Fourier Series — L2 Theory and Convergence

⚡ Key Takeaways

The Fourier series of an $L^2$ function converges to it in the $L^2$ norm — always, unconditionally — because $\{e^{inx}/\sqrt{2\pi}\}_{n\in\mathbb{Z}}$ is a complete orthonormal basis of $L^2([-\pi,\pi])$ . This is the Riesz–Fischer theorem in disguise. Pointwise/uniform convergence is a separate, harder, and conditional story (Carleson's theorem). Foundational to harmonic analysis, signal processing, and spectral methods; the $L^2$ theory crystallized 1900–1910 (Fischer, Riesz, Parseval), pointwise a.e. convergence resolved by Carleson in 1966.

💡 The Core Insight

$L^2([-\pi,\pi])$ is a separable Hilbert space, and the trigonometric system $\{e_n(x) = \tfrac{1}{\sqrt{2\pi}}e^{inx}\}$ is an orthonormal basis of it. Once you accept this, the entire convergence theory is the abstract Hilbert-space theory of orthonormal bases: partial sums are orthogonal projections onto finite-dimensional subspaces, Bessel's inequality becomes Parseval's equality, and "convergence" means norm convergence — which holds for free. No regularity of $f$ is needed. All the genuine difficulty in Fourier analysis lives outside the $L^2$ norm.

DefinitionFourier Coefficients and Partial Sums

For $f \in L^1([-\pi,\pi])$ (in particular $f \in L^2$ , since the interval has finite measure), the $n$ -th Fourier coefficient is

1

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

Derivation hint: This is the inner product $\langle f, e_n\rangle$ with $e_n = \tfrac{1}{\sqrt{2\pi}}e^{inx}$ , divided by $\sqrt{2\pi}$ ; it is the coefficient that minimizes mean-square error onto $e_n$ . The $N$ -th symmetric partial sum is

2

\quad S_N f(x) = \sum_{n=-N}^{N} \hat f(n)\,e^{inx}.

Derivation hint: Orthogonal projection of $f$ onto $\mathrm{span}\{e^{inx}: |n|\le N\}$ expanded in the basis $\{e^{inx}\}$ . Here $L^2([-\pi,\pi])$ carries the inner product and norm

3

\quad \langle f,g\rangle = \frac{1}{2\pi}\int_{-\pi}^{\pi} f(x)\overline{g(x)}\,dx, \qquad \|f\|_2 = \langle f,f\rangle^{1/2}.

Derivation hint: The normalization $\tfrac{1}{2\pi}$ makes $\{e^{inx}\}$ orthonormal: $\langle e^{imx}, e^{inx}\rangle = \delta_{mn}$ , by direct integration.

TheoremRiesz–Fischer /

L^2

Convergence

The system $\{e^{inx}\}_{n\in\mathbb{Z}}$ is a complete orthonormal basis of $L^2([-\pi,\pi])$ (with inner product [3]). Consequently, for every $f \in L^2([-\pi,\pi])$ :

4

\quad \lim_{N\to\infty} \|f - S_N f\|_2 = 0,

Derivation hint: Completeness $\Leftrightarrow$ the only function orthogonal to all $e^{inx}$ is $0$ a.e.; combine with Bessel to force the partial-sum tail to zero. and Parseval's identity holds:

5

\quad \|f\|_2^2 = \sum_{n=-\infty}^{\infty} |\hat f(n)|^2 = \frac{1}{2\pi}\int_{-\pi}^{\pi}|f(x)|^2\,dx.

Derivation hint: Take $N\to\infty$ in $\|f\|_2^2 = \|S_Nf\|_2^2 + \|f-S_Nf\|_2^2$ (Pythagoras), using [4]. Moreover, the map $f \mapsto (\hat f(n))_{n\in\mathbb{Z}}$ is a unitary isomorphism $L^2([-\pi,\pi]) \cong \ell^2(\mathbb{Z})$ . In particular, for any sequence $(c_n)\in\ell^2(\mathbb{Z})$ there exists $f\in L^2$ with $\hat f(n) = c_n$ .

ProofCompleteness ⟹ L² Convergence and Parseval

Step 1 — Best approximation. Fix $N$ and let $V_N = \mathrm{span}\{e^{inx}: |n|\le N\}$ . Among all $g\in V_N$ , the partial sum $S_N f$ minimizes $\|f-g\|_2$ . Indeed for $g = \sum_{|n|\le N} c_n e^{inx}$ ,

\|f-g\|_2^2 = \|f\|_2^2 - 2\,\mathrm{Re}\sum_{|n|\le N}\overline{c_n}\hat f(n) + \sum_{|n|\le N}|c_n|^2,

which (completing the square per coordinate) is minimized exactly at $c_n = \hat f(n)$ . Hence $S_Nf$ is the orthogonal projection of $f$ onto $V_N$ and $f - S_N f \perp V_N$ .

