Experimental Evaluation of a State-Of-The-Art Grounder

Many state-of-the-art declarative systems use a ground-and-solve approach, where the problem statement, expressed in a high-level language, is first grounded into a low-level representation. Next, a solver is used to search a solution for the low-level representation. In order to prevent a combinatorial blowup of the grounding, many intelligent techniques have been developed. In this paper we study in detail three such techniques (Lifted Unit Propagation, Grounding With Bounds, and Reduced Grounding) to get a better insight in their individual merits and their interactions. Our experiments take as benchmarks all the NP problems of the previous Answer Set Programming (ASP) competitions. The experiments are performed with IDP 3 and all tools needed to run them are made publicly available. The first experiment discusses the impact of the three techniques on the "efficiency" of the grounding step, and on each other. In a second set of experiments we show that a reduction in the grounding size as a result of the application of these grounding techniques does not reduce the search space. We give an in-depth analysis of our results and discuss what this means for the development of grounding techniques for declarative systems.