Giuseppe Perelli

Reasoning About Strategies: On the Model-Checking Problem

In open systems verification, to formally check for reliability, one needs an appropriate formalism to model the interaction between agents and express the correctness of the system no matter how the environment behaves. An important contribution in this context is given by modal logics for strategic ability, in the setting of multiagent games, such as A<scp>tl</scp>, A<scp>tl</scp>*, and the like. Recently, Chatterjee, Henzinger, and Piterman introduced <i>Strategy Logic</i>, which we denote here by CHP-S<scp>l</scp>, with the aim of getting a powerful framework for reasoning explicitly about strategies. CHP-S<scp>l</scp> is obtained by using first-order quantifications over strategies and has been investigated in the very specific setting of two-agents turned-based games, where a nonelementary model-checking algorithm has been provided. While CHP-S<scp>l</scp> is a very expressive logic, we claim that it does not fully capture the strategic aspects of multiagent systems. In this article, we introduce and study a more general strategy logic, denoted S<scp>l</scp>, for reasoning about strategies in multiagent concurrent games. As a key aspect, strategies in S<scp>l</scp> are not intrinsically glued to a specific agent, but an explicit binding operator allows an agent to bind to a strategy variable. This allows agents to share strategies or reuse one previously adopted. We prove that S<scp>l</scp> strictly includes CHP-S<scp>l</scp>, while maintaining a decidable model-checking problem. In particular, the algorithm we propose is computationally not harder than the best one known for CHP-S<scp>l</scp>. Moreover, we prove that such a problem for S<scp>l</scp> is N<scp>on</scp>E<scp>lementary</scp>. This negative result has spurred us to investigate syntactic fragments of S<scp>l</scp>, strictly subsuming A<scp>tl</scp>*, with the hope of obtaining an elementary model-checking problem. Among others, we introduce and study the sublogics S<scp>l</scp>[<scp>ng</scp>], S<scp>l</scp>[<scp>bg</scp>], and S<scp>l</scp>[1<scp>g</scp>]. They encompass formulas in a special prenex normal form having, respectively, nested temporal goals, Boolean combinations of goals, and, a single goal at a time. Intuitively, for a goal, we mean a sequence of bindings, one for each agent, followed by an L<scp>tl</scp> formula. We prove that the model-checking problem for S<scp>l</scp>[1<scp>g</scp>] is 2E<scp>xp</scp>T<scp>ime</scp>-<scp>complete</scp>, thus not harder than the one for A<scp>tl</scp>*. In contrast, S<scp>l</scp>[<scp>ng</scp>] turns out to be N<scp>on</scp>E<scp>lementary</scp>-hard, strengthening the corresponding result for S<scp>l</scp>. Regarding S<scp>l</scp>[<scp>bg</scp>], we show that it includes CHP-S<scp>l</scp> and its model-checking is decidable with a 2E<scp>xp</scp>T<scp>ime</scp>lower-bound. It is worth enlightening that to achieve the positive results about S<scp>l</scp>[1<scp>g</scp>], we introduce a fundamental property of the semantics of this logic, called <i>behavioral</i>, which allows to strongly simplify the reasoning about strategies. Indeed, in a nonbehavioral logic such as S<scp>l</scp>[<scp>bg</scp>] and the subsuming ones, to satisfy a formula, one has to take into account that a move of an agent, at a given moment of a play, may depend on the moves taken by any agent in another counterfactual play.

