Adversary and Partition Search

Adversary search refers to the issues that arise when trying to program computers to play games of strategy, like chess and bridge. Partition search is a technique that can be used to substantially reduce the portion of a game tree that needs to be expanded by a computer player. Conventional techniques store positions that have been evaluated along with their values, but partition search that incorporates dependency analysis, so that groups of positions with identical values can be identified as the search proceeds. By storing a group of positions instead of just a single one, we increase the number of positions that can be evaluated by table lookup instead of by search, reducing the running time of the associated algorithm.

CIRL

Partition search was discovered at CIRL, and we continue to work on extensions to the method. It is also one of the main technical ideas underlying GIB.

Pointers

Partition Search
We introduce a new form of game search called partition search that incorporates dependency analysis, allowing substantial reductions in the portion of the tree that needs to be expanded. Both theoretical results and experimental data are presented. For the game of bridge, partition search provides approximately as much of an improvement over existing methods as alpha-beta pruning provides over minimax. Written by Matt Ginsberg and appeared in Proc. AAAI in 1997. Compressed postsript document.

Parent areas:
Systematic search

Implementation:
GIB

Back to the CIRL Overview Page.