Supreme Local Lyapunov Exponent (SLLE)

• SLLE is a rigorous indicator defined as the supremum of local finite-time Lyapunov exponents, capturing the worst-case expansion rate in chaotic dynamics.
• It provides a local, instantaneous metric that refines global averaged stability measures by quantifying transient instabilities in dynamical systems.
• SLLE is applied in synchronization, compressive sensing, and turbulence to guarantee stability, ensure reconstructability, and bound energy transfer rates.

The Supreme Local Lyapunov Exponent (SLLE) is a rigorous, trajectory-independent indicator characterizing the worst-case finite-time expansion rate of infinitesimal perturbations in deterministic dynamical systems. Defined as the supremum of local (finite-time) Lyapunov exponents over all points of an attractor or relevant region in phase space, the SLLE quantifies the extreme local instability that may arise, providing a critical refinement over global (infinite-time averaged) Lyapunov exponents for applications requiring guarantees of absence of bursts, desynchronization, or reconstructability in chaotic and turbulent systems (Chen et al., 2012, Chen et al., 2013, Divitiis, 2017, Waldner et al., 2010).

1. Definition and Mathematical Framework

Let $ẋ = f(x)$ describe an $n$-dimensional smooth autonomous system with flow $x(t)$. Linearizing around a trajectory yields the variational equation $\delta ẋ = Df(x(t)) \delta x$, with fundamental solution matrix $H(x₀,Δt)$ over interval $[0,Δt]$. The local Lyapunov exponent (LLE) in the $i$th direction at $x₀$ is

$$λ_{\mathrm{loc},i}(x₀,Δt) = \frac{1}{Δt} \ln σ_i(H(x₀,Δt)),$$

where $σ_i$ is the $n$0th singular value of $n$1. For a compact, invariant set $n$2 (e.g., a strange attractor), the Supreme Local Lyapunov Exponent is

$n$3

The largest SLLE is of principal interest and can be written as

$n$4

In turbulence, the SLLE is defined via the supremum of the finite-scale instantaneous separation-growth rate over all initial orientations (Divitiis, 2017, Waldner et al., 2010).

2. Key Properties and Theoretical Significance

• Trajectory Independence: Since the supremum is taken over the entire attractor (or support), the SLLE is independent of the particular reference trajectory, characterizing the maximal instantaneous expansion possible anywhere in the accessible phase space (Chen et al., 2012).
• Finite-Time, Worst-Case Instability: While classical global Lyapunov exponents quantify average asymptotic rates, the SLLE is a finite-time, local, and extremal metric—capturing the strongest expansion rate in a prescribed time window (Chen et al., 2012, Chen et al., 2013).
• Bounding Relationship with Global Exponents: Oseledec’s multiplicative ergodic theorem yields $n$5 for almost all $n$6. Since $n$7 is the supremum over $n$8, $n$9, and typically $x(t)$0 from above as $x(t)$1 (Chen et al., 2012).

3. Numerical Computation Methods

• Sampling Along Trajectories: Generate a long reference trajectory on the attractor. Slide a window of length $x(t)$2, compute the LLE at each window position using standard QR or SVD integrators of the variational equation, and take the maximum value obtained as an estimate of $x(t)$3. As the window shift $x(t)$4 and the total trajectory length increases, this estimate converges from below to the true supremum (Chen et al., 2012, Chen et al., 2013).
• Direct Ellipsoidal Methods: For instantaneous (“local in time”) rates, compute the symmetric part $x(t)$5, where $x(t)$6 is the Jacobian. The largest eigenvalue of $x(t)$7 at each $x(t)$8 yields the pointwise SLLE, i.e., the strongest local separation rate (Waldner et al., 2010).

