Fourier series Study Guide

📖 Core Concepts Fourier series – expresses a periodic function \(f(x)\) as an infinite sum of sinusoids (harmonics) with frequencies that are integer multiples of the fundamental frequency. Fourier coefficients – numbers that weight each sine or cosine term. Real form uses \(an,\;bn\); complex form uses \(cn\). Orthogonal basis – \(\{\cos nx,\;\sin nx\}{n=0}^\infty\) are mutually orthogonal on \([-\pi,\pi]\); this orthogonality lets us isolate each coefficient via integration. Convergence types – Pointwise (Dirichlet conditions): series converges to \(f(x)\) at continuous points and to the midpoint of the jump at discontinuities. Uniform & absolute: guaranteed when \(f\) is continuously differentiable (or satisfies a Hölder condition). Parseval (Mean‑square) theorem – total energy of \(f\) equals the sum of squares of its coefficients. Gibbs phenomenon – overshoot near a jump discontinuity that never vanishes, no matter how many terms are added. Even/odd decomposition – even part \(\Rightarrow\) cosine‑only series; odd part \(\Rightarrow\) sine‑only series. --- 📌 Must Remember Real‑form coefficients \[ an=\frac{1}{\pi}\int{-\pi}^{\pi} f(x)\cos nx\,dx,\qquad bn=\frac{1}{\pi}\int{-\pi}^{\pi} f(x)\sin nx\,dx \] Complex‑form coefficient \[ cn=\frac{1}{2\pi}\int{-\pi}^{\pi} f(x)\,e^{-inx}\,dx \] Orthogonality integrals \[ \int{-\pi}^{\pi}\cos mx\cos nx\,dx= \begin{cases} \pi,& m=n\neq0\\[2pt] 2\pi,& m=n=0\\[2pt] 0,& m\neq n \end{cases} \] (same pattern for \(\sin\) terms). Dirichlet conditions – piecewise continuity + piecewise continuous derivative \(\Rightarrow\) pointwise convergence to \(f\) (or its average at jumps). Parseval – \(\displaystyle \frac{1}{2\pi}\int{-\pi}^{\pi}|f(x)|^{2}dx=\sum{n=-\infty}^{\infty}|cn|^{2}\). Riemann–Lebesgue lemma – \(cn\to0\) as \(|n|\to\infty\) for any integrable periodic \(f\). Derivative property (complex form) – if \(f'(x)\) exists, its coefficients are \( (jn)cn\). Even function ⇒ all \(bn=0\); odd function ⇒ all \(an=0\). --- 🔄 Key Processes Compute coefficients (analysis) Choose form (real or complex). Plug \(f(x)\) into the appropriate integral formula. Exploit symmetry: drop sine terms for even \(f\), cosine terms for odd \(f\). Reconstruct function (synthesis) Insert the computed \(an,bn\) (or \(cn\)) into the series sum. Use partial sums \(SN(x)=\sum{n=-N}^{N}cne^{inx}\) for approximations. Apply derivative property Differentiate term‑by‑term: multiply each \(cn\) by \(jn\). Convolution of periodic signals Compute Fourier coefficients of each signal. Multiply corresponding coefficients: \((fg)n = cn^{(f)}\,cn^{(g)}\). Inverse‑transform to obtain the convolution in the time domain. --- 🔍 Key Comparisons Real (sine‑cosine) vs. Complex (exponential) form Real: separates even/odd parts, uses two real coefficients per harmonic. Complex: single complex coefficient per harmonic; algebraically cleaner for manipulation. Even function vs. Odd function Even: only cosine terms (\(bn=0\)). Odd: only sine terms (\(an=0\)). Pointwise vs. Uniform convergence Pointwise (Dirichlet): may converge to the average at jumps. Uniform: convergence is the same speed everywhere; requires stronger smoothness. Dirichlet test vs. Dirichlet‑Jordan test Dirichlet: piecewise continuity + piecewise continuous derivative. Dirichlet‑Jordan: relaxes derivative condition to bounded variation (still ensures convergence to the average). --- ⚠️ Common Misunderstandings “Fourier series always converges to the original function.” – Only under Dirichlet (or stronger) conditions; at jumps it converges to the midpoint. Ignoring Gibbs overshoot. – The ringing is real; it does not disappear with more terms. Mixing up coefficient prefactors. – Real form uses \(1/\pi\); complex form uses \(1/(2\pi)\). Assuming Parseval holds for any series. – Valid only for square‑integrable periodic functions and the actual Fourier coefficients. Treating even/odd symmetry as “no need to compute any coefficients.” – Only the corresponding set (sine or cosine) drops out; the other set may still be non‑zero. --- 🧠 Mental Models / Intuition Lego‑block picture – each sine or cosine is a LEGO brick of a specific size (frequency) and colour (amplitude/phase). The Fourier series builds the target shape by stacking the right bricks. Orthogonal axes – think of \(\cos nx\) and \(\sin nx\) as independent directions in a high‑dimensional space; the coefficient is the “shadow” (dot product) of the function onto that direction. Gibbs as ringing – a sudden edge forces the series to “wiggle” to approximate the jump, similar to trying to draw a square with a finite number of smooth curves. --- 🚩 Exceptions & Edge Cases Discontinuous functions – converge only to the midpoint of the jump; exhibit Gibbs ringing. Non‑smooth (non‑differentiable) functions – may converge pointwise but not uniformly; coefficient decay is slower (≈\(1/n\)). Functions that are not piecewise continuous – Fourier series may diverge or fail to represent the function meaningfully. Time scaling vs. coefficient scaling – scaling the argument \(x\) changes the spacing of harmonic frequencies, not the magnitude of coefficients directly. --- 📍 When to Use Which Real form – when the signal is purely real and you need to exploit even/odd symmetry (e.g., power‑spectral analysis). Complex exponential form – for algebraic manipulation, differentiation, convolution, or when working with complex‑valued signals. Amplitude‑phase form – when amplitude and phase of each harmonic are of primary interest (e.g., modulation analysis). Dirichlet conditions test – before assuming pointwise convergence; if they fail, consider alternative representations (e.g., Fourier transform). Parseval – to compute total signal energy without reconstructing the time‑domain function. --- 👀 Patterns to Recognize Only cosine terms → even function; only sine terms → odd function. Coefficient magnitude decays rapidly → underlying function is smooth. \(cn\) \(1/n\) → function has a jump discontinuity. Overshoot 9 % of jump height near a discontinuity → Gibbs phenomenon is present. Zero \(a0\) → function has zero average (mean value). --- 🗂️ Exam Traps Wrong coefficient prefactor – mixing \(1/\pi\) with \(1/(2\pi)\) leads to off‑by‑factor‑2 errors. Including \(n=0\) in the sine sum – \(\sin 0x = 0\); the term should be omitted. Assuming series converges to \(f\) at a jump – the correct value is \(\frac{f(x^+)+f(x^-)}{2}\). Choosing the real form for a complex‑valued problem – you’ll miss the negative‑frequency terms. Neglecting the factor \(j n\) when differentiating – derivative coefficients are multiplied, not unchanged. Confusing even/odd symmetry with zero average – an even function can have a non‑zero \(a0\). ---