5.5Diagonalisation and multiplicities
The idea
A matrix is diagonalisable when it has n independent eigenvectors. Then A = PDP⁻¹, and powers become easy.
Think of it like this: Changing to the right pair of glasses: in the eigenvector basis the matrix is just n independent stretches.
Why it exists
It explains when repeated eigenvalues are harmless, and gives Aᵏ = PDᵏP⁻¹.
How it works
- Algebraic multiplicity: how often λ is a root of the characteristic equation.
- Geometric multiplicity: the dimension of its eigenspace (always ≤ algebraic).
- Diagonalisable ⇔ geometric = algebraic for every eigenvalue (n distinct eigenvalues are enough; real symmetric always works).
- Then P has the eigenvectors as columns, D the eigenvalues, and Aᵏ = PDᵏP⁻¹.
Worked example: The shear [[2, 1],[0, 2]]
λ = 2 twice (algebraic 2) but only one independent eigenvector (geometric 1): it is not diagonalisable.
Now play with it
- →Compare the 3×3 symmetric matrix (repeated eigenvalue, still diagonalisable) with the Jordan block.
- →Raise a diagonalisable matrix to a power and check against direct multiplication.
| -1/2 | 1 |
| 1 | 1 |
| 2 | 0 |
| 0 | 5 |
| -2/3 | 2/3 |
| 2/3 | 1/3 |
| 2094 | 1031 |
| 2062 | 1063 |
| 2094 | 1031 |
| 2062 | 1063 |
| eigenvalue λ | algebraic mult. | geometric mult. | eigenvectors (basis of the eigenspace) |
|---|---|---|---|
| 2 | 2 | 1 | (1, 0, 0) |
| 3 | 1 | 1 | (0, 0, 1) |
Check yourself
Which 2 × 2 matrix is NOT diagonalisable?
Exam notes (the dense version)
A is diagonalisable if A = PDP⁻¹ with D diagonal, which happens exactly when A has n linearly independent eigenvectors. Each eigenvalue has an algebraic multiplicity (its multiplicity as a root) and a geometric multiplicity (the dimension of its eigenspace, which is ≤ the algebraic one). Diagonalisable ⇔ geometric = algebraic for every eigenvalue. n distinct eigenvalues are enough; real symmetric matrices are always diagonalisable. Then Aᵏ = PDᵏP⁻¹.
Previous year questions: Eigenvalues & Eigenvectors (14)
PYQs are grouped by topic, so this is the full set for Eigenvalues & Eigenvectors; try the ones that match this subtopic. Open all 14 on the topic page.