P2P-Based Scalable Execution Platform for Algorithmically Transitive Network

“Algorithmically Transitive Network” (ATN) is a novel computational model based on a data-flow network, consisting of the following operations: a forward propagation propelled with node firing and token creation, a backward propagation caused by evaluating differential coefficients, and a topological alteration taken place by autonomous agents. In the research of the ATN, a simulation run on some parallel processing scheme is essential. As a flexible and powerful implementation scheme, the paper employs a P2P based distributed platform, and describes the mechanisms for simulation and P2P deployment of the ATN. The implemented platform has the following three features: flexible allocation of ATN nodes to the physical resources, unified description of communication between nodes, and several methods to realize high parallelism. The proposed scheme is also helpful to verify applicability of the employed P2P system.