Dynamics of MultiAgent Systems

‘Multiagent system’ (MAS) is a generic concept covering a wide variety of phenomena in the technolog- ical society, such as computer networks, robotic systems, autonomous cars, virtual organisations, elec- tronic markets, and other distributed software systems. A MAS consists of several ‘agents’ (e.g., software systems, robots, algorithmic trading systems) that act autonomously and intelligently in a common en- vironment, which is governed by some organisational rules or system norms, in pursuit of individual or collective goals, interacting with the environment and with each other by communication, planning and executing actions and strategies.

MAS often have a modular and dynamical nature: agents are represented by relatively autonomous modules and can enter or leave the system and can change their internal states, roles, goals, functionalities and abilities to act and interact within the system. The common environment in which agents operate can change due to their (inter)actions or its own dynamics. Finally, the organisation rules and system norms that govern the agents’ behaviour can change due to the dynamics of external system requirements and norms. This inherent dynamicity makes it quite challenging to properly specify, design and verify the behaviour of such a system.

While there has been extensive work on specification and verification of MAS with fixed struc- ture (fixed number of agents, fixed agents’ capabilities, communication protocols, knowledge, system norms, etc.), no systematic research on dynamical MAS has been performed yet. What is urgently needed is a systematic study of dynamics of MAS and of the extensions and adaptations of the existing techniques that are necessary to model and reason about them.

The main goals of this workshop will be to discuss, explore and systematise the major dynamical aspects of MAS, to propose methodologies and frameworks for their modelling and to initiate the de- velopment of feasible approaches and methods for their formal specification, automated verification and algorithmic synthesis.

The concept of MAS ranges over theoretical and engineering issues in computer science and artificial intelligence, but is also increasingly relevant in social sciences and humanities, and many of the main challenges related to dynamical aspects of MAS occur in these fields. The workshop will therefore bring together experts mainly from Computer Science and AI, with sufficient background and knowledge in social sciences and humanities, to explore and discuss the dynamics of MAS.