Heuristic search revisited

Abstract In this paper, we present general models for estimating time and space requirements of heuristic search algorithms. We also present an empirical study on a set of four well-known heuristic search algorithms. This study is useful in two ways. On one hand, it gives a general framework for comparing heuristic search algorithms from various respects including time, space, solution quality, and search effectiveness. On the other hand, it provides additional independent empirical results collected from different domains. These results can be used to strengthen other results obtained by other researchers in the area. Furthermore, the obtained results are machine independent in the sense that the CPU time is separated from the timing parameters. This enables us to give quantitative arguments on heuristic search algorithms for any architecture.