Singular Values and Norms Explained with Worked Examples

The spectral norm of a matrix is its largest singular value , and the Frobenius norm is . The condition number is .

Write 2x or 2*x to multiply, ^ for powers, and ; between simultaneous equations. You can also type factor, expand, derivative of, integrate … from a to b, limit … as x->0 or 15% of 80.

Try an example

Before you start

This page is easier if you are comfortable with:

What Singular values and norms means

The spectral norm is the biggest factor by which the matrix can stretch a vector, found from the singular value decomposition. The Frobenius norm treats the matrix as a long vector and uses the magnitude formula, giving the same value as the root of the sum of squared singular values.

The condition number measures how sensitive the solution of is to small errors. A large ratio means the matrix is nearly singular, close to having a zero determinant.

For an orthogonal matrix all singular values are 1, so the condition number is 1: the best possible. See low-rank approximation for errors measured in these norms.

What is it used for?

Singular values measure how much a matrix can stretch a vector. Matrix norms are built from them.

How to use singular values and norms

  1. Find the singular values.
  2. The spectral norm is the largest one.
  3. The Frobenius norm is the root of the sum of squares of the entries (or of the singular values).
  4. Divide largest by smallest for the condition number.

Graph

Singular values 3 and 1: spectral norm 3, Frobenius norm √10σ₁ = 3σ₂ = 1Frobenius ≈ 3.16200.511.522.533.5400.511.522.533.54Singular values 3 and 1: spectral norm 3, Frobenius norm √10xyGraph with x from 0 to 4 and y from 0 to 4. Shows bars (3); point σ₁ = 3 at (1, 3); point σ₂ = 1 at (2, 1); point Frobenius ≈ 3.162 at (3, 3.16228).

The spectral norm is the largest bar; the Frobenius norm combines all the singular values.

Worked examples

These examples are worked by hand, step by step, so you can compare your own working.

Example 1: the norms of a diagonal matrix

Find the spectral norm and condition number of .

Worked example
Answer
∥A∥2=3, κ=1.5\|A\|_{2}=3,\ \kappa=1.5
  1. Singular values
  2. Condition number

Example 2: Frobenius norm

Find the Frobenius norm of .

Worked example
Answer
50≈7.07\sqrt{50}\approx7.07
  1. Entries
  2. Via singular values

Example 3: a rank-one matrix

Find the Frobenius norm of .

Worked example
Answer
22
  1. Entries

Special cases

A few situations come up again and again. Knowing them saves time and helps you spot answers that cannot be right.

σ = (5, 2)

spectral norm

5.

same

Frobenius √(25 + 4)

5.3852.

same

condition number 5/2

2.5.

A = (1 2; 3 4)

Frobenius √(1 + 4 + 9 + 16)

5.4772.

Word problems

Spectral norm = σ₁; Frobenius norm = √(Σa²ᵢⱼ) = √(Σσ²ᵢ); condition number = σ_max/σ_min.

Spectral

Largest singular value.

  1. 5
  2. Check: ✓5

Frobenius

√(σ₁² + σ₂²).

  1. √29
  2. Check: ✓5.3852

Condition number Higher

5 ÷ 2.

  1. 2.5
  2. Check: ✓2.5

Direct Frobenius Higher

Entries 1, 2, 3, 4.

  1. √30
  2. Check: ✓5.4772

Common mistakes

Spectral norm = largest eigenvalue

eigenvalue ✗

largest singular value ✓

Not eigenvalues of A.

Frobenius = Σ|a|

sum ✗

root of the sum of squares ✓

Formula.

Condition number σ_min/σ_max

inverted ✗

σ_max/σ_min ✓

≥ 1.

Condition number < 1

possible ✗

always ≥ 1 ✓

Ratio.

Exam tips

  1. ‖A‖₂ = σ₁.
  2. ‖A‖_F = √Σσ².
  3. κ = σ_max/σ_min ≥ 1.
  4. Large κ = ill-conditioned.
  5. Orthogonal Q: κ = 1.

Practice questions

Try each one on paper, then open the answer. Use the solver above to see the full working.

1. Frobenius of σ = (3, 4) Foundation

5

2. Condition number for σ = (10, 2) Foundation

5

3. Frobenius of (2 0; 0 3), 4 dp Foundation

3.6056

4. Condition number for σ = (8, 8) Higher

1

5. Frobenius of (1 1; 1 1) Higher

2

6. Condition number for σ = (6, 0.5) Higher

12

Practice generator

Unlimited fresh questions with instant marking. Choose a level, work it out on paper, then type your answer.

Solve: …

Key terms

Spectral norm
σ₁.
Frobenius norm
√Σa².
Condition number
σ_max/σ_min.
Ill-conditioned
Large κ.
Singular value
σ.
Matrix norm
Size of a matrix.

Frequently asked questions

What is the spectral norm?

The largest singular value.

What is the Frobenius norm?

√(Σaᵢⱼ²) = √(Σσᵢ²).

What is the condition number?

σ_max/σ_min.

What does a large condition number mean?

Solutions are sensitive to errors.

Which page covers the SVD?

singular value decomposition.

Which page covers vector magnitude?

magnitude of a vector.

More in Vectors and matrices and the formula sheets.