Step 2 — Pythagoras and Bessel. Since $f - S_Nf \perp S_Nf$ ,

\|f\|_2^2 = \|S_N f\|_2^2 + \|f-S_Nf\|_2^2 = \sum_{|n|\le N}|\hat f(n)|^2 + \|f-S_Nf\|_2^2.

Dropping the nonnegative remainder gives Bessel's inequality $\sum_{|n|\le N}|\hat f(n)|^2 \le \|f\|_2^2$ , so $\sum_n |\hat f(n)|^2 < \infty$ .

Step 3 — Completeness. We must show $\|f-S_Nf\|_2\to 0$ . Equivalently, the only obstruction is whether the $e^{inx}$ span a dense subspace. By the Stone–Weierstrass theorem, trigonometric polynomials are uniformly dense in the continuous $2\pi$ -periodic functions $C(\mathbb{T})$ ; uniform density on a finite-measure interval implies $L^2$ -density. And $C(\mathbb{T})$ is dense in $L^2(\mathbb{T})$ . So trigonometric polynomials are dense in $L^2$ .

Step 4 — Density ⟹ convergence. Given $\varepsilon>0$ , choose a trig polynomial $P$ of degree $M$ with $\|f-P\|_2<\varepsilon$ . For $N\ge M$ , $P\in V_N$ , and since $S_N f$ is the best $V_N$ -approximation (Step 1),

\|f - S_N f\|_2 \le \|f-P\|_2 < \varepsilon.

Thus $\|f-S_Nf\|_2\to 0$ , establishing [4].

Step 5 — Parseval. Let $N\to\infty$ in the Pythagoras identity of Step 2; the remainder vanishes by [4], yielding [5].

Step 6 — Unitarity / surjectivity (Riesz–Fischer). Given $(c_n)\in\ell^2$ , the partial sums $g_N = \sum_{|n|\le N} c_n e^{inx}$ form a Cauchy sequence in $L^2$ because $\|g_N - g_M\|_2^2 = \sum_{M<|n|\le N}|c_n|^2 \to 0$ . By completeness of $L^2$ (Riesz–Fischer's metric core), $g_N\to g$ in $L^2$ , and $\hat g(n) = c_n$ by continuity of the inner product. $\qquad\blacksquare$

∎

⚠️ Two distinct "completenesses"

The proof uses two completeness facts with the same name. (a) $L^2$ is a complete metric space (Cauchy sequences converge) — this gives surjectivity in Step 6. (b) The trigonometric system is complete as an orthonormal family (no nonzero vector orthogonal to all of it) — this gives [4]. Do not conflate them; both are essential and they are logically independent statements.

mathematician · Frigyes Riesz | 1880–1956

Riesz, working in Hungary, and Ernst Fischer in Austria independently proved in 1907 the theorem now bearing both names: $L^2$ is complete and isomorphic to $\ell^2$ via Fourier coefficients. This was a foundational moment — it justified Hilbert's emerging abstract theory of square-summable sequences by anchoring it in concrete function spaces, and it made "convergence in mean" the natural mode for Fourier series. Riesz went on to lay much of the groundwork for functional analysis (the Riesz representation theorem, $L^p$ duality, compact operators).

🧠 Intuition

Think of $L^2$ as an infinite-dimensional Euclidean space and $\{e^{inx}\}$ as its coordinate axes. The Fourier coefficient $\hat f(n)$ is literally the component of the "vector" $f$ along the $n$ -th axis. Parseval [5] is the Pythagorean theorem: the squared length of $f$ equals the sum of squares of its components. $L^2$ convergence [4] just says that if you keep more axes, the projection $S_N f$ gets closer to $f$ — exactly as in $\mathbb{R}^k$ , projecting onto more coordinate axes recovers the vector. The reason this is "easy" while pointwise convergence is "hard": the $L^2$ norm only sees average discrepancy. A reconstructed signal can overshoot wildly at a single point (Gibbs phenomenon) yet have tiny mean-square error, because the overshoot occupies vanishing measure. The norm forgives pointwise sins; pointwise convergence does not.

Figure

Orthogonal projection:

S_N f

is the closest point in

V_N

to

f

; the error is perpendicular

ExampleSawtooth: coefficients, Parseval, and a series sum

Given. $f(x) = x$ on $[-\pi,\pi]$ , extended $2\pi$ -periodically. Compute $\hat f(n)$ , verify $L^2$ membership, and use Parseval to evaluate $\sum_{n\ge1} 1/n^2$ .

