research-article

Symplectic Runge--Kutta Schemes for Adjoint Equations, Automatic Differentiation, Optimal Control, and More

Author:

J. M. Sanz-SernaAuthors Info & Claims

SIAM Review, Volume 58, Issue 1

Pages 3 - 33

https://doi.org/10.1137/151002769

Published: 01 January 2016 Publication History

Abstract

The study of the sensitivity of the solution of a system of differential equations with respect to changes in the initial conditions leads to the introduction of an adjoint system, whose discretization is related to reverse accumulation in automatic differentiation. Similar adjoint systems arise in optimal control and other areas, including classical mechanics. Adjoint systems are introduced in such a way that they exactly preserve a relevant quadratic invariant (more precisely, an inner product). Symplectic Runge--Kutta and partitioned Runge--Kutta methods are defined through the exact conservation of a differential geometric structure, but may be characterized by the fact that they preserve exactly quadratic invariants of the system being integrated. Therefore, the symplecticness (or lack of symplecticness) of a Runge--Kutta or partitioned Runge--Kutta integrator should be relevant to understanding its performance when applied to the computation of sensitivities, to optimal control problems, and in other applications requiring the use of adjoint systems. This paper examines the links between symplectic integration and those applications and presents in a new, unified way a number of results currently scattered among or implicit in the literature. In particular, we show how some common procedures, such as the direct method in optimal control theory and the computation of sensitivities via reverse accumulation, imply, probably unbeknownst to the user, “hidden” integrations with symplectic partitioned Runge--Kutta schemes.

References

[1]

L. Abia and J. M. Sanz-Serna, Partitioned Runge-Kutta methods for separable Hamiltonian problems, Math. Comp., 60 (1993), pp. 617--634.

[2]

V. I. Arnold, Mathematical Methods of Classical Mechanics, 2nd ed., Springer, New York, 1989.

[3]

P. B. Bochev and C. Scovel, On quadratic invariants and symplectic structure, BIT, 34 (1994), pp. 337--345.

[4]

J. F. Bonnans and J. Laurent-Varin, Computation of order conditions for symplectic partitioned Runge--Kutta schemes with application to optimal control, Numer. Math., 103 (2006), pp. 1-10.

[5]

J. C. Butcher, Numerical Methods for Ordinary Differential Equations, 2nd ed., J. Wiley, Chichester, UK, 2008.

[6]

D. G. Cacuci, Sensitivity theory for nonlinear systems. I. Nonlinear functional analysis approach, J. Math. Phys., 22 (1981), pp. 2794--2802.

[7]

M. Chyba, E. Hairer, and G. Vilmart, The role of symplectic methods in optimal control, Optim. Control Appl. Methods, 30 (2009), pp. 367--382.

[8]

G. J. Cooper, Stability of Runge-Kutta methods for trajectory problems, IMA J. Numer. Anal., 7 (1987), pp. 1--13.

[9]

A. L. Dontchev, W. W. Hager, and V. M. Veliov, Second-order Runge--Kutta approximations in control constrained optimal control, SIAM J. Numer. Anal., 38 (2000), pp. 202--226.

[10]

S. Ervedoza and E. Zuazua, Numerical Approximation of Exact Control Waves, Springer, New York, 2013.

[11]

A. Farrés, J. Laskar, S. Blanes, F. Casas, J. Makazaga, and A. Murua, High precision symplectic integrators for the solar system, Celestial Mechanics and Dynamical Astronomy, 116 (2013), pp. 141--174.

[12]

W. Feller, An Introduction to Probability Theory and Its Applications, Vol. 2, J. Wiley, New York, 1971.

[13]

J. Frank and S. Zhuk, Symplectic Möbius integrators for LQ optimal control problems, in Proceedings of the 53rd IEEE Conference on Decision and Control, Los Angeles, CA, IEEE, 2014, pp. 6377--6382.

[14]

M. B. Giles and N. A. Pierce, An introduction to the adjoint approach to design, Flow, Turbulence and Combustion, 65 (2000), pp. 393--415.

[15]

A. Griewank, A mathematical view of automatic differentiation, Acta Numer., 12 (2003), pp. 321--398.

[16]

D. F. Griffiths and J. M. Sanz-Serna, On the scope of the method of modified equations, SIAM J. Sci. Statist. Comput., 7 (1986), pp. 994--1008.

[17]

W. W. Hager, Runge-Kutta methods in optimal control and the transformed adjoint system, Numer. Math., 87 (2000), pp. 247--282.

[18]

E. Hairer, Backward analysis of numerical integrators and symplectic methods, Ann. Numer. Math., 1 (1994), pp. 107--132.

[19]

E. Hairer, Ch. Lubich, and G. Wanner, Geometric Numerical Integration, 2nd ed., Springer, Berlin, 2006.

[20]

E. Hairer, A. Murua, and J. M. Sanz-Serna, The nonexistence of symplectic multiderivative Runge-Kutta methods, BIT, 34 (1994), pp. 80--87.

