Logical Characterizations of PSPACE

We study the expressive power in the finite of the logic Fixed-Point+Counting, the extension of first-order logic which is obtained through adding both the fixed-point constructor and the ability to count.To this end an isomorphism preserving (‘generic’) model of computation is introduced whose PTime restriction exactly corresponds to this level of expressive power, while its PSpace restriction corresponds to While+Counting. From this model we obtain a normal form which shows a rather clear separation of the relational vs. the arithmetical side of the algorithms involved.In parallel, we study the relations of Fixed-Point+Counting with the infinitary logics and the corresponding pebble games.The main result, however, involves the concept of an arithmetical invariant. By this we mean a functor taking every finite relational structure to an expansion of (an initial segment of) the standard arithmetical structure. In particular its values are linearly ordered structures. We establish the existence of a family of arithmetical invariants with the following properties:• The invariants themselves can be evaluated in polynomial time.• A class of finite relational structures is definable in Fixed-Point+Counting if and only if membership can be decided in polynomial time on the basis of the values of one of the invariants.• The invariant r classifies all finite relational structures exactly up to equivalence with respect to the logic We also give a characterization of Fixed-Point+Counting in terms of sequences of formulae in the : It corresponds exactly to the polynomial time computable families (φn)n ∈ ω in these logics.Towards a positive assessment of the expressive power of Fixed-Point+Counting, it is shown that the natural extension of fixed-point logic by Lindström quantifiers, which capture all the PTime computable properties of cardinalities of definable predicates, is strictly weaker than what we get here. This implies in particular that every extension of fixed-point logic by means of monadic Lindström quantifiers, which stays within PTime, must be strictly contained in Fixed-Point+Counting.

Towards finite homomorphism-homogeneous relational structures

In the constraint satisfaction problem CSP(H) corresponding to a finite relational structure H, the aim is to decide, given a relational structure G, whether there exists a homomorphism from G to H. In (Bulatov, 2003), we proved that if H is a conservative structure, then it can be associated with a complete edge-3-colored graph whose vertex set is the universe of H. The complexity and a solution algorithm for CSP(H) strongly depend on certain properties of the associated graph. In this paper we show how a similar edge-3-colored graph can be defined for an arbitrary finite relational structure H. Then we study properties of the defined graph and find a solution algorithm for CSP(H), where G(H) satisfies some restrictions. The latter result substantially generalizes the results (2000,2002,1998,1997) concerning max-closed constraints and constraints with a 2-semilattice, semigroup or conservative groupoid polymorphism. Finally, we complete the study of the complexity of maximal constraint languages started in (Bulatov et al., 2001).

A number of efficient methods for evaluating first-order and monadic-second order queries on finite relational structures are based on tree-decompositions of structures or queries. We systematically study these methods.In the first part of the article, we consider arbitrary formulas on tree-like structures. We generalize a theorem of Courcelle [1990] by showing that on structures of bounded tree-width a monadic second-order formula (with free first- and second-order variables) can be evaluated in time linear in the structure size plus the size of the output.In the second part, we study tree-like formulas on arbitrary structures. We generalize the notions of acyclicity and bounded tree-width from conjunctive queries to arbitrary first-order formulas in a straightforward way and analyze the complexity of evaluating formulas of these fragments. Moreover, we show that the acyclic and bounded tree-width fragments have the same expressive power as the well-known guarded fragment and the finite-variable fragments of first-order logic, respectively.

We show that if a finite relational structure has a near unanimity polymorphism, then the constraint satisfaction problem with that structure as its fixed template has bounded pathwidth duality, putting the problem in nondeterministic logspace. This generalizes the analogous result of Dalmau and Krokhin for majority polymorphisms and lends further support to a conjecture suggested by Larose and Tesson.

We provide a canonical construction of conformal covers for finite hypergraphs and present two immediate applications to the finite model theory of relational structures. In the setting of relational structures, conformal covers serve to construct guarded bisimilar companion structures that avoid all incidental Gaifman cliques—thus serving as a partial analogue in finite model theory for the usually infinite guarded unravellings. In hypergraph theoretic terms, we show that every finite hypergraph admits a bisimilar cover by a finite conformal hypergraph. In terms of relational structures, we show that every finite relational structure admits a guarded bisimilar cover by a finite structure whose Gaifman cliques are guarded. One of our applications answers an open question about a clique constrained strengthening of the extension property for partial automorphisms (EPPA) of Hrushovski, Herwig and Lascar. A second application provides an alternative proof of the finite model property (FMP) for the clique guarded fragment of first-order logic CGF, by reducing (finite) satisfiability in CGF to (finite) satisfiability in the guarded fragment, GF.

