Skip to main content

Section 14.4 Solving Linear Systems

We’ve explored what linear systems are and how to write them in compact matrix form. Now it’s time to actually solve them. (Later in the chapter, we’ll also learn to visualize their behavior in the phase plane.)
We’ll start with a method that mirrors how we solved single constant-coefficient equations: We guess exponential solutions. This guess transforms the differential system into an algebraic problem β€” one that naturally leads to eigenvalues and eigenvectors.

Subsection Guessing Exponential Solutions

Recall how we solved single first-order linear equations like \(y' = ky\) by guessing solutions of the form \(y(t) = Ce^{kt}\text{.}\) We’ll try the same idea for systems.
Consider the system:
\begin{equation} \begin{array}{l} \dfrac{dx}{dt} = 3x + 4y\\ \dfrac{dy}{dt} = -4x + 3y \end{array}\tag{14.1} \end{equation}
Let’s guess that \(x(t)\) and \(y(t)\) share the same exponential rate \(r\text{:}\)
\begin{equation*} x(t) = Ce^{rt}, \qquad y(t) = De^{rt}. \end{equation*}

πŸ“

Differentiating gives:
\begin{equation*} x' = Cr e^{rt}, \qquad y' = Dr e^{rt}. \end{equation*}
Substituting into the system:
\begin{alignat*}{3} Cr e^{rt} \amp {}={} \amp 3Ce^{rt} \amp {}+{} \amp 4De^{rt}\\ Dr e^{rt} \amp {}={} \amp -4Ce^{rt} \amp {}+{} \amp 3De^{rt} \end{alignat*}
Divide through by \(e^{rt}\) (which is never zero):
\begin{alignat*}{3} Cr \amp {}={} \amp 3C \amp {}+{} \amp 4D\\ Dr \amp {}={} \amp -4C \amp {}+{} \amp 3D \end{alignat*}
This leaves an algebraic system involving the unknowns: \(C\text{,}\) \(D\text{,}\) and \(r\text{.}\) It turns out that these unknowns have a very special relationship. To see this, flip the system around:
\begin{alignat*}{3} 3{\DLBb C} \amp {}+{} \amp 4{\DLGb D} \amp {}={} \amp r{\DLBb C} \\ -4{\DLBb C} \amp {}+{} \amp 3{\DLGb D} \amp {}={} \amp r{\DLGb D} \text{.} \end{alignat*}
Now, convert this into a \(2 \times 2\) matrix system:
\begin{equation*} \begin{bmatrix} 3 \amp 4\\ -4 \amp 3 \end{bmatrix} \begin{bmatrix} {\DLBb C}\\ {\DLGb D} \end{bmatrix} = r \begin{bmatrix} {\DLBb C}\\ {\DLGb D} \end{bmatrix} \end{equation*}
This is the eigenvalue equation \(A \vec{v} = r \vec{v}\) for the coefficient matrix
\begin{equation*} A = \begin{bmatrix} 3 \amp 4 \\ -4 \amp 3 \end{bmatrix}. \end{equation*}
This tells us something powerful:
  • Linear systems often have solutions that are exponential functions.
  • The allowed exponential rates \(r\) are the eigenvalues of \(A\text{.}\)
  • The corresponding coefficients \(C\) and \(D\) form the eigenvectors.

🌌 Example 304. Solving a Simple Linear System.

