Perron–Frobenius Theorem Explained with Worked Examples

A matrix with all entries positive has a largest eigenvalue that is real and positive, with an eigenvector that has all positive entries; every other eigenvalue is smaller in size. It describes the long-term behaviour of repeated multiplication.

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 Perron–Frobenius theorem means

Repeatedly multiplying a vector by a positive matrix pushes it towards the direction of the dominant eigenvector, because the dominant eigenvalue grows fastest (compare the powers in diagonalisation). This explains steady states in population models and rankings of web pages.

For a matrix whose rows each add to 1 (a Markov chain), the dominant eigenvalue is 1 and the eigenvector gives the steady state. Finding the eigenvalues uses the characteristic equation and the vector comes from the eigenvector equation.

The theorem also bounds the eigenvalues between the smallest and largest row sums, and it links to the matrix-tree theorem and other results about graphs and networks.

What is it used for?

For a positive matrix there is one dominant eigenvalue with a positive eigenvector, which controls the long-term behaviour of repeated multiplication.

How to use perron–frobenius theorem

  1. Check all entries are positive.
  2. Find the eigenvalues; the largest in size is real and positive.
  3. Find its eigenvector, which can be chosen with all entries positive.
  4. Use it for long-term behaviour.

Graph

Powers of the matrix [[2,1],[1,2]] are dominated by λ = 33⁴ = 8100.511.522.533.544.55020406080100120Powers of the matrix [[2,1],[1,2]] are dominated by λ = 3nλⁿλ = 3λ = 1Graph with x from 0 to 5 and y from 0 to 130. Shows the curve λ = 3; the curve λ = 1; point 3⁴ = 81 at (4, 81).

The largest eigenvalue grows far faster than the other, so it decides the direction after many steps.

Worked examples

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

Example 1: a symmetric positive matrix

Find the Perron eigenvalue of .

Worked example
Answer
λ=3, v=(1,1)\lambda=3,\ \mathbf{v}=(1,1)
  1. Eigenvalues
  2. Eigenvector for 3

Example 2: an all-ones matrix

Find the Perron eigenvalue of .

Worked example
Answer
λ=2, v=(1,1)\lambda=2,\ \mathbf{v}=(1,1)
  1. Trace 2, determinant 0

Example 3: a Markov matrix

Show that has largest eigenvalue 1.

Worked example
Answer
λ=1 and 0.3\lambda=1\ \text{and}\ 0.3
  1. Rows add to 1, so (1,1) is an eigenvector
  2. Other eigenvalue from the trace

Special cases

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

A = (2 1; 1 2)

dominant eigenvalue

3.

A = (1 2; 3 2)

dominant eigenvalue

4.

same

row sums range

3 to 5.

Markov (0.5 0.5; 0.2 0.8)

dominant eigenvalue

1.

Word problems

A positive matrix has a real positive dominant eigenvalue with an all-positive eigenvector; all other eigenvalues are smaller in modulus. It lies between the smallest and largest row sums.

Dominant

(2 1; 1 2): λ² − 4λ + 3.

  1. (4 + 2)/2
  2. Check: ✓3

Dominant

(1 2; 3 2): tr 3, det −4.

  1. (3 + 5)/2
  2. Check: 4 ✓4

Row-sum bound Higher

Row sums 3 and 5; λ = 4 is between.

  1. 3 ≤ 4 ≤ 5
  2. Check: ✓4

Stochastic matrix Higher

Rows sum to 1.

  1. 1
  2. Check: ✓1

Common mistakes

Dominant eigenvalue may be negative

negative ✗

positive for a positive matrix ✓

Theorem.

Applies to any matrix

any ✗

non-negative, irreducible ✓

Condition.

Eigenvector has mixed signs

mixed ✗

all positive ✓

Theorem.

Other eigenvalues can be equal in size

equal ✗

strictly smaller for positive matrices ✓

Spectral gap.

Exam tips

  1. Positive matrix: λ₁ > 0.
  2. Eigenvector all positive.
  3. Row sums bracket λ₁.
  4. Markov chains: λ₁ = 1.
  5. Power iteration converges.

Practice questions

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

1. Dominant eigenvalue of (3 1; 1 3) Foundation

4

2. Dominant eigenvalue of (2 2; 2 2) Foundation

4

3. Dominant eigenvalue of (1 1; 1 1) Foundation

2

4. Smaller row sum of (1 2; 3 2) Higher

3

5. Larger row sum of (1 2; 3 2) Higher

5

6. Dominant eigenvalue of (0.9 0.1; 0.3 0.7) Higher

1

Practice generator

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

Solve: …

Key terms

Perron–Frobenius theorem
Dominant eigenvalue of positive matrices.
Dominant eigenvalue
Largest in modulus.
Positive matrix
All entries > 0.
Stochastic matrix
Rows sum to 1.
Spectral gap
λ₁ vs the rest.
Power iteration
Repeated multiplication.

Frequently asked questions

What does the Perron–Frobenius theorem say?

A positive matrix has a real positive dominant eigenvalue with a positive eigenvector.

Where does λ₁ lie?

Between the smallest and largest row sums.

What is λ₁ for a stochastic matrix?

1.

Why does it matter?

It describes long-term behaviour of Markov chains and populations.

Which page covers eigenvectors?

eigenvector equation.

Which page covers the matrix-tree theorem?

matrix tree theorem.

More in Vectors and matrices and the formula sheets.