Multi-Agent Systems

by Albert Major

Core Concept: Emergence

Definition and Theory

• Emergence is complexity arising from simplicity. It is the phenomenon where a system possesses properties that its individual parts do not have ("The whole is greater than the sum of its parts").
• "More is Different": When many simple units interact under specific rules, they create something new (a higher-order entity).
• Examples of Scale:
  • Water: Individual water molecules are not wet; "wetness" is an emergent property of many interacting molecules.
  • Biology: Atoms form molecules -> proteins -> cells -> organs -> individuals -> societies.

The Ant Colony Example

• Distributed Control: An ant colony has no central leader ("mastermind"). No single ant knows the state of the whole colony.
• Simple Rules: Ants follow basic local rules, such as secreting chemicals (pheromones) to signal their current job (worker, caretaker, soldier, gatherer).
• Adaptation via Interaction: If a predator kills the gatherers, the colony starves unless roles change.
  • Worker ants encounter fewer gatherers (based on chemical signals).
  • Once the encounter rate drops below a critical point, the worker switches its role to become a gatherer.
• Result: Balance is restored through random individual interactions, creating a robust, intelligent system from "stupid" parts.

Multi-Agent Systems (MAS)

Definition of an Agent

• An entity with perception and action capabilities.
• It senses the environment and acts upon it.
• Examples: Humans, animals, robots, autonomous vehicles, or software processes.

Definition of a Multi-Agent System

• A system or group of potentially interacting agents sharing a common environment.