Papers by Wiebe Van Der Hoek
Epistemic Logic for AI and Computer Science, 1995
Springer eBooks, 2000
Abstract: We extend our general approach to characterizing information to multi-agent systems.In ... more Abstract: We extend our general approach to characterizing information to multi-agent systems.In particular, we provide a formal description of an agent's knowledge containing exactlythe information conveyed by some (honest) formula j. Only knowing is important for dynamicagent systems in two ways. First of all, one wants to compare different states of knowledgeof an agent and, secondly, for agent a's decisions,
Logic Journal of the IGPL, May 18, 2009
We define formalisms to reason about Coalitional Games (CGs), in which one can express what coali... more We define formalisms to reason about Coalitional Games (CGs), in which one can express what coalitions of agents can achieve. We start with Quantified CGs (QCGs), in which each agent has some goals he wants to satisfy, which may change over time. Then we focus on CGs themselves. Although CGs can be well analysed in a formalism close to Pauly's Coalition Logic, in QCGs, when having preferences, some differences become apparent.
European Conference on Artificial Intelligence, Aug 27, 2012
Social laws-sets of constraints imposed on the behaviour of agents within a multi-agent system wi... more Social laws-sets of constraints imposed on the behaviour of agents within a multi-agent system with the goal of some desirable overall behaviour resulting-are an important mechanism for coordinating multi-agent behaviour. When considering social laws in human environments, the inspiration for social laws in multiagent systems, we argue that a key design principle is least change. That is, social laws are more likely to be accepted and adopted, and hence successful, if they are conservative, in the sense that they represent the smallest change possible from the pre-existing status quo that is required to effect the desired objective. Our aim in the present paper is to introduce, formalise, and investigate the notion of a conservative social law for multi-agent systems. To make the idea of a conservative social law precise, we formalise the notion of a distance metric for social laws, and discuss a range of possible properties for such metrics. We then formulate the conservative social law problem, (i.e., the problem of constructing an effective social law that requires the least change according to this metric), discuss some possible interpretations of distance in this context, and discuss some issues surrounding conservative social laws.
Lecture Notes in Computer Science, 2001
3APL is an agent programming language based on the concept of an intelligent agent. An intelligen... more 3APL is an agent programming language based on the concept of an intelligent agent. An intelligent agent is a computational entity with a mental state consisting of its beliefs and goals. The operational semantics of the language 3APL is specified by a formal semantics in terms of a socalled transition system. An operational semantics allows operational reasoning about agents, but does not allow for a compositional style of reasoning based on the structure of the agent itself. For this purpose, in this paper we construct a denotational semantics which corresponds to the operational semantics and provides the basis for a semantics of a programming logic for (part of) 3APL. The programming logic is a variant of a modal logic with operators for reasoning about the actions and the beliefs of an agent. Our results clarify the relation between more practical approaches to agents, represented by agent programming languages, and more theoretical work on agents, represented by socalled agent logics.
Lecture Notes in Computer Science, 2004
Lecture Notes in Computer Science, 1997
ABSTRACT
Lecture Notes in Computer Science, 2005
HAL (Le Centre pour la Communication Scientifique Directe), 2014
A well-studied phenomenon in network theory are optimal schedules to distribute information by on... more A well-studied phenomenon in network theory are optimal schedules to distribute information by one-to-one communication between nodes. One can take these communicative actions to be 'telephone calls', and this process of spreading information is known as gossiping [4]. It is typical to assume a global scheduler who simply executes a possibly non-deterministic protocol. Such a protocol can be seen as consisting of a sequence of instructions "first, agent a calls b, then c, next, d calls b ... ". We investigate epistemic gossip protocols, where an agent a will call another agent not because it is so instructed but based on its knowledge or ignorance of the factual information that is distributed over the network. Such protocols therefore don't need a central schedular, but they come at a cost: they may take longer to terminate than non-epistemic, globally scheduled, protocols. We describe various epistemic protocols, we give their logical properties, and we model them in a number of ways.
