Linearization Intuition in Phase Plane Analysis

students, in many differential equations, the hardest part is understanding how a system behaves near an equilibrium point. A phase plane can show the big picture, but sometimes the curves are too complicated to read directly. That is where linearization comes in ✨. The main idea is simple: near a point of interest, a nonlinear system often behaves a lot like a linear system. This gives us a powerful shortcut for predicting motion, stability, and local behavior.

In this lesson, you will learn how linearization helps us study equilibrium points, why it is useful in phase plane analysis, and how to interpret the result. By the end, you should be able to explain the core intuition, connect it to stability, and use it to reason about systems without solving everything exactly.

Why linearization matters

Many real-world systems are nonlinear. That means terms like $x^2$, $xy$, $\sin(x)$, or $e^x$ appear in the differential equations. Nonlinear systems are often impossible to solve exactly in a simple closed form. But near an equilibrium point, the system may look almost linear. That local approximation can reveal a lot about the nearby trajectories.

Think about a ball in a bowl 🏀. If the ball is near the bottom, the shape of the bowl is gentle and smooth. Very close to the bottom, the bowl can be approximated by a parabola, which is a simple curve. In the same way, a nonlinear system can be approximated by its tangent behavior near an equilibrium point.

In phase plane analysis, we usually want to know:

Where are the equilibrium points?
Are they stable or unstable?
What do nearby trajectories do?

Linearization helps answer these questions by replacing the nonlinear system with a simpler linear one near the equilibrium.

Suppose we have a system written as

$$\frac{dx}{dt}=f(x,y), \qquad \frac{dy}{dt}=g(x,y).$$

An equilibrium point $(x_0,y_0)$ satisfies

$$f(x_0,y_0)=0, \qquad g(x_0,y_0)=0.$$

To study the system near $(x_0,y_0)$, we look at small changes $u=x-x_0$ and $v=y-y_0$. The idea is to rewrite the system in terms of these small deviations and keep only the first-order terms.

The intuition behind “local linear behavior”

The most important idea in linearization is that very small changes are governed mainly by the first derivatives. In single-variable calculus, you may know the tangent-line approximation

$$f(x)\approx f(a)+f'(a)(x-a).$$

This says that near $x=a$, the function behaves almost like a line. Linearization for systems is the same idea, but with two variables and matrices.

If $f$ and $g$ are differentiable, then near $(x_0,y_0)$ we approximate:

$$f(x,y)\approx f(x_0,y_0)+f_x(x_0,y_0)(x-x_0)+f_y(x_0,y_0)(y-y_0),$$

$$g(x,y)\approx g(x_0,y_0)+g_x(x_0,y_0)(x-x_0)+g_y(x_0,y_0)(y-y_0).$$

Because $(x_0,y_0)$ is an equilibrium, the constant terms are $0$. So the system becomes approximately

$$\frac{du}{dt}=f_x(x_0,y_0)u+f_y(x_0,y_0)v,$$

$$\frac{dv}{dt}=g_x(x_0,y_0)u+g_y(x_0,y_0)v.$$

This is a linear system. Its coefficient matrix is called the Jacobian matrix:

$$J(x,y)=\begin{pmatrix} f_x(x,y) & f_y(x,y) \\ g_x(x,y) & g_y(x,y) \end{pmatrix}.$$

At the equilibrium point, the linearization is given by $J(x_0,y_0)$.

What does this mean intuitively? The Jacobian measures how each component of the system changes when $x$ or $y$ changes a little. So the linearization tells us the local “push” the system gives to nearby points.

How linearization fits into phase plane analysis

Phase plane analysis studies trajectories in the $xy$-plane instead of viewing $x$ and $y$ as functions of time separately. Each point in the plane represents a state of the system. The direction field and solution curves show how the state evolves.

When we linearize at an equilibrium point, we are zooming in on a tiny neighborhood around that point 🔍. The linearized system usually has a much easier phase portrait than the original nonlinear system. Because linear systems are well understood, we can classify the equilibrium using eigenvalues of the Jacobian matrix.

For a $2\times 2$ matrix, the eigenvalues tell us whether nearby trajectories move toward the equilibrium, away from it, or around it. In particular:

If both eigenvalues have negative real parts, the equilibrium is locally stable.
If at least one eigenvalue has positive real part, the equilibrium is unstable.
If eigenvalues are purely imaginary or one has zero real part, the linearization may be inconclusive for the nonlinear system.

This connection is why linearization is central in phase plane analysis. It links the geometry of trajectories to algebraic information from derivatives.

For example, consider a nonlinear system whose Jacobian at an equilibrium has eigenvalues $\lambda_1=-2$ and $\lambda_2=-5$. The linearized system predicts that solutions near the equilibrium decay toward it. In a phase plane sketch, trajectories spiral in or move straight in depending on the eigenvectors and whether the eigenvalues are real or complex. The exact nonlinear system may differ farther away, but close to the equilibrium, the local behavior is similar.

