DIFFERENTIAL EQUATIONS • SYSTEMS OF DIFFERENTIAL EQUATIONS

Phase Portraits & Trajectories — Phase Portraits and Trajectories

Visualize how systems of differential equations evolve by tracing solution curves in the phase plane.

Historical Context & Motivation

For centuries, scientists described change using single equations — one equation tracking one variable over time. But the real world is rarely that simple. A predator population affects its prey, and the prey population in turn affects the predators. Two chemical concentrations influence each other simultaneously. To capture this interconnected behavior, mathematicians developed systems of differential equations, where multiple quantities change together. The challenge then became: how do you see what such a system does without solving every equation by hand?

The answer emerged through a powerful geometric idea. Instead of plotting each variable separately against time, mathematicians plotted the variables against each other. The resulting picture — a phase portrait — reveals the long-term behavior of a system at a glance: does it spiral inward to a stable state, explode outward, or cycle endlessly?

1880s
Poincaré's Qualitative Methods
Henri Poincaré pioneered the idea of studying differential equations geometrically rather than seeking exact formulas. He introduced the phase plane as a visual tool for understanding two-variable systems.
1892
Lyapunov's Stability Theory
Aleksandr Lyapunov formalized the concept of stability, classifying whether solutions near an equilibrium point stay nearby or drift away — a cornerstone of phase portrait analysis.
1920s–30s
Ecological & Engineering Applications
Lotka and Volterra used phase portraits to model predator–prey dynamics, while engineers applied them to electrical circuits and control systems.
1960s–present
Computational Phase Portraits
Computers enabled the rapid generation of phase portraits, making them essential visualization tools in physics, biology, economics, and beyond.

The central question that phase portraits address is this: given a system of differential equations, can we predict where every possible starting condition leads — without solving the equations algebraically? As you will see, the answer is a resounding yes, and the tool for doing so is both elegant and surprisingly intuitive.

Core Principles & Definitions

Before diving into diagrams, you need a handful of key ideas. A system of two first-order differential equations tracks how two quantities — call them x and y — change over time. The phase plane is the xy-plane where each point represents a specific state of the system. As time passes, the state traces out a curve called a trajectory. The collection of many trajectories from different starting points forms the phase portrait.

1

Phase Plane

A coordinate plane where each axis represents one of the system's variables (not time). Every point (x, y) is a possible state of the system.
2

Trajectory (Solution Curve)

A curve in the phase plane traced by a solution as time advances. Each initial condition yields a unique trajectory. Arrows indicate the direction of increasing time.
3

Equilibrium Point

A point where dx/dt = 0 and dy/dt = 0 simultaneously. The system is at rest here. Also called a fixed point or critical point.
4

Direction Field

At every point in the phase plane, the system defines a velocity vector. Plotting many such vectors creates a direction field that guides trajectory shapes.
5

Stability

An equilibrium is stable if nearby trajectories approach it, and unstable if they move away.
KEY TAKEAWAY
Think of the phase plane as a weather map for a system. Each point is a location, and the direction field is like wind arrows showing which way everything is "blowing." A trajectory is the path a leaf would follow if you dropped it into that wind. An equilibrium point is the calm eye of a storm — a place where nothing moves. A phase portrait is the full weather map with many leaf-paths drawn in, revealing the overall flow pattern at a glance.

Visual Explanation — Anatomy of a Phase Portrait

The diagram below shows a typical phase portrait for a stable spiral (also called a spiral sink). Notice how every trajectory curves inward toward the equilibrium at the origin. The arrows on each curve indicate the direction of motion as time increases. Trajectories that start far from the origin still eventually wind their way in, illustrating asymptotic stability: the system settles toward the equilibrium state in the long run.

A stable spiral phase portrait. The pink dot marks the equilibrium at the origin. Each colored curve is a trajectory spiraling inward. Arrows show the direction of motion as time increases.

Several important features stand out. First, no two trajectories ever cross each other — this is a consequence of the uniqueness theorem for differential equations. If two trajectories crossed at a point, that point would have two different "futures," which is impossible for a well-defined system. Second, the winding (spiraling) pattern tells you the system involves oscillatory behavior — the state rotates as it changes. Third, the inward drift tells you the oscillations are decaying — the system is losing energy over time and settling toward the equilibrium.

Mathematical Framework

The simplest systems studied through phase portraits are linear systems with constant coefficients. These involve two first-order equations where the rate of change of each variable depends on both variables in a straightforward (linear) way. The key numbers that determine the shape of every trajectory are the trace and determinant of the system's coefficient matrix — two summary numbers you can compute directly from the four constants in the equations.