Solve the system:
\begin{align*} \frac{dx}{dt} \amp = x + y\\ \frac{dy}{dt} \amp = 4x + y \end{align*}
Solution.
Write in matrix form \(\frac{d\vec{X}}{dt} = A\vec{X}\) with \(A = \begin{bmatrix}1 \amp 1 \\ 4 \amp 1\end{bmatrix}\text{.}\)
Guess \(\vec{X}(t) = \vec{v}e^{rt}\text{.}\) Substituting gives \(A\vec{v} = r\vec{v}\text{.}\)
Find eigenvalues of \(A\) by solving \(\det(A - rI) = 0\text{.}\)
\begin{equation*} \det \begin{bmatrix} 1-r \amp 1 \\ 4 \amp 1-r \end{bmatrix} = (1-r)^2 - 4 = r^2 - 2r - 3. \end{equation*}
Set equal to zero: \(r^2 - 2r - 3 = 0\text{.}\) Factor:
\begin{equation*} (r - 3)(r + 1) = 0 \quad \Rightarrow \quad r_1 = 3, \ r_2 = -1. \end{equation*}
Find eigenvectors for each eigenvalue.
  • For r = 3: Solve \((A - 3I)\vec{v} = 0\text{.}\)
    \begin{equation*} \begin{bmatrix} -2 \amp 1 \\ 4 \amp -2 \end{bmatrix} \begin{bmatrix} v_1 \\ v_2 \end{bmatrix} = \begin{bmatrix} 0 \\ 0 \end{bmatrix}. \end{equation*}
    From the first row: \(-2v_1 + v_2 = 0 \Rightarrow v_2 = 2v_1\text{.}\) Choose \(v_1 = 1\text{,}\) giving eigenvector \(\vec{v}_1 = \begin{bmatrix}1 \\ 2\end{bmatrix}\text{.}\)
  • For r = -1: Solve \((A + I)\vec{v} = 0\text{.}\)
    \begin{equation*} \begin{bmatrix} 2 \amp 1 \\ 4 \amp 2 \end{bmatrix} \begin{bmatrix} v_1 \\ v_2 \end{bmatrix} = \begin{bmatrix} 0 \\ 0 \end{bmatrix}. \end{equation*}
    From the first row: \(2v_1 + v_2 = 0 \Rightarrow v_2 = -2v_1\text{.}\) Choose \(v_1 = 1\text{,}\) giving eigenvector \(\vec{v}_2 = \begin{bmatrix}1 \\ -2\end{bmatrix}\text{.}\)
Write the general solution:
\begin{equation*} \vec{X}(t) = C_1 \vec{v}_1 e^{3t} + C_2 \vec{v}_2 e^{-t}. \end{equation*}
Substitute the eigenvectors:
\begin{equation*} \begin{bmatrix} x(t) \\ y(t) \end{bmatrix} = C_1 \begin{bmatrix} 1 \\ 2 \end{bmatrix} e^{3t} + C_2 \begin{bmatrix} 1 \\ -2 \end{bmatrix} e^{-t}. \end{equation*}
Or written componentwise:
\begin{equation*} x(t) = C_1 e^{3t} + C_2 e^{-t}, \qquad y(t) = 2C_1 e^{3t} - 2C_2 e^{-t}. \end{equation*}
The constants \(C_1\) and \(C_2\) will be found from initial conditions.

🌌 Example 305. Complex Eigenvalues and Spirals.

Consider the system:
\begin{align*} \frac{dx}{dt} \amp = 2x - 5y\\ \frac{dy}{dt} \amp = 5x + 2y \end{align*}
Solve the system and describe the behavior of solutions.
Solution.
Write \(A = \begin{bmatrix}2 \amp -5 \\ 5 \amp 2\end{bmatrix}\text{.}\)
Find eigenvalues by solving \(\det(A - rI) = 0\text{.}\)
\begin{equation*} \det \begin{bmatrix} 2-r \amp -5 \\ 5 \amp 2-r \end{bmatrix} = (2-r)^2 + 25 = r^2 - 4r + 29. \end{equation*}
Set equal to zero: \(r^2 - 4r + 29 = 0\text{.}\) Using the quadratic formula:
\begin{equation*} r = \frac{4 \pm \sqrt{(-4)^2 - 4(29)}}{2} = \frac{4 \pm \sqrt{-100}}{2} = 2 \pm 5i. \end{equation*}
The eigenvalues are \(2 \pm 5i\) β€” complex conjugates. The real part (\(2\)) indicates that solutions grow exponentially, and the imaginary part (\(5i\)) indicates that they also rotate: the system’s trajectories spiral outward.
Find an eigenvector for \(r = 2 + 5i\) (the other is its conjugate). Solving \((A - rI)\vec{v} = 0\) gives an eigenvector \(\vec{v} = \begin{bmatrix} 1 \\ -i \end{bmatrix}\text{.}\)
Write the solution.
We form the real solution by taking the real and imaginary parts:
\begin{equation*} \begin{aligned} x(t) \amp = e^{2t}\left( C_1 \cos(5t) + C_2 \sin(5t) \right) \\ y(t) \amp = e^{2t}\left( C_1 \sin(5t) - C_2 \cos(5t) \right) \end{aligned} \end{equation*}
The exponential factor \(e^{2t}\) causes growth; the sine and cosine produce rotation. The system’s phase portrait shows spirals moving outward from the origin β€” a connection we make precise in Connecting Eigenvalues to Phase Portraits.

Checkpoint 306. πŸ€”πŸ’­ Solving Linear Systems Reading Questions.

(a) πŸ“–β“ Eigenvalues and Behavior.

Suppose the eigenvalues of a system are \(r = -2 \pm 3i\text{.}\) What does this tell you about the solutions?
  • They spiral inward toward the origin.
  • They spiral outward away from the origin.
  • The negative real part means decay, not growth.
  • They move straight toward the origin with no oscillation.
  • The imaginary part implies rotation (spiraling).
You have attempted of activities on this page.