Agent behaviors and coordinative mechanism

Abstract

Purpose

The purpose of this paper is to improve the behaviors coordination mechanism, to maintain the system's long time‐scale and stable competitive capability, when the agents in the system focus on cooperating with each other.

Design/methodology/approach

Effort level for every agent, whose dynamics can be described as a stochastic partial differential equation, and the incentive of effort as the control of the corresponding agent, are introduced to describe agents' behavior abstracted. The cooperative stochastic differential game model is constructed: first, the optimal resolve trajectory mapping with profit maximization of the system are obtained, then the transitory imputation coupled with effort initial state of the system by introducing dynamic Shapley value imputation method. Based on the results obtained, the profit distribution strategies and the equilibration incentive compensation mechanism are given, due to the evolution law of the payoff and the state variable.

Findings

It is concluded that: the transitory compensation to agent for efforts and incentive, which can be changed with the system state at current and in history and in future changed, would guarantee the realization of the Shapley value imputation throughout the game horizon.

Originality/value

In this paper, the interactivity between agents in the system is considered first. The dynamical Shapley imputation mechanism and the transitory compensatory mechanism are provided to make the imputation more stable and feasible.

Citation

Zheng, X., Xu, X. and Cui Luo, C. (2012), "Agent behaviors and coordinative mechanism", Kybernetes, Vol. 41 No. 10, pp. 1586-1603. https://doi.org/10.1108/03684921211276765