Comparison of Two Differential Evolution Approaches for Combinatorial Double Auctions

摘要

Combinatorial double auctions improve the inefficiency of single-side auctions. However, the winner determination problem (WDP) is computationally challenging in combinatorial double auctions. Differential Evolution provides an approach to cope with the complexity for solving optimization problems. In this paper, the effectiveness of applying two DE based algorithms to solve the WDP in combinatorial double auctions will be studied. The first approach searches solutions in the continuous solution space using the standard DE and transforms the solution to discrete solutions to evaluate the fitness function values. The second approach searches solutions directly in the discrete solution space using binary operations in DE. The effectiveness of these two algorithms is compared by conduct several experiments.