Framework for performance comparison of Lambert algorithms

Abstract

This paper proposes a framework to compare the performances of different Lambert algorithms in a systematic way. Minimal parameters that can uniquely determine the iterator of each algorithm are identified then the cost to execute a single iteration of the algorithm is calculated. A set of parameters to create experimental problem instances which can determine a unique set of input parameters for each algorithm is selected. Numerical experiments are conducted using the determined set and provide the numbers of iterations for solved problem instances. The experimental results, combined with the cost per iteration, can be used to create the performance dominance map for considered algorithms. Results of a case study for performance comparison of two popular Lambert algorithms are presented to demonstrate the effectiveness of the proposed framework.