We propose a new parallel search algorithm – A! – based on cooperating A* search agents, concurrency and a secondary tiebreaking heuristic. The search agents in A! share information asynchronously and trade some of their independence for additional search focus and a more global view of the search task. A! is inherently nondeterministic due to the implicit randomness of instruction scheduling, but given a consistent primary heuristic, it still finds optimal solutions for the single-source shortest path problem (SSSP). A! combines into a single cooperative search algorithm the breadth available in parallel execution and the depth-first orientation of both locally and globally informed search.
We experimentally show that A! outperforms both vanilla A* and an explicitly randomized, noncooperative parallel A* variant. We present an empirical study on cooperation benefits and scalability in the classic 15-puzzle context. The results imply that cooperation and concurrency can successfully be harnessed in algorithm design, inviting further inquiry into algorithms of this kind.
IOS Press, Inc.
6751 Tepper Drive
Clifton, VA 20124
Tel.: +1 703 830 6300
Fax: +1 703 830 2300 email@example.com
(Corporate matters and books only) IOS Press c/o Accucoms US, Inc.
For North America Sales and Customer Service
West Point Commons
Lansdale PA 19446
Tel.: +1 866 855 8967
Fax: +1 215 660 5042 firstname.lastname@example.org