Course contentsShow
← Mathematics
Lesson 2195 of 3,12145. Linear Algebra — Determinants and EigenvaluesFree lesson

When Diagonalization Fails

Identifying defective matrices that lack enough linearly independent eigenvectors to diagonalize.

When Diagonalization Fails

What You'll Learn

You'll learn to identify defective matrices—matrices that cannot be diagonalized because they lack a full set of linearly independent eigenvectors.

The Core Problem

Not every matrix can be diagonalized, even if you can find all its eigenvalues. Remember that diagonalization requires constructing a matrix P from n linearly independent eigenvectors for an n×n matrix. Sometimes, a matrix simply doesn't have enough independent eigenvectors to form P.

What Makes a Matrix Defective?

A matrix is defective when its geometric multiplicity (the dimension of an eigenspace) is less than its algebraic multiplicity (how many times an eigenvalue appears as a root of the characteristic polynomial).

Example: Consider a 3×3 matrix with eigenvalue λ = 5 appearing three times (algebraic multiplicity = 3). If the eigenspace for λ = 5 only has dimension 1 or 2, you cannot find 3 linearly independent eigenvectors. The matrix is defective and cannot be diagonalized.

The Repeated-Eigenvalue Warning Sign

Defectiveness typically occurs with repeated eigenvalues. If all eigenvalues are distinct, you're guaranteed n independent eigenvectors. But repeated eigenvalues require investigation: does each eigenspace have enough independent vectors?

Analogy: Think of building a 3-person committee (your matrix P). If you need 3 people but only 2 qualified candidates exist, you can't form the committee—similarly, you can't form P without enough eigenvectors.

What Happens Instead?

When diagonalization fails, you may need more advanced techniques like Jordan canonical form (beyond this lesson's scope), or work directly with the matrix without diagonalizing it.

Key Takeaway: A matrix is defective and cannot be diagonalized when the geometric multiplicity of at least one eigenvalue is less than its algebraic multiplicity—meaning you don't have enough linearly independent eigenvectors to form the matrix P.