Systems: Complex Eigenvalues — Solving Systems with Complex Eigenvalues (Spirals and Centers) - Differential Equations

SECTION 1

Historical Context & Motivation

When mathematicians and scientists first began studying systems of differential equations, they quickly realized that some systems don't settle down to a steady state or shoot off to infinity. Instead, these systems oscillate — they cycle, spiral, and rotate. Think of a pendulum swinging back and forth, or two competing animal populations that rise and fall in a repeating pattern. The mathematics behind these behaviors turns out to involve complex eigenvalues, which are eigenvalues that contain the imaginary unit i (where i² = −1).

The development of this theory stretches across several centuries. From Euler's groundbreaking formula connecting exponentials and trigonometry, to Poincaré's geometric vision of differential equations, each breakthrough added a piece to the puzzle. Understanding this history helps us see why complex eigenvalues aren't just an algebraic curiosity — they're the key to modeling oscillatory behavior in physics, biology, and engineering.

1748 • Euler's Formula

Leonhard Euler publishes the famous identity e^iθ = cos θ + i sin θ, connecting exponential functions with trigonometric functions. This formula is the mathematical bridge that lets us convert complex eigenvalue solutions into sines and cosines.

1850s • Matrix Theory Takes Shape

Arthur Cayley and James Joseph Sylvester formalize the algebra of matrices and introduce the concept of eigenvalues (originally called 'characteristic roots'). This gives mathematicians a systematic tool for analyzing systems of linear equations.

1881 • Poincaré's Phase Portraits

Henri Poincaré introduces the geometric approach to differential equations, classifying solution curves into spirals, centers, nodes, and saddles. His phase portrait method lets us visualize how systems evolve over time.

1920s–1960s • Control Theory & Engineering

Engineers apply complex eigenvalue analysis to electrical circuits, mechanical vibrations, and control systems. The real part of a complex eigenvalue becomes a key indicator of whether oscillations grow, shrink, or stay constant — critical for designing stable aircraft and bridges.

The central question this lesson addresses is: when the eigenvalues of a 2 × 2 system are complex numbers α ± βi, how do we write the general solution in terms of real-valued functions, and what does that solution look like geometrically? By the end, you'll be able to determine whether a system produces spirals or centers just by inspecting its eigenvalues.

SECTION 2

Core Principles & Definitions

Before diving into the solution process, let's establish the foundational ideas you'll need. A system of linear differential equations can be written in the compact form x′ = Ax, where A is a 2 × 2 matrix of constants and x is a vector of unknown functions. The behavior of this system is governed by the eigenvalues of A — the special numbers λ that satisfy det(A − λI) = 0. When the discriminant of the characteristic equation is negative, the eigenvalues come in conjugate pairs: λ = α + βi and λ̄ = α − βi.

Complex Eigenvalues Come in Pairs

For a matrix with real entries, complex eigenvalues always appear as conjugate pairs: α + βi and α − βi. You never get just one complex eigenvalue by itself.

α Controls Growth or Decay

The real part α determines whether solutions spiral inward (α < 0), spiral outward (α > 0), or form closed orbits (α = 0). Think of it as a 'damping' or 'amplifying' factor.

β Controls Oscillation Speed

The imaginary part β determines how fast the solution rotates. A larger |β| means faster oscillation — the system cycles more rapidly.

Spirals vs. Centers

When α ≠ 0, trajectories are spirals (either inward or outward). When α = 0, trajectories are closed ellipses called centers — the oscillation neither grows nor decays.

✦

KEY TAKEAWAY

Think of complex eigenvalues like a person running on a spiral staircase. The real part α tells you whether you're going up (expanding outward), going down (collapsing inward), or staying on the same floor (a center). The imaginary part β tells you how fast you're running around the staircase. If α = 0, you're just running in circles on one level — that's a center.

SECTION 3

Visualizing Spirals and Centers

The most powerful way to understand complex eigenvalue systems is through phase portraits — plots in the x₁x₂-plane that show how the state of the system evolves over time. Each curve in the phase portrait is called a trajectory, and the direction arrows show which way the system moves as time increases. The diagram below shows three different behaviors determined by the sign of α.

Left: When α < 0, trajectories spiral inward toward the origin (stable spiral). Center: When α = 0, trajectories form closed ellipses — a center. Right: When α > 0, trajectories spiral outward from the origin (unstable spiral).

In the diagram above, notice how the direction arrows always curve rather than travel in straight lines. This curving motion is the hallmark of complex eigenvalues. In contrast, systems with real eigenvalues produce straight-line trajectories (nodes and saddle points). The stable spiral on the left represents a system where energy is being dissipated — like a pendulum with friction that gradually comes to rest. The center in the middle represents a frictionless system — like an ideal pendulum that swings forever. The unstable spiral on the right represents a system where energy is being added — like an oscillation that grows out of control.