What makes ATL* decidable? a decidable fragment of strategy logic

Strategy Logic (Sl, for short) has been recently introduced by Mogavero, Murano, and Vardi as a formalism for reasoning explicitly about strategies, as first-order objects, in multi-agent concurrent games. This logic turns out to be very powerful, strictly subsuming all major previously studied modal logics for strategic reasoning, including Atl, Atl*, and the like. The price that one has to pay for the expressiveness of Sl is the lack of important model-theoretic properties and an increased complexity of decision problems. In particular, Sl does not have the bounded-tree model property and the related satisfiability problem is highly undecidable while for Atl* it is 2ExpTime-complete. An obvious question that arises is then what makes Atl* decidable. Understanding this should enable us to identify decidable fragments of Sl. We focus, in this work, on the limitation of Atl* to allow only one temporal goal for each strategic assertion and study the fragment of Sl with the same restriction. Specifically, we introduce and study the syntactic fragment One-Goal Strategy Logic (Sl[1g], for short), which consists of formulas in prenex normal form having a single temporal goal at a time for every strategy quantification of agents. We show that Sl[1g] is strictly more expressive than Atl*. Our main result is that Sl[1g] has the bounded tree-model property and its satisfiability problem is 2ExpTime-complete, as it is for Atl*.

Checking Interval Properties of Computations

Model checking is a powerful method widely explored in formal verification. Given a model of a system, e.g., a Kripke structure, and a formula specifying its expected behaviour, one can verify whether the system meets the behaviour by checking the formula against the model. Classically, system behaviour is expressed by a formula of a temporal logic, such as LTL and the like. These logics are “point-wise” interpreted, as they describe how the system evolves state-by-state. However, there are relevant properties, such as those constraining the temporal relations between pairs of temporally extended events or involving temporal aggregations, which are inherently “interval-based”, and thus asking for an interval temporal logic. In this paper, we give a formalization of the model checking problem in an interval logic setting. First, we provide an interpretation of formulas of Halpern and Shoham’s interval temporal logic HS over finite Kripke structures, which allows one to check interval properties of computations. Then, we prove that the model checking problem for HS against finite Kripke structures is decidable by a suitable small model theorem, and we provide a lower bound to its computational complexity.

Synthesis with Rational Environments

Synthesis is the automated construction of a system from its specification. The system has to satisfy its specification in all possible environments. The environment often consists of agents that have objectives of their own. Thus, it makes sense to soften the universal quantification on the behavior of the environment and take the objectives of its underlying agents into an account. Fisman et al. introduced rational synthesis: the problem of synthesis in the context of rational agents. The input to the problem consists of temporal-logic formulas specifying the objectives of the system and the agents that constitute the environment, and a solution concept (e.g., Nash equilibrium). The output is a profile of strategies, for the system and the agents, such that the objective of the system is satisfied in the computation that is the outcome of the strategies, and the profile is stable according to the solution concept; that is, the agents that constitute the environment have no incentive to deviate from the strategies suggested to them.

Synthesis with rational environments

Synthesis is the automated construction of a system from its specification. The system has to satisfy its specification in all possible environments. The environment often consists of agents that have objectives of their own. Thus, it makes sense to soften the universal quantification on the behavior of the environment and take the objectives of its underlying agents into an account. Fisman et al. introduced rational synthesis: the problem of synthesis in the context of rational agents. The input to the problem consists of temporal logic formulas specifying the objectives of the system and the agents that constitute the environment, and a solution concept (e.g., Nash equilibrium). The output is a profile of strategies, for the system and the agents, such that the objective of the system is satisfied in the computation that is the outcome of the strategies, and the profile is stable according to the solution concept; that is, the agents that constitute the environment have no incentive to deviate from the strategies suggested to them. In this paper we continue to study rational synthesis. First, we suggest an alternative definition to rational synthesis, in which the agents are rational but not cooperative. We call such problem strong rational synthesis. In the strong rational synthesis setting, one cannot assume that the agents that constitute the environment take into account the strategies suggested to them. Accordingly, the output is a strategy for the system only, and the objective of the system has to be satisfied in all the compositions that are the outcome of a stable profile in which the system follows this strategy. We show that strong rational synthesis is 2ExpTime-complete, thus it is not more complex than traditional synthesis or rational synthesis. Second, we study a richer specification formalism, where the objectives of the system and the agents are not Boolean but quantitative. In this setting, the objective of the system and the agents is to maximize their outcome. The quantitative setting significantly extends the scope of rational synthesis, making the game-theoretic approach much more relevant. Finally, we enrich the setting to one that allows coalitions of agents that constitute the system or the environment.

