The Design of a Multi-Agent Framework for Distributed Hyper-Linking-Based Theorem Proving

Abstract

A multi-agent framework for distributed theorem proving of the hyper-linking method is proposed in this paper. Distributed theorem proving at the clause-level can be achieved by collaborating multi-agents. In this framework, each computer collaborates on proving theorems without stopping the jobs currently running. The hyper-linking proof procedure is wrapped as software agents, which can be partially performed for a specic task. Each computer, referred to as proof house, involved in this framework can work solely or collaborate with other registered proof houses to complete the proof of the given problems. Agents for monitoring the trac of environment, the performance and liveness of each proof houses, updating and sharing proof experiences, and tuning parameters of proof houses are designed. The architecture and the design concepts are addressed.