NETMaths
The bookUnit 3 · Ordinary Differential Equations57 / 83

Stability and phase portraits

Why this is asked: Classify a 2×2 linear system from trace and determinant alone; for nonlinear systems linearise and use Hartman–Grobman — but remember the centre case is the one linearisation cannot decide.

Classification of ẋ = Ax in the plane

Let T = trace A,D=detA,Δ=T24DA, D = \det A, \Delta = T^{2} - 4D.

Condition Type Stability
D < 0 saddle unstable
D>0,Δ>0,T<0D > 0, \Delta > 0, T < 0 stable node asymptotically stable
D>0,Δ>0,T>0D > 0, \Delta > 0, T > 0 unstable node unstable
D>0,Δ<0,T<0D > 0, \Delta < 0, T < 0 stable spiral asymptotically stable
D>0,Δ<0,T>0D > 0, \Delta < 0, T > 0 unstable spiral unstable
D > 0, T = 0 centre stable, not asymptotically
D = 0 degenerate (a line of equilibria)

Summary: asymptotically stable ⇔ T < 0 and D > 0 ⇔ both eigenvalues have negative real part.

Nonlinear systems

Linearise at an equilibrium: J = Df(x*).

  • Hartman–Grobman: if no eigenvalue of J has zero real part (a hyperbolic equilibrium), the nonlinear phase portrait is locally equivalent to the linear one.
  • The centre case is undecidable by linearisation: ẋ =yx3= -y - x^{3}, ẏ =xy3= x - y^{3} has a linear centre but is actually asymptotically stable, while flipping the cubic signs makes it unstable.

Lyapunov's direct method

Find V with V(x*) = 0, V > 0 nearby:

  • V̇ ≤ 0 ⇒ stable; V̇ < 0 ⇒ asymptotically stable; V̇ > 0 ⇒ unstable.

Works where linearisation fails. For a conservative system, the energy is the natural V.

Key takeaways

  • T < 0 and D > 0 ⇔ asymptotically stable; D < 0 ⇔ saddle.
  • Hyperbolic equilibria are settled by linearisation; centres are not.
  • Lyapunov functions handle the non-hyperbolic cases.

See it move

The trace–determinant planeinteractive

Drag a point across the plane and watch the portrait change: saddle, node, spiral, centre — a map instead of a list of rules.

Unlock interactive visuals

The trap here

“Linearisation determines stability at every equilibrium” — false

=yx3= -y - x^{3}, ẏ =xy3= x - y^{3} at the origin

The linearisation is a centre (eigenvalues ±i), but V=x2+y2V = x^{2} + y^{2} gives V̇ =2(x4+y4)<0= -2(x^{4} + y^{4}) < 0: asymptotically stable. Non-hyperbolic equilibria escape Hartman–Grobman.

More on this →

Check yourself

For the system ẋ = Ax with A = [[−2, 1], [1, −2]], the origin is

Next: First-order PDE: Lagrange, Charpit, characteristics

Create a free account to keep your place and have this feed your study plan.

Open this in the full syllabus view · Unit 3