Symbolic Model Checking for Dynamic Epistemic Logic Johan van Benthem 1,2 , Jan van Eijck 1,3 , Malvin Gattinger 1 , and Kaile Su 4,5 1 Institute for Logic, Language & Computation (ILLC), University of Amsterdam 2 Department of Philosophy, Stanford University 3 Centrum Wiskunde & Informatica, Amsterdam 4 Institute for Integrated and Intelligent Systems, Griffith University 5 Department of Computer Science, Jinan University Abstract. Dynamic Epistemic Logic (DEL) can model complex infor- mation scenarios in a way that appeals to logicians. However, existing DEL implementations are ad-hoc, so we do not know how the frame- work really performs. For this purpose, we want to hook up with the best available model-checking and SAT techniques in computational logic. We do this by first providing a bridge: a new faithful representation of DEL models as so-called knowledge structures that allow for symbolic model checking. Next, we show that we can now solve well-known bench- mark problems in epistemic scenarios much faster than with existing DEL methods. Finally, we show that our method is not just a matter of implementation, but that it raises significant issues about logical repre- sentation and update. 1 Introduction We bring together two strains in the area of epistemic model checking. On one side, there are many frameworks for symbolic model checking on interpreted systems using temporal logics [24,30]. On the other hand, there are explicit model checkers for variants of Dynamic Epistemic Logic (DEL) like DEMO [15] with inferior performance but superior usability as they allow specification in dynamic languages directly. The goal of our work is to connect the two worlds of symbolic model checking and DEL in order to gain new insights on both sides. Existing work on model checking DEL mainly focuses on specific examples, for example the Dining Cryptographers [28], the Sum and Product riddle [26] or Russian Cards [12]. Given these specific approaches, a general approach to symbolic model checking the full DEL language is desirable. A first step is [30] which presents symbolic model checking for temporal logics of knowledge. How- ever, it does not cover announcements or other dynamics. The framework here extends these ideas with dynamic operators and a twist on the semantics. Our knowledge structures are similar in spirit to hypercubes from [25], but of a different type: We do not use interpreted systems and temporal relations are not part of our models. Hence also our language does not contain temporal operators but primitives for epistemic events like announcements. c  Springer-Verlag Berlin Heidelberg 2015 W. van der Hoek et al. (Eds.): LORI 2015, LNCS 9394, pp. 366–378, 2015. DOI: 10.1007/978-3-662-48561-3_30