The Matrix Cookbook

Learn / The Matrix Cookbook

03 - Inverses

Advanced formula selection, assumption checking, and exact application for Inverses, book pages 17-23.

Learning path

0%

0 of 3 sections marked complete · about 25 minutes

Learning objectives

What you will be able to explain

  • Basic inverse algebra
  • Recognize the Woodbury identity
  • Sherman-Morrison scalar case
  • Implications involving inverses
  • Choose a convergent inverse expansion
  • Generalized inverse conditions

Section 01

Basic inverse algebra to Sherman-Morrison scalar case

01

Basic inverse algebra

Inverse distributes over products only with reversed order, not over sums.

Guided checkpoint

Assume all needed inverses exist.

Source: Sec. 3.1, p. 17

02

Recognize the Woodbury identity

The matrix inversion lemma turns a low-rank update into a smaller central inverse.

Guided checkpoint

Which form inverts A+UCVA+UCV using an inverse in the smaller update space?

Source: Sec. 3.2, pp. 18-19

03

Sherman-Morrison scalar case

Sherman-Morrison gives IuvT/(1+vTu)I-uv^T/(1+v^Tu); the entry is 12/3=1/31-2/3=1/3.

Guided checkpoint

Let A=I2A=I_2, u=[1,0]Tu=[1,0]^T, v=[2,0]Tv=[2,0]^T. Compute the top-left entry of (A+uvT)1(A+uv^T)^{-1}.

Source: Sec. 3.2, pp. 18-19

Section 02

Implications involving inverses to Generalized inverse conditions

01

Implications involving inverses

Combine definitions with A2=AA^2=A: multiplying by A1A^{-1} yields A=IA=I.

Guided checkpoint

Audit each statement.

Source: Sec. 3.3, p. 20

02

Choose a convergent inverse expansion

The Neumann series is the matrix analogue of the geometric series and requires a convergence condition.

Guided checkpoint

When the spectral radius of AA is below one, which series equals (IA)1(I-A)^{-1}?

Source: Sec. 3.4, p. 20

03

Generalized inverse conditions

The four Penrose equations uniquely characterize the pseudoinverse.

Guided checkpoint

Match each Moore-Penrose condition.

Source: Secs. 3.5-3.6, pp. 21-23

Section 03

Pseudoinverse consequences

01

Pseudoinverse consequences

Every matrix has a Moore-Penrose pseudoinverse; the associated products are orthogonal projectors.

Guided checkpoint

Let A=USigmaVTA=U Sigma V^T be an SVD.

Source: Sec. 3.6, pp. 21-23

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.