Chapter 13 — Further Reading

Annotated pointers for going deeper on the column space, the null space, and the first half of Strang's four-fundamental-subspaces picture. The three "anchor" textbooks below are referenced throughout this book; we map each chapter to the relevant sections so you can read in parallel. Section numbers follow the most widely circulated editions and may shift slightly between printings.

The three anchor textbooks

  • Gilbert Strang, Introduction to Linear Algebra (5th ed.), §3.1–3.3 and §3.5 — the essential companion to this chapter. Strang is the originator of the four-fundamental-subspaces pedagogy, and this chapter is written in his spirit. §3.1 (spaces of vectors and the column space), §3.2 (the null space and the solving of $A\mathbf{x}=\mathbf{0}$), and §3.3 (the complete solution to $A\mathbf{x}=\mathbf{b}$ — particular plus special solutions) cover exactly our §13.2–13.8, in the same order, with the same emphasis on reading bases off the RREF. §3.5 introduces the four subspaces together (our §13.9, completed in Chapter 14). If you read only one outside source for this material, read these sections of Strang; his framing is the modern standard. His MIT OpenCourseWare lectures (below) narrate the same pages.
  • Sheldon Axler, Linear Algebra Done Right (4th ed.), §3B (Null Spaces and Ranges). Axler is the rigorous, proof-first complement, and he develops the abstract versions of this chapter's two subspaces: the null space ($\operatorname{null} T$) and the range ($\operatorname{range} T$) of a linear map $T$, free of matrices. His Fundamental Theorem of Linear Maps ($\dim V = \dim\operatorname{null} T + \dim\operatorname{range} T$) is the coordinate-free rank–nullity theorem that Chapter 14 will prove. Math majors should read §3B alongside the proof in our §13.4 and the Math-Major Sidebar on kernel and image; it is the cleanest path to seeing why these constructions matter for all vector spaces, not just $\mathbb{R}^n$.
  • Stephen Boyd & Lieven Vandenberghe, Introduction to Applied Linear Algebra (VMLS), Chapters 5–8. The applied, data-oriented view. Their treatment of linear independence, the range of a matrix mapping, and the solution sets of linear equations connects this chapter directly to least squares (their Part III) and to the data applications. Best matched to the CS/data-science learning path and to the feature-redundancy theme. Freely and legally downloadable as a PDF from the authors — see below.

Free online resources

  • MIT OpenCourseWare, 18.06 Linear Algebra (Gilbert Strang), Lectures 6–9. The definitive video companion. Lecture 6 (column space and null space), Lecture 7 (solving $A\mathbf{x}=\mathbf{0}$, pivot and free variables), Lecture 8 (solving $A\mathbf{x}=\mathbf{b}$, the complete solution), and Lecture 9 (independence, basis, dimension) map almost one-to-one onto this chapter. Strang draws the four-subspaces diagram by hand and explains, in his unmistakable way, why the pivot columns of the original matrix give the column-space basis. Full video, transcripts, and problem sets, free.
  • 3Blue1Brown, Essence of Linear Algebra, "The column space, null space, and rank" segments (within the chapters on linear transformations and on inverse matrices/rank). Grant Sanderson animates exactly the funnel picture of §13.9 — the input space folding, a direction collapsing to zero (the null space), and the image landing on a lower-dimensional flat (the column space). His treatment of rank as "the number of dimensions in the output" is the geometric reading we used in §13.6.2. Watch before re-reading the worked example if the collapse picture has not yet clicked.
  • Khan Academy, Linear Algebra, "Matrix transformations" and "Alternate coordinate systems" units. Gentler, exercise-rich coverage of column space, null space, and solving $A\mathbf{x}=\mathbf{b}$, with immediate auto-graded practice. Good for shoring up the ⭐ and ⭐⭐ exercises before the harder tiers.
  • Boyd & Vandenberghe, VMLS free PDF and Python companion. The full textbook and its numpy-friendly companion notebooks are posted by the authors at no cost. The companion code shows rank and range computations, reinforcing the C-track exercises and the toolkit functions of §13.10.

On the applications in this chapter

  • Chemical reaction balancing as a null space (Case Study 1). Any introductory chemistry text covers balancing by inspection; for the linear-algebra treatment, search for "balancing chemical equations with linear algebra" or see the stoichiometry sections of a chemical-engineering text. The generalization to reaction networks — where $\dim N(S)$ counts independent reactions — is the subject of chemical reaction network theory and of flux-balance analysis in systems biology, both of which analyze the null space of a (large) stoichiometric matrix. Conservation laws as null-space rows recur in physics; for nuclear-decay balancing the same idea applies to baryon/lepton number and charge.
  • Reachability and controllability (Case Study 2). The column-space-of-the-controllability-matrix story is standard in any first control-theory course; look for the Kalman rank condition and the controllability matrix $\mathcal{C} = [\,B\ AB\ \cdots\ A^{n-1}B\,]$. The Cayley–Hamilton fact that powers beyond $A^{n-1}B$ add nothing is proved via the characteristic polynomial of Chapter 24 [verify]. The dual notion, observability (which states can be inferred from outputs), is the same idea applied to the transpose system — a left-null-space / row-space story you will be equipped to read after Chapter 14.
  • Feature spaces and redundancy (the ML thread). The column-space-as-feature-space framing, and the way redundant features land in the null space, is the entry point to feature spaces in machine learning; the broader goal of finding a small column space that captures most of a dataset is dimensionality reduction, made precise by the SVD of Chapter 30. For the statistical face of a nontrivial null space — collinear predictors — see any regression text's treatment of multicollinearity, and Chapter 38's discussion of the condition number.

A note on where this is going

The column space and null space are half of a single structure. The most important next reading is Strang §3.5 and §4.1, which complete the four-fundamental-subspaces picture (adding the row space $C(A^{\mathsf{T}})$ and the left null space $N(A^{\mathsf{T}})$) and state the rank–nullity theorem — Chapter 14 of this book. If you want the abstract version first, Axler §3B proves the Fundamental Theorem of Linear Maps directly. And if you want to see why the four subspaces pair up at right angles — the deepest payoff of all — skim ahead to the orthogonality of Part IV and the four-subspaces-via-SVD picture of Chapter 30. Strang's recurring diagram, which you met here in its two-subspace form, is the single image worth carrying through the rest of the book.