Powers of Diagonalizable Matrices
What you'll learn: How to compute high powers of a matrix A^n quickly and efficiently using diagonalization.
Why This Matters
Computing something like A¹⁰⁰ by multiplying A by itself 100 times is computationally nightmarish. But if A is diagonalizable, there's a brilliant shortcut that reduces the work to just raising numbers to the 100th power—a trivial task.
The Core Idea
Recall that a diagonalizable matrix A can be written as:
A = PDP⁻¹
where D is diagonal (containing eigenvalues) and P contains the corresponding eigenvectors as columns.
Here's the key insight: when you want to compute A², watch what happens:
A² = (PDP⁻¹)(PDP⁻¹) = PD(P⁻¹P)DP⁻¹ = PDDP⁻¹ = PD²P⁻¹
The middle terms P⁻¹P cancel to give the identity matrix! This pattern continues for any power:
A^n = PD^nP⁻¹
Since D is diagonal, D^n is incredibly easy to compute—just raise each diagonal entry to the nth power:
If D = diag(λ₁, λ₂, λ₃), then D^n = diag(λ₁ⁿ, λ₂ⁿ, λ₃ⁿ)
Example Walk-Through
Suppose A = PDP⁻¹ where D = diag(2, 3). To find A¹⁰:
- Compute D¹⁰ =
diag(2¹⁰, 3¹⁰)=diag(1024, 59049) - Then A¹⁰ = PD¹⁰P⁻¹
Just three matrix multiplications versus nine for repeated multiplication!
The Analogy
Think of P and P⁻¹ as "coordinate transformations." In the eigenvector coordinate system, your matrix is diagonal (simple), so taking powers is trivial. Then you transform back to the original coordinates.
Key Takeaway: For diagonalizable matrices, computing A^n = PD^nP⁻¹ reduces the problem to raising individual eigenvalues to the nth power—exponentially faster than repeated multiplication.