📐 Reading a 2×2 Coefficient Matrix
A 2×2 matrix is just a compact way to organize four numbers. If your system is dx/dt = ax + by and dy/dt = cx + dy, the coefficient matrix A stores those four constants: the top row is [a, b] and the bottom row is [c, d]. Two special numbers summarize what the matrix does — the trace (T = a + d, the sum of the diagonal entries) and the determinant (D = ad − bc). You only need these two numbers to classify the phase portrait — no further matrix algebra is required.
COMPONENT FORM
dx/dt = ax + by, dy/dt = cx + dy
The rate of change of each variable depends on the current values of both variables. This coupling is what makes the system interesting. The constants a, b, c, d are given numbers that you read directly from the problem.
TRACE AND DETERMINANT
T = a + d, D = ad − bc
T = a + d is the trace (add the two diagonal entries of the matrix) and D = ad − bc is the determinant (multiply the diagonals and subtract). These two numbers classify every possible phase portrait.
DISCRIMINANT
Δ = T² − 4D
The discriminant tells you whether the growth rates are real numbers or involve oscillation. If Δ > 0: two distinct real growth rates (nodes or saddle). If Δ = 0: one repeated rate. If Δ < 0: complex rates (spirals or centers). The sign of T then determines stability: T < 0 means trajectories approach the equilibrium; T > 0 means they move away.
💡 Why Trace and Determinant Matter
You do not need to solve the entire system to sketch its phase portrait. The trace T and determinant D of the coefficient matrix act like a DNA code: they tell you whether trajectories spiral, converge to straight lines, diverge, or oscillate. Negative trace (T < 0) means trajectories approach the origin (stable). Positive trace (T > 0) means they flee (unstable). A negative discriminant (Δ < 0) means they rotate and spiral.

Classifying Equilibrium Points

Two numbers — the trace T and the determinant D — are all you need to classify the shape of the phase portrait. The diagram below maps out the major types based on where (T, D) falls in the plane. This classification is the single most useful tool for reading a phase portrait.

The trace–determinant plane classifies all 2D linear systems. The horizontal axis is the trace T = a + d (sum of diagonal entries). The vertical axis is the determinant D = ad − bc. The dashed parabola T² = 4D separates nodes (outside the parabola, Δ > 0) from spirals and centers (inside the parabola, Δ < 0). Saddle points live below the T-axis (D < 0).
Classification of equilibrium points by trace, determinant, and discriminant
TypeConditions (T, D, Δ)Trajectory ShapeStability
Stable NodeT < 0, D > 0, Δ > 0Straight-line paths converging to originStable
Unstable NodeT > 0, D > 0, Δ > 0Straight-line paths diverging from originUnstable
Saddle PointD < 0Hyperbolic curves — some approach, some fleeUnstable
Stable SpiralT < 0, D > 0, Δ < 0Spiraling inward toward originStable
Unstable SpiralT > 0, D > 0, Δ < 0Spiraling outward from originUnstable
CenterT = 0, D > 0Closed elliptical orbitsNeutrally stable

Worked Example

Let us classify the equilibrium and sketch the phase portrait for a concrete system. Consider:

GIVEN SYSTEM
dx/dt = −x + 2y, dy/dt = −2x − y
The four constants are a = −1, b = 2, c = −2, d = −1. We can organize these as the coefficient matrix A = [−1, 2; −2, −1], but all the calculations below use just these four numbers directly.
Classify and Sketch the Phase Portrait
1
Step 1 — Identify the Four ConstantsRead the four constants from the equations. Here dx/dt = −x + 2y gives a = −1 and b = 2. The equation dy/dt = −2x − y gives c = −2 and d = −1.
a = −1, b = 2, c = −2, d = −1
2
Step 2 — Compute the Trace and DeterminantThe trace is the sum of the diagonal constants: T = a + d = −1 + (−1) = −2. The determinant is T = ad − bc = (−1)(−1) − (2)(−2) = 1 − (−4) = 5.
T = −2, D = 5
3
Step 3 — Compute the DiscriminantThe discriminant is Δ = T² − 4D = (−2)² − 4(5) = 4 − 20 = −16. Since Δ < 0, the system involves oscillatory (rotating) behavior — trajectories will spiral rather than move in straight lines.
Δ = −16 < 0 → oscillatory (spiral or center)
4
Step 4 — Use the Trace to Determine StabilitySince Δ < 0 we have a spiral or center. To decide which, look at the trace. T = −2 < 0, which means the system is losing energy over time — trajectories decay toward the equilibrium. If T were positive, they would grow outward. If T = 0, they would form closed loops (a center).
T = −2 < 0 → trajectories spiral inward (stable)
5
Step 5 — Classify and Describe the Phase PortraitSince Δ < 0 (oscillatory) and T < 0 (decaying), the equilibrium is a stable spiral. Trajectories spiral inward toward the origin as time goes on. The amount of spiraling is related to |Δ| — a larger |Δ| means faster rotation relative to the decay rate.
Stable spiral — trajectories spiral inward toward the origin

Strengths & Limitations of Phase Portrait Analysis

Phase portraits are remarkably powerful, but like any tool, they have boundaries. Understanding both their strengths and their limitations will help you decide when to use this approach and when you need something more.

