Appendix J — Applications Catalog
One of this book's three promises is real applications in every chapter — never an appendix bolted on at the end, but woven into the exposition, the case studies, and the "Real-World Application" callouts. This appendix is the index to all of them. Use it two ways: to find where a particular application is treated, or to find which applications illustrate a concept you are studying.
The catalog is organized by field. Each entry names the application, the chapter where it appears, and the linear-algebra concept it rests on. Many applications recur across several chapters as the same example deepens — PageRank, image compression, and the qubit are the book's three great recurring anchors, and you will see them resurface below at escalating levels of sophistication.
J.1 Machine learning & AI
| Application | Chapter(s) | Linear-algebra concept |
|---|---|---|
| Recommendation systems (taste as a direction in "taste space") | 1, 18, 31, 33 | Vectors; cosine similarity; low-rank approximation; matrix factorization |
| Word and image embeddings (meaning as a vector/direction) | 2, 33 | Vectors; inner products; embedding spaces |
| Neural-network layers as matrix multiplication | 33 | Matrix–vector products; composition of linear maps |
| Matrix-factorization recommenders (Netflix-style latent factors) | 1, 31, 33 | Low-rank SVD approximation; latent factors |
| Data layout for ML (samples × features matrices) | 7 | Matrices as data; row/column conventions |
| Why models train slowly — the loss "bowl" and its conditioning | 28, 38 | Positive-definite quadratic forms; condition number |
| Training as optimization on a quadratic surface | 28 | Positive definiteness; curvature/Hessian |
| Compressing feature spaces / dimensionality reduction | 31, 32 | SVD; PCA |
| Whitening and PCA preprocessing of data | 21, 32 | Orthogonal matrices; covariance eigendecomposition |
| Latent Semantic Analysis (meaning from a term–document matrix) | 30 | SVD |
| Classification and data fitting (Boyd-style) | 33 | Least squares |
The recurring "wow." The same SVD that compresses an image in Chapter 31 powers the recommender in Chapters 31 and 33 and the PCA of Chapter 32. Learn it once, use it everywhere is not a slogan here — it is literally the same factorization, $A=U\Sigma V^{\mathsf{T}}$, doing all of these jobs.
J.2 Computer graphics & video games
| Application | Chapter(s) | Linear-algebra concept |
|---|---|---|
| The linear algebra inside a video game (60-frames-a-second pipeline) | 1, 12 | Linear transformations; the rendering pipeline |
| Spinning a sprite; rotating a 2D object | 7 | Rotation matrices |
| Composing transformations in a game engine | 8 | Matrix multiplication as composition |
| The order of operations in a graphics pipeline | 8, 12 | Non-commutativity of matrix products |
| Homogeneous coordinates; rotation/scaling/translation together | 12 | Affine maps via an extra dimension |
| Rendering a single frame; the model–view–projection chain | 12 | Composed transformations; projection |
| Skeletal animation and scene graphs | 12 | Hierarchies of composed transformations |
| Augmented reality placing a virtual object in a room | 12 | Transformations; computer-vision pose |
| CAD and architectural drawing (deliberate flattening) | 12 | Projection onto a plane |
| Color and the RGB basis; color spaces as coordinate systems | 6, 15 | Basis, span, independence; change of basis |
| Undoing a color transform | 9 | Matrix inverse |
| Curve and font design (Bézier-style polynomial curves) | 5 | Polynomial vector spaces |
| The shape of every letter (polynomial spaces behind fonts/animation) | 5 | Function/polynomial spaces |
| Shadows and contact (projecting geometry onto a surface) | 19 | Orthogonal projection |
| Pointing the camera; orthonormal camera frames | 20 | Gram–Schmidt; orthonormal bases |
| Triangle area and orientation in computational geometry | 11 | Determinant as signed area |
| Which way is "out"? Orientation and winding | 11 | Sign of the determinant |
| Procedural animation and physics engines (oscillation) | 26 | Complex eigenvalues as rotation |
| Real-time numerical reliability in rendering pipelines | 38 | Conditioning; stability |
J.3 Quantum mechanics & physics
| Application | Chapter(s) | Linear-algebra concept |
|---|---|---|
| The qubit (a quantum state as a unit vector) | 1, 5, 34 | Vectors; complex inner-product spaces |
| Quantum logic gates | 21, 40 | Orthogonal/unitary matrices |
| Observables as eigenvalue problems | 23, 27 | Eigenvalues/eigenvectors; Hermitian operators |
| Hermitian operators and real measured values | 27, 34 | Spectral theorem; $A^{*}=A$ |
| The qubit's inner product as a probability amplitude | 34 | Hermitian inner product |
| Quantum computing and chemistry | 27, 40 | Spectral decomposition; unitary evolution |
| Rotating a vector field | 7 | Linear maps on $\mathbb{R}^n$ |
| Crystal lattices and tetrahedron volume | 11 | Determinant as signed volume |
| Natural frequencies / normal modes of a vibrating structure | 23, 24 | Eigenvalues as frequencies |
| Principal stresses in a loaded beam (the stress tensor) | 27 | Symmetric matrices; spectral theorem |
| The heat equation; solving PDEs by eigen-expansion | 22 | Orthogonal function bases |
| Tensors in deep learning and physics | 40 | Multilinear algebra |
The qubit's arc. The quantum qubit is teased in Chapters 1 and 5 as "a vector that can be in two states at once," made precise as a unitary transformation in Chapter 21, connected to Hermitian observables in Chapter 27, and given its full home in the complex inner-product (Hilbert) space of Chapter 34 — with a final word in Chapter 40. Follow it to see one application mature alongside the mathematics.
