Various parallel methods exist for the numerical integration of the ordinary differential equations used to describe satellite motion. Several parallel integration methods were tested and compared to two of the most efficient serial methods in use today. Tests were completed for two cases, the normalized two-body problem and a single GPS satellite including nominal force models. In the high accuracy regime, none of the parallel methods showed significant speed-up compared to the serial methods for the two-body problem. The five processor, 10th order PIRKN method using a Gauss-Legendre corrector was the most promising method tested. It exhibited a speed-up of 1.18 at twelve digits of accuracy and 1.19 at fourteen digits of accuracy compared to the summed form of a Störmer-Cowell Predictor-Corrector method, but no speed-up compared to a general Class II Predictor-Corrector method. For the GPS orbit problem, none of the PIRK methods approached the performance of the fastest serial method, a general Class II Predictor-Corrector method. Variations of the Parallel Extrapolation method performed as well as but did not greatly exceed the performance of the fastest serial method.
Parallel Numerical Integration Methods for Orbital Motion
J of Astronaut Sci
The Journal of the Astronautical Sciences ; 48 , 4 ; 517-536
2000-12-01
20 pages
Article (Journal)
Electronic Resource
English
Parallel Numerical Integration Methods for Orbital Motion
Online Contents | 2000
|Parallel Numerical Integration Methods for Orbital Motion
Online Contents | 2000
|Parallel Numerical Integration Methods for Orbital Motion (AAS 00-151)
British Library Conference Proceedings | 2000
|Parallel Integration of Perturbed Orbital Motion
AIAA | 2019
|