably normalized matrix coefficients of irreducible unitary representations form an orthonormal basis of $L^2(G)$ . For noncommutative groups, these representations can have dimension greater than one, so scalar Fourier coefficients are replaced by matrix-valued data. (math.columbia.edu)

Connections and applications

Harmonic analysis supplies estimates and decompositions used in partial differential equations, number theory, complex analysis, and geometric measure theory. Its Fourier and wavelet methods also underpin parts of signal processing and approximation theory. The emphasis varies: applications may require reconstruction and efficient representation, while theoretical problems often focus on boundedness, cancellation, or regularity. (math.ucla.edu)

References

Lecture Notes 1 for 247Amath.ucla.edu
A Technical Survey of Harmonic Analysismath.ucla.edu
Notes on Harmonic Analysis Part I: The Fourier Transformarxiv.org
Notes on harmonic analysis Part II: the Fourier Seriesarxiv.org
Harmonic Analysis on Compact Lie Groups: The Peter-Weyl Theoremmath.columbia.edu
Math 247A (Harmonic Analysis)math.ucla.edu
Math 247A: Fourier Analysismath.ucla.edu