SECTION 4

Mathematical Framework

Let's set up the mathematics step by step. We start with a 2 × 2 system x′ = Ax where A is a constant matrix. Our goal is to find the eigenvalues of A, and when those eigenvalues turn out to be complex, to express the general solution using only real-valued functions (sines, cosines, and exponentials).

CHARACTERISTIC EQUATION

det(A − λI) = 0 → λ² − (tr A)λ + det A = 0

Here, tr A is the trace (sum of diagonal entries) and det A is the determinant. Complex eigenvalues occur when (tr A)² − 4(det A) < 0.

COMPLEX EIGENVALUE FORM

λ = α ± βi where α = (tr A)/2, β = √(4 det A − (tr A)²) / 2

The value α is the real part and β is the imaginary part (always taken as positive).

Once you find the complex eigenvalue λ = α + βi, you compute the corresponding eigenvector v by solving (A − λI)v = 0. This eigenvector will also have complex entries, so you can write it as v = a + bi, where a and b are real vectors. The trick is to form two real-valued solutions from this single complex eigenvector.

GENERAL REAL SOLUTION

x(t) = c₁ eᵅᵗ(a cos βt − b sin βt) + c₂ eᵅᵗ(a sin βt + b cos βt)

Here, a = Re(v) is the real part of the eigenvector, b = Im(v) is the imaginary part, and c₁, c₂ are arbitrary constants determined by initial conditions.

💡

Why Euler's Formula Matters Here

The general solution formula comes from applying Euler's formula: e^iβt = cos βt + i sin βt. When you multiply the complex exponential e^(α+βi)t by the complex eigenvector and then separate real and imaginary parts, you get two linearly independent real solutions. The factor e^αt provides growth or decay, while cos βt and sin βt provide the oscillation.

SECTION 5

Classifying Behavior: Spirals vs. Centers

Now that we have the general solution formula, we can classify the behavior of any 2 × 2 linear system with complex eigenvalues. The classification depends entirely on α, the real part. The table below summarizes the three cases, and the diagram that follows provides a visual decision flowchart.

Condition	Behavior	Stability	Physical Example
α < 0 (negative real part)	Stable spiral — trajectories spiral inward	Asymptotically stable: solutions → 0 as t → ∞	Damped pendulum, shock absorber
α = 0 (purely imaginary)	Center — trajectories form closed ellipses	Stable but NOT asymptotically: solutions stay bounded but don't approach 0	Ideal frictionless pendulum, LC circuit
α > 0 (positive real part)	Unstable spiral — trajectories spiral outward	Unstable: solutions → ∞ as t → ∞	Feedback howl in a microphone, nuclear chain reaction

Classification of complex eigenvalue systems

This flowchart shows the decision process: first check whether eigenvalues are complex, then examine the sign of the real part α to classify the system as a stable spiral, center, or unstable spiral.

One detail that many students overlook is the direction of spiraling. Whether trajectories spiral clockwise or counterclockwise depends on the specific entries of the matrix A, not just the sign of α. To determine the direction, you can plug a specific point (like the point (1, 0)) into the equation x′ = Ax and see which way the velocity vector points. If the velocity pushes the trajectory counterclockwise, the spiral goes counterclockwise, and vice versa.

SECTION 6

Worked Example

Let's work through a complete example. Consider the system x′ = Ax where:

GIVEN MATRIX

A = [ 1, −2 ; 1, −1 ]

This means x₁′ = x₁ − 2x₂ and x₂′ = x₁ − x₂.

Solving a System with Complex Eigenvalues

1
Step 1 — Find the Characteristic Equation

We compute det(A − λI) = 0. The matrix A − λI is [ (1−λ), −2 ; 1, (−1−λ) ]. The determinant is (1−λ)(−1−λ) − (−2)(1) = −1 − λ + λ + λ² + 2 = λ² + 1.

Characteristic equation: λ² + 1 = 0
2
Step 2 — Find the Eigenvalues

Solving λ² + 1 = 0 gives λ² = −1, so λ = ±i. In our notation, α = 0 and β = 1. Since α = 0, this system will produce a center with closed elliptical orbits.

Eigenvalues: λ = ±i (α = 0, β = 1)
3
Step 3 — Find the Eigenvector for λ = i

We solve (A − iI)v = 0. The matrix A − iI is [ (1−i), −2 ; 1, (−1−i) ]. From the second row: v₁ + (−1−i)v₂ = 0, so v₁ = (1+i)v₂. Choosing v₂ = 1, we get v = [ 1+i, 1 ]. We separate into real and imaginary parts: v = [1, 1] + i[1, 0]. So a = [1, 1] and b = [1, 0].