Step 1 — Coefficients. $\hat f(0) = \frac{1}{2\pi}\int_{-\pi}^\pi x\,dx = 0$ (odd integrand). For $n\ne 0$ , integrate by parts:

\hat f(n) = \frac{1}{2\pi}\int_{-\pi}^{\pi} x e^{-inx}\,dx = \frac{1}{2\pi}\left[\frac{x e^{-inx}}{-in}\right]_{-\pi}^{\pi} + \frac{1}{2\pi in}\int_{-\pi}^\pi e^{-inx}dx.

The integral term vanishes; the boundary term gives $\hat f(n) = \frac{1}{2\pi}\cdot\frac{2\pi(-1)^{n}}{-in}\cdot(\text{sign bookkeeping}) = \frac{(-1)^{n}\, i}{n}$ . So $|\hat f(n)| = 1/|n|$ .

Step 2 — $L^2$ norm. $\|f\|_2^2 = \frac{1}{2\pi}\int_{-\pi}^\pi x^2\,dx = \frac{1}{2\pi}\cdot\frac{2\pi^3}{3} = \frac{\pi^2}{3}.$ Finite, so $f\in L^2$ . ✓

Step 3 — Parseval [5].

\frac{\pi^2}{3} = \sum_{n\ne 0}\frac{1}{n^2} = 2\sum_{n=1}^\infty \frac{1}{n^2}.

Answer.

\boxed{\;\sum_{n=1}^{\infty}\frac{1}{n^2} = \frac{\pi^2}{6}\;}

The Fourier series converges to $x$ in $L^2$ , even though pointwise it converges to $0$ at $x=\pm\pi$ (the average of the jump) — a vivid reminder that $L^2$ convergence ignores point values.

ExampleA pure

\ell^2

coefficient sequence with no classical function feel

Given. Define $c_n = \frac{1}{n}$ for $n\ge 1$ and $c_n = 0$ otherwise. Is there an $L^2$ function with these as Fourier coefficients? Estimate $\|f\|_2$ .

Step 1 — $\ell^2$ check. $\sum_n |c_n|^2 = \sum_{n=1}^\infty 1/n^2 = \pi^2/6 < \infty$ , so $(c_n)\in\ell^2(\mathbb{Z})$ . ✓

Step 2 — Riesz–Fischer surjectivity. By Theorem [4]/[5] (Step 6 of the proof), there exists $f\in L^2([-\pi,\pi])$ with $\hat f(n)=c_n$ , namely the $L^2$ -limit of $g_N = \sum_{n=1}^N \frac{1}{n}e^{inx}$ .

Step 3 — Norm via Parseval.

\|f\|_2^2 = \sum_{n}|c_n|^2 = \frac{\pi^2}{6} \approx 1.6449.

Answer.

\boxed{\;\|f\|_2 = \frac{\pi}{\sqrt 6}\approx 1.2825\;}

Note: this $f$ is not given by an elementary closed form, yet it is a perfectly legitimate $L^2$ function — the existence is guaranteed abstractly by completeness, with no formula required. (In fact $\sum \frac1n e^{inx}$ relates to $-\log(1-e^{ix})$ , but the point stands: $\ell^2$ coefficients ⟺ $L^2$ function, unconditionally.)

⚠️

L^2

convergence does NOT imply pointwise convergence

$\|f-S_Nf\|_2\to0$ guarantees convergence in mean square only. It implies a subsequence $S_{N_k}f\to f$ pointwise a.e., but not the full sequence. There exist continuous functions whose Fourier series diverge at a point (du Bois-Reymond, 1873) and integrable functions whose series diverge everywhere (Kolmogorov, 1926) — yet all such $f$ in $L^2$ still satisfy [4]. Never write " $S_Nf(x)\to f(x)$ " as a consequence of the Riesz–Fischer theorem.

CounterexampleGibbs phenomenon —

L^2

fine, pointwise overshoot persists

Take the square wave $f = \mathrm{sgn}(x)$ on $[-\pi,\pi]$ . Its partial sums $S_N f$ converge to $f$ in $L^2$ (since $f\in L^2$ , automatic by [4]). But near the jump at $x=0$ , the partial sums overshoot the limit by a fixed factor $\approx 8.95\%$ of the jump height, for every $N$ — the overshoot does not shrink, it merely migrates toward the discontinuity. The maximum of $S_Nf$ converges to $\frac{2}{\pi}\int_0^\pi \frac{\sin t}{t}\,dt \approx 1.0895$ times the half-jump. The $L^2$ norm is undisturbed because the overshoot region has width $O(1/N)\to 0$ : it contributes $O(1/N)\to 0$ to $\|f-S_Nf\|_2^2$ . Mean-square convergence and uniform convergence are genuinely different.

