Chapter 22 — Further Reading

Annotated pointers for going deeper, mapped to the standard texts and to free resources. The throughline to keep in mind: Fourier analysis is projection onto an orthogonal basis, so the most illuminating treatments are the ones that present it as linear algebra rather than as a bag of integral formulas.

Core linear algebra texts

  • Gilbert Strang, Introduction to Linear Algebra (5th/6th ed.), §§8.1–8.5 and §10. Strang is the right first stop because he tells exactly the story of this chapter: he treats functions as vectors with the inner product $\int fg$, presents the sines and cosines as an orthogonal basis, and frames Fourier coefficients as projections. His discussion of the discrete Fourier transform and the FFT (the $N\log N$ algorithm of §22.8) is especially clear and connects directly to the orthogonal-matrix viewpoint of Chapter 21. If you read one supplementary source, read this.

  • Gilbert Strang, Computational Science and Engineering, Chapter 4. A fuller, application-driven treatment of Fourier series, the DFT, the FFT, and their use in solving differential equations — the heat-equation motivation noted in §22.7. Heavier on signals and PDEs; excellent for physics and engineering readers who want to see the eigenfunction connection (sines and cosines as eigenfunctions of differentiation) developed in depth.

  • Sheldon Axler, Linear Algebra Done Right (4th ed.), Chapter 6 (Inner Product Spaces). Axler builds inner product spaces, orthonormal bases, and orthogonal projection abstractly and rigorously, which is the proper foundation for the "functions are vectors" leap of §22.2. He does not dwell on Fourier series specifically, but his treatment of projection onto an orthonormal basis is exactly the theorem this chapter applies. The natural companion for the proof track and a clean lead-in to the abstract inner product spaces of Chapter 34.

Optimization and applications

  • Stephen Boyd and Lieven Vandenberghe, Introduction to Applied Linear Algebra: Vectors, Matrices, and Least Squares (VMLS), Chapters 5 and 12. Boyd and Vandenberghe's free textbook (available online) is the best place to see orthonormal bases and least squares treated as the practical, computational tools they are. Their framing of projection and least squares is the same one that makes truncation-as-best-approximation (§22.6) obvious. Pair it with their companion language Julia/Python notebooks to compute spectra and reconstructions yourself.

Signal processing, for the application side

  • Steven W. Smith, The Scientist and Engineer's Guide to Digital Signal Processing (free online). A famously accessible, intuition-first treatment of the DFT, the FFT, and real-world signal processing. Light on linear-algebra abstraction but superb on the engineering of the ideas in §22.7–22.8 — windowing, leakage, and why the FFT matters. Read alongside Strang to connect the geometry to the practice.

  • Alan Oppenheim and Ronald Schafer, Discrete-Time Signal Processing. The standard graduate signal-processing reference, for readers who want the full rigor of sampling, the DFT, and filter design. More than this chapter requires, but the definitive next step if signals become your field.

On convergence and the Gibbs phenomenon

  • Any rigorous real-analysis or Fourier-analysis text (e.g., Stein & Shakarchi, Fourier Analysis). For the precise convergence theory behind §22.6 — pointwise vs. mean-square convergence, the conditions under which the series equals $f$, and a careful derivation of the Gibbs constant $\approx 1.17898$. This is where the [verify]-flagged history (Fourier's 1807 memoir, the early skepticism, Gibbs's later quantification) is treated authoritatively, and where the completeness theorem previewed in the Math-Major Sidebar is proved.

The calculus connection

This chapter leans on ideas from integral calculus — the integral as the inner product, and the infinite sum as a limit of partial sums. If the convergence of an infinite series feels shaky, revisit the calculus treatment of series convergence first; the Fourier series is just such a series with basis functions as terms, and the same notions of partial sum, convergence, and limit apply. The tools for actually plotting spectra and reconstructions — overlaying partial sums, drawing the Gibbs overshoot, displaying a frequency bar chart — are developed in the signals and plotting material, which pairs naturally with the matplotlib figures of §22.6 and the coding exercises.

Free online resources

  • MIT OpenCourseWare 18.06 (Strang's Linear Algebra). The lectures on orthogonality, projection, and the Fourier/DFT material are the video companion to this chapter and to all of Part IV.
  • 3Blue1Brown, But what is the Fourier Transform? and Essence of Linear Algebra. Geometric, animation-first intuition. The Fourier video visualizes "winding a signal around a circle" — a vivid picture of the complex-exponential projection of §22.9 — and the linear-algebra series reinforces the projection-and-basis viewpoint that this whole chapter rests on.