How Many Eigenvalues Does a Matrix Have — and Why?

One of the most fundamental questions in linear algebra is: how many eigenvalues does a matrix have? The answer lies in the nature of the matrix and the algebra behind it. Let's explore the complete reasoning and its implications.

✅ Quick Answer

A square matrix of size \( n \times n \) has exactly \( n \) eigenvalues, counting multiplicities.

These eigenvalues may be:

• All real (for some matrices)
• Some real, some complex (even if the matrix is real)
• All complex (if the matrix is complex)

🧠 Why Is This True?

To find eigenvalues, we solve the characteristic equation:

\[ \det(A - \lambda I) = 0 \]

This gives a polynomial of degree \( n \) in \( \lambda \), known as the characteristic polynomial.

By the Fundamental Theorem of Algebra, any degree-\( n \) polynomial has exactly \( n \) complex roots (real or complex), when counting multiplicities. Therefore, an \( n \times n \) matrix always has \( n \) eigenvalues.

📌 Multiplicities: Algebraic vs. Geometric

It’s important to distinguish between two types of multiplicity when talking about eigenvalues:

Type	Definition
Algebraic Multiplicity	How many times an eigenvalue appears as a root of the characteristic polynomial
Geometric Multiplicity	The number of linearly independent eigenvectors corresponding to that eigenvalue

🧮 Example: A 3×3 Matrix

Suppose you have a matrix \( A \in \mathbb{R}^{3 \times 3} \). Its characteristic equation might look like:

\[ \lambda^3 - 4\lambda^2 + 5\lambda - 2 = 0 \]

This third-degree polynomial will have 3 roots (real or complex), meaning:

• 3 real eigenvalues
• or 1 real and 2 complex conjugates
• or a repeated root (e.g., \( \lambda = 1 \) twice and \( \lambda = 2 \) once)

In all cases: three eigenvalues total, because the matrix is 3×3.

📘 Summary Table

Matrix Size	Number of Eigenvalues	Notes
\( n \times n \)	Exactly \( n \)	Including multiplicities, in the complex field
2×2 Real Matrix	2	May be real or complex conjugates
Defective Matrix	Less than \( n \) independent eigenvectors	Still has \( n \) eigenvalues algebraically

🔁 Final Thoughts

Every square matrix has a full set of eigenvalues — exactly as many as its dimension. The nature of these eigenvalues (real or complex, distinct or repeated) tells you much about the matrix’s structure, stability, and behavior. Whether you're analyzing a system of equations, a Markov chain, or performing PCA, eigenvalues are essential to understanding the matrix's true effect.

If a Matrix Has Two Eigenvalues, Are There Only Two Directions of Stretch or Squash?

This is a deep and insightful question — and with a slight refinement, it leads to a powerful geometric understanding of matrix transformations in 2D. Let's unpack it.

✅ Refined Understanding

If a 2×2 matrix has two distinct eigenvalues and is diagonalizable, then there are exactly two independent directions (the eigenvectors) where the matrix acts by only stretching or squashing, without rotating.

In those two directions, the matrix scales vectors by the eigenvalues \( \lambda_1 \) and \( \lambda_2 \). These directions remain unchanged (except for their length and possibly sign) when the matrix is applied.

🔁 What Happens Geometrically?

When you apply a matrix to a vector, it usually changes both its direction and magnitude. But:

• Eigenvectors are special directions where only the length changes, not the direction.
• Eigenvalues are the factors by which those vectors are stretched or compressed.

So, for a 2×2 matrix with two distinct eigenvalues:

\[ A \vec{v}_1 = \lambda_1 \vec{v}_1, \quad A \vec{v}_2 = \lambda_2 \vec{v}_2 \]

Along \( \vec{v}_1 \) and \( \vec{v}_2 \), the transformation is pure scaling — these are the “axes of pure action.” Any other vector is a combination of \( \vec{v}_1 \) and \( \vec{v}_2 \), and its transformation is a blend of these two stretchings, often resulting in rotation or shear.

📌 When This Is Not True

There are exceptions where having two eigenvalues doesn’t guarantee two independent directions of pure scaling:

Situation	What Happens	Example Matrix
Repeated eigenvalue, but one eigenvector (defective)	Only one direction is invariant under scaling	\( \begin{bmatrix} 1 & 1 \\ 0 & 1 \end{bmatrix} \)
Complex eigenvalues	No real direction of scaling — transformation involves rotation	\( \begin{bmatrix} 0 & -1 \\ 1 & 0 \end{bmatrix} \)

So the presence of two eigenvalues does not always imply two distinct directions of pure stretch/squash in real 2D space. But it does when:

• The eigenvalues are distinct
• The matrix is diagonalizable

💡 Visualization Insight

Imagine pressing on a sheet of elastic material. In most directions, the shape bends or twists. But in exactly two special directions (the eigenvectors), the material stretches or compresses straight — no twisting involved. Those are the matrix’s eigen-directions.

🧠 Summary

Yes, we can say that if a 2×2 matrix has two distinct eigenvalues and is diagonalizable, then it has exactly two directions — the eigenvectors — where it stretches or squashes the space without rotating it. These are the “pure action” directions of the matrix.

In all other directions, the matrix transformation is a composite of the actions along the eigenvectors, which can involve rotation, skewing, or flattening.