Comparing the advantages and disadvantages of phase portrait analysis
StrengthsLimitations
Reveal global behavior at a glance without solving the system algebraically.Limited to two-variable systems for easy visualization (3D phase space is harder to interpret).
Classify stability and type of equilibrium quickly using just the trace and determinant.Do not tell you the exact solution x(t) and y(t) — only the qualitative shape of solutions.
Work for nonlinear systems too (by approximating the system with a linear one near each equilibrium).The linear approximation can fail at borderline cases (for example, when the trace is exactly zero in a nonlinear system).
The uniqueness theorem guarantees trajectories never cross, simplifying the picture.Time information is lost — you cannot read how fast the system moves along a trajectory from the portrait alone.
KEY TAKEAWAY
Phase portraits are like satellite photos of traffic flow: they show you which roads lead to the city center (stable equilibria) and which lead away (unstable), and whether traffic spirals through roundabouts or shoots straight down highways — but they do not tell you the speed of any individual car. When you need exact positions at exact times, you still need to solve the equations, but when you want the big picture, the phase portrait is unbeatable.

Connection to Nonlinear Systems & Advanced Theory

Everything discussed so far applies to linear systems, but the real world is overwhelmingly nonlinear. The good news is that near an equilibrium point, a nonlinear system often behaves like a simpler linear approximation — this idea is called linearization. Just as you can approximate a curved road with a straight line when you zoom in close enough, you can approximate a nonlinear system with a linear one near each equilibrium. You then compute the trace and determinant of that local linear approximation to classify the behavior — exactly the same method as before.

Linear vs. nonlinear phase portrait analysis
FeatureLinear SystemsNonlinear Systems
Coefficient matrixConstant matrix A — same everywhere in the planeLocal linear approximation — changes at each equilibrium
Number of equilibriaOnly one (the origin), if D ≠ 0Can have multiple equilibria throughout the plane
Global behaviorFully determined by trace and determinantMay include limit cycles, chaos, and other complex structures
Classification methodTrace–determinant analysis of ALinearize at each equilibrium, then apply trace–determinant to the local approximation

In more advanced courses, you will encounter phenomena that have no linear counterpart: limit cycles (isolated closed trajectories that nearby solutions approach or flee), bifurcations (qualitative changes in the portrait as a parameter varies), and even chaos in three or more dimensions. The trace–determinant classification you learned here is the essential first step toward all of these topics.

Practice Problems

PROBLEM 1CONCEPTUAL
Explain why two different trajectories in a phase portrait can never cross each other. What theorem guarantees this?
PROBLEM 2BASIC CALCULATION
Given the system dx/dt = 3x, dy/dt = −2y, compute the trace T and determinant D, and classify the equilibrium at the origin.
PROBLEM 3INTERMEDIATE
For the system dx/dt = −3x + 4y, dy/dt = −2x + y, compute the trace T, determinant D, and discriminant Δ. Determine whether the equilibrium is a node, spiral, center, or saddle, and state whether it is stable or unstable.
PROBLEM 4APPLIED
A simple model of two competing species has populations x(t) and y(t) near an equilibrium, governed (after linearization) by dx/dt = −x − y, dy/dt = −x − y. Compute T and D, classify the equilibrium, and interpret what it means biologically.
PROBLEM 5CRITICAL THINKING
A linear system has trace T = 0 and determinant D > 0. Its phase portrait consists of closed elliptical orbits (a center). Explain why you cannot rely on the linear approximation alone to confirm that a nonlinear system also has a center at the same equilibrium. What could go wrong?

Summary

A phase portrait is a collection of trajectories drawn in the phase plane — a coordinate system where each axis represents one of the system's variables, not time. For a linear system dx/dt = ax + by, dy/dt = cx + dy, the behavior of every trajectory is determined by just two numbers computed from the four constants: the trace T = a + d and the determinant D = ad − bc. The discriminant Δ = T² − 4D then tells you whether trajectories spiral (Δ < 0) or move along straight-line-like paths (Δ > 0).

The six main classifications are: stable nodes (T < 0, D > 0, Δ > 0 — paths converge along straight-like curves), unstable nodes (T > 0, D > 0, Δ > 0 — paths diverge), saddle points (D < 0 — always unstable), stable spirals (T < 0, Δ < 0 — spiral inward), unstable spirals (T > 0, Δ < 0 — spiral outward), and centers (T = 0, D > 0 — closed orbits, neutrally stable). Near equilibria of nonlinear systems, you can apply the same trace–determinant method to a local linear approximation to classify each equilibrium's behavior.

Varsity Tutors • Differential Equations • Phase Portraits & Trajectories