J.4 Data science & statistics
| Application | Chapter(s) | Linear-algebra concept |
|---|---|---|
| Linear regression (predicting house prices, etc.) | 14, 17 | Least squares as projection onto $C(A)$ |
| Calibrating a sensor (voltage → temperature) | 17, 19 | Least squares |
| Degrees of freedom behind regression | 14 | Rank–nullity |
| Multicollinearity and a regression that "blew up" | 38 | Conditioning; near-dependence |
| Stable regression via QR (when the normal equations lie) | 20 | QR factorization; numerical stability |
| Cosine similarity for document/search relevance | 18 | Dot product; angle in $n$ dimensions |
| "People like you also liked" (collaborative filtering) | 18 | Cosine similarity |
| Principal Component Analysis (exploratory data analysis) | 16, 27, 32 | Covariance eigendecomposition; SVD |
| The shape of data: covariance, Mahalanobis distance, outliers | 28 | Positive-definite quadratic forms |
| Confidence ellipses in statistics | 28 | Quadratic forms; covariance |
| Eigenfaces and face recognition | 32 | PCA on image data |
| Genes mirror geography (population structure via PCA) | 32 | PCA |
| Hidden market factors (factor models in finance) | 32 | PCA; factor models |
| Rank as data compression | 14, 31 | Rank; low-rank approximation |
| Latent factors in recommender systems | 31, 33 | SVD low-rank structure |
| Factor models and explained variance | 19 | Projection; variance decomposition |
| Best low-rank approximation of a data matrix | 19, 31 | Eckart–Young; SVD |
J.5 Signals & communications
| Application | Chapter(s) | Linear-algebra concept |
|---|---|---|
| Audio mixing (waveforms add by superposition) | 1 | Linearity / superposition |
| Sound as a vector in a function space (the equalizer in your pocket) | 5 | Function vector spaces |
| Signal compression / sparse representation | 5 | Bases; spans |
| Fourier series (decomposing a signal into pure tones) | 22 | Orthogonal function bases; coefficients as projections |
| MP3 / AAC audio compression (throwing away inaudible sound) | 22 | Orthogonal-basis projection; thresholding |
| JPEG and the Discrete Cosine Transform on 8×8 blocks | 22 | Orthonormal-basis transform |
| Digital electronics and signal integrity (square-wave content) | 22 | Fourier analysis |
| The everyday spectrum (radio, light, audio bands) | 22 | Spectral / frequency decomposition |
| Denoising a signal by projecting out noise | 19 | Orthogonal projection |
| Orthonormal bases in signal processing | 20 | Gram–Schmidt; QR |
| Control systems and signal processing (oscillatory response) | 26 | Complex eigenvalues |
| Error-correcting codes (encoding as a linear map) | 35 | Kernel and image of a linear map |
| Background removal in video (separating motion from stillness) | 31 | Low-rank + sparse SVD decomposition |
J.6 Economics & finance
| Application | Chapter(s) | Linear-algebra concept |
|---|---|---|
| Leontief input–output model of an economy | 3, 9 | Linear systems; the Leontief inverse |
| Input–output economics (balancing production) | 4 | Gaussian elimination |
| Equilibrium in a production economy | 9 | Matrix inverse |
| Unique market equilibrium | 11 | $\det \ne 0 \Leftrightarrow$ invertible |
| State transitions / Markov steps in economics | 7, 8 | Stochastic matrices; matrix powers |
| A subscription business as a Markov chain | 8, 25 | Transition matrices; steady state |
| The long run of a labor market (steady states, mixing rates) | 25 | Diagonalization; dominant eigenvalue |
| Boom and bust / business cycles | 26 | Complex eigenvalues as oscillation |
| A quant's pricing pipeline (many scenarios, one factorization) | 10 | LU decomposition reused across right-hand sides |
| Factor models of the market | 19, 32 | Projection; PCA |
| Cyclic dynamics in economics | 26 | Complex eigenvalues |
J.7 Engineering
| Application | Chapter(s) | Linear-algebra concept |
|---|---|---|
| Circuit analysis (currents via Kirchhoff's laws) | 3, 4 | Linear systems; Gaussian elimination |
| Currents in a circuit as a linear system | 3 | Systems of equations |
| The four subspaces of an electrical network | 14 | Column/null/row/left-null spaces |
