Solving the Partner Units Configuration...

The partner units problem (PUP) is an acknowledgedhard benchmark problem for the logic programming communitywith various industrial application fields. The state-of-the-art heuristicfor the PUP is the QuickPup heuristic. Unfortunately, complexdomain-dependent heuristics like QuickPup could not be realizedwithin a declarative solving framework like constraint or answerset programming. A new hybrid technique called constraint answerset programming (CASP) offers the possibility to realize declarativeframeworks in which it is possible to also express complex heuristicslike QuickPup. In this paper we present the CASP solver ASCASS(A Simple Constraint Answer Set Solver) which providesnovel methods for defining and exploiting problem-dependent searchheuristics. Beyond the possibility of using already built-in problemindependentheuristics, ASCASS allows on the ASP level the defi-nition of problem-dependent variable selection, value selection andpruning strategies which guide the search of the CP solver. Due tothe new possibilities for representing and exploiting complex domainheuristics in ASCASS, we show how to encode the PUP andrealize QuickPup in ASCASS. An evaluation reveals that due tothe QuickPup heuristic, which is not expressible in any other ASPor CASP approach, ASCASS outperforms state-of-the-art ASP andCASP solvers on the tested PUP instances.