Cayley–Hamilton Theorem Explained

The Cayley–Hamilton theorem says something surprising: every square matrix satisfies its own characteristic equation. Compute a matrix's characteristic polynomial p(λ), then plug the matrix A itself into it, and you get the zero matrix: p(A) = 0. It matters because this single fact lets you find matrix inverses and high powers without long multiplications.

The theorem works for any square matrix of any size. All you need is the characteristic polynomial, built from a simple determinant.

Key Concepts

Square matrix: a matrix with equal rows and columns, written n×n.

Identity matrix I: 1s on the diagonal, 0s elsewhere — behaves like the number 1 in multiplication.

Characteristic polynomial p(λ): p(λ) = det(A − λI). For a 2×2 matrix it is a quadratic.

Eigenvalues: the roots of p(λ) = 0.

The theorem: p(A) = 0, where the constant term multiplies I and 0 is the zero matrix.

Formulas

For 2×2 A = [[a, b], [c, d]]: p(λ) = λ² − (trace)λ + (det), where trace = a + d and det = ad − bc. The theorem gives A² − (trace)A + (det)I = 0.

Inverse: A⁻¹ = ((trace)I − A) / (det), valid when det ≠ 0. Powers: the same relation reduces A³, A⁴, … to combinations of A and I. Limits: A must be square; the inverse formula fails when det = 0 (singular matrix).

Worked Example

Take A = [[2, 1], [1, 2]]. Step 1: trace = 2 + 2 = 4; det = (2)(2) − (1)(1) = 3. Step 2: p(λ) = λ² − 4λ + 3 = 0, roots λ = 1, 3. Step 3: substitute: A² − 4A + 3I = 0. Step 4: A² = [[5, 4], [4, 5]]. Step 5: [[5,4],[4,5]] − [[8,4],[4,8]] + [[3,0],[0,3]] = [[0,0],[0,0]]. The zero matrix confirms the theorem. Bonus: A⁻¹ = (4I − A)/3 = [[2/3, −1/3], [−1/3, 2/3]] — check by multiplying with A.

Common Misconceptions

Misconception 1: "It only works for 2×2 matrices." It holds for every square n×n matrix; the polynomial is just degree n.

Misconception 2: "p(A) = 0 means the number 0." The zero is the zero matrix of A's size, not the scalar 0.

Misconception 3: "The inverse formula works when det = 0." Dividing by zero is impossible — singular matrices genuinely have no inverse, though p(A) = 0 still holds.

Real-World Applications

Engineers use it for powers of transition matrices in weather and population models, physicists for rotations and quantum operators, and graphics programmers for undoing transformations. Control theory uses characteristic polynomials to test system stability.

The Cayley–Hamilton theorem turns λ² − 4λ + 3 = 0 into A² − 4A + 3I = 0 — a fast route to inverses and powers for any square matrix.