Stability

Introduction

Stability in systems of ordinary differential equations (ODEs) concerns the behavior of solutions near equilibrium points. It determines whether small perturbations grow, decay, or remain bounded over time. Stability analysis is crucial for understanding long-term system behavior in physics, engineering, biology, and economics.

"Stability is the cornerstone of understanding dynamical systems and predicting their future states." -- Stephen Wiggins

Equilibrium Points

Definition

Equilibrium points (or fixed points) satisfy f(x_0) = 0 for system dx/dt = f(x). Solutions starting at equilibrium remain constant.

Classification

Equilibria can be isolated or form sets. Common types: nodes, saddles, foci, centers, depending on linearization eigenvalues.

Existence

Existence depends on system properties, continuity, and domain constraints. Analytic or numerical methods locate equilibria.

Types of Stability

Stability in the Sense of Lyapunov

Equilibrium x_0 is stable if for every ε>0, there exists δ>0 so that |x(0)-x_0|<δ implies |x(t)-x_0|<ε for all t≥0.

Asymptotic Stability

Stable and solutions converge to x_0 as t → ∞. Stronger than Lyapunov stability.

Instability

Equilibrium is unstable if it is not stable. Small perturbations lead solutions away from x_0.

Linearization Method

Concept

Approximate nonlinear system near equilibrium by linear system using Jacobian matrix at x_0.

Procedure

Compute Jacobian J = Df(x_0), solve linear system dy/dt = Jy.

Limitations

Linearization predicts local stability if eigenvalues have nonzero real parts. Fails for center or degenerate cases.

dx/dt = f(x), f: ℝⁿ → ℝⁿx₀ equilibrium, f(x₀) = 0J = Df(x₀) = matrix of partial derivatives at x₀Linearized: dy/dt = JyAnalyze eigenvalues of J

Lyapunov Stability Theory

Lyapunov Functions

Scalar function V(x), positive definite near equilibrium, with negative definite derivative along trajectories, implies stability.

Lyapunov's Direct Method

Does not require solution of ODE. Construct V(x) to infer stability properties.

LaSalle's Invariance Principle

Extends Lyapunov theory to cases where V̇ ≤ 0. Identifies invariant sets for asymptotic stability.

Jacobian Matrix and Eigenvalues

Definition

Jacobian matrix J = (∂f_i/∂x_j) evaluated at equilibrium provides linear approximation.

Eigenvalues and Stability

Real parts of eigenvalues decide stability: all negative → asymptotically stable; any positive → unstable.

Complex Eigenvalues

Complex conjugate pairs with negative real parts imply spiral sink; positive real parts imply spiral source.

J = [ ∂f₁/∂x₁ ∂f₁/∂x₂ ... ∂f₁/∂xₙ ][ ∂f₂/∂x₁ ∂f₂/∂x₂ ... ∂f₂/∂xₙ ][ ... ... ... ][ ∂fₙ/∂x₁ ∂fₙ/∂x₂ ... ∂fₙ/∂xₙ ]Eigenvalues λ₁, λ₂, ..., λₙStability: Re(λ_i) < 0 ∀ i → stable ∃ i: Re(λ_i) > 0 → unstable

Phase Portraits and Stability

Visual Representation

Phase portraits show trajectories in state space, illustrating stability behavior around equilibria.

Attractors and Repellors

Stable equilibria act as attractors; unstable ones as repellors; neutral centers form closed orbits.

Examples

Node: all trajectories approach or diverge along eigenvectors; saddle: trajectories approach in some directions and diverge in others.

Asymptotic Stability

Definition

Equilibrium is asymptotically stable if stable and limₜ→∞ x(t) = x₀ for trajectories starting sufficiently close.

Criteria

Linear system: eigenvalues of Jacobian have strictly negative real parts. Nonlinear: Lyapunov function with negative definite derivative.

Physical Interpretation

System returns to equilibrium after small disturbance, dissipates energy or perturbation over time.

Instability

Definition

Equilibrium is unstable if not stable: arbitrarily small perturbations lead to solutions diverging from equilibrium.

Indicators

Eigenvalue with positive real part in linearization. Lyapunov function cannot be found with required properties.

Consequences

System exhibits divergence, oscillations with growing amplitude, or chaotic behavior near equilibrium.

Limit Cycles and Stability

Limit Cycle Definition

Closed isolated trajectory in phase space. Solutions nearby approach or diverge from it as time progresses.

Stability of Limit Cycles

Stable limit cycle: attracts neighboring trajectories. Unstable: repels. Semi-stable: attracts from one side only.

Methods of Analysis

Poincaré maps, Floquet theory, and perturbation methods assess limit cycle stability.

Example: Van der Pol oscillatorEquation: d²x/dt² - μ(1 - x²) dx/dt + x = 0Limit cycle exists for μ > 0Stability determined via Floquet multipliers

Stability Criteria

Routh-Hurwitz Criterion

Algebraic test for all roots of characteristic polynomial to have negative real parts.

Lyapunov's Direct Method

Construction of positive definite functions with negative definite derivatives.

Hartman-Grobman Theorem

Local topological equivalence of nonlinear system near hyperbolic equilibrium to its linearization.

Examples and Applications

Linear System Stability

System: dx/dt = Ax, where eigenvalues of A determine stability. Example: 2D system with eigenvalues -1 and -2 is asymptotically stable.

Nonlinear Pendulum

Equilibrium at rest stable; inverted pendulum unstable. Stability analyzed by linearization and energy methods.

Population Models

Equilibria correspond to steady states. Stability determines persistence or extinction.

Example: Logistic equationdx/dt = r x (1 - x/K)Equilibria at x=0 (unstable), x=K (stable if r>0)

Control Systems

Stability crucial for feedback design. Pole placement and Lyapunov methods used for analysis and synthesis.

Mechanical Systems

Stability of equilibria corresponds to stable configurations, vibration analysis involves eigenvalue computation.

References

• H. K. Khalil, Nonlinear Systems, 3rd ed., Prentice Hall, 2002, pp. 259-320.
• S. H. Strogatz, Nonlinear Dynamics and Chaos, Westview Press, 2015, pp. 94-150.
• J. Hale, Ordinary Differential Equations, 2nd ed., Wiley-Interscience, 1980, pp. 120-175.
• L. Perko, Differential Equations and Dynamical Systems, 3rd ed., Springer, 2001, pp. 100-160.
• V. I. Arnold, Geometrical Methods in the Theory of Ordinary Differential Equations, Springer, 1988, pp. 45-90.