Closed mariomulansky closed 11 years ago
Hmm, for the current stepper the formulas for the construction of a1, and a2 are not given in this paper.
Implement SB3A with m=4. Here, all coefficients are analytically known.
Done, tests have still to be implemented.
The current high-order symplectic stepper is not suitable for high-precision types as it constructs parameters from double. A symplectic stepper with arguments constructed from integers should be added, i think such a stepper can be found e.g. in R. McLachlan: "ON THE NUMERICAL INTEGRATION OF ORDINARY DIFFERENTIAL EQUATIONS BY SYMMETRIC COMPOSITION METHODS" , 1995