Eigenvalues and Eigenvectors of 2x2 Matrices

Introduction

Welcome to the next lesson in your study of linear algebra. In our previous lessons, we explored how matrices transform space and how we can use the inverse of a matrix to reverse a transformation and solve systems of equations like $A x = b$ . We saw that every vector $x$ is mapped to a new vector $b$ .

Today, we'll focus on a fascinating question: during a transformation, are there any vectors that don't change their direction? That is, vectors that are simply stretched, shrunk, or flipped, but remain on the same line through the origin. These special vectors are called eigenvectors, and the factor by which they are scaled is the corresponding eigenvalue.

This lesson will teach you how to compute eigenvalues and eigenvectors of a 2x2 matrix. This is not just a mathematical exercise; understanding a transformation's principal axes (its eigenvectors) is fundamental to many advanced AI techniques, such as dimensionality reduction with Principal Component Analysis (PCA) and analyzing the behavior of complex systems.

1. The Core Idea: Vectors that Stay on their Span

The formal definition of an eigenvector $v$ and its corresponding eigenvalue $λ$ for a matrix $A$ is:

A v = λ v

This equation states that when the matrix $A$ acts on the vector $v$ , the result is the same as simply scaling $v$ by a number $λ$ . The vector $v$ must be non-zero, otherwise the equation is trivially true for any $λ$ .

To build a strong intuition for what this means geometrically, let's begin with a video from 3Blue1Brown's excellent series.

Eigenvectors and eigenvalues | Chapter 14, Essence of linear algebra

This video provides a powerful visual explanation of eigenvectors as the vectors that remain on their own span (the line they define) during a linear transformation.

Please watch the segment from 01:19 to 05:15. Focus on the core idea: most vectors are knocked off their span by a transformation, but some special vectors (eigenvectors) are just stretched or squished, staying on the same line. The amount they are stretched by is the eigenvalue.

2. Finding the Eigenvalues: The Characteristic Equation

Now that you have the geometric intuition, how do we find these special values and vectors algebraically? Your preference for a theory-first approach will be well-served here, as the derivation is elegant and builds directly on concepts like determinants from our previous lessons.

Let's rearrange the core equation:

Start with the definition: $A v = λ v$
Subtract $λ v$ from both sides: $A v - λ v = 0$
To factor out $v$ , we need to express $λ v$ as a matrix-vector product. We can do this using the identity matrix $I$ : $A v - λ I v = 0$
Now, factor out $v$ :
$(A - λ I) v = 0$

This equation is key. We are looking for a non-zero vector $v$ that solves this. You might recognize this from our work on matrix inverses: a matrix-vector product can only equal the zero vector for a non-zero vector if the matrix itself is singular. A singular matrix is one that squishes space into a lower dimension, and a key property of a singular matrix is that its determinant is zero.

Therefore, to find the eigenvalues $λ$ , we need to find the values that make the matrix $(A - λ I)$ singular. This leads us to the characteristic equation:

det (A - λ I) = 0

The solutions to this equation are the eigenvalues of matrix $A$ .

The following video segment provides a great visual summary of this derivation.

Eigenvectors and eigenvalues | Chapter 14, Essence of linear algebra

The same 3Blue1Brown video explains this derivation, connecting the algebra to the geometric idea of a transformation collapsing space onto a line or point.

Now watch from 05:15 to 09:28. This segment shows how the equation A\mathbf{v} = \lambda\mathbf{v} leads directly to the requirement that \det(A - \lambda I) = 0.

3. The Computational Method: A Step-by-Step Guide

We now have a complete, two-part procedure for finding eigenvalues and eigenvectors.

Find the Eigenvalues ( $λ$ ): Compute the determinant of $(A - λ I)$ . This will give you a polynomial in $λ$ (the "characteristic polynomial"). Solve the characteristic equation $det (A - λ I) = 0$ to find the roots, which are the eigenvalues.
Find the Eigenvectors ( $v$ ): For each eigenvalue $λ$ you found, substitute it back into the equation $(A - λ I) v = 0$ and solve for the vector $v$ . The non-zero solutions form the set of eigenvectors for that eigenvalue.

Let's walk through an example. Consider the matrix:

A = [1411]

Step 1: Find Eigenvalues
First, we set up the characteristic equation $det (A - λ I) = 0$ :

det ([1411] - λ [1001]) = 0

det [1 - λ 4 1 1 - λ] = 0

Now, we compute the determinant of this 2x2 matrix:

(1 - λ) (1 - λ) - (1) (4) = 0

λ^{2} - 2 λ + 1 - 4 = 0

λ^{2} - 2 λ - 3 = 0

We solve this quadratic equation by factoring:

(λ - 3) (λ + 1) = 0

The eigenvalues are $λ_{1} = 3$ and $λ_{2} = - 1$ .

Let's reinforce this by watching a clear, narrated example.

Finding Eigenvalues and Eigenvectors

The video 'Finding Eigenvalues and Eigenvectors' by Professor Dave Explains provides a clear, step-by-step example of this process.

Watch the segment from 04:38 to 06:31. He walks through the exact process we just outlined: setting up \det(A - \lambda I) = 0 and solving the quadratic to find the eigenvalues for this very same matrix.

