Math AA HL: Eigenvalues and Eigenvectors

Understanding eigenvalues and eigenvectors is not just a procedural exercise in linear algebra; it is a gateway to mastering how matrices, which represent complex transformations, behave at their core. These concepts are fundamental for analyzing systems that evolve over time, such as in physics and economics, and are essential for simplifying complex matrix operations through diagonalization. In your IB Math AA HL course, you will learn to find these special scalars and vectors, interpret them geometrically, and apply them to solve real-world problems, including systems of differential equations.

The Core Idea: Scaling Directions

Before diving into calculations, grasp the fundamental concept. When a square matrix $A$ acts on a vector $v$ via multiplication, it typically rotates and stretches the vector, changing its direction. An eigenvector is a special nonzero vector $v$ that, when transformed by $A$, does not change its direction. It may only be scaled (stretched or shrunk) by a specific factor. That scaling factor is the corresponding eigenvalue, often denoted by $\lambda$ (lambda).

Formally, for an $n\times n$ matrix $A$, a nonzero vector $v$ is an eigenvector if it satisfies the equation: $$A\mathbf{v}=\lambda \mathbf{v}$$ The scalar $\lambda$ is the eigenvalue associated with eigenvector $v$.

Geometrically, imagine a transformation represented by $A$. An eigenvector points along a line that remains unchanged under that transformation. Every vector on that line is simply stretched or compressed by the factor $\lambda$. If $\lambda$ is positive, the direction is preserved; if negative, it is reversed; and if $\lambda =0$, the eigenvector is mapped to the zero vector. This interpretation is key to visualizing matrix behavior.

Finding Eigenvalues: The Characteristic Equation

To find eigenvalues, we rearrange the eigenvector equation: $A\mathbf{v}=\lambda \mathbf{v}$ becomes $A\mathbf{v}-\lambda \mathbf{v}=0$, or $(A-\lambda I)\mathbf{v}=0$, where $I$ is the identity matrix.

For nonzero solutions $\mathbf{v}$ to exist, the matrix $(A-\lambda I)$ must be singular—its determinant must be zero. This leads to the characteristic equation: $$\det (A-\lambda I)=0$$

The determinant of $(A-\lambda I)$ yields a polynomial in $\lambda$, called the characteristic polynomial. The roots of this polynomial are the eigenvalues of $A$.

Worked Example: Let $A=\left(\begin{array}{cc}3& 1\\ 0& 2\end{array}\right)$. Find its eigenvalues.

1. Form $A-\lambda I=\left(\begin{array}{cc}3-\lambda & 1\\ 0& 2-\lambda \end{array}\right)$.
2. Compute the determinant: $\det (A-\lambda I)=(3-\lambda )(2-\lambda )-(1)(0)=(3-\lambda )(2-\lambda )$.
3. Set the characteristic polynomial equal to zero: $(3-\lambda )(2-\lambda )=0$.
4. Solve for $\lambda$: ${\lambda }_1=3$, ${\lambda }_2=2$.

Thus, the eigenvalues are $\lambda =3$ and $\lambda =2$.

Finding Eigenvectors: Solving $(A-\lambda I)\mathbf{v}=0$

Once an eigenvalue $\lambda$ is known, its corresponding eigenvectors are found by solving the homogeneous system of linear equations $(A-\lambda I)\mathbf{v}=0$. The solutions form the eigenspace for that $\lambda$.

Continuing our example, find the eigenvectors for $A=\left(\begin{array}{cc}3& 1\\ 0& 2\end{array}\right)$.

*For ${\lambda }_1=3$:*

1. Substitute into $(A-\lambda I)$: $A-3I=\left(\begin{array}{cc}0& 1\\ 0& -1\end{array}\right)$.
2. Solve $\left(\begin{array}{cc}0& 1\\ 0& -1\end{array}\right)\left(\begin{array}{c}x\\ y\end{array}\right)=\left(\begin{array}{c}0\\ 0\end{array}\right)$.

This gives the equations $0x+1y=0$ and $0x-1y=0$, both simplifying to $y=0$.

1. $x$ is a free variable. Therefore, eigenvectors are of the form $\left(\begin{array}{c}t\\ 0\end{array}\right)$ for any $t\ne 0$. A basis for this eigenspace is $\left(\begin{array}{c}1\\ 0\end{array}\right)$.

*For ${\lambda }_2=2$:*

1. $A-2I=\left(\begin{array}{cc}1& 1\\ 0& 0\end{array}\right)$.
2. Solve $\left(\begin{array}{cc}1& 1\\ 0& 0\end{array}\right)\left(\begin{array}{c}x\\ y\end{array}\right)=\left(\begin{array}{c}0\\ 0\end{array}\right)$.

This gives $1x+1y=0$, or $x=-y$.

1. $y$ is a free variable. So eigenvectors are of the form $\left(\begin{array}{c}-t\\ t\end{array}\right)$. A basis is $\left(\begin{array}{c}-1\\ 1\end{array}\right)$.

Diagonalization: Simplifying Matrix Powers

A square matrix $A$ is diagonalizable if it can be expressed in the form $A=PD{P}^{-1}$, where $D$ is a diagonal matrix containing the eigenvalues of $A$, and $P$ is an invertible matrix whose columns are the corresponding eigenvectors of $A$.