Eigenvector: v = [1+i, 1], giving a = [1, 1] and b = [1, 0]
4
Step 4 — Write Two Real Solutions

Using the formula with α = 0 (so e^αt = e⁰ = 1), the two real solutions are: x₁(t) = a cos t − b sin t = [1,1] cos t − [1,0] sin t = [cos t − sin t, cos t]. And x₂(t) = a sin t + b cos t = [1,1] sin t + [1,0] cos t = [sin t + cos t, sin t].

x₁(t) = [cos t − sin t, cos t], x₂(t) = [sin t + cos t, sin t]
5
Step 5 — Write the General Solution

The general solution is x(t) = c₁ x₁(t) + c₂ x₂(t). Written out in component form:

x(t) = c₁[cos t − sin t, cos t] + c₂[sin t + cos t, sin t]
6
Step 6 — Interpret the Result

Since α = 0, there is no exponential growth or decay — the solutions are purely oscillatory with period 2π. Every solution traces out a closed ellipse in the x₁x₂-plane. The origin is a center. The constants c₁ and c₂ determine which ellipse the trajectory follows and where along the ellipse it starts.

Classification: CENTER — closed elliptical orbits

SECTION 7

Comparing Complex vs. Real Eigenvalue Systems

It helps to contrast complex eigenvalue systems with the real eigenvalue systems you've already studied. The table below highlights the key differences and similarities.

Feature	Real Eigenvalues	Complex Eigenvalues
Eigenvalue form	λ₁, λ₂ are real numbers	λ = α ± βi (conjugate pair)
Solution functions	Pure exponentials: e^λt	Exponentials × trig: e^αt cos βt, e^αt sin βt
Phase portrait shape	Straight-line trajectories (nodes, saddle points)	Curving trajectories (spirals, ellipses)
Oscillation?	No — purely monotone growth/decay	Yes — oscillation is always present
Number of independent solutions	2 (from 2 eigenvectors, or generalized)	2 (real and imaginary parts of one complex solution)
Stability condition	Both eigenvalues negative	Real part α < 0

Comparing systems with real vs. complex eigenvalues

✦

KEY TAKEAWAY

Real eigenvalues produce systems that behave like a car driving straight toward or away from a destination — no turning, no oscillation. Complex eigenvalues produce systems that behave like a car driving on a roundabout: there's always a rotational component. The imaginary part creates the rotation, and the real part determines whether the car is slowly spiraling into the center of the roundabout, spiraling out, or just going around in a perfect circle.

SECTION 8

Connection to Nonlinear Systems & Advanced Theory

The linear systems we've studied in this lesson are important in their own right, but they're also the foundation for analyzing more complex, nonlinear systems. In a nonlinear system, the matrix A is replaced by functions that depend on x₁ and x₂. Near an equilibrium point, you can linearize the nonlinear system (essentially zooming in very close), and the resulting linear approximation has the same eigenvalue-based classification we've been using.

Feature	Linear Systems (This Lesson)	Nonlinear Systems (Advanced)
Equation form	x′ = Ax (constant matrix)	x′ = f(x) (general vector function)
Spiral/center classification	Determined entirely by eigenvalues of A	Determined by eigenvalues of the Jacobian matrix at each equilibrium
Centers	Always produce perfectly closed orbits	May be spirals instead — linearization can be inconclusive for α = 0
Spirals	Spiral in or out forever	May approach limit cycles (stable oscillations)
Exact solutions	Always possible with eigenvalue method	Rarely possible — typically need numerical or qualitative methods

Linear vs. nonlinear system analysis

One important caution for the future: when linearization gives purely imaginary eigenvalues (α = 0), the linear analysis predicts a center, but the actual nonlinear system might be a spiral. This is because the higher-order terms that we ignored during linearization can tip the balance. In contrast, when α ≠ 0, the linearization reliably predicts the behavior of the nonlinear system near the equilibrium. This distinction becomes crucial in courses on nonlinear dynamics and chaos theory.

SECTION 9

Practice Problems

Question 1: CONCEPTUAL

The eigenvalues of a 2 × 2 system are λ = −3 ± 5i. Without solving the system, describe the type of phase portrait and the stability of the origin. Explain your reasoning.

