01 / 21

Classifying
Critical Points

Learn to identify and classify nodes, saddles, spirals, and centers using eigenvalues of the Jacobian matrix in systems of differential equations.

Differential Equations

What Is a Critical Point?

A critical point (also called an equilibrium point) of a system \(\mathbf{x}' = \mathbf{f}(\mathbf{x})\) is any point \(\mathbf{x}_0\) where \(\mathbf{f}(\mathbf{x}_0) = \mathbf{0}\). At a critical point, the system is at rest: no change occurs over time.

Why Classify Critical Points?

Stability
Does the system return to equilibrium after a small disturbance, or diverge away?
Behavior
Do trajectories spiral in, rush straight toward the point, or oscillate forever?
Applications
Population models, circuits, mechanical systems: all depend on the character of equilibria.

The Phase Portrait Zoo

Stable Node Unstable Node Saddle Point Spiral Center Star Node

The Eigenvalue Method

For the linear system \(\mathbf{x}' = A\mathbf{x}\) with coefficient matrix \(A\), the eigenvalues \(\lambda_1, \lambda_2\) of \(A\) completely determine the type and stability of the critical point at the origin.

Classification Rules

1
Real, same sign, distinct: \(\lambda_1 \neq \lambda_2\), both negative \(\Rightarrow\) stable node; both positive \(\Rightarrow\) unstable node.
2
Real, opposite signs: \(\lambda_1 < 0 < \lambda_2\) \(\Rightarrow\) saddle point (always unstable).
3
Complex: \(\lambda = \alpha \pm \beta i\). If \(\alpha < 0\) \(\Rightarrow\) stable spiral; if \(\alpha > 0\) \(\Rightarrow\) unstable spiral.
4
Pure imaginary: \(\lambda = \pm \beta i\) (i.e. \(\alpha = 0\)) \(\Rightarrow\) center (stable but not asymptotically stable).
5
Repeated real: \(\lambda_1 = \lambda_2 = \lambda\). If \(A = \lambda I\) \(\Rightarrow\) star node; otherwise \(\Rightarrow\) degenerate (improper) node.

The \(\tau\)-\(\Delta\) Plane

Spirals / Centers Stable Nodes Unstable Nodes Saddle Points Centers
\(\tau\)
\(\Delta\)
\(\tau^2 = 4\Delta\)
(\( \Delta < 0 \))
\(\tau < 0\)
\(\tau > 0\)
(\( \tau = 0 \))

Nodes vs Saddles

Stable Node
\(\lambda_1, \lambda_2 < 0\), both real. All trajectories approach the origin. \(\Delta > 0\), \(\tau < 0\).
Unstable Node
\(\lambda_1, \lambda_2 > 0\), both real. All trajectories flee the origin. \(\Delta > 0\), \(\tau > 0\).
Saddle Point
\(\lambda_1 < 0 < \lambda_2\). Trajectories approach along one eigenvector, flee along the other. \(\Delta < 0\).

Spirals vs Centers

Stable Spiral
\(\lambda = \alpha \pm \beta i\) with \(\alpha < 0\). Trajectories spiral inward. The real part causes decay.
Unstable Spiral
\(\lambda = \alpha \pm \beta i\) with \(\alpha > 0\). Trajectories spiral outward. The real part causes growth.
Center
\(\lambda = \pm \beta i\) (pure imaginary). Closed orbits around the origin. No decay or growth.

How to Classify Step by Step

1
Compute \(\tau = \text{tr}(A) = a_{11} + a_{22}\) and \(\Delta = \det(A) = a_{11}a_{22} - a_{12}a_{21}\).
2
Check \(\Delta\): if \(\Delta < 0\), it is a saddle point. Stop here.
3
Compute the discriminant \(D = \tau^2 - 4\Delta\).
4
If \(D > 0\): real distinct eigenvalues. Use sign of \(\tau\): negative \(\Rightarrow\) stable node, positive \(\Rightarrow\) unstable node.
5
If \(D < 0\): complex eigenvalues. \(\tau < 0\) \(\Rightarrow\) stable spiral; \(\tau = 0\) \(\Rightarrow\) center; \(\tau > 0\) \(\Rightarrow\) unstable spiral.
6
If \(D = 0\): repeated eigenvalue \(\lambda = \tau/2\). Check if \(A = \lambda I\) for star node vs degenerate node.

