The aim of being able to reason about quantities, time or space has been the main objective of the many efforts on the integration of propositional planning with extensions to handle different theories. Planning modulo theories are an approximation inspired by satisfiability modulo theories that generalize the integration of arbitrary theories with propositional planning. Parallel plans are crucial to reduce plan lengths and hence the time needed to reach a feasible plan in many approaches. Parallelization of actions relies on the (...) notion of interference, which is usually determined syntactically at compile time. In this paper we define a semantic notion of interference between actions in PMT. Apart from being strictly stronger than any syntactic notion of interference, we show how semantic interference can be easily and efficiently checked by calling an off-the-shelf SMT solver at compile time, constituting a technique orthogonal to the solving method. (shrink)

Maximum Satisfiability problem is an optimization variant of the Satisfiability problem denoted as MAX-SAT. The aim of this problem is to find Boolean variable assignment that maximizes the number of satisfied clauses in the Boolean formula. In case the number of variables per clause is equal or greater than three, then this problem is considered NP-complete. Hence, many researchers have developed techniques to deal with MAX-SAT. In this paper, we investigate the impact of different hybrid versions of binary harmony search (...) algorithm on solving MAX 3-SAT problem. Therefore, we propose two novel hybrid binary HS algorithms. The first hybridizes Flip heuristic with HS, and the second uses Tabu search combined with Flip heuristic. Furthermore, a distinguished feature of our proposed approaches is using an objective function that is updated dynamically based on the stepwise adaptation of weights mechanism to evaluate the MAX-SAT solution using the proposed hybrid versions. The performance of the proposed approaches is evaluated over standard MAX-SAT benchmarks, and the results are compared with six evolutionary algorithms and three stochastic local search algorithms. The obtained results are competitive and show that the proposed novel approaches are effective. (shrink)