Splitting algorithm of valuation algebra and its application in automaton representation of semiring valued constraint problems

Abstract

A splitting algorithm is developed for solving single-query projection problems in valuation algebras. This method is based on a generalized combination theorem. It is shown that by using this new kind of combination property, a given single-query projection problem can be broken into pieces of subprojection problems which might be solved simultaneously by different computational resources. At last, as an application of splitting algorithms, we develop an optimized procedure for automaton representation of semiring valued constraint problems.