Physical review. E, Statistical physics, plasmas, fluids, and related interdisciplinary topics · 1993 · 28 citations · 3 references
Natural SciencesMathematicsPhysicsDynamical SystemsSystem DynamicsTime Series AnalysisContinuous Dynamical SystemsLyapunov ExponentsNumerical AlgorithmsNumerical AnalysisNumerical Methods For Differential EquationsNumerical StabilityBenettin AlgorithmFundamental MatrixLargest Lyapunov ExponentLyapunov Exponent ComputationLyapunov FunctionsLyapunov Stability AnalysisNumerical Stability AnalysisNumerical Stability In Dynamics
The Lyapunov exponent algorithm of Benettin et al. [C.R. Acad. Sci. Paris 286A, 431 (1978); Meccanica 15, 9 (1980)] and of Shimada and Nagashima [Prog. Theor. Phys. 61, 1605 (1979)] is converted to a continuous time algorithm. The Gram-Schmidt orthonormalization process is incorporated into the differential equations, making orthogonalization continuous. Differential equations for the k-dimensional fiducial volumes are also derived. It is further shown that this algorithm is a factorization of the fundamental matrix \ensuremath{\Phi} in the form \ensuremath{\Phi}(t)=E(t)W(t)${\mathit{E}}^{\mathrm{\ensuremath{-}}1}$(${\mathit{t}}_{0}$), where E(t) is orthonormal and W(t) is upper triangular. Numerical stability of the algorithm is considered, and it is shown that the standard Gram-Schmidt process can be used to stabilize the (possibly) unstable equations of motion for the orthonormal basis vectors. A numerical example is presented.
3
Deterministic Nonperiodic Flow
Edward N. Lorenz · Journal of the Atmospheric Sciences · 1963 · 19.3K citations · Full text