[21]

E. Hairer, S. P. Nø rsett, and G. Wanner, Solving Ordinary Differential Equations I: Nonstiff Problems, 2nd ed., Springer, Berlin, 1993.

[22]

E. Hairer, and G. Wanner, Solving Ordinary Differential Equations II: Stiff and Differential-Algebraic Problems, 2nd ed., Springer, Berlin, 1996.

[23]

S. Lall and M. West, Discrete variational Hamiltonian mechanics, J. Phys. A, 39 (2006), pp. 5509--5519.

[24]

F. M. Lasagni, Canonical Runge-Kutta methods, Z. Angew. Math. Phys., 39 (1988), pp. 952--953.

[25]

B. Leimkuhler and S. Reich, Simulating Hamiltonian Dynamics, Cambridge University Press, Cambridge, UK, 2005.

[26]

J. E. Marsden and M. West, Discrete mechanics and variational integrators, Acta Numer., 10 (2001), pp. 357--514.

[27]

A. Murua, On order conditions for partitioned symplectic methods, SIAM J. Numer. Anal., 34 (1997), pp. 2204--2211.

[28]

S. Ober-Blöbaum, O. Junge, and J. E. Marsden, Discrete mechanics and optimal control: An analysis, ESAIM Control Optim. Calc. Var., 17 (2011), pp. 322--352.

[29]

I. M. Ross, A road for optimal control: The right way to commute, Ann. New York Acad. Sci., 1065 (2006), pp. 210--231.

[30]

A. Sandu, D. N. Daescu, G. R. Carmichael, and T. Chai, Adjoint sensitivity analysis of regional air quality models, J. Comput. Phys., 204 (2005), pp. 222--252.

[31]

J. M. Sanz-Serna, Runge-Kutta schemes for Hamiltonian systems, BIT, 28 (1988), pp. 877--883.

[32]

J. M. Sanz-Serna, Geometric integration, in The State of the Art in Numerical Analysis, I. S. Duff and G. A. Watson, eds., Clarendon Press, Oxford, 1997, pp. 121--143.

[33]

J. M. Sanz-Serna and L. Abia, Order conditions for canonical Runge--Kutta schemes, SIAM J. Numer. Anal., 28 (1991), pp. 1081--1096.

[34]

J. M. Sanz-Serna and M. P. Calvo, Numerical Hamiltonian Problems, Chapman and Hall, London, 1994.

[35]

R. R. Scherer and H. Türke, Reflected and transposed Runge-Kutta methods, BIT, 23 (1983), pp. 262--266.

[36]

T. Schlick, Molecular Modelling and Simulation: An Interdisciplinary Guide, 2nd ed., Springer, New York, 2010.

[37]

Z. Sirkes and E. Tziperman, Finite difference of adjoint or adjoint of finite difference?, Monthly Weather Rev., 49 (1997), pp. 5--40.

[38]

E. T. Sontag, Mathematical Control Theory, Deterministic Finite Dimensional Systems, 2nd ed., Springer, New York, 1998.

[39]

Y. B. Suris, Preservation of symplectic structure in the numerical solution of Hamiltonian systems, in Numerical Solution of Differential Equations, S. S. Filippov, ed., Akad. Nauk. SSSR, Inst. Prikl. Mat., Moscow, 1988, pp. 138--144 (in Russian).

[40]

Y. B. Suris, Hamiltonian methods of Runge-Kutta type and their variational interpretation, Math. Model., 2 (1990), pp. 78--87 (in Russian).

[41]

E. Trélat, Contrôle Optimal: Théorie et Applications, Vuibert, Paris, 2005.

[42]

J. Zabczyk, Mathematical Control Theory: An Introduction, Birkhäuser, Boston, 1995.

Cited By

Hofmann SBorzì A(2024)Pointwise Error Estimates of Numerical Solutions to Linear Quadratic Optimal Control ProblemsJournal of Scientific Computing10.1007/s10915-024-02484-799:1Online publication date: 11-Mar-2024
https://dl.acm.org/doi/10.1007/s10915-024-02484-7
Amodio PBrugnano LIavernaro F(2023)(Spectral) Chebyshev collocation methods for solving differential equationsNumerical Algorithms10.1007/s11075-022-01482-w93:4(1613-1638)Online publication date: 13-Jan-2023
https://dl.acm.org/doi/10.1007/s11075-022-01482-w
Lang JSchmitt B(2023)Exact Discrete Solutions of Boundary Control Problems for the 1D Heat EquationJournal of Optimization Theory and Applications10.1007/s10957-022-02154-4196:3(1106-1118)Online publication date: 9-Jan-2023
https://dl.acm.org/doi/10.1007/s10957-022-02154-4
Show More Cited By

Index Terms

Symplectic Runge--Kutta Schemes for Adjoint Equations, Automatic Differentiation, Optimal Control, and More

Index terms have been assigned to the content through auto-classification.

