• Advanced
  • About 16 min
  • 5 practice questions

Powers of matrices and inverses from matrix equations

Find high powers of a matrix from a pattern, use idempotent, involutory and nilpotent matrices, and get A⁻¹ directly from an equation such as A² − 5A + 7I = O.

If something looks wrong, please tell us.

The Practice course computed A², checked matrix polynomials and noticed that powers can cycle. Advanced questions go further: they ask for A^50 or a general Aⁿ, or they give a polynomial equation and ask for A⁻¹ without computing any cofactors. Both rest on one idea: once A is tied to I by a simple relation, every power and the inverse follow by algebra.

Patterns in powers

Compute A² and A³, guess the pattern, and confirm it with one more power (a full proof would use induction). For A = [1 a; 0 1], A² = [1 2a; 0 1] and in general Aⁿ = [1 na; 0 1], so [1 2; 0 1]^10 = [1 20; 0 1]. For the all-ones 3 × 3 matrix J, each element of J² is 1 + 1 + 1 = 3, so J² = 3J, J³ = 9J and Jⁿ = 3ⁿ⁻¹J.

Worked example 1: a general power

Let A = [3 −4; 1 −1]. A² = [9 − 4 −12 + 4; 3 − 1 −4 + 1] = [5 −8; 2 −3] and A³ = A²A = [15 − 8 −20 + 8; 6 − 3 −8 + 3] = [7 −12; 3 −5]. The pattern is Aⁿ = [1 + 2n −4n; n 1 − 2n]: n = 2 gives [5 −8; 2 −3] and n = 3 gives [7 −12; 3 −5]. So A^10 = [21 −40; 10 −19]. A quick consistency check: |A| = −3 + 4 = 1, so every power must have determinant 1, and (1 + 2n)(1 − 2n) + 4n² = 1.

Three special kinds of matrix

Matrices defined by an equation
NameDefining equationConsequence
IdempotentA² = AAⁿ = A for every n ≥ 1; I − A is idempotent too
InvolutoryA² = IA⁻¹ = A; even powers are I, odd powers are A
NilpotentA^k = O for some kI + A is invertible, and |A| = 0

With an idempotent A, the binomial expansion works because A commutes with I: (I + A)ⁿ = I + C(n, 1)A + C(n, 2)A² + ... = I + (2ⁿ − 1)A, since every power of A is A and the coefficients other than the first add up to 2ⁿ − 1. So (I + A)³ = I + 7A.

For a nilpotent matrix, the inverse of I + N is a finite series. Take N = [0 1 2; 0 0 3; 0 0 0]. N² has a single non-zero element, 3 in row 1, column 3, and N³ = O. Then (I + N)(I − N + N²) = I + N³ = I, so (I + N)⁻¹ = I − N + N² with no determinant needed. Any strictly upper triangular n × n matrix has Nⁿ = O.

Inverse from a polynomial equation

Every 2 × 2 matrix satisfies A² − (a + d)A + |A| I = O. More generally, if a matrix satisfies A² + pA + qI = O with q ≠ 0, then A(A + pI) = −qI, so A⁻¹ = −(A + pI)/q. Multiply the equation by A⁻¹; do not divide by A.

Worked example 2: A⁻¹ without cofactors

A = [3 1; −1 2] satisfies A² − 5A + 7I = O (Practice course). Rearranging, A(5I − A) = 7I, so A⁻¹ = (5I − A)/7 = (1/7)[2 −1; 1 3]. The usual formula agrees: |A| = 7 and adj A = [2 −1; 1 3]. The polynomial route pays off for 3 × 3 matrices, where a relation such as A³ = 6A² − 11A + 6I gives A⁻¹ = (A² − 6A + 11I)/6 in one line.

Worked example 3: high powers from the polynomial

If A² = 2A + 3I, then A³ = 2A² + 3A = 2(2A + 3I) + 3A = 7A + 6I, and A⁴ = 7A² + 6A = 7(2A + 3I) + 6A = 20A + 21I. Each power reduces to the form xA + yI. A = [1 2; 2 1] satisfies this relation (its diagonal sum is 2 and its determinant is −3), and A⁻¹ = (A − 2I)/3 = [−1/3 2/3; 2/3 −1/3].

If A² + pA + qI = O with q ≠ 0, then A⁻¹ = −(A + pI)/q. Idempotent: (I + A)ⁿ = I + (2ⁿ − 1)A. Nilpotent with N³ = O: (I + N)⁻¹ = I − N + N².

— Rule to remember

Key takeaways

Find a pattern in A, A², A³ and check it with the determinant. Recognise idempotent, involutory and nilpotent matrices and use their equations to collapse powers. Turn any polynomial relation into an inverse by factoring out A, and reduce high powers to the form xA + yI.

Check your understanding

  1. 1. What is [1 2; 0 1]^10?

    1. [1 1024; 0 1]
    2. [10 20; 0 10]
    3. [1 2; 0 1]
    4. [1 20; 0 1]
    Show answer

    Answer: D. [1 a; 0 1]ⁿ = [1 na; 0 1], so the power is [1 20; 0 1].

  2. 2. A is idempotent (A² = A). What is (I + A)³?

    1. I + 3A
    2. I + 7A
    3. I + 8A
    4. 8A
    Show answer

    Answer: B. (I + A)³ = I + 3A + 3A² + A³ = I + 3A + 3A + A = I + 7A.

  3. 3. A = [2 3; 1 2] satisfies A² − 4A + I = O. What is A⁻¹?

    1. [−2 3; 1 −2]
    2. [2 3; 1 2]
    3. [6 −3; −1 6]
    4. [2 −3; −1 2]
    Show answer

    Answer: D. A(4I − A) = I, so A⁻¹ = 4I − A = [2 −3; −1 2]. Check: |A| = 1 and adj A gives the same matrix.

  4. 4. A = [3 −4; 1 −1]. What is A^10?

    1. [21 −40; 10 −9]
    2. [30 −40; 10 −10]
    3. [21 −40; 10 −19]
    4. [19 −40; 10 −21]
    Show answer

    Answer: C. Aⁿ = [1 + 2n −4n; n 1 − 2n], so A^10 = [21 −40; 10 −19]. Its determinant is −399 + 400 = 1, as it must be.

  5. 5. N = [0 1 2; 0 0 3; 0 0 0]. What is the smallest k with N^k = O?

    1. 2
    2. 4
    3. 3
    4. No power of N is O
    Show answer

    Answer: C. N² = [0 0 3; 0 0 0; 0 0 0] is not O, but N³ = N²N = O.

Privacy · Terms · Refunds · Delivery