Abstract

In this paper we attempt to analyze the problem of centrally allocating tasks to selfinterested agents in environments characterized by high uncertainty. The analysis is based on equilibrium dynamics, taking into account the agents ’ short-term and long-term considerations. The model reflects a rich realism by allowing both tasks and agents to dynamically arrive in the environment. This is correlated with real life scenarios where agents take part in a continuous process while having only partial information concerning the different world states and other agents ’ capabilities. Based on the comprehensive analysis of the equilibrium we present innovative algorithms that significantly facilitate the extraction of the agents ’ strategies for any specific settings. The findings are illustrated computationally and compared with parallel mechanisms for allocating the tasks in a closed environment and in a centralized enforceable manner. 1.

Citations