MODULE 01
Linear Algebra
Vector spaces through SVD. The language every model in this curriculum is written in, built from first principles with worked numeric examples.
20 lessons~8h reading
- 0118 min
Vectors and Vector Spaces
BeginnerComing soonVectors as arrows, lists and functions; the eight axioms of a vector space and why they matter.
- 0222 min
Subspaces, Span and Basis
BeginnerComing soonWhich subsets are themselves vector spaces, how span builds them, and why a basis is the minimal description.
Assumes: Vectors and Vector Spaces
- 0324 min
Linear Independence, Rank and Nullity
IntermediateComing soonTesting dependence by elimination, the rank–nullity theorem, and reading rank off a matrix.
Assumes: Subspaces, Span and Basis
- 0420 min
Matrices and Their Operations
BeginnerComing soonMatrices as linear maps: multiplication four different ways, transpose, trace and block structure.
- 0526 min
Special Matrices
IntermediateComing soonDiagonal, triangular, symmetric, orthogonal, projection, idempotent, nilpotent and partitioned matrices, and the properties each guarantees.
Assumes: Matrices and Their Operations
- 0620 min
Systems of Linear Equations
BeginnerComing soonExistence and uniqueness of solutions, consistency, and the geometry of under- and over-determined systems.
Assumes: Matrices and Their Operations
- 0724 min
Gaussian Elimination
BeginnerComing soonRow reduction to echelon and reduced row echelon form, pivoting, and operation counts.
Assumes: Systems of Linear Equations
- 0822 min
LU Decomposition
IntermediateComing soonFactorising A into lower and upper triangular parts, partial pivoting, and why LU beats repeated elimination.
Assumes: Gaussian Elimination
- 0922 min
Determinants
BeginnerComing soonDeterminants as signed volume: cofactor expansion, elimination-based computation, and the product rule.
Assumes: Gaussian Elimination
- 1024 min
Matrix Inverse and Pseudoinverse
IntermediateComing soonInvertibility conditions, Gauss–Jordan inversion, adjugate formula, and the Moore–Penrose pseudoinverse.
Assumes: Determinants
- 1124 min
The Four Fundamental Subspaces
IntermediateComing soonColumn space, null space, row space and left null space, and the orthogonality relations that connect them.
Assumes: Linear Independence, Rank and Nullity
- 1226 min
Orthogonality and Projections
IntermediateComing soonInner products, orthogonal complements, and the projection matrix as the idempotent operator onto a subspace.
Assumes: The Four Fundamental Subspaces
- 1324 min
Gram–Schmidt and QR Decomposition
AdvancedComing soonTurning any basis orthonormal, the numerical instability of classical Gram–Schmidt, and QR for least squares.
Assumes: Orthogonality and Projections
- 1428 min
Least Squares and the Normal Equations
IntermediateComing soonDeriving the normal equations three ways — calculus, geometry and projection — the foundation of linear regression.
Assumes: Orthogonality and Projections
- 1528 min
Eigenvalues and Eigenvectors
IntermediateComing soonThe characteristic polynomial, geometric vs algebraic multiplicity, and what eigenvectors mean geometrically.
Assumes: Determinants
- 1626 min
Diagonalisation and Similarity
AdvancedComing soonWhen a matrix is diagonalisable, similarity transforms, matrix powers, and the spectral theorem for symmetric matrices.
Assumes: Eigenvalues and Eigenvectors
- 1724 min
Quadratic Forms and Definiteness
AdvancedComing soonWriting quadratic forms as xᵀAx, classifying definiteness via eigenvalues and leading minors, and the link to convexity.
Assumes: Diagonalisation and Similarity
- 1832 min
Singular Value Decomposition
AdvancedComing soonEvery matrix factors as UΣVᵀ: derivation, geometric reading, low-rank approximation and the Eckart–Young theorem.
Assumes: Diagonalisation and Similarity · The Four Fundamental Subspaces
- 1922 min
Norms and the Condition Number
AdvancedComing soonVector and matrix norms, equivalence relations between them, and why conditioning decides numerical stability.
Assumes: Singular Value Decomposition
- 2020 min
Linear Algebra in NumPy
BeginnerComing soonTranslating every operation in this module into NumPy, plus the broadcasting and dtype traps that cause silent bugs.
Assumes: Matrices and Their Operations