A leader-based parallel cross entropy algorithm for MCP

The Cross Entropy algorithm is a new search method for combinatorial problem. However, it needs considerable computational time to achieve good solution quality. To make the Cross Entropy algorithm faster, this paper proposes a leader-based cooperative parallel algorithm. Unlike the widely used coarse-grained parallelization strategy, our method has a leader , which can move around freely, and several controlled followers. To evaluate the performance of the algorithm, we implement our algorithm using OpenMPI on MIMD architecture, and has applied it on 25 selected MCP benchmark problems. The speedup and efficiency is analyzed, and the results obtained are compared with those obtained by four other best heuristic algorithms, GLS , EdgeAC+LS, EA/G and RLS. Categories and Subject Descriptors G.2.1 [Discrete Mathematics]: Combinatorics—Combinatorial algorithms; I.2.11 [Artificial Intelligence]: Distributed Artificial Intelligence—Parallel heuristic methods General Terms Algorithms Ke...

• Type: Conference
• Year: 2007
• Where: GECCO
• Authors: Zhanhua Bai, Qiang Lv
• Keywords: Cooperative Parallel Algorithm, Cross Entropy Algorithm, GECCO 2007, Parallel Algorithm