4. Applications in Synchronization, Sensing, and Turbulence

• Impulsive Synchronization: In impulsively coupled chaotic systems, establishing global exponential stability (no bursts) of the synchronization manifold requires the largest SLLE of the error dynamics over one impulse period $x(t)$9 to be strictly negative. Specifically, $\delta ẋ = Df(x(t)) \delta x$0 is a necessary and sufficient condition for forever-burst-free synchronization, permitting impulsive intervals orders of magnitude larger than classical Lyapunov conditions while guaranteeing stability (Chen et al., 2012).
• Chaotic Analog-to-Information Conversion: For nonlinear compressive sensing architectures using chaotic modulation, reconstructability of sparse signals from sub-Nyquist samples is guaranteed if the largest SLLE of the error system is negative. This ensures exponential convergence of synchronization error, and thus correct parameter recovery in inverse problems (Chen et al., 2013).
• Turbulence and Energy Cascade: In homogeneous isotropic turbulence, the SLLE is identified as the maximum possible instantaneous growth rate for particle separations. Entropy-maximization under the “fully developed chaos” hypothesis yields a uniform distribution for finite-scale local exponents, relating the supremum $\delta ẋ = Df(x(t)) \delta x$1 directly to moments of the increment statistics and providing a closure coefficient in the von Kármán–Howarth and Corrsin equations for the energy cascade (Divitiis, 2017).

5. Analytical Results and Example Systems

• Lorenz System: For specific parameters (e.g., $\delta ẋ = Df(x(t)) \delta x$2, $\delta ẋ = Df(x(t)) \delta x$3, $\delta ẋ = Df(x(t)) \delta x$4), the minimal period $\delta ẋ = Df(x(t)) \delta x$5 sets the reference interval for SLLE calculation. While the classical largest conditional Lyapunov exponent becomes negative for $\delta ẋ = Df(x(t)) \delta x$6, computationally the SLLE criterion reveals true absence of bursts only for $\delta ẋ = Df(x(t)) \delta x$7 with $\delta ẋ = Df(x(t)) \delta x$8. Increasing the interval ($\delta ẋ = Df(x(t)) \delta x$9) relaxes this bound toward the CLE-based threshold (Chen et al., 2012, Chen et al., 2013).
• Liu System: Analogous analysis shows the negative SLLE criterion matching observed reconstruction intervals and confirming the tightness of the bound in practical chaotic modulation applications (Chen et al., 2013).
• Turbulent Flow: In turbulent flows, the SLLE $H(x₀,Δt)$0 at scale $H(x₀,Δt)$1 is given by

$H(x₀,Δt)$2

where $H(x₀,Δt)$3 is the turbulent kinetic energy and $H(x₀,Δt)$4 is the longitudinal velocity correlation. The mean exponent is $H(x₀,Δt)$5, with the SLLE providing the coefficient for the rate of energy transfer (Divitiis, 2017).

6. Comparisons to Other Lyapunov Exponent Notions

• Oseledec and RK Exponents: Oseledec local exponents conflate divergence and acceleration, producing strong fluctuations and requiring time averaging for interpretability. Rateitschak–Klages (RK) exponents, using a constrained frame orthogonal to the flow, improve this but still involve finite recovery times and integration. The SLLE, especially via the symmetric ellipsoid method, is frame-independent, zero-recovery, and captures local, instantaneous sources of divergence directly (Waldner et al., 2010).
• Local versus Supreme Exponents: The SLLE is strictly an upper bound for any local or global Lyapunov exponent on the attractor, capturing the most unstable direction and time, making it uniquely suitable for worst-case analyses relevant in engineering and physical systems.

7. Interpretational Remarks and Generalizations

The SLLE concept provides a unified tool for both low-dimensional chaos and high-dimensional turbulence, encapsulating the worst-case local instability and acting as a sharp, computable criterion for stability, reconstructability, and predictability. Its role as the “key coefficient” for the closure of energy transfer equations in turbulence and as a stability criterion in synchronization theory cements its theoretical and practical relevance. The SLLE construction, grounded in supremal analysis over attractors or phase-space regions, underpins its robustness across disparate nonlinear phenomena (Chen et al., 2012, Chen et al., 2013, Divitiis, 2017, Waldner et al., 2010).