Jump to content

Paranoid algorithm

From Wikipedia, the free encyclopedia

In combinatorial game theory, the paranoid algorithm is an algorithm that aims to improve the alpha-beta pruning capabilities of the maxn algorithm by making the player p chosen to maximize the score "paranoid" of the other players by assuming they are cooperating to minimize p's score, thus minimizing any n-player game to a two-player game by making the opposing player the sum of the other player's scores. This returns the game to a zero-sum game and makes it analyzable via any optimization techniques usually applied in pair with the minimax theorem.[1] It performs notably faster than the maxn algorithm because of those optimizations.[2]

See also

[edit]

References

[edit]
  1. ^ Sturtevant, Nathan; Korf, Richard (30 July 2000). "On Pruning Techniques for Multi-Player Games" (PDF). AAAI-00 Proceedings: 201–207.
  2. ^ Sturtevant, Nathan (2003). "A Comparison of Algorithms for Multi-player Games". Lecture Notes in Computer Science. Berlin, Heidelberg: Springer Berlin Heidelberg. pp. 108–122. doi:10.1007/978-3-540-40031-8_8. ISBN 978-3-540-20545-6.