Quasi-interpretations are an useful tool to control resources usage of term rewriting systems, either time or space. They not only combine well with path orderings and provide characterizations of usual complexity classes but also give hints in order to optimize the program. Moreover, the existence of a quasi-interpretation is decidable.In this paper, we present some more characterizations of complexity classes using quasi-interpretations. We mainly focus on small space-bounded complexity classes. On one hand, by restricting quasi-interpretations to sums (that is allowing only affine quasi-interpretations), we obtain a characterization of LinSpace. On the other hand, a strong tiering discipline on programs together with quasi-interpretations yield a characterization of LogSpace.Lastly, we give two new characterizations of Pspace: in the first, the quasi-interpretation has to be strictly decreasing on each rule and in the second, some linearity constraints are added to the system but no assumption concerning the termination proof is made.KeywordsTuring MachineFunction SymbolTransitive ClosurePolynomial SpaceCovering GraphThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Inductive definitions over finite structures

In this article, we consider the interplay of generalized quantifiers and built-in relations over finite structures, in particular, in the range of logics capturing the circuit complexity classes \(\mathrm{AC^{0}}\) and \(\mathrm{TC^{0}}\) . It is well known that for capturing \(\mathrm{AC^{0}}\) first-order logic has to be equipped with order and, e.g., predicates for addition and multiplication, whereas for \(\mathrm{TC^{0}}\) generalized quantifiers such as majority quantifiers are necessary. The sharp division between the classes \(\mathrm{AC^{0}}\) and \(\mathrm{TC^{0}}\) can be explained by the fact that \(\mathrm{AC^{0}}\) is not closed under restricting \(\mathrm{AC^{0}}\) -computable queries into simple subsequences of the input, whereas \(\mathrm{TC^{0}}\) is closed under such relativization as its queries can be expressed in terms of first-order formulas using universe-independent generalized quantifiers and order as the only built-in relation. In the terminology of abstract logics, the above means that logics capturing \(\mathrm{AC^{0}}\) do not have the relativization property, and hence, they are not regular logics unlike the logics capturing \(\mathrm{TC^{0}}\) . This weakness of \(\mathrm{AC^{0}}\) has been also elaborated in the line of research on the Crane Beach Conjecture. The conjecture (which was refuted by Barrington et al.) was that if a language \( L \) has a neutral letter, then \( L \) can be defined in \(\operatorname{FO}_{\mathcal{A}}\) , first-order logic with the collection of all numerical built-in relations \(\mathcal{A}\) , if and only if \( L \) can be already defined in \(\operatorname{FO}_{\leq}\) . Our approach is two-fold. First, we study universe-independent cardinality quantifiers \(\operatorname{\mathsf{Q}}\) defined by a parameter set \(S\subseteq\mathbb{N}\) and formulate a combinatorial criterion for \( S \) implying that all languages in \(\mathrm{DLOGTIME}\) -uniform \(\mathrm{TC^{0}}\) can be defined in \(\operatorname{FO}_{\leq}(\operatorname{\mathsf{Q}})\) . For instance, this criterion is satisfied if \( S \) is the range of some polynomial with positive integer coefficients of degree at least two. Second, by adapting the key properties of abstract logics to accommodate built-in relations, we define the regular interior \(\operatorname{\mathcal{R}-int}(\mathcal{L})\) (the largest regular \(\mathcal{L}^{*}\) such that \(\mathcal{L}^{*}\subseteq\mathcal{L}\) ) and regular closure \(\operatorname{\mathcal{R}-cl}(\mathcal{L})\) (the least regular \(\mathcal{L}^{*}\) such that \(\mathcal{L}\subseteq\mathcal{L}^{*}\) ), of a logic \(\mathcal{L}\) with built-in relations, and show that the Crane Beach Conjecture can be interpreted as a statement concerning the regular interior of \(\mathcal{L}\) . By extending the results of Barrington et al., we further show that if \(\mathcal{B}=\{+\}\) , or \(\mathcal{B}\) contains only unary relations besides \(\leq\) , then \(\operatorname{\mathcal{R}-int}(\operatorname{FO}_{\mathcal{B}})\equiv \operatorname{FO}_{\leq}\) . In contrast, our results from the first part of the article imply that if \(\mathcal{B}\) contains \(\leq\) and the range of a polynomial of degree at least two, then \(\operatorname{\mathcal{R}-cl}(\operatorname{FO}_{\mathcal{B}})\) includes all languages in \(\mathrm{DLOGTIME}\) -uniform \(\mathrm{TC^{0}}\) .