Example 1: Stable Node

Classify the critical point at the origin for the system with coefficient matrix:

1
\(\tau = -3 + (-2) = \) \(-5\), \(\Delta = (-3)(-2) - (1)(0) = \) \(6\)
2
\(\Delta = 6 > 0\), so not a saddle.
3
\(D = (-5)^2 - 4(6) = 25 - 24 = \) \(1 > 0\): real, distinct eigenvalues.
4
\(\tau = -5 < 0\) \(\Rightarrow\) Stable Node. Eigenvalues: \(\lambda_1 = -3\), \(\lambda_2 = -2\).

Example 2: Saddle Point

1
\(\tau = 1 + (-2) = \) \(-1\), \(\Delta = (1)(-2) - (3)(2) = -2 - 6 = \) \(-8\)
2
\(\Delta = -8 < 0\) \(\Rightarrow\) Saddle Point. Done: no further steps needed.

Example 3: Stable Spiral

1
\(\tau = -1 + (-1) = \) \(-2\), \(\Delta = (-1)(-1) - (-4)(2) = 1 + 8 = \) \(9\)
2
\(\Delta = 9 > 0\): not a saddle.
3
\(D = (-2)^2 - 4(9) = 4 - 36 = \) \(-32 < 0\): complex eigenvalues.
4
\(\tau = -2 < 0\) \(\Rightarrow\) Stable Spiral. Eigenvalues: \(\lambda = -1 \pm 2\sqrt{2}\,i\).

Example 4: A Center

1
\(\tau = 0 + 0 = \) \(0\), \(\Delta = (0)(0) - (-3)(3) = \) \(9\)
2
\(\Delta = 9 > 0\): not a saddle. \(D = 0 - 36 = -36 < 0\): complex eigenvalues.
3
\(\tau = 0\) exactly \(\Rightarrow\) Center. Eigenvalues: \(\lambda = \pm 3i\). Closed elliptical orbits.

Common Mistakes

✕ Confusing Stability
A center is stable but NOT asymptotically stable. Trajectories stay bounded but never converge to the origin.
✕ Forgetting the Saddle Shortcut
If \(\Delta < 0\), it is always a saddle: no need to compute \(\tau^2 - 4\Delta\). Many students waste time continuing the algorithm.
✕ Sign Errors in \(\det(A)\)
Remember: \(\det(A) = a_{11}a_{22} - a_{12}a_{21}\). Miscomputing the cross terms flips the entire classification.

Summary Table

Eigenvalues Type Stability Key Condition
Real, both \(< 0\) Stable Node Asymptotically Stable \(\Delta > 0, \tau < 0, D \ge 0\)
Real, both \(> 0\) Unstable Node Unstable \(\Delta > 0, \tau > 0, D \ge 0\)
Real, opposite signs Saddle Unstable \(\Delta < 0\)
Complex, \(\alpha < 0\) Stable Spiral Asymptotically Stable \(D < 0, \tau < 0\)
Complex, \(\alpha > 0\) Unstable Spiral Unstable \(D < 0, \tau > 0\)
Pure imaginary Center Stable (not asymp.) \(D < 0, \tau = 0\)

Question 1 of 5

The matrix \(A = \begin{pmatrix} 2 & 0 \\ 0 & 5 \end{pmatrix}\) has eigenvalues \(\lambda_1 = 2\) and \(\lambda_2 = 5\). What type of critical point does the origin have?

Question 2 of 5

For \(A = \begin{pmatrix} 0 & -5 \\ 5 & 0 \end{pmatrix}\), the eigenvalues are \(\lambda = \pm 5i\). What is the critical point classification?

Question 3 of 5

Given \(\tau = \text{tr}(A) = -4\) and \(\Delta = \det(A) = -5\), what type of critical point is the origin?

Question 4 of 5

If the eigenvalues of a system are \(\lambda = -2 \pm 3i\), the critical point is:

Question 5 of 5

Which condition on the trace-determinant plane separates nodes from spirals?