A Repeated-Update Problem in the DeltaBlue Algorithm

Abstract

We observe a repeated-update problem in the process of updating walkabout strengths of the DeltaBlue algorithm, which is known as a fast constraint solving method based on local propagation. According to the basic references on the DeltaBlue algorithm, the time complexity of the planning phase is described as O(MN) for a constraint problem such that the number of constraints is N and the maximum number of methods a constraint has is M . We, however, point out that the time complexity is not O(MN) using a very simple example. In this example, the time complexity is exponential order for N . Then we present a corrected DeltaBlue algorithm whose time complexity is O(EN ²) for a constraint problem such that the number of constraints is N and the maximum number of variables a constraint has is E . Finally we measure the performance of the corrected DeltaBlue algorithm using two benchmarks.