Parallel Patterns for Agent-Based Evolutionary Computing

Abstract

Computing applications such as metaheuristics-based optimization can greatly benefit from multi-core architectures available on modern supercomputers. In this paper, we describe an easy and efficient way to implement certain population-based algorithms (in the discussed case, multi-agent computing system) on such runtime environments. Our solution is based on an Erlang software library which implements dedicated parallel patterns. We provide technological details on our approach and discuss experimental results.

Concurrent agent‐based evolutionary computations as adaptive dataflows

“…Modeling multi‐agent evolutionary computations as static dataflows was investigated within the ParaPhrase Project, where the authors proposed a kind of virtual dataflow machine programmed with the use of a set of predefined “operators” (each corresponding to one parallel computing pattern). In this approach, the computation (dataflow) is specified upfront and remains unchanged at runtime, and the rate of elements within the flow is constant.…”

“…This paper introduces a new execution model for agent‐based computing systems based on variable‐rate dynamic dataflows. When compared with previous works, the new model has more “degrees of freedom” (eg, different queuing policies, varying granularity of the data in the flow). As a result, it leaves more space for optimization both at the programming phase and at runtime.…”