Practical Tractability of CSPS by Higher Level Consistency and Tree Decomposition

One fundamental research result in the area of Constraint Processing (CP) is a condition that guarantees problem tractability by relating the consistency level of a Constraint Satisfaction Problem (CSP) to the structure of the problem. In our research, we propose to build effective problem-solving strategies that exploit the above-mentioned result in practice. To this end, our investigations target two fundamental mechanisms in CP: (1) Consistency properties and their algorithms and (2) Backtrack search in a tree decomposition. In particular, we propose a new consistency property whose level is controlled by a parameter, and present algorithms for enforcing it. Then, we investigate strategies for backtrack search that apply those algorithms in localized contexts defined by a tree decomposition of the constraint network.