Skip to main content
Log in

Variable steps for reversible integration methods

Variable Schrittweiten für reversible Integrationsverfahren

  • Published:
Computing Aims and scope Submit manuscript

Abstract

Conventional variable-step implementation of symplectic or reversible integration methods destroy the symplectic or reversible structure of the system. We show that to preserve the symplectic structure of a method the step size has to be kept almost constant. For reversible methods variable steps are possible but the step size has to be equal for “reflected” steps. We demonstrate possible ways to construct reversible variable step size methods. Numerical experiments show that for the Kepler problem the new methods perform better than conventional variable step size methods or symplectic constant step size methods. In particular they exhibit linear growth of the global error (as symplectic methods with constant step size).

Zusammenfassung

Werden symplektische oder reversible Integrationsverfahren mit herkömmlichen Schrittweitensteuerungen verwendet, so geht die symplektische, bzw. die reversible Struktur des Problems verloren. In dieser Arbeit wird gezeigt, dass die symplektische Struktur des Verfahrens nur dann erhalten wird, wenn die Schrittweite fast konstant bleibt. Für reversible Verfahren sind echt variable Schrittweiten möglich, die Schrittweite muss jedoch für “gespiegelte” Schritte gleich sein. Es werden verschiedene Wege aufgezeigt, um reversible, variable Schrittweiten zu konstruieren. Numerische Experimente zeigen, dass für das Keplerproblem die neuen Methoden den herkömmlichen Schrittweitensteuerungen oder den symplektischen Verfahren mit konstanter Schrittweite überlegen sind. Insbesondere wächst der globale Fehler linear.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Similar content being viewed by others

References

  1. Calvo, M. P., Sanz-Serna, J. M.: Variable steps for symplectic integrators. In: Numerical analysis (Griffiths, D. F., Watson, G. A., eds.), pp. 34–38.

  2. Channel, P. J., Scovel, C.: Symplectic integration of Hamiltonian systems. Nonlinearity3, 231–259 (1990).

    Google Scholar 

  3. Hairer, E.: letter from 2/11/1988.

  4. Hairer, E., Nørsett, S., Wanner, G.: Solving ordinary differential equations I. Nonstiff Problems. London: Longman 1992. Berlin, Heidelberg, New York, Tokyo: Springer 1993.

    Google Scholar 

  5. Kirchgraber, U.: Multistep methods are essentially one-step methods. Numer. Math.48, 85–90 (1986).

    Google Scholar 

  6. Lasagni, F.: Canonical Runge-Kutta methods. ZAMP39, 952–953 (1988).

    Google Scholar 

  7. Lasagni, F.: Integration methods for Hamiltonian differential equations. Unpublished manuscript (1990).

  8. McLachlan, R. I., Scovel, C.: A survey of open problems in symplectic integration. Preprint (1993).

  9. Moser, J.: Stable and random motion in dynamical systems. New Jersey: Princeton University Press 1973.

    Google Scholar 

  10. Okunbor, D., Skeel, R. D.: An explicit Runge-Kutta-Nyström method is canonical if and only if its adjoint is explicit. SIAM J. Number. Anal.29, 521–527 (1992).

    Google Scholar 

  11. Sanz-Serna, J. M.: Runge-Kutta schemes for Hamiltonian systems. BIT28, 877–883 (1988).

    Google Scholar 

  12. Sanz-Serna, J. M.: Symplectic integrators for Hamiltonian problems: an overview. Acta Numer. 243-286 (1992).

  13. Sanz-Serna, J. M., Calvo, M. P.: Numerical Hamiltonian problems. London: Chapman & Hall 1994.

    Google Scholar 

  14. Sevryuk, M. B.: Reversible systems. Berlin, Heidelberg, New York, Tokyo: Springer 1986 (Lecture Notes in Mathematics, Vol. 1211).

    Google Scholar 

  15. Stoffer, D.: On reversible and canonical integration methods. Research Report 88-05, Seminar für angewandte Mathemaktik, ETH Zürich (1988).

  16. Stoffer, D.: General linear methods: connection to one-step methods and invariant curves. Numer. Math.64, 395–407 (1993).

    Google Scholar 

  17. Stoffer, D.: On the global error of linear multistep methods. In: Geometric behaviour of numerical integration methods, submitted as Habilitationsschrift, ETH-Zürich, 1994.

  18. Stoffer, D., Nipp, K.: Invariant curves for variable step size integrators. BIT31, 169–180 (1991).

    Google Scholar 

  19. Suris, Y. B.: Canonical transformations generated by methods of Runge-Kutta type for the numerical integration of the systemx″=−∂U/∂x. Zh. Vychisl. Mat. i Mat. Fiz.29, 202–211 (1989) (in Russian).

    Google Scholar 

  20. Tang, Y.-F.: The necessary condition for a Runge-Kutta scheme to be symplectic for Hamiltonian systems. Comp. Math. Appl.26, 13–20 (1993).

    Google Scholar 

  21. Yoshida, M.: Recent progress in the theory and application of symplectic integrators. Celestial Mech. Dynam. Astronom.56, 27–43 (1993).

    Google Scholar 

Download references

Author information

Authors and Affiliations

Authors

Rights and permissions

Reprints and permissions

About this article

Cite this article

Stoffer, D. Variable steps for reversible integration methods. Computing 55, 1–22 (1995). https://doi.org/10.1007/BF02238234

Download citation

  • Received:

  • Revised:

  • Issue Date:

  • DOI: https://doi.org/10.1007/BF02238234

AMS Subject Classifications

Key words

Navigation