← Mathematics for Machine Learning

Learn / Mathematics for Machine Learning

04 - Matrix Decompositions

Advanced mastery of Chapter 4, covering every named section and exercise-style synthesis across book pages 98-138.

Learning path

0%

0 of 4 sections marked complete · about 42 minutes

Learning objectives

What you will be able to explain

  • Determinant and trace
  • A 2x22x2 spectral audit
  • Eigenvalues and eigenvectors
  • Cholesky prerequisites
  • Eigendecomposition and diagonalization
  • SVD anatomy

Section 01

Determinant and trace to Eigenvalues and eigenvectors

01

Determinant and trace

Determinant measures invertibility/volume; trace aggregates diagonal or spectral values.

Guided checkpoint

For square matrices.

Source: Sec. 4.1, pp. 99-105

02

A 2x22x2 spectral audit

Trace is 7 and determinant is 12−2=1012-2=10.

Guided checkpoint

For A=[[4,1],[2,3]]A=[[4,1],[2,3]], compute trace and determinant.

Source: Secs. 4.1-4.2, pp. 99-114

03

Eigenvalues and eigenvectors

Orthogonality follows for symmetric matrices, not arbitrary ones.

Guided checkpoint

Judge each claim.

Source: Sec. 4.2, pp. 105-114

Section 02

Cholesky prerequisites to SVD anatomy

01

Cholesky prerequisites

Positive definiteness yields real positive pivots.

Guided checkpoint

Which matrix admits the standard A=LLTA=LL^T Cholesky factorization?

Source: Sec. 4.3, pp. 114-115

02

Eigendecomposition and diagonalization

Diagonalization requires a complete eigenbasis.

Guided checkpoint

For A=PDP−1A=PDP^{-1}.

Source: Sec. 4.4, pp. 115-119

03

SVD anatomy

SVD exists for rectangular matrices and exposes orthogonal input/output directions.

Guided checkpoint

For A=UΣVTA=U \Sigma V^T, match role.

Source: Sec. 4.5, pp. 119-129

Section 03

SVD and eigensystems to Best rank-k approximation

01

SVD and eigensystems

SVD connects two symmetric eigendecompositions.

Guided checkpoint

Let A=UΣVTA=U \Sigma V^T.

Source: Sec. 4.5, pp. 119-129

02

Low-rank approximation error

Eckart-Young truncates the final singular value; squared error is 121^2.

Guided checkpoint

A matrix has singular values 5,3,1. What is the squared Frobenius error of its best rank-2 approximation?

Source: Sec. 4.6, pp. 129-134

03

Best rank-k approximation

The leading singular modes capture maximal energy.

Guided checkpoint

Which construction minimizes Frobenius error among rank-kk matrices?

Source: Sec. 4.6, pp. 129-134

Section 04

Matrix phylogeny to Exercise-style eigenvalue inference

01

Matrix phylogeny

The phylogeny organizes decompositions by matrix structure and purpose.

Guided checkpoint

Match decomposition to its main structural requirement.

Source: Sec. 4.7, pp. 134-135

02

Further-reading bridge

Algorithmic stability and conditioning extend the exact formulas.

Guided checkpoint

Which subject most directly studies stable numerical computation of decompositions?

Source: Sec. 4.8, p. 135

03

Exercise-style eigenvalue inference

Triangular eigenvalues are diagonal entries; product is the determinant 2(−1)4=−82(-1)4=-8.

Guided checkpoint

A triangular matrix has diagonal entries 2,-1,4. What is the product of its eigenvalues?

Source: Chapter 4 exercises, pp. 137-138

Knowledge check

Turn understanding into recall.

The quiz now follows the same concepts in scored form. You can return to this lesson from the quiz whenever a gap appears.

Start the quiz →Back to course