A formal series approach to averaging: Exponentially small error estimates
Philippe Chartier Ander Murua Jesús María Sanz-Serna
The techniques, based on formal series and combinatorics, used nowadays to analyze numerical integrators may be applied to perform high-order averaging in oscillatory periodic or quasi-periodic dynamical systems. When this approach is employed, the averaged system may be written in terms of (i) scalar coefficients that are universal, i.e. independent of the system under consideration and (ii) basis functions that may be written in an explicit, systematic way in terms of the derivatives of the Fourier coefficients of the vector field being averaged. The coefficients may be recursively computed in a simple fashion. We show that this approach may be used to obtain exponentially small error estimates, as those first derived by Neishtadt. All the constants that feature in the estimates have a simple explicit expression.
keywords: Averaging exponentially small error estimates B-series high-order averaging quasi-stroboscopic averaging Lie algebras Lie groups. Magnus expansions Chen-Fliess series

Year of publication

Related Authors

Related Keywords

[Back to Top]