Weighted-sequence problem: ASP vs CASP and declarative vs problem-oriented solving

Search problems with large variable domains pose a challenge to current answer-set programming (ASP) systems as large variable domains make grounding take a long time, and lead to large ground theories that may make solving infeasible. To circumvent the "grounding bottleneck" researchers proposed to integrate constraint solving techniques with ASP in an approach called constraint ASP (CASP). In the paper, we evaluate an ASP system clingo and a CASP system clingcon on a handcrafted problem involving large integer domains that is patterned after the database task of determining the optimal join order. We find that search methods used by clingo are superior to those used by clingcon, yet the latter system, not hampered by grounding, scales up better. The paper provides evidence that gains in solver technology can be obtained by further research on integrating ASP and CSP technologies.