Modal logic
The Propositional Mu-Calculus is Elementary
Proceedings of the 11th Colloquium on Automata, Languages and Programming
Modal Characterisation Theorems over Special Classes of Frames
LICS '05 Proceedings of the 20th Annual IEEE Symposium on Logic in Computer Science
On the definability of simulability and bisimilarity by finite epistemic models
CLIMA'11 Proceedings of the 12th international conference on Computational logic in multi-agent systems
Tangled modal logic for spatial reasoning
IJCAI'11 Proceedings of the Twenty-Second international joint conference on Artificial Intelligence - Volume Volume Two
Non-finite axiomatizability of dynamic topological logic
ACM Transactions on Computational Logic (TOCL)
| Hi-index | 0.00 |
We show that given a finite, transitive and reflexive Kripke model 驴 W, 驴, 驴 驴 驴 驴 and $${w \in W}$$ , the property of being simulated by w (i.e., lying on the image of a literalpreserving relation satisfying the `forth' condition of bisimulation) is modally undefinable within the class of S4 Kripke models. Note the contrast to the fact that lying in the image of w under a bisimulation is definable in the standard modal language even over the class of K4 models, a fairly standard result for which we also provide a proof.We then propose a minor extension of the language adding a sequent operator $${\natural}$$ (`tangle') which can be interpreted over Kripke models as well as over topological spaces. Over finite Kripke models it indicates the existence of clusters satisfying a specified set of formulas, very similar to an operator introduced by Dawar and Otto. In the extended language $${{\sf L}^+ = {\sf L}^{\square\natural}}$$ , being simulated by a point on a finite transitive Kripke model becomes definable, both over the class of (arbitrary) Kripke models and over the class of topological S4 models.As a consequence of this we obtain the result that any class of finite, transitive models over finitely many propositional variables which is closed under simulability is also definable in L +, as well as Boolean combinations of these classes. From this it follows that the μ-calculus interpreted over any such class of models is decidable.