Step 2: Find Eigenvectors
Now we find the eigenvector for each eigenvalue.

For $λ_{1} = 3$ :
We solve $(A - 3 I) v = 0$ .

[1 - 3 4 1 1 - 3] [v_{1} v_{2}] = [00] ⟹ [- 2 4 1 - 2] [v_{1} v_{2}] = [00]

Both rows give the same equation: $- 2 v_{1} + v_{2} = 0$ , or $v_{2} = 2 v_{1}$ . This means any vector where the second component is twice the first is an eigenvector. A simple choice is $v_{1} = [12]$ .

For $λ_{2} = - 1$ :
We solve $(A - (- 1) I) v = 0$ .

[1 - (- 1) 4 1 1 - (- 1)] [v_{1} v_{2}] = [00] ⟹ [2412] [v_{1} v_{2}] = [00]

This gives the equation $2 v_{1} + v_{2} = 0$ , or $v_{2} = - 2 v_{1}$ . A simple choice is $v_{2} = [1 - 2]$ .

The following video segment walks through this exact process.

Finding Eigenvalues and Eigenvectors

Now for the second part: finding the eigenvectors. Professor Dave provides a very clear walkthrough for both eigenvalues.

Watch from 06:31 to 11:27. Observe how he takes each eigenvalue (\lambda=3 and \lambda=-1), plugs it back into (A - \lambda I)\mathbf{v} = \mathbf{0}, and solves the resulting system. Note that there's a whole line of eigenvectors for each \lambda; we just pick a simple representative.

The image below provides a clean summary of a similar calculation, which you can use as a reference.

Calculating Eigenvalues and Eigenvectors of a 2x2 Matrix — This image provides a clear, handwritten-style summary of the entire process for a 2x2 matrix: setting up and solving the characteristic equation for the eigenvalues, and then solving the system $(A - \lambda I)\mathbf{v} = \mathbf{0}$ for the eigenvectors.

4. A Polymath's Insight: The Trace and Determinant Shortcut

For someone aiming for deep understanding, it's valuable to see connections between different concepts. For any 2x2 matrix, there is a beautiful relationship between its eigenvalues and two properties we already know: the trace and the determinant.

Let's look at the characteristic polynomial for a general 2x2 matrix $A = [a c b d]$ :

det (A - λ I) = (a - λ) (d - λ) - b c = λ^{2} - (a + d) λ + (a d - b c) = 0

Notice two things:

The term $(a + d)$ is the trace of the matrix (the sum of the diagonal elements).
The term $(a d - b c)$ is the determinant of the matrix.

From algebra, we also know that for any quadratic equation $λ^{2} + B λ + C = 0$ with roots $λ_{1}$ and $λ_{2}$ , the sum of the roots is $λ_{1} + λ_{2} = - B$ and the product is $λ_{1} λ_{2} = C$ .

Comparing these facts gives us a powerful shortcut:

Sum of eigenvalues = Trace of the matrix: $λ_{1} + λ_{2} = a + d$
Product of eigenvalues = Determinant of the matrix: $λ_{1} λ_{2} = a d - b c$

This means you can often find the eigenvalues more quickly and check your work. For our example matrix $A = [1411]$ :

$trace (A) = 1 + 1 = 2$
$det (A) = (1) (1) - (1) (4) = - 3$

Our calculated eigenvalues were $λ_{1} = 3$ and $λ_{2} = - 1$ . Let's check:

Sum: $3 + (- 1) = 2$ . Matches the trace.
Product: $(3) (- 1) = - 3$ . Matches the determinant.

This provides a more profound connection between the properties of a matrix and its fundamental scaling behavior.

A quick trick for computing eigenvalues

This relationship between the trace, determinant, and eigenvalues is an elegant shortcut. The following article from 3Blue1Brown explains this 'quick trick'.

Read the text starting from the heading '1. The trace of a matrix...' down to 'Take a moment to see how you can derive...'. This section clearly states the two key facts: that the trace equals the sum of eigenvalues, and the determinant equals their product.

Conclusion

In this lesson, you've learned the complete computational procedure for finding the eigenvalues and eigenvectors of a 2x2 matrix, starting from the theoretical definition and ending with a practical calculation.

Key Takeaways:

Eigenvectors are non-zero vectors that only change in length, not direction, when a linear transformation is applied. Eigenvalues are the scaling factors.
The core relationship is defined by the equation $A v = λ v$ .
Eigenvalues are found by solving the characteristic equation $det (A - λ I) = 0$ .
Eigenvectors are found by solving the system of equations $(A - λ I) v = 0$ for each eigenvalue $λ$ .
For any 2x2 matrix, the sum of its eigenvalues equals its trace, and the product of its eigenvalues equals its determinant.

Preview of the Next Lesson:
We have now mastered the computation of eigenvalues and eigenvectors. In the next lesson, we will delve deeper into their geometric interpretation. We will explore what eigenvalues and eigenvectors tell us about transformations like rotations, shears, and projections, and why they form a more "natural" coordinate system for understanding the action of a matrix. This geometric view is essential for appreciating their role in advanced AI and data analysis.