Repeated Eigenvalues and Generalized Eigenvectors
What you'll learn: How to solve systems of linear ODEs when the coefficient matrix has repeated eigenvalues that don't produce enough independent eigenvectors.
The Problem with Repeated Eigenvalues
When solving a system x' = Ax using the eigenvalue method, we normally need n linearly independent eigenvectors for an n×n matrix. But sometimes a repeated eigenvalue (appearing twice or more) only gives us one eigenvector instead of two. We can't build the full solution with just one eigenvector!
What Are Generalized Eigenvectors?
A generalized eigenvector is a "stand-in" vector that fills the gap when we don't have enough regular eigenvectors. For a repeated eigenvalue λ with only one eigenvector v, we find a generalized eigenvector w by solving:
(A - λI)w = v
Think of it like this: a regular eigenvector is "captured" by A at eigenvalue λ. A generalized eigenvector isn't quite captured itself, but it maps to something that is captured (the eigenvector).
Building the General Solution
For a 2×2 system with repeated eigenvalue λ and eigenvector v:
- First solution: x₁(t) = e^(λt)v
- Second solution (using generalized eigenvector w): x₂(t) = e^(λt)(tv + w)
The general solution is then x(t) = c₁x₁(t) + c₂x₂(t).
Notice the t term in the second solution—it comes from the variation of parameters approach and accounts for the "deficiency" in eigenvectors.
When Do You Need This?
Check the geometric multiplicity (number of independent eigenvectors) versus the algebraic multiplicity (how many times λ appears as a root). If geometric < algebraic, you need generalized eigenvectors.
Key Takeaway: When repeated eigenvalues don't yield enough eigenvectors, generalized eigenvectors—found by solving (A - λI)w = v—complete the solution basis, introducing polynomial terms multiplied by the exponential.