On this page
Before you start
This page is easier if you are comfortable with:
What Symmetric matrices: real eigenvalues, orthogonal eigenvectors (spectral theorem) means
For A = ( 2 1 1 2 ) the eigenvalues are 1 and 3 with eigenvectors ( 1 , − 1 ) and ( 1 , 1 ) , whose scalar product is zero. Normalising them gives the columns of an orthogonal matrix Q .
Non-symmetric matrices need not behave this way: a rotation matrix has complex eigenvalues (see eigenvalues of a 2 × 2 matrix ). The nice behaviour of symmetric matrices is why they appear in quadratic forms and statistics.
The method is the same as diagonalisation, but with the bonus that P − 1 = P T once the eigenvectors are normalised, so no inverse needs to be computed.
What is it used for? Symmetric matrices have real eigenvalues and perpendicular eigenvectors. This makes them especially well-behaved and common in applications.
Principal axes of stress, inertia and data. Quadratic forms and conic sections.Statistics, where covariance matrices are symmetric.
How to use symmetric matrices: real eigenvalues, orthogonal eigenvectors (spectral theorem)
Check that A = A T . Find the eigenvalues from the characteristic equation ; they will be real. Find an eigenvector for each; those for distinct eigenvalues are perpendicular. Normalise them to build Q .
Graph Eigenvectors (1, 1) and (1, −1) of a symmetric matrix are perpendicular origin -3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3 Eigenvectors (1, 1) and (1, −1) of a symmetric matrix are perpendicular x y λ = 3 λ = 1 Graph with x from -3 to 3 and y from -3 to 3. Shows the curve λ = 3; the curve λ = 1; point origin at (0, 0). The two eigen-directions meet at a right angle.
Worked examples
These examples are worked by hand, step by step, so you can compare your own working.
Example 1: a 2 × 2 symmetric matrix Find the eigenvalues and eigenvectors of ( 2 1 1 2 ) .
Worked example
Answer
λ = 1 , 3 ; ( 1 , − 1 ) , ( 1 , 1 ) \lambda=1,3;\ \ (1,-1),(1,1) λ = 1 , 3 ; ( 1 , − 1 ) , ( 1 , 1 ) Eigenvalues
λ 2 − 4 λ + 3 = 0 ⇒ 1 , 3
Eigenvectors
( 1 , − 1 ) and ( 1 , 1 )
Dot product
1 − 1 = 0
Example 2: another symmetric matrix Find the eigenvalues of ( 1 2 2 1 ) .
Worked example
Answer
λ = 3 , − 1 \lambda=3,\ -1 λ = 3 , − 1 Trace 2, determinant −3
λ 2 − 2 λ − 3 = 0
Factorise
( λ − 3 ) ( λ + 1 ) = 0
Example 3: a non-symmetric contrast Find the eigenvalues of ( 0 − 1 1 0 ) .
Worked example
Answer
λ = ± i (not real) \lambda=\pm i\ \text{(not real)} λ = ± i (not real) Trace 0, determinant 1
λ 2 + 1 = 0
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) eigenvalues
3 and 1.
eigenvectors (1,1), (1,−1) dot product
0.
unit eigenvector entry 1/√2
0.7071.
A = (3 1; 1 3) eigenvalues
4 and 2.
Word problems
A = Aᵀ has real eigenvalues, perpendicular eigenvectors for distinct eigenvalues, and A = QDQᵀ with Q orthogonal.
Larger eigenvalue tr 4, det 3.
(4 + 2)/2 Check: ✓ 3
Perpendicular (1, 1)·(1, −1).
1 − 1 Check: ✓ 0
Normalised Higher Divide by √2.
1/√2 Check: ✓ 0.7071
Another matrix Higher (3 1; 1 3): a + b.
3 + 1 Check: ✓ 4
Common mistakes
Eigenvalues may be complex complex ✗
always real ✓
Symmetric.
Eigenvectors always perpendicular always ✗
for distinct eigenvalues ✓
Repeated ones can be chosen so.
Q⁻¹ computed compute ✗
Qᵀ ✓
Orthogonal.
Applies to any matrix any ✗
real symmetric ✓
Condition.
Exam tips
A = Aᵀ ⇒ real λ.
Distinct λ ⇒ perpendicular v.
A = QDQᵀ.
Normalise to get Q.
(a b; b a): λ = a ± b.
Practice questions
Try each one on paper, then open the answer. Use the solver above to see the full working.
1. Larger eigenvalue of (5 2; 2 5) Foundation 7
2. Smaller eigenvalue of (5 2; 2 5) Foundation 3
3. Dot product (1, 1)·(1, −1) Foundation 0
4. Larger eigenvalue of (4 3; 3 4) Higher 7
5. Unit eigenvector entry 1/√2, 4 dp Higher 0.7071
6. Trace of (6 1; 1 6) Higher 12
Practice generator Unlimited fresh questions with instant marking. Choose a level, work it out on paper, then type your answer.
Foundation Higher Mixed New question
Solve: …
The practice generator needs JavaScript.
Key terms
Symmetric matrix A = Aᵀ. Spectral theorem Orthogonal diagonalisation. Orthogonal matrix QᵀQ = I. Eigenvalue Real here. Eigenvector Mutually perpendicular. Orthonormal basis Eigenvector basis.
Frequently asked questions
What do symmetric matrices guarantee? Real eigenvalues and orthogonal eigenvectors.
What is A = QDQᵀ? The orthogonal diagonalisation.
Are eigenvectors for equal eigenvalues perpendicular? They can be chosen to be.
What are the eigenvalues of (a b; b a)? a + b and a − b.
Which page covers diagonalisation? diagonalisation.
Which page covers quadratic forms? quadratic forms.