A Graph-Based Method for Improving GSAT

GSAT is a randomized greedy local repair procedure that was introduced for solving propositional satis ability and constraint satisfaction problems. We present an improvement to GSAT that is sensitive to the problem's structure. When the problem has a tree structure the algorithm is guaranteed to nd a solution in linear time. For non-tree networks, the algorithm designates a subset of nodes, called cutset, and executes a regular GSAT algorithm on this set of variables. On all the rest of the variables it executes a specialized local search algorithm for trees. This algorithm nds an assignment that, like GSAT, locally minimizes the sum of unsatis ed constraints and also globally minimizes the number of con icts in every tree-like subnetwork. We will present results of experiments showing that this new algorithm outperforms regular GSAT on sparse networks whose cycle-cutset size is bounded by 30 of the nodes.

• Type: Conference
• Year: 1996
• Where: AAAI
• Authors: Kalev Kask, Rina Dechter
• Keywords: AAAI 1996, Algorithm, Intelligent Agents, Local Search Algorithm, Regular Gsat Algorithm