Optimal solutions in systems consisting of rational agents

Abstract

Rational agents and systems consisting of such agents are considered. An agent is an autonomous object that has sources of information about the environment (for example, physical sensors) and affects this environment (for example, with the help of actuators). A rational agent is an agent who, in order to achieve a set goal, is able to act effectively, that is, to use behavioral strategies that are close to optimal. It is assumed that there is a utility function, which is defined on the set of possible sequences of actions of the agent (agents, if a system of agents is considered) and takes values in the set of real numbers. A rational agent acts in such a way as to maximize the utility function. A system of rational agents is a system consisting of rational agents that have a common goal and act in an optimal way to achieve it. Agents act using the optimal (or close to optimal) solution of the extreme problem formulated for the system. The work gives examples of systems consisting of rational agents. The main groups of problems and the corresponding mathematical methods of their solution related to the management of agents and systems of rational agents are determined: cooperation of agents (creation of a system of agents), planning and coordination of actions of agents, search of placement of the system of agents, recognition. Cooperation is necessary when no single agent has sufficient experience, resources, and information to solve a problem, but different agents have the expertise and capabilities to solve different parts of the problem. Planning is the development of a method of action of agents and the entire system in the future depending on the situations that may arise, the choice of an effective method of action, optimal distribution of resources. Coordination is such an organization of the actions of various agents that make up the system, which ensures the efficiency of this system. The tasks of finding the optimal placement of agent system and the task of recognizing the state of the environment are important. Examples of such problems are given and mathematical methods of their solution are indicated.