📝 SummarXiv

Abstract

We propose a new method to prove the partitioned Runge--Kutta methods with symplectic conditions for determinate and stochastic Hamiltonian systems are symplectic. We utilize Gr\"obner basis technology which is the one of symbolic computation method based on computer algebra theory and geometrical mechanical proving theory. In this approach, from determinate Hamilton's equations, we get the relations of partial differentials which are regarded as polynomials of plenty variables marked indeterminates. Then, we compute the Gr\"obner basis of above polynomials, and the normal form of symplectic expression, which is as the middle expression, with respect to the Gr\"obner basis. Then, we compute the Gr\"obner basis of symplectic conditions and the normal form of the middle expression with respect to above Gr\"obner basis, and get that the normal form is zero, which complete the proof. We also develop this procedure to the stochastic Hamiltonian systems case and get similar result. In this paper, the new try provide us a new idea to prove the structure-preservation laws of another numerical methods, including the energy conservation law, the momentum conservation law and so on.