Reasoning about Strategies: on the Satisfiability Problem

Strategy Logic (SL, for short) has been introduced by Mogavero, Murano, and Vardi as a useful formalism for reasoning explicitly about strategies, as first-order objects, in multi-agent concurrent games. This logic turns out to be very powerful, subsuming all major previously studied modal logics for strategic reasoning, including ATL, ATL*, and the like. Unfortunately, due to its high expressiveness, SL has a non-elementarily decidable model-checking problem and the satisfiability question is undecidable, specifically Sigma_1^1. In order to obtain a decidable sublogic, we introduce and study here One-Goal Strategy Logic (SL[1G], for short). This is a syntactic fragment of SL, strictly subsuming ATL*, which encompasses formulas in prenex normal form having a single temporal goal at a time, for every strategy quantification of agents. We prove that, unlike SL, SL[1G] has the bounded tree-model property and its satisfiability problem is decidable in 2ExpTime, thus not harder than the one for ATL*.

Solving Parity Games Using an Automata-Based Algorithm

Parity games are abstract infinite-round games that take an important role in formal verification. In the basic setting, these games are two-player, turn-based, and played under perfect information on directed graphs, whose nodes are labeled with priorities. The winner of a play is determined according to the parities (even or odd) of the minimal priority occurring infinitely often in that play. The problem of finding a winning strategy in parity games is known to be in UPTime \(\cap \) CoUPTime and deciding whether a polynomial time solution exists is a long-standing open question. In the last two decades, a variety of algorithms have been proposed. Many of them have been also implemented in a platform named PGSolver. This has enabled an empirical evaluation of these algorithms and a better understanding of their relative merits.

Multi-agent Path Planning in Known Dynamic Environments

We consider the problem of planning paths of multiple agents in a dynamic but predictable environment. Typical scenarios are evacuation, reconfiguration, and containment. We present a novel representation of abstract path-planning problems in which the stationary environment is explicitly coded as a graph (called the arena) while the dynamic environment is treated as just another agent. The complexity of planning using this representation is pspace-complete. The arena complexity (i.e., the complexity of the planning problem in which the graph is the only input, in particular, the number of agents is fixed) is np-hard. Thus, we provide structural restrictions that put the arena complexity of the planning problem into ptime(for any fixed number of agents). The importance of our work is that these structural conditions (and hence the complexity results) do not depend on graph-theoretic properties of the arena (such as clique- or tree-width), but rather on the abilities of the agents.

Nash Equilibria in Concurrent Games with Lexicographic Preferences.

We study concurrent games with finite-memory strategies where players are given a Büchi and a mean-payoff objective, which are related by a lexicographic order: a player first prefers to satisfy its Büchi objective, and then prefers to minimise costs, which are given by a mean-payoff function. In particular, we show that deciding the existence of a strict Nash equilibrium in such games is decidable, even if players’ deviations are implemented as infinite memory strategies.

Binding Forms in First-Order Logic

Aiming to pinpoint the reasons behind the decidability of some complex extensions of modal logic, we propose a new classification criterion for sentences of first-order logic, which is based on the kind of binding forms admitted in their expressions, i.e., on the way the arguments of a relation can be bound to a variable. In particular, we describe a hierarchy of four fragments focused on the Boolean combinations of these forms, showing that the less expressive one is already incomparable with several first-order limitations proposed in the literature, as the guarded and unary negation fragments. We also prove, via a novel model-theoretic technique, that our logic enjoys the finite-model property, Craigs interpolation, and Beths definability. Furthermore, the associated model-checking and satisfiability problems are solvable in PTime and Sigma_3^P, respectively.