A quantitative rate that characterizes the average exponential divergence or convergence of infinitesimally close trajectories in a dynamical system, typically defined as the time-asymptotic limit of the growth rate of small perturbations measured by the logarithm of a tangent map or variational flow.

Principle

The Lyapunov exponent is the long-time average of the instantaneous exponential growth rates of infinitesimal perturbations along a trajectory; formally it is given by the limit of (1/t) log ||DΦ^t(v)|| for a tangent vector v under the linearized flow, when that limit exists.

Demonstration

Consider an iterated one-dimensional map on an interval where a trajectory x_n is perturbed by δx_n: the maximal Lyapunov exponent is computed from successive linearizations as the time-average of log |f'(x_n)|. For parameter values producing chaotic behavior this average is positive, indicating exponential separation of nearby initial conditions.

Misapplication

Interpreting a short-time transient increase of perturbations as the asymptotic Lyapunov exponent, or computing exponents using an inappropriate norm or insufficient integration time, can produce misleading results; similarly, applying tangent-linear formulas to nondifferentiable maps or using finite differences without convergence checks is incorrect.

Consequence

A positive maximal Lyapunov exponent implies sensitive dependence on initial conditions and limits long-term predictability; a negative exponent implies exponential convergence to an attractor or stable solution. The spectrum of Lyapunov exponents constrains invariant manifolds and fractal dimensions of attractors.

Reversal

Reversing the sign of a Lyapunov exponent corresponds to replacing divergence by convergence: a system with a positive exponent becomes contracting if parameters or dynamics are changed so that corresponding exponents become negative; finite-time exponents can change sign along trajectories even when asymptotic exponents do not.

Boundary

Lyapunov exponents are defined for systems with well-defined linearization along trajectories (smooth flows or differentiable maps) and for norms on tangent spaces; they may be ill-defined or require generalized definitions for nondifferentiable, discontinuous, or purely stochastic systems without a deterministic skeleton.

Semantic Tension

The maximal Lyapunov exponent is often conflated with measures of complexity such as metric entropy or with finite-time growth rates; unlike entropy, Lyapunov exponents quantify exponential rates of infinitesimal separation and require linearization, not only statistical unpredictability.

Synthesis

A Lyapunov exponent is the asymptotic exponential growth rate of infinitesimal perturbations along a trajectory; it organizes stability and predictability by translating local linearized stretching or contraction into global statements about divergence, attractors, and sensitivity to initial conditions.