Show answer

Since the eigenvalues are complex (they contain i), the phase portrait will involve curving trajectories rather than straight lines. The real part is α = −3, which is negative, so the system is a stable spiral. Trajectories spiral inward toward the origin as t → ∞. The origin is asymptotically stable because the factor e^−3t causes all solutions to decay to zero. The imaginary part β = 5 determines the oscillation speed.

Question 2: BASIC CALCULATION

Find the eigenvalues of the matrix A = [ 0, −4 ; 1, 0 ]. Classify the origin as a spiral or center.

Show answer

The characteristic equation is det(A − λI) = (0 − λ)(0 − λ) − (−4)(1) = λ² + 4 = 0. Solving gives λ² = −4, so λ = ±2i. The eigenvalues are purely imaginary with α = 0 and β = 2. Since α = 0, the origin is a center. All trajectories are closed ellipses, and the system oscillates with period π (since the angular frequency is β = 2, the period is 2π/β = π).

Question 3: INTERMEDIATE

Solve the system x′ = Ax where A = [ 2, −5 ; 1, −2 ]. Find the eigenvalues, an eigenvector, and write the general solution in real form.

Show answer

The characteristic equation is (2−λ)(−2−λ) − (−5)(1) = λ² − 4 + 5 = λ² + 1 = 0, giving λ = ±i (α = 0, β = 1). For λ = i, we solve (A − iI)v = 0: the first row gives (2−i)v₁ − 5v₂ = 0, so v₁ = 5v₂/(2−i) = 5(2+i)/5 = 2+i. Choosing v₂ = 1: v = [2+i, 1]. Thus a = [2, 1] and b = [1, 0]. The general solution is x(t) = c₁([2,1]cos t − [1,0]sin t) + c₂([2,1]sin t + [1,0]cos t) = c₁[2cos t − sin t, cos t] + c₂[2sin t + cos t, sin t]. Since α = 0, this is a center with purely oscillatory solutions.

Question 4: APPLIED

A damped spring-mass system can be modeled as x′ = [ 0, 1 ; −4, −2 ] x, where x₁ represents position and x₂ represents velocity. Find the eigenvalues and determine whether the mass oscillates as it returns to equilibrium. If so, describe what happens physically.

Show answer

The characteristic equation is (−λ)(−2−λ) − (1)(−4) = λ² + 2λ + 4 = 0. Using the quadratic formula: λ = (−2 ± √(4−16))/2 = (−2 ± √(−12))/2 = −1 ± i√3. The eigenvalues are complex with α = −1 and β = √3 ≈ 1.73. Since α = −1 < 0, the system is a stable spiral. Physically, the mass oscillates back and forth (due to the imaginary part β = √3) while the amplitude of oscillation gradually decreases (due to e^−t from α = −1). The mass eventually comes to rest at the equilibrium position — this is underdamped oscillation.

Question 5: CRITICAL THINKING

Consider the matrix A = [ a, −1 ; 1, a ] where a is a real constant. For what values of a does the system x′ = Ax have a center? For what values is it a stable spiral? An unstable spiral? Prove that the eigenvalues are always complex regardless of a.

Show answer

The characteristic equation is (a−λ)² − (−1)(1) = (a−λ)² + 1 = 0. Expanding: λ² − 2aλ + a² + 1 = 0. The discriminant is (2a)² − 4(a²+1) = 4a² − 4a² − 4 = −4, which is always negative regardless of a. This proves the eigenvalues are always complex. Using the quadratic formula: λ = (2a ± √(−4))/2 = a ± i. So α = a and β = 1. When a = 0, we have a center. When a < 0, the system is a stable spiral. When a > 0, it is an unstable spiral. The key insight is that the discriminant is constant (−4), so no value of a can produce real eigenvalues. The parameter a directly controls the real part α, smoothly transitioning the system from stable spiral through center to unstable spiral.

SUMMARY

Lesson Summary

Summary

When a 2 × 2 system x′ = Ax has a negative discriminant in its characteristic equation, the eigenvalues form a conjugate pair λ = α ± βi. The general solution combines exponential functions (from the real part α) with sine and cosine functions (from the imaginary part β), producing oscillatory behavior. The formula x(t) = c₁e^αt(a cos βt − b sin βt) + c₂e^αt(a sin βt + b cos βt) uses the real and imaginary parts of the eigenvector to construct two linearly independent real solutions.

The classification of the phase portrait depends on the sign of α: α < 0 produces a stable spiral (trajectories spiral inward, asymptotically stable), α = 0 produces a center (closed elliptical orbits, stable but not asymptotically), and α > 0 produces an unstable spiral (trajectories spiral outward, unstable). These patterns model real-world oscillations — from damped pendulums to electrical circuits — and form the foundation for analyzing nonlinear systems through linearization.