Erich Grädel

On the Restraining Power of Guards

Guarded fragments of first-order logic were recently introduced by Andreka, van Benthem and Nemeti; they consist of relational first-order formulae whose quantifiers are appropriately relativized by atoms. These fragments are interesting because they extend in a natural way many propositional modal logics, because they have useful model-theoretic properties and especially because they are decidable classes that avoid the usual syntactic restrictions (on the arity of relation symbols, the quantifier pattern or the number of variables) of almost all other known decidable fragments of first-order logic. Here, we investigate the computational complexity of these fragments. We prove that the satisfiability problems for the guarded fragment ( GF ) and the loosely guarded fragment ( LGF ) of first-order logic are complete for deterministic double exponential time. For the subfragments that have only a bounded number of variables or only relation symbols of bounded arity, satisfiability is E xptime -complete. We further establish a tree model property for both the guarded fragment and the loosely guarded fragment, and give a proof of the finite model property of the guarded fragment. It is also shown that some natural, modest extensions of the guarded fragments are undecidable.

On the decision problem for two-variable first-order logic

We identify the computational complexity of the satisfiability problem for FO2, the fragment of first-order logic consisting of all relational first-order sentences with at most two distinct variables. Although this fragment was shown to be decidable a long time ago, the computational complexity of its decision problem has not been pinpointed so far. In 1975 Mortimer proved that FO2 has thefinite-modelproperty, which means that if an FO2-sentence is satisfiable, then it has a finite model. Moreover, Mortimer showed that every satisfiable FO2-sentence has a model whose size is at most doubly exponential in the size of the sentence. In this paper, we improve Mortimers bound by one exponential and show that every satisfiable FO2-sentence has a model whose size is at most exponential in the size of the sentence. As a consequence, we establish that the satisfiability problem for FO2 is NEXPTIME-complete. ?

Two-variable logic with counting is decidable

We prove that the satisfiability and the finite satisfiability problems for C/sup 2/ are decidable. C/sup 2/ is first-order logic with only two variables in the presence of arbitrary counting quantifiers 3/sup /spl ges/m/,m/spl ges/1. It considerably extends L/sup 2/ plain first-order with only two variables, which is known to be decidable by a result of Mortimers. Unlike L/sup 2/, C/sup 2/ does not have the finite model property.

Guarded fixed point logic

Guarded fixed point logics are obtained by adding least and greatest fixed points to the guarded fragments of first-order logic that were recently introduced by H. Andreka et al. (1998). Guarded fixed point logics can also be viewed as the natural common extensions of the modal p-calculus and the guarded fragments. We prove that the satisfiability problems for guarded fixed point logics are decidable and complete for deterministic double exponential time. For guarded fixed point sentences of bounded width, the most important case for applications, the satisfiability problem is EXPTIME-complete.

On logics with two variables

Abstract This paper is a survey and systematic presentation of decidability and complexity issues for modal and non-modal two-variable logics. A classical result due to Mortimer says that the two-variable fragment of first-order logic, denoted FO2, has the finite model property and is therefore decidable for satisfiability. One of the reasons for the significance of this result is that many propositional modal logics can be embedded into FO2. Logics that are of interest for knowledge representation, for the specification and verification of concurrent systems and for other areas of computer science are often defined (or can be viewed) as extensions of modal logics by features like counting constructs, path quantifiers, transitive closure operators, least and greatest fixed points, etc. Examples of such logics are computation tree logic CTL, the modal μ-calculus Lμ, or popular description logics used in artificial intelligence. Although the additional features are usually not first-order constructs, the resulting logics can still be seen as two-variable logics that are embedded in suitable extensions of FO2. Typically, the applications call for an analysis of the satisfiability and model checking problems of the logics employed. The decidability and complexity issues for modal and non-modal two-variables logics have been studied quite intensively in the last years. It has turned out that the satisfiability problems for two-variable logics with full first-order quantification are usually much harder (and indeed highly undecidable in many cases) than the satisfiability problems for corresponding modal logics. On the other side, the situation is different for model checking problems. The model checking problem of a modal logic has essentially the same complexity as the model checking problem of the corresponding two variable logic with full quantification.

Metafinite model theory

Motivated by computer science challenges, we suggest to extend the approach and methods of finite model theory beyond finite structures.

Dependence and Independence

We introduce an atomic formula

Finite Presentations of Infinite Structures:Automata and Interpretations

{\vec{y} \bot_{\vec{x}}\vec{z}}