Overview

AgE 3 is a rewrite of a distributed, multi-agent platform developed by Intelligent Information Systems Group with a focus on providing a simple, modern and lightweight computational environment.

AgE has been firstly created as a platform tightly connected to the concept of Evolutionary Multi Agent Systems. With the third version we decided to split the platform into two separate parts:

• core – which provide a distribution, parallelization and various services to the computation,
• compute – which is the part being executed as a computation by the core.

This separation greatly simplified both layers and made it possible to support other computational models. AgE 3 is build with the focus on simplicity and extensibility. It is possible to start creating its own compute module with a single class and then add more complicated logic and use more advanced services provided by core.

Architecture

AgE 3 is a distributed platform based on a cluster of nodes. Each node is a separate JVM which can be run on a single or multiple hosts. Nodes perform an automatic neighbourhood discovery and initialize a common, global state of the cluster. They provide a unified view of resources for the compute level.

A compute module is a computation that can be run on AgE cluster. In the most basic form, it is a single class implementing Runnable interface. In more complicated scenarios it can be any collections of classes and dependencies which has a Runnable entry point.

AgE provides some means to control and inspect the state of the cluster and computations. The basic one is the console that provides a Nashorn-based (JavaScript-based) command line interface.