This decomposition is incredibly powerful. For instance, computing high powers of $A$ becomes trivial: $$A^n=(PD{P}^{-1}{)}^n=P{D}^n{P}^{-1}$$ Since $D$ is diagonal, $D^n$ is simply each diagonal entry (eigenvalue) raised to the $n$th power.

A matrix is diagonalizable if and only if it has a full set of $n$ linearly independent eigenvectors. This is guaranteed if all eigenvalues are distinct.

Applying to our example: We have eigenvalues ${\lambda }_1=3,{\lambda }_2=2$ with corresponding eigenvectors $\left(\begin{array}{c}1\\ 0\end{array}\right)$ and $\left(\begin{array}{c}-1\\ 1\end{array}\right)$. We can form: $$P=\left(\begin{array}{cc}1& -1\\ 0& 1\end{array}\right),D=\left(\begin{array}{cc}3& 0\\ 0& 2\end{array}\right)$$ You can verify that $A=PD{P}^{-1}$ and that $A^{10}=P{D}^{10}{P}^{-1}=P\left(\begin{array}{cc}{3}^{10}& 0\\ 0& 2^{10}\end{array}\right)P^{-1}$, a much simpler calculation.

Application: Solving Systems of Linear Differential Equations

One of the most important applications is solving a system of first-order linear differential equations of the form: $$\frac{d\mathbf{x}}{dt}=A\mathbf{x}$$ where $\mathbf{x}(t)$ is a vector of functions and $A$ is a constant matrix.

The diagonalization process decouples this intertwined system. Let $\mathbf{x}=P\mathbf{y}$, where $P$ is the eigenvector matrix. Substituting: $$P\frac{d\mathbf{y}}{dt}=AP\mathbf{y}⟹\frac{d\mathbf{y}}{dt}=P^{-1}AP\mathbf{y}=D\mathbf{y}$$ Since $D$ is diagonal, the system becomes: $$\frac{d{y}_1}{dt}={\lambda }_1{y}_1,\frac{d{y}_2}{dt}={\lambda }_2{y}_2,...$$ These are simple, decoupled equations with solutions $y_i(t)=C_i{e}^{{\lambda }_{i}t}$. The general solution for the original system is then $\mathbf{x}(t)=P\mathbf{y}(t)$, which is a linear combination of terms of the form $e^{\lambda t}\mathbf{v}$, where $(\lambda ,\mathbf{v})$ are eigenvalue-eigenvector pairs.

This elegantly shows how the long-term behavior of the system (growth, decay, oscillation) is governed by the eigenvalues, while the eigenvectors determine the "modes" or patterns of behavior.

Common Pitfalls

1. Algebraic Errors in the Characteristic Polynomial: A single sign error when computing $\det (A-\lambda I)$ will yield incorrect eigenvalues. Always double-check your arithmetic, especially with larger matrices. Use the trace and determinant as a quick check: for a $2\times 2$ matrix, the sum of eigenvalues equals the trace, and their product equals the determinant.

1. Confusing Eigenvectors with Eigenspaces: An eigenvector is not unique; any nonzero scalar multiple of an eigenvector is also an eigenvector for the same eigenvalue. When asked to "find the eigenvectors," you are typically finding a basis for the eigenspace. Do not make the mistake of thinking $\left(\begin{array}{c}1\\ 0\end{array}\right)$ is the only eigenvector for $\lambda =3$ in our example; $\left(\begin{array}{c}-5\\ 0\end{array}\right)$ is equally valid.

1. Assuming Diagonalizability is Guaranteed: Not all matrices are diagonalizable. A matrix may fail if it has repeated eigenvalues but does not have enough linearly independent eigenvectors (a defective matrix). You must verify you can form a full set of independent eigenvectors before writing $A=PD{P}^{-1}$.

1. Misapplying to Non-Square Matrices: Eigenvalues and eigenvectors are only defined for square matrices. The equation $A\mathbf{v}=\lambda \mathbf{v}$ requires $A$ to be $n\times n$ and $\mathbf{v}$ to be $n\times 1$.

Summary

• Eigenvalues ($\lambda$) and eigenvectors ($\mathbf{v}$) satisfy the defining equation $A\mathbf{v}=\lambda \mathbf{v}$, where an eigenvector's direction is invariant under the transformation by $A$, only scaled by $\lambda$.
• You find eigenvalues by solving the characteristic equation $\det (A-\lambda I)=0$. You find corresponding eigenvectors by solving the homogeneous system $(A-\lambda I)\mathbf{v}=0$.
• Diagonalization, expressed as $A=PD{P}^{-1}$, decomposes a matrix using its eigenvalues (in $D$) and eigenvectors (in $P$). This simplifies computing matrix powers and analyzing matrix functions.
• A key application is solving systems of linear differential equations $\frac{d\mathbf{x}}{dt}=A\mathbf{x}$, where the general solution is built from terms $e^{\lambda t}\mathbf{v}$, directly linking the system's dynamics to the eigenvalues and eigenvectors of $A$.
• Mastering these concepts provides a powerful toolkit for simplifying and analyzing linear transformations across pure and applied mathematics.