Nonlinear Dynamics: Fixed Points and Stability in Flows

The Poincaré-Bendixson Theorem states that for a nonlinear autonomous system dx/dt = p(x,y), dy/dt = q(x,y) with continuous first-order partial derivatives in a domain D, if there exists a bounded subdomain R of D containing no critical points and a half-path C+ entirely within R, then either C+ is itself a closed path or its limit set L(C+) is a closed path that C+ approaches spirally from inside or outside, guaranteeing the existence of at least one periodic orbit (limit cycle) within R.

Limit cycles are isolated closed trajectories in planar dynamical systems. The Poincaré-Bendixson theorem states that for a trajectory entering a bounded region with finitely many singular points, the omega-limit set is either a single equilibrium point or a closed cycle. Bendixson's criterion provides a necessary condition for the existence of closed trajectories: if the divergence of the vector field maintains constant sign throughout a simply connected region, no closed trajectory can exist. The proof uses Green's theorem to show that the integral of the divergence along a closed trajectory must be zero, contradicting the constant sign assumption. The Poincaré map (or monodromy map) is a fundamental tool for analyzing trajectories near closed orbits. It maps points on a transversal curve to their images after one complete revolution around the orbit. The fixed points of this map correspond to the closed orbits themselves. The map is monotonic, meaning it cannot increase then decrease or vice versa, which follows from the fact that trajectories of smooth vector fields cannot intersect. The behavior of trajectories near a limit cycle is classified by the Poincaré map: (1) Stable limit cycle when the map is monotonic decreasing and crosses the diagonal transversally, with trajectories approaching from both inside and outside; (2) Unstable limit cycle when the map is monotonic increasing and crosses the diagonal transversally, with trajectories moving away from both sides; (3) Semi-stable limit cycle when the map is tangent to the diagonal, with trajectories approaching from one side and moving away from the other. Semi-stable cycles are non-robust and disappear under small perturbations, while stable and unstable cycles are robust.

For vector fields with finite equilibrium points on the plane or sphere, the Poincaré-Bendixson theorem classifies omega limit sets into three types: equilibrium points, periodic orbits (limit cycles), or separatrix polygons (trajectories connecting equilibrium points). A limit cycle is an isolated periodic orbit that attracts or repels nearby trajectories. A separatrix polygon consists of heteroclinic and homoclinic orbits connecting equilibrium points. This theorem provides a complete classification of long-term behaviors for planar systems.

An omega limit point is a point a such that there exists a sequence tn → +∞ with x(tn) → a. The omega limit set is the collection of all such points. If a trajectory is bounded and its omega limit set is contained in the domain, then the omega limit set is non-empty, bounded, closed, connected, and consists of entire trajectories. In dimension N = 2, if the omega limit set is bounded, non-empty, and contains no equilibrium points, then it is a single closed trajectory (Poincaré-Bendixson theorem). In higher dimensions, omega limit sets can be more complex, such as a sphere, not a single trajectory.

This video explains analytical methods to prove the existence or nonexistence of limit cycles in two-dimensional dynamical systems. To rule out closed orbits, one can use index theory (which requires that any closed orbit must encircle fixed points whose indices sum to +1), recognize gradient systems (which cannot have periodic orbits), construct Lyapunov functions (continuously differentiable scalar functions that are positive definite and have negative derivative along trajectories), or apply Dulac's criterion (which uses a scalar function g such that the divergence of g times the vector field has one sign throughout a simply connected region). To establish the existence of a limit cycle, the Poincaré-Bendixson theorem states that if a closed bounded region contains no fixed points and contains a trajectory that remains within it forever, then that region must contain a closed orbit or a trajectory spiraling toward one; this is typically proven by constructing a trapping region where the vector field points inward along all boundaries.

Global asymptotic stability requires that all trajectories starting from any initial condition converge to the equilibrium. A Lyapunov function V(x) is a positive definite function (V(x) > 0 for x ≠ equilibrium, V(equilibrium) = 0) whose time derivative is negative definite (dV/dt < 0 for x ≠ equilibrium). For the logistic equation dx/dt = x(1-x), the function V = x - 1 - ln(x) serves as a Lyapunov function, proving that x* = 1 is globally asymptotically stable for x > 0.