⚠️ Bessel vs. Parseval

Bessel's inequality $\sum|\hat f(n)|^2 \le \|f\|_2^2$ holds for any orthonormal system. Equality (Parseval, [5]) holds iff the system is complete. For an incomplete system (e.g. drop $e^{i0x}$ from the basis), Bessel is strict for functions with a nonzero mean. The leap from inequality to equality is the content of completeness — it is not automatic.

connection

The $L^2$ theory sits at a crossroads. It is the prototypical instance of Orthonormal Bases in Hilbert Spaces: every separable Hilbert space looks like $\ell^2$ , and Fourier series is the canonical realization. Via the Plancherel Theorem it generalizes from the circle to $\mathbb{R}$ , where the Fourier transform is the unitary map $L^2(\mathbb{R})\to L^2(\mathbb{R})$ . Abstractly it is a special case of the Spectral Theorem — the $e^{inx}$ are simultaneous eigenfunctions of translation (and of $-d^2/dx^2$ ), and Fourier series diagonalizes these operators. The convergence dichotomy connects to Lp Convergence of Fourier Series (M. Riesz: $S_Nf\to f$ in $L^p$ for $1<p<\infty$ , false at $p=1,\infty$ ) and to Carleson's Theorem on a.e. convergence. The coefficient decay rate ties to Sobolev Spaces: $f\in H^s$ iff $\sum (1+n^2)^s|\hat f(n)|^2<\infty$ .

🔬 Where this appears

⏳ History

- 1807 / 1822 — Joseph Fourier asserts (in his heat-conduction work, published fully in Théorie analytique de la chaleur) that arbitrary functions admit trigonometric expansions, provoking controversy among Lagrange, Laplace, and Poisson over convergence.

- 1829 — Dirichlet gives the first rigorous pointwise convergence theorem for piecewise-monotone functions with finitely many discontinuities.

- 1873 — Paul du Bois-Reymond constructs a continuous function whose Fourier series diverges at a point, shattering naive optimism about pointwise convergence.

- 1899 — J. Willard Gibbs (and earlier Wilbraham, 1848) explains the persistent overshoot near jumps — the Gibbs phenomenon.

- 1900s — Hilbert and Schmidt develop the abstract theory of

\ell^2

and integral equations, setting the stage for Hilbert space.

- 1907 — Frigyes Riesz and Ernst Fischer independently prove the Riesz–Fischer theorem:

L^2

is complete and isometric to

\ell^2

via Fourier coefficients, establishing unconditional mean-square convergence.

- 1926 — Andrey Kolmogorov constructs an

L^1

function whose Fourier series diverges everywhere — emphatically separating

L^1

from

L^2

behavior.

- 1966 — Lennart Carleson proves that the Fourier series of any

L^2

function converges almost everywhere, resolving a conjecture open since Luzin (1915).

- 1968 — Richard Hunt extends Carleson's theorem to

L^p

for all

p>1

; Kahane–Katznelson (1966) show a.e. convergence can fail on any prescribed measure-zero set, pinning down the sharp result.

open-problem

The $L^2$ convergence theory is complete and closed — there are no open problems in [4]/[5] itself; they are theorems. The frontier lies in adjacent refinements. (a) Sharp constants and rates: the exact dependence of $\|f-S_Nf\|_2$ on smoothness classes interacts with multiplier theory and remains subtle for endpoint and weighted spaces. (b) Convergence in higher dimensions: for $\mathbb{T}^d$ , $d\ge2$ , spherical partial sums $\sum_{|n|\le R}$ exhibit the notorious Bochner–Riesz problem — for which $L^p$ and which summation order does convergence hold? This is open in full generality for $d\ge3$ and connects to the Kakeya and restriction conjectures, among the deepest unsolved problems in harmonic analysis. (c) Carleson-type results in higher dimensions and on general groups remain only partially understood. So while the Hilbert-space core is settled, replacing the $L^2$ norm by $L^p$ or by pointwise/multidimensional notions reopens some of the hardest questions in modern analysis.

Fourier Series — L2 Theory and Convergence

Formal Statement

Why It Holds / Derivation

Building Intuition

Worked Examples

Common Misconceptions

Connections

Real-World Applications

Historical Development

Subtleties & Limits

See Also