Generalizing Partial Order and Dynamic Backtracking

RecentlyFtwo new backtracking algorithmsF dynamic backtracking (DB)and partial order dynamicbacktracking (PDB) have been presented. These algorithms have the property to be additive on disjoint subproblemsand yet use only polynomial space. Unlike DBFPDBonly imposes a partial search order and therefore appears to have morefreedomthan DBto explore the search space. HoweverFbothalgorithms are not directly comparable in terms of flexibility. In this paper we present new backtracking algorithms that are obtained by relaxing the ordering conditions of PDB.This gives them additional flexibility while still being additive on disjoint subproblems. In particularF we show that our algorithms generalize both DBand PDB.

• Type: Conference
• Year: 1998
• Where: AAAI
• Authors: Christian Bliek
• Keywords: AAAI 1998, DB)and Partial Order, Dynamic Backtracking, Intelligent Agents, Partial Search Order