The notion of logical reducibilities is derived from the idea of interpretations between theories. It was used by Lovász and Gács [LG77] and Immerman [Imm87] to give complete problems for certain complexity classes and hence establish new connections between logical definability and computational complexity.However, the notion is also interesting in a purely logical context. For example, it is helpful to establish nonexpressibility results.We say that a class of τ-structures is a >complete problem for a logic under L-reductions if it is definable in [τ] and if every class definable in can be ”translated” into by L-formulae (cf. §4).We prove the following theorem:1.1. Theorem. There are complete problemsfor partial fixed-point logic andfor inductive fixed-point logic under quantifier-free reductions.The main step of the proof is to establish a new normal form for fixed-point formulae (which might be of some interest itself). To obtain this normal form we use theorems of Abiteboul and Vianu [AV91a] that show the equivalence between the fixed-point logics we consider and certain extensions of the database query language Datalog.In [Dah87] Dahlhaus gave a complete problem for least fixed-point logic. Since least fixed-point logic equals inductive fixed-point logic by a well-known result of Gurevich and Shelah [GS86], this already proves one part of our theorem.However, our class gives a natural description of the fixed-point process of an inductive fixed-point formula and hence sheds some light on completely different aspects of the logic than Dahlhaus's construction, which is strongly based on the features of least fixed-point formulae.

Capturing complexity classes by fragments of second-order logic

The expressive power of first order logic with generalized quantifiers over finite ordered structures is studied. The following problem is addressed: Given a family Q of generalized quantifiers expressing a complexity class C, what is the expressive power of first order logic FO(Q) extended by the quantifiers in Q? From previously studied examples, one would expect that FO(Q) captures L/sup C/, i.e., logarithmic space relativized by an oracle in C. We show that this is not always true. However, we derive sufficient conditions on complexity class C such that FO(and) captures L/sup C/. These conditions are fulfilled by a large number of relevant complexity classes, in particular, for example, by NP. As an application of this result, it follows that first order logic extended by Henkin quantifiers captures L/sup NP/. This answers a question raised by Blass and Gurevich. Furthermore we show that for many families Q of generalized quantifiers (including the family of Henkin quantifiers), each FO(Q)-formula can be replaced by an equivalent FO(Q)-formula, with only two occurrences of generalized quantifiers.

Reflexive digraphs with near unanimity polymorphisms

We show that if a class K of finite relational structures is closed under quasi‐substructures, then there is no saturated K‐generic structure that is superstable but not ω ‐stable (© 2010 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)

This work deals with the concept of relativization. My focus is on the acquisition of restrictive relative clause structure by Brazilian Portuguese speakers learning English as a foreign language. This kind of structure was chosen because it is considered one of the main structural devices used in the languages of the world for encoding recursion; it can vary considerably in terms of its distribution and syntactic form. The present work aims to spot some strategies used by learners to solve specific tasks, here called challenges. These challenges invite the use of structures that alternate subject-subject (SS), subject-object (SO), subject-object (OS), object-object (OO) combinations. It is characterized by a syntactic diversity that happens according to the position of the characters (small ball-shaped faces like the ones in some video games) in the challenges. As alternative for the relativization, participants in the activity used strategies such as use of passive voice in one sentence, use of time adverb (time expressions), use of coordinate clause, independent sentences. The participants were 3rd grade high school students aged between 15 and 17 years old. Aspects like age and experience abroad were taken into account. First of all, it was applied a pre-test with English teachers, holders of college degrees, and students from a so-called post advanced level of an English language school in order to validate the instrument (the challenges). Among the strategies, the additive structure (conjunction ‘and’) was the most recurrent, followed closely by relativization. The results showed that these learners make use of their cognitive background in order to develop alternative ways to fulfill the tasks assigned in each challenge. The theoretical part was principally based in AXT 1994.