Chapter 3 — Further Reading

An annotated map to where the books and free resources treat systems of linear equations — the geometry of solution sets, the row and column pictures, and the conditions for none / one / infinitely many. You need none of these to follow this book; they are second voices for when an idea is worth hearing twice. (Reminder: the solving algorithm is Chapter 4's subject; below, focus on the sections about what a system means and what its solutions look like.)

The companion textbooks

  • Gilbert Strang, Introduction to Linear Algebra (Wellesley–Cambridge Press), Chapters 1–2. The single best companion to this chapter. Strang opens the whole subject with exactly our two pictures — he calls them the "row picture" and the "column picture" — and his treatment of $A\mathbf{x}$ as a combination of the columns is the seed of everything in Part III. His MIT Lecture 1 (below) is titled "The Geometry of Linear Equations" and is the natural next hour after reading this chapter. Best fit for: CS, data science, and engineering readers; the row/column framing here is borrowed directly from Strang.

  • Sheldon Axler, Linear Algebra Done Right (Springer; 4th edition freely readable online). Axler treats systems as a consequence of the theory of linear maps rather than as the starting point, so his discussion of "when does $A\mathbf{x} = \mathbf{b}$ have a solution?" arrives later and in the language of range and null space (our column space and null space). For the conditions in Section 3.7 — consistency, uniqueness, and the particular-plus-homogeneous structure — Axler's chapters on linear maps give the cleanest proofs. Best fit for: math majors; revisit alongside Chapters 13–14 and 35 once you have met the four subspaces.

  • Stephen Boyd & Lieven Vandenberghe, Introduction to Applied Linear Algebra (Cambridge; free PDF at vmls-book.stanford.edu). The most application-driven of the three. Its chapters on linear equations and on least squares are perfect alongside this chapter's case studies — the input–output and network-flow models, and the over-determined systems that np.linalg.lstsq handles. Boyd is also excellent on the practical meaning of "no exact solution" (you fit the closest one). Best fit for: readers who want to see systems doing real work in data fitting, economics, and engineering immediately.

The essential video companion

  • ★ 3Blue1Brown (Grant Sanderson), Essence of Linear Algebra (free, on YouTube). Watch the episodes "Linear combinations, span, and basis" and "Inverse matrices, column space, and null space" alongside this chapter. The first animates the column picture — how scaled column vectors combine to reach a target — better than any static figure can; the second shows geometrically why a singular transformation makes some right-hand sides unreachable (no solution) and others over-reachable (infinitely many), which is precisely the dichotomy of Section 3.7. Best fit for: everyone; indispensable for seeing the column picture move.

Free and open resources

  • MIT OpenCourseWare 18.06, Linear Algebra (Gilbert Strang), Lecture 1 — "The Geometry of Linear Equations." Free video. Strang draws the row picture and the column picture for the same small system, live on the board, in the first fifteen minutes — the ideal companion to Sections 3.2–3.3. Search "MIT 18.06 Lecture 1."
  • Khan Academy, Systems of equations and Linear Algebra. Free, exercise-driven coverage of solving and interpreting small systems; good for shoring up the high-school-level row picture before this chapter's deeper treatment.
  • Immersive Math, Immersive Linear Algebra (immersivemath.com). Free online book with interactive figures; its chapter on Gaussian elimination opens with the geometry of intersecting planes, matching this chapter's anchor image.
  • numpy documentation — numpy.linalg.solve, numpy.linalg.matrix_rank, and numpy.linalg.lstsq. The reference pages for the three functions used throughout this chapter. The lstsq page in particular explains what is returned for rank-deficient systems — useful background for the Computational Note in Section 3.4.

On the applications and history

  • Wassily Leontief, "Input–Output Economics" (Scientific American, 1951), and any intermediate macroeconomics text's chapter on input–output analysis, for the model behind Case Study 1 and the chapter's input-output models in economics link. Leontief received the 1973 Nobel Memorial Prize in Economic Sciences for this work.
  • Any introductory circuits text (e.g. Fundamentals of Electric Circuits, Sadiku & Alexander), the chapters on Kirchhoff's laws and nodal/mesh analysis, for Case Study 2. These show that every resistive circuit reduces to a linear system, and that the "superposition theorem" of circuit theory is the Chapter 1 superposition principle applied to $A\mathbf{x} = \mathbf{b}$.
  • The MacTutor History of Mathematics Archive (mathshistory.st-andrews.ac.uk) for checking the chapter's [verify]-flagged history: the Nine Chapters on the Mathematical Art and its early elimination method, Gauss's orbit computations, and the Rouché–Capelli consistency criterion.

How to read alongside this book

If you are a CS/data-science reader: read Strang Chapters 1–2, watch the two 3Blue1Brown episodes above, and skim Boyd's least-squares chapter so you understand what lstsq is doing. If you are a math major: keep Axler nearby and treat Section 3.7 and its Math-Major Sidebar as the bridge to his range/null-space machinery — but watch the 3Blue1Brown column-space video first, so the proofs land on a picture. If you are in physics/engineering: watch MIT 18.06 Lecture 1, work Case Study 2 (circuits) carefully, and notice how Kirchhoff's two conservation laws each become one row of $A\mathbf{x} = \mathbf{b}$.