The solution for the branching factor of the alpha-beta pruning algorithm and its optimality
Communications of the ACM
On the branching factor of the alpha-beta pruning algorithm
Artificial Intelligence
Hi-index | 0.01 |
This paper contains a probabilistic analysis of the performance of the game-searching SSS^* algorithm which has been shown to be superior to @a-@b. Necessary and sufficient conditions for node expansion are established, and an expression for the average number of nodes expanded is derived. The branching factor of SSS^* is shown to coincide with that of @a-@b, thus rendering the two algorithms asymptotically equivalent. Numerical comparison of the expected complexities of the two algorithms is finally carried out over a wide spectrum of search depths and branching degrees. The latter shows that the savings in the number of positions evaluated by SSS^* relative to that of @a-@b is rather limited and is not enough to offset the increase in other computational resources.