| Structural engineering: one stiffness matrix, many loads | 10 | LU factorization reuse |
| Circuit simulation across thousands of timesteps | 10 | LU decomposition |
| Will a truss stand or wobble? (the four subspaces of a truss) | 14 | Rank; null space (mechanisms) |
| Degrees of freedom in a sensor network | 14 | Rank; dimension |
| Degrees of freedom in robotics | 6, 15 | Dimension; independence |
| Where is the robot's hand? (two-joint arm) | 7 | Rotation matrices; composition |
| Spacecraft and robot orientation; rigid orientation in $\mathrm{SO}(3)$ | 21 | Orthogonal matrices; rotations |
| Rotated frames in robotics and graphics | 16 | Change of basis |
| Structural engineering and stability | 28 | Positive-definite matrices |
| The pseudoinverse: solving the unsolvable | 30 | SVD pseudoinverse |
| When a bridge simulation loses its digits | 38 | Condition number; stability |
| Two coupled tanks (mixing, equilibrium) | 37 | Matrix exponential; systems of ODEs |
| The RLC circuit and the damping spectrum | 26, 37 | Eigenvalues deciding oscillation vs. decay |
| Critical damping (why a defective matrix matters) | 36 | Jordan form; defective eigenvalues |
J.8 Biology, chemistry & networks
| Application | Chapter(s) | Linear-algebra concept |
|---|---|---|
| Balancing chemical equations | 4, 13 | Row reduction; the null space as a conservation law |
| Crystal-lattice and molecular volume | 11 | Determinant as volume |
| Counting the dimension of a molecule (vibrational modes) | 15 | Dimension; degrees of freedom |
| Population growth rates (Leslie matrices) | 24 | Eigenvalues as long-term growth rate |
| Cyclic dynamics in ecology | 26 | Complex eigenvalues |
| Genes mirror geography (population genetics) | 32 | PCA |
| Counting connections in a social network | 8 | Powers of an adjacency matrix |
| Graph diffusion and network analysis | 27 | Symmetric (Laplacian) eigendecomposition |
| PageRank (ranking the web) | 3, 23, 29 | Stochastic matrices; dominant eigenvector; power iteration |
| Markov chains everywhere | 29 | Stochastic matrices; steady state |
| Ranking sports teams on uneven schedules | 29 | Eigenvector centrality |
| Which scientific papers matter? (citation networks) | 29 | Eigenvector influence |
| A defective customer pipeline (cohorts piling up) | 36 | Jordan blocks |
PageRank's arc. The book's flagship application is seeded in Chapter 3 (the web as a giant linear system) and Chapter 23 (a teaser that it is secretly an eigenvector problem), then solved in Chapter 29 as the dominant eigenvector of a stochastic matrix, computed by power iteration — the very routine you build in the from-scratch toolkit. The capstone (Chapter 39) lets you rank a real citation network with it.
J.9 Cross-cutting: cryptography, computing, and numerics
| Application | Chapter(s) | Linear-algebra concept |
|---|---|---|
| Hill-cipher cryptography (encrypt/decrypt with a matrix) | 9 | Matrix multiplication and inverse |
| Reachable states in control theory | 13 | Column space as "what you can steer to" |
| Algorithm analysis by solving a recurrence | 25 | Diagonalization (closed-form powers) |
| The Jacobian and change of variables in probability | 11 | Determinant as a volume-scaling factor |
| Numerical reliability in scientific computing | 30, 38 | SVD; condition number |
| The derivative and integral as linear operators | 35 | Abstract linear maps on function spaces |
| Orthogonal polynomials in numerical computing | 34 | Inner-product spaces; orthogonality |
How the catalog reflects the book's themes
Scan the Concept columns and a pattern jumps out: a small number of ideas — linear systems, projection, eigenvalues, and the SVD — appear over and over across completely unrelated fields. That is the book's fourth theme made visible: linear algebra is the most applied branch of pure mathematics precisely because the same handful of structures governs recommenders and bridges, qubits and citation networks, audio compression and market equilibria. The fields rotate; the mathematics stays the same. When you have finished the book, this appendix doubles as a menu: pick a field you care about, and you already own the tools to read everything in its row.