Example 1: a simple nonlinear system

Consider

$$\frac{dx}{dt}=x-x^3,$$

$$\frac{dy}{dt}=-y.$$

The equilibrium points satisfy

$$x-x^3=0, \qquad -y=0.$$

So $x=0$ or $x=\pm 1$, and $y=0$. The equilibria are $(0,0)$, $(1,0)$, and $(-1,0)$.

Now find the Jacobian:

$$J(x,y)=\begin{pmatrix} 1-3x^2 & 0 \\ 0 & -1 \end{pmatrix}.$$

At $(0,0)$, we get

$$J(0,0)=\begin{pmatrix} 1 & 0 \\ 0 & -1 \end{pmatrix}.$$

The eigenvalues are $\lambda_1=1$ and $\lambda_2=-1$. One direction is repelling and the other is attracting, so $(0,0)$ is a saddle point and therefore unstable.

At $(1,0)$, we get

$$J(1,0)=\begin{pmatrix} -2 & 0 \\ 0 & -1 \end{pmatrix}.$$

Both eigenvalues are negative, so trajectories near $(1,0)$ move toward the equilibrium. The same is true at $(-1,0)$ because

$$J(-1,0)=\begin{pmatrix} -2 & 0 \\ 0 & -1 \end{pmatrix}.$$

This example shows how linearization quickly identifies local stability without solving the nonlinear system exactly.

Example 2: why the approximation is only local

Now consider a system such as

$$\frac{dx}{dt}=y-x^3,$$

$$\frac{dy}{dt}=-x-y.$$

Near $(0,0)$, the terms $x^3$ and higher-order effects are small compared with the linear terms. The Jacobian is

$$J(x,y)=\begin{pmatrix} -3x^2 & 1 \\ -1 & -1 \end{pmatrix},$$

so at $(0,0)$,

$$J(0,0)=\begin{pmatrix} 0 & 1 \\ -1 & -1 \end{pmatrix}.$$

The linearized system describes the behavior near the origin. But if $x$ and $y$ become large, the nonlinear term $x^3$ can dominate and the linear approximation may no longer match the true trajectories. That is why linearization is not a global description of the whole phase plane.

A good way to think about this is a map of a city 🗺️. A small neighborhood map is accurate for walking around downtown, but it does not show the whole country. Linearization is the neighborhood map near an equilibrium.

When linearization works and when it can fail

Linearization gives the best results when the equilibrium is hyperbolic. That means none of the eigenvalues of the Jacobian have real part $0$. In this case, the nonlinear system usually has the same local qualitative behavior as the linearized system.

If an eigenvalue has real part $0$, then linearization may not tell the whole story. For example, a center in the linear system may become a spiral or something more complicated in the nonlinear system. In such cases, higher-order terms matter more, and the linear approximation alone is not enough.

This distinction is important in phase plane analysis. The goal is not just to compute a matrix, but to interpret what the matrix says about motion near equilibrium points.

How to use linearization in practice

When studying a system, students, a practical workflow is:

Find equilibrium points by solving

$$f(x,y)=0, \qquad g(x,y)=0.$$

Compute the Jacobian matrix

$$J(x,y)=\begin{pmatrix} f_x & f_y \\ g_x & g_y \end{pmatrix}.$$

Evaluate the Jacobian at each equilibrium point.

Find the eigenvalues of the resulting matrix.

Use the eigenvalues to classify local behavior and stability.

This procedure is often enough to sketch the local phase portrait and understand whether trajectories approach, leave, or curve around an equilibrium.

A useful connection to remember is that the nonlinear system and its linearization share the same equilibrium point. The linear system does not replace the original problem, but it gives a reliable first look near the equilibrium.

Conclusion

Linearization intuition is one of the most important ideas in phase plane analysis because it turns a hard nonlinear problem into a manageable local linear one. By focusing on small changes near an equilibrium, we use derivatives and the Jacobian matrix to predict nearby motion. The eigenvalues of the linearization help determine stability and the shape of local trajectories.

students, if you remember only one thing, remember this: near an equilibrium point, a smooth nonlinear system often behaves like its linear approximation. That is why linearization is such a powerful tool. It gives a clear, mathematically precise way to understand local behavior without needing an exact solution.

Study Notes

A phase plane shows solutions as curves in the $xy$-plane.
An equilibrium point satisfies $f(x_0,y_0)=0$ and $g(x_0,y_0)=0$.
Linearization approximates a nonlinear system near an equilibrium using first-order terms.
The Jacobian matrix is $J(x,y)=\begin{pmatrix} f_x & f_y \\ g_x & g_y \end{pmatrix}$.
At an equilibrium, the linearized system uses $J(x_0,y_0)$.
Eigenvalues of the Jacobian help determine local stability.
If all eigenvalues have negative real parts, the equilibrium is locally stable.
If any eigenvalue has positive real part, the equilibrium is unstable.
Linearization is a local tool, not a global one.
Hyperbolic equilibria are where linearization is usually most reliable.
If the real part of an eigenvalue is $0$, linearization may be inconclusive.
The big idea is that nearby nonlinear behavior often looks like a linear system.