Lecture 8 - Thu 2020 09 24 - Virtual % ============================================================================== Continue prior lecture: Flows in the circle and bifurcations. % % ============================================================================== Start with Phase Plane -1- Introduce system. x_t = f(x, y) and y_t = g(x, y). -2- Orbits/trajectories: curves in the plane. -3- Assume f and g smooth. Therefore trajectories do not cross! -4- Want to classify possible phase plane portraits. Strong restriction imposed by 3 and plane geometry: Jordan curve theorem leads to self-trapping of trajectories. Makes classification possible. -5- Critical points; linearization ... delY_t = A delY [A 2x2 matrix and delY = [delx, dely] vector of perturbations to C.P.] Classification: Write characteristic polynomial in terms of: \Delta = determinant = \lambda_1 * \lambda_2 \tau = trace = \lambda_1 + \lambda_2 0 = det(A-\lambda) = \lambda^2 - tau*\lambda + \Delta = (\lambda - \tau/2)^2 + \Delta - \tau^2/4 Draw regions in the \Delta-\tau plane [\tau vertical abscissa]. Describe: saddle, nodes, spirals + improper nodes = star + degenerate nodes + centers. Saddles: contradiction? Trajectories should not cross, but they seem to. Uh? Stable and unstable nodes AND stable and unstable spirals. Transition set-ups [not structurally stable]. Do examples. Star: double eigenvalue with two eigenvectors. Degenerate node: double eigenvalue with single eigenvector. Center. Note that going across a transition is a bifurcation! Attracting and repelling critical points. Point out "degeneracy" when linear theory fails for \Delta = 0. Meta-theorem. When is linear picture not structurally stable. Do pictures. STABILITY JARGON: Lyapunov stable; Attracting; AND Asymptotically stable. Example of attracting but not Lyapunov stable. % % ============================================================================== EOF