arXiv (Cornell University), Jun 23, 2016
Gossip protocols aim at arriving, by means of point-to-point or group communications, at a situat... more Gossip protocols aim at arriving, by means of point-to-point or group communications, at a situation in which all the agents know each other's secrets. We consider distributed gossip protocols which are expressed by means of epistemic logic. We provide an operational semantics of such protocols and set up an appropriate framework to argue about their correctness. Then we analyze specific protocols for complete graphs and for directed rings. Epistemic Protocols for Distributed Gossip 2.1 Syntax We loosely use the syntax of the language CSP (Communicating Sequential Processes) of [11] that extends the guarded command language of [6] by disjoint parallel composition and commands for synchronous communication. CSP was realized in the distributed programming language OCCAM (see INMOS [12]). The main difference is that we use as guards epistemic formulas and as communication primitives calls that do not require synchronization. Also, the syntax of our distributed programs is very limited. In order to define gossip protocols we introduce in turn calls and epistemic guards. Throughout the paper we assume a fixed finite set A of at least three agents. We assume that each agent holds exactly one secret and that there exists a bijection between the set of agents and the set of secrets. We denote by P the set of all secrets (for propositions). Furthermore, it is assumed that each secret carries information identifying the agent to whom that secret belongs. 2.1.1 Calls Each call concerns two agents, the caller (a below) and the agent called (b). We distinguish three modes of communication of a call: push-pull, written as ab or (a, b). During this call the caller and the called agent learn each other's secrets, push, written as a ⊲ b. After this call the called agent learns all the secrets held by the caller, pull, written as a ⊳ b. After this call the caller learns all the secrets held by the called agent. Variables for calls are denoted by c, d. Abusing notation we write a ∈ c to denote that agent a is one of the two agents involved in the call c (e.g., for c := ab we have a ∈ c and b ∈ c). Calls in which agent a is involved are denoted by c a. 2.1.2 Epistemic guards Epistemic guards are defined as formulas in a simple modal language with the following grammar: φ ::= F a p | ¬φ | φ ∧ φ | K a φ , where p ∈ P and a ∈ A. Each secret is viewed as a distinct symbol. We denote the secret of agent a by A, the secret of agent b by B and so on. We denote the set of so defined formulas by L and we refer to its members as epistemic formulas or epistemic guards. We read F a p as 'agent a is familiar with the secret p' (or 'p belongs to the set of secrets a knows about') and K a φ as 'agent a knows that formula φ is true'. So this language is an epistemic language where atoms consist of 'knowing whether' statements about propositional atoms, if we view secrets as Boolean variables. Atomic expressions in L concern only who knows what secrets. As a consequence the language cannot express formally the truth of a secret p. This level of abstraction suffices for the purposes of the current paper. However, expressions F a p could be given a more explicit epistemic reading in terms of 'knowing whether'. That is, 'a is familiar with p' can be interpreted (on a suitable Kripke model) as 'a knows whether the secret p is true or not'. This link is established in [3].
Mathematical Social Sciences, Sep 1, 1999
In this paper we study the notion of group knowledge in a modal epistemic context. Starting with ... more In this paper we study the notion of group knowledge in a modal epistemic context. Starting with the standard de nition of this kind of knowledge on Kripke models, we show that this de nition gives rise to some quite counter-intuitive behaviour. Firstly, using a strong notion of derivability, w e show that group knowledge in a state can always, but trivially be derived from each of the agents' individual knowledge. In that sense, group knowledge is not really implicit, but rather explicit knowledge of the group. Thus, a weaker notion of derivability seems to be more adequate. However, adopting this more`local view', we argue that group knowledge need not be distributed over the members of the group: we give an example in which the traditional concept of group knowledge is stronger than what can be derived from the individual agents' knowledge. We then propose two additional properties on Kripke models: we show that together they are su cient to guarantee`distributivity', while, when leaving one out, one may construct models that do not ful l this principle.