Lyapunov stability theory determines equilibrium point stability using energy-like functions called Lyapunov functions. For a nonlinear system ẋ = f(x), the origin is an equilibrium point. Alexander M. Lyapunov proposed this method to assess stability without solving system trajectories. The approach uses generalized energy functions to determine whether an equilibrium is stable, asymptotically stable, or unstable based on how the function changes over time.

Lyapunov stability theory provides a method to determine the stability of nonlinear dynamical systems by constructing a scalar function V(x) (called a Lyapunov function) that is positive definite and has a negative semi-definite time derivative along system trajectories; if V̇(x) < 0 for all x ≠ 0, the equilibrium is asymptotically stable, while V̇(x) ≤ 0 with V̇(x) = 0 only at the equilibrium point also guarantees asymptotic stability, offering an alternative to eigenvalue analysis for proving system stability.

The Lyapunov stability theorem provides mathematical conditions for determining the stability of equilibrium points in nonlinear dynamical systems using a scalar function called a Lyapunov function. For an equilibrium point x*, if there exists a positive definite function v(x) (with v(x*)=0 and v(x)>0 elsewhere) whose time derivative v_dot(x) is negative semi-definite (≤0), then x* is stable; if v_dot(x) is negative definite (<0), then x* is locally asymptotically stable with a region of attraction defined by the domain of v(x); and if the domain is the entire state space and v(x) approaches infinity as ||x|| increases, then x* is globally asymptotically stable.

Lyapunov's theorem states that for an equilibrium point at the origin of a dynamical system x' = f(x), if there exists a continuously differentiable positive definite function V(x) (called a Lyapunov function) such that its time derivative V̇(x) is negative semi-definite (≤ 0), then the equilibrium is stable; if V̇(x) is negative definite (< 0 for all x ≠ 0), then the equilibrium is asymptotically stable. The proof involves constructing a level set Ω_β contained within a ball B_ε around the origin, showing this set is positively invariant, and using the continuity of V at the origin to establish the existence of a ball B_Δ such that trajectories starting inside remain within B_ε. For asymptotic stability, assuming the limit of V(x(t)) is positive leads to a contradiction by integrating V̇ over a compact set where it achieves a negative minimum value.

Strange attractors are mathematical objects combining two properties: they attract nearby orbits (stability) while exhibiting chaotic motion (aperiodicity and sensitive dependence). The Henon map (a 2D discrete mapping with parameters A and B) demonstrates this through its fractal structure when plotted in phase space. The Lorenz equations (three differential equations with parameters sigma, rho, beta) similarly produce strange attractors when solved numerically. Both systems show orbits that cycle periodically but with seemingly random amplitude variations, revealing how simple deterministic rules can generate complex, structured attractors.

This lecture segment introduces chaos theory through the Lorenz system, demonstrating how small changes in initial conditions lead to vastly different outcomes. The instructor shows how a cubic volume of initial conditions contracts onto a strange attractor with fractional dimension, illustrating the hallmark properties of chaotic systems including sensitivity to initial conditions and the stretching/folding mechanism that characterizes chaotic dynamics.

Attractors are sets of states toward which systems evolve regardless of starting conditions. Point attractors represent equilibrium (system settles to rest), limit cycles represent periodic behavior (system repeats in loops), and strange attractors represent chaotic behavior (system moves within bounded regions without repeating). Strange attractors are fractal structures with fractional dimensions (Lorenz attractor ~2.06), existing between integer dimensions. They never intersect themselves because that would imply a single point having two different futures, which is impossible in deterministic systems.

The Lorenz system (three coupled ODEs with two nonlinear terms) and Rössler system (simpler with one nonlinear term) demonstrate chaos through parameter-dependent behavior. With appropriate parameters, all initial conditions are attracted to strange attractors—fractal objects of dimension less than the full phase space. These systems show how deterministic equations produce complex, unpredictable behavior, with homoclinic orbits causing the instability that leads to chaos.