abstract

Modified Chebyshev Picard Iteration (MCPI) is a numerical method for approximating solutions of Ordinary Differential Equations (ODEs). MCPI uses Picard Iteration with Orthogonal Chebyshev Polynomial basis functions to recursively update approximate time histories of system states. Unlike stepping numerical integrators, such as explicit RungeKutta methods, MCPI approximates large segments of the trajectory by evaluating the forcing function at multiple nodes along the current approximation during each iteration. Importantly, the Picard sequence theoretically converges to the solution over large time intervals if the forces are continuous and once differentiable. Orthogonality of the basis functions and a vectormatrix formulation allow for low overhead cost, efficient iterations, and parallel evaluation of the forcing function. Despite these advantages MCPI only achieves a geometric rate of convergence. Depending on the quality of the starting approximation, MCPI sometimes requires more function evaluations than competing methods; for parallel applications, this is not a serious drawback, but may be for some serial applications.
To improve efficiency, the Terminal Convergence Approximation Modified Chebyshev Picard Iteration (TCAMCPI) was developed. TCAMCPI takes advantage of the property that once moderate accuracy of the approximating trajectory has been achieved, the subsequent displacement of nodes asymptotically approaches zero. Applying judicious approximation methods to the force function at each node in the terminal convergence iterations is shown to dramatically reduce the computational cost to achieve accurate convergence.
To illustrate this approach we consider highorder sphericalharmonic gravity for high accuracy orbital propagation. When combined with a starting approximation from the 2body solution TCAMCPI, is shown to outperform 2 current stateofpractice integration methods for astrodynamics. This paper presents the development of TCAMCPI, and its implementation for orbit propagation. The precision and efficiency performance for TCAMCPI, in comparison to two stateofpractice numerical integration methods (RungeKutta 78 and Gauss Jackson 8th order multistep algorithm), are studied for typical orbits.