← Differential Equations

Unit 12: Nonlinear Systems and Phase-Plane Analysis

Trajectories flowing in the phase plane, equilibria read straight off eigenvalues, and what linearization can and can't tell you about a nonlinear system.

9 topics · 32 questions

Each topic is a path, not a menu: work down it in order and every question builds on the one above. Click any of them to watch the answer draw itself.

9.1 The Phase Plane

  1. How it worksHow does a system's velocity field assign an arrow to every point in the phase plane, sketching out trajectories?
  2. How it worksHow do you tell a system's equilibrium points apart from everywhere else just by looking at the phase plane?
  3. Going furtherWhat happens to the phase portrait's arrows near a point where the velocity field is exactly zero?
  4. Worked exampleFind the equilibrium points of the system x' = y, y' = -x + x3 by setting both rates to zero.
Your turn1 of 2
Find the nonzero equilibrium x-value(s) of the system x' = y, y' = -7x + x3 (other than the origin), by setting both rates to zero.

9.1 Classifying Equilibria from Eigenvalues

  1. How it worksHow do the sign and type of a linear system's eigenvalues sort every equilibrium into a node, saddle, spiral, or center?
  2. Worked exampleFind the equilibrium type at the origin of x' = Ax for A = [[-2, 1], [1, -2]] using its eigenvalues.
  3. Going furtherWhat happens to an equilibrium's classification as a system's eigenvalues are continuously varied from negative real parts to positive?
  4. Ties togetherHow does classifying an equilibrium by its eigenvalues here connect to solving the same system's equations explicitly?
Your turn1 of 2
For A = [[-2, 2.12], [2.12, -2]] (symmetric, off-diagonal entries equal), find the two eigenvalues, a+b and a-b. Report the larger one.

9.1 Nodes

  1. How it worksHow do two real eigenvalues of the same sign force every nearby trajectory toward or away from a node?
  2. Worked exampleSketch the phase portrait near the origin for x' = Ax with A = [[-1, 0], [0, -4]], a node.
  3. Common mistakeWhy don't trajectories near a node cross the eigenvector lines even though they curve to approach them?
  4. Going furtherWhat happens to a node's trajectories when the two real eigenvalues are equal instead of merely the same sign?
Your turn1 of 2
For the diagonal system x' = Ax with A = [[-3, 0], [0, -3]] (a node, both eigenvalues the same sign), find the ratio of eigenvalues d/a, which controls how sharply trajectories bend toward the slower eigendirection.

9.1 Saddle Points

  1. How it worksHow do one positive and one negative eigenvalue create the incoming and outgoing directions that define a saddle?
  2. Worked exampleSketch the phase portrait near the origin for x' = Ax with A = [[2, 0], [0, -3]], a saddle point.
  3. Common mistakeWhy isn't a saddle point considered stable even though one special direction leads straight into it?
Your turn1 of 2
For the diagonal system x' = Ax with A = [[4, 0], [0, -2]] where a > 0 and d < 0 (a saddle point), find the product of the eigenvalues, a·d (always negative for a saddle).

9.1 Spirals and Centers

  1. How it worksHow does the sign of the real part of a complex eigenvalue pair decide whether trajectories spiral in, spiral out, or form closed loops?
  2. Worked exampleFind the equilibrium type of x' = Ax for A = [[-1, 2], [-2, -1]] using its complex eigenvalues.
  3. Going furtherWhat happens to a spiral equilibrium's trajectories as the real part of its eigenvalues shrinks toward exactly zero?
Your turn1 of 2
For A = [[-2, 2], [-2, -2]], find the modulus |lambda| = √(p2+q2) of its complex eigenvalues p plus or minus qi.

9.1 Why a Center Is the Fragile, Borderline Case

  1. Common mistakeWhy doesn't linearizing a nonlinear system near an equilibrium reliably tell you whether that equilibrium is truly a center?
  2. Going furtherWhat happens to a linear system's closed orbits around a center the instant a tiny bit of damping is added?
  3. Ties togetherHow does a center's fragility here connect back to why undamped free vibration sits at the boundary of the three damping cases?
Your turn1 of 2
A linear system has a center with pure imaginary eigenvalues plus or minus 4i. A tiny damping term is added, shifting the eigenvalues to -0.224 plus or minus 4i. Find the new real part's magnitude, epsilon = 0.224, as a fraction of the frequency w: epsilon/w.

9.2 Linearizing a Nonlinear System Near an Equilibrium

  1. How it worksHow does the Jacobian matrix of partial derivatives at an equilibrium build the best linear approximation to a nonlinear system there?
  2. Worked exampleFind the Jacobian and classify the equilibrium at (0,0) for the nonlinear system x' = x - x2 - xy, y' = 2y - xy - y2.
  3. Going furtherWhat happens to the accuracy of a linearization as you move farther away from the equilibrium point it was built at?
Your turn1 of 2
For the nonlinear system x' = 3x - x2 - xy, y' = 3y - xy - y2, find the Jacobian's trace at the equilibrium (0,0): a+b (since the Jacobian there is [[a,0],[0,b]]).

9.5 The Predator-Prey System

  1. How it worksHow do the Lotka-Volterra equations turn a predator's need to eat and a prey's need to reproduce into a coupled system?
  2. Worked exampleFind the nonzero equilibrium of the predator-prey system dx/dt = x(3 - y), dy/dt = -y(2 - x).
  3. Worked exampleFind the linearization of the predator-prey system dx/dt = x(3 - y), dy/dt = -y(2 - x) at its nonzero equilibrium.
  4. Ties togetherHow does the predator-prey system's closed orbits around its equilibrium connect back to the center case in linear systems?
Your turn1 of 2
Find the sum of the coordinates (x+y) at the nonzero equilibrium of the predator-prey system dx/dt = x(4 - y), dy/dt = -y(4 - x).

9.4 The Competing Species System

  1. How it worksHow do the nullclines of a competing species system carve the phase plane into regions with different long-term outcomes?
  2. Worked exampleFind all four equilibria of the competing species system dx/dt = x(1 - x - y), dy/dt = y(0.75 - y - 0.5x).
  3. Going furtherWhat happens to the competing species system's long-term outcome when the competition coefficients are lowered enough?
  4. Common mistakeWhy doesn't a competing species system always end with one species driving the other completely extinct?
Your turn1 of 2
For the competing species system dx/dt = x(1 - x - y), dy/dt = y(0.779 - y - 0.56x) with c not equal to 1, find the x-coordinate of the coexistence equilibrium: x = (K-1)/(c-1).