Recommendations

Implicit-Explicit Runge--Kutta Schemes for Numerical Discretization of Optimal Control Problems

Implicit-explicit (IMEX) Runge--Kutta methods play a major rule in the numerical treatment of differential systems governed by stiff and nonstiff terms. This paper discusses order conditions and symplecticity properties of a class of IMEX Runge--Kutta ...
Automatic differentiation of explicit Runge-Kutta methods for optimal control

This paper considers the numerical solution of optimal control problems based on ODEs. We assume that an explicit Runge-Kutta method is applied to integrate the state equation in the context of a recursive discretization approach. To compute the ...
Almost symplectic Runge-Kutta schemes for Hamiltonian systems

Symplectic Runge-Kutta schemes for the integration of general Hamiltonian systems are implicit. In practice, one has to solve the implicit algebraic equations using some iterative approximation method, in which case the resulting integration scheme is ...

Comments

Information & Contributors

Information

Published In

cover image SIAM Review

SIAM Review Volume 58, Issue 1

DOI:10.1137/siread.58.1

Issue’s Table of Contents

© 2016, Society for Industrial and Applied Mathematics.

Publisher

Society for Industrial and Applied Mathematics

United States

Publication History

Published: 01 January 2016

Author Tags

Author Tags

Qualifiers

Research-article

Contributors

Other Metrics

View Article Metrics

Bibliometrics & Citations

Bibliometrics

Article Metrics

10
Total Citations
View Citations
0
Total Downloads

Downloads (Last 12 months)0
Downloads (Last 6 weeks)0

Reflects downloads up to 06 Oct 2024

Other Metrics

View Author Metrics

Citations

Cited By

Hofmann SBorzì A(2024)Pointwise Error Estimates of Numerical Solutions to Linear Quadratic Optimal Control ProblemsJournal of Scientific Computing10.1007/s10915-024-02484-799:1Online publication date: 11-Mar-2024
https://dl.acm.org/doi/10.1007/s10915-024-02484-7
Amodio PBrugnano LIavernaro F(2023)(Spectral) Chebyshev collocation methods for solving differential equationsNumerical Algorithms10.1007/s11075-022-01482-w93:4(1613-1638)Online publication date: 13-Jan-2023
https://dl.acm.org/doi/10.1007/s11075-022-01482-w
Lang JSchmitt B(2023)Exact Discrete Solutions of Boundary Control Problems for the 1D Heat EquationJournal of Optimization Theory and Applications10.1007/s10957-022-02154-4196:3(1106-1118)Online publication date: 9-Jan-2023
https://dl.acm.org/doi/10.1007/s10957-022-02154-4
Wang BJiang Y(2023)Structure-Preserving Algorithms with Uniform Error Bound and Long-time Energy Conservation for Highly Oscillatory Hamiltonian SystemsJournal of Scientific Computing10.1007/s10915-023-02178-695:3Online publication date: 17-Apr-2023
https://dl.acm.org/doi/10.1007/s10915-023-02178-6
Bencomo MChan J(2023)Discrete Adjoint Computations for Relaxation Runge–Kutta MethodsJournal of Scientific Computing10.1007/s10915-023-02102-y94:3Online publication date: 1-Feb-2023
https://dl.acm.org/doi/10.1007/s10915-023-02102-y
Matsubara TMiyatake YYaguchi TRanzato MBeygelzimer ADauphin YLiang PVaughan J(2021)Symplectic adjoint method for exact gradient of neural ODE with minimal memoryProceedings of the 35th International Conference on Neural Information Processing Systems10.5555/3540261.3541850(20772-20784)Online publication date: 6-Dec-2021
https://dl.acm.org/doi/10.5555/3540261.3541850
Jay L(2021)Symplecticness conditions of some low order partitioned methods for non-autonomous Hamiltonian systemsNumerical Algorithms10.1007/s11075-020-00898-686:2(495-514)Online publication date: 1-Feb-2021
https://dl.acm.org/doi/10.1007/s11075-020-00898-6
Hühnerbein RSavarino FPetra SSchnörr C(2021)Learning Adaptive Regularization for Image Labeling Using Geometric AssignmentJournal of Mathematical Imaging and Vision10.1007/s10851-020-00977-263:2(186-215)Online publication date: 1-Feb-2021
https://dl.acm.org/doi/10.1007/s10851-020-00977-2
Ito SMatsuda TMiyatake Y(2021)Adjoint-based exact Hessian computationBIT10.1007/s10543-020-00833-061:2(503-522)Online publication date: 1-Jun-2021
https://dl.acm.org/doi/10.1007/s10543-020-00833-0
Zhang XTu LWu JLi S(2019)Trajectory Planning Based on Optimal Control and Exact DerivativesIntelligent Robotics and Applications10.1007/978-3-030-27529-7_49(580-591)Online publication date: 8-Aug-2019
https://dl.acm.org/doi/10.1007/978-3-030-27529-7_49

View Options

View options

Get Access

Login options

Check if you have access through your login credentials or your institution to get full access on this article.

Full Access

Get this Publication

Media

Figures

Other

Tables

View Issue’s Table of Contents