Artificial Intelligence, Sep 1, 1999
International Journal of Foundations of Computer Science, Jun 1, 1992
We discuss issues of expressibility and completeness of the logic of implicit knowledge (I) and “... more We discuss issues of expressibility and completeness of the logic of implicit knowledge (I) and “everybody’s knowledge” (E), as introduced in a system with a number m of epistemic agents by Halpern & Moses. The operator E is defined as a conjunction and corresponds semantically to the union of the m accessibility relations. Dually, the operator I is semantically associated with an intersection, but it is, surprisingly, not equivalent with a disjunction. From the view of Kripke structures there is a related asymmetry: although union can be modally defined, intersection cannot! We discuss consequences (in terms of (in)expressibility, correspondence and completeness) of this property for the epistemic logic under consideration and also present an extension of modal logic in which intersection is expressible.
arXiv (Cornell University), Feb 3, 2018
In this contribution we present arbitrary arrow update model logic (AAUML). This is a dynamic epi... more In this contribution we present arbitrary arrow update model logic (AAUML). This is a dynamic epistemic logic or update logic. In update logics, static/basic modalities are interpreted on a given relational model whereas dynamic/update modalities induce transformations (updates) of relational models. In AAUML the update modalities formalize the execution of arrow update models, and there is also a modality for quantification over arrow update models. Arrow update models are an alternative to the well-known action models. We provide an axiomatization of AAUML. The axiomatization is a rewrite system allowing to eliminate arrow update modalities from any given formula, while preserving truth. Thus, AAUML is decidable and equally expressive as the base multi-agent modal logic. Our main result is to establish arrow update synthesis: if there is an arrow update model after which ϕ, we can construct (synthesize) that model from ϕ. We also point out some pregnant differences in update expressivity between arrow update logics, action model logics, and refinement modal logic.
Artificial Intelligence, 2017
In this paper we introduce arbitrary arrow update logic (AAUL). The logic AAUL takes arrow update... more In this paper we introduce arbitrary arrow update logic (AAUL). The logic AAUL takes arrow update logic, a dynamic epistemic logic where the accessibility relations of agents are updated rather than the set of possible worlds, and adds a quantifier over such arrow updates. We investigate the relative expressivity of AAUL compared to other logics, most notably arbitrary public announcement logic (APAL). Additionally, we show that the model checking problem for AAUL is PSPACE-complete. Finally, we introduce a proof system for AAUL, and prove it to be sound and complete.
International Game Theory Review
From the text: This special issue of the International Game Theory Review is on Logic and the Fou... more From the text: This special issue of the International Game Theory Review is on Logic and the Foundations of Game and Decision Theory. It contains a selection of papers presented at the 10th Conference on Logic and the Foundations of Game and Decision Theory (LOFT10), which took place at the University of Sevilla (Spain), July 18-20, 2012. On top of this, the issue contains an additional paper and a short note that fall within the theme of LOFT. While this special issue collects papers with a focus on game theory, a second set of papers that are more logic-oriented can be found in a special issue of the Journal of Philosophical Logic (Vol. 42, No. 6, 2013).
Multi-Agent Systems, 2015
We implement a framework to evaluate epistemic gossip protocols. Gossip protocols spread informat... more We implement a framework to evaluate epistemic gossip protocols. Gossip protocols spread information within a network of agents by pairwise communications. This tool, Epistemic Gossip Protocol (EGP), is applied to epistemic gossip protocols presented in [1]. We introduce a programming language for epistemic gossip protocols. We describe an interpreter for this language, together with a model generator and model checker, for a dynamic model of the protocol. The tool EGP outputs key dynamic properties of such protocols, thus facilitating the process of protocol design and planning. We conclude with some experimental results.
Uploads
Papers by Wiebe Van Der Hoek