Previous editions‎ > ‎PADABS 2016‎ > ‎

Program 2016

PDF 
  • 14:00-16:00 First Session: Welcome to PADABS and paper session
    • Welcome (Vittorio Scarano, Gennaro Cordasco, Paul Richmond and Carmine Spagnuolo)
    • Robert Chisholm, Paul Richmond, and Steve Maddock, “A Standardised Benchmark for Assessing the Performance of Fixed Radius Near Neighbours”
      • Abstract: Fixed radius near neighbours (FRNNs) lies at the centre of many agent based models, whereby agents require awareness of their local peers. Due to its central role, handling of FRNNs is often a limiting factor of performance. However without a standardised metric to assess the handling of FRNNs, contributions to the field lack the rigorous appraisal necessary to expose their relative benefits.This paper presents a standardised specification of a multi agent based benchmark model. The benchmark model provides a means for the objective assessment of FRNNs performance, through the comparison of implementations. Results collected from implementations of the benchmark model under three agent based modelling frameworks show the CPU bound performance to scale linearly with agent population, in contrast the GPU accelerated framework only became linear after maximal device utilisation around 150,000 agents. The performance of each of the assessed frameworks was also found unaffected by changes to the rate of agent movement.
    • Michele Carillo, Gennaro Cordasco, Flavio Serrapica, Carmine Spagnuolo, Przemysaw Szufel, and Luca Vicidomini, “D-MASON on the Cloud: an Experience with Amazon Web Services”
      • Abstract: D-MASON framework is a parallel version of the MASON library for writing and running Agent-based simulations – a class of models which, by simulating the behavior of multiple agents, aims to emulate and/or predict complex phenomena. D-MASON has been conceived to harness the amount of unused computing power available in common installations like educational laboratory. Then the focus moved to dedicated installation, such as massively parallel machines or supercomputing centers. In this paper, D-M ASON takes another step forward and now it can be used on a cloud environment. The goal of the paper is twofold. Firstly, we are going to present D-MASON on the cloud – an D-MASON extension which, starting from a IaaS (Infrastructure as a Service) abstraction, provides a SIMulation-as-a-Service (SIMaaS) abstraction that simplifies the process of setting up and running distributed simulations in the cloud. Secondly, an additional goal of the paper is to assess computational and economical efficiency of running distributed multi-agent simulations on the Amazon Web Services EC2 instances. The computational speed and costs of an EC2 cluster will be compared against an on-site HPC cluster.
    • Alessandro Pellegrini, Cristina Montañola-Sales, Francesco Quaglia and Josep Casanovas-Garcia, “Load-Sharing Policies in Parallel Simulation of Agent-Based Demographic Models”
      • Abstract: Execution parallelism in Agent-Based Simulation (ABS) is mandatory to deal with complex/large-scale. This fact raises the need for run-time environments able to fully exploit hardware parallelism, while jointly offering ABS-suited programming abstractions. In this paper, we target last-generation general-purpose Parallel Discrete Event Simulation (PDES) platforms for multicore systems. We discuss a programming model to support both implicit (direct in-place access) and explicit (message passing) interactions across concurrent Logical Processes (LPs). For this PDES platform, we discuss different load-sharing policies combining event rate and implicit/explicit LPs' interactions. The study is based on a synthetic test case, representative of a class of agent-based models, run on an off-the-shelf machine equipped with 32 CPU-cores.
    • Aaron Howell and Paul Brenner, “Computational Considerations for a Global Human Well­being Simulation”
      • Abstract: Global scale human simulations have application in diverse fields such as economics, anthropology and marketing. The sheer number of agents however makes them extremely sensitive to variations in algorithmic complexity resulting in potentially prohibitive computational resource costs. In this paper we show that the computational capability of modern servers has increased to the point where billions of individual agents can be modeled on moderate institutional resources and (in a few years) on high end consumer systems. We close with the proposition of future frameworks to enable collaborative modelling of the global population.
  • 16:00-16:30 break
  • 16:30-17:30 Second Session: Tutorial, Panel and Round Table
    • Tutorial on D-Mason 3.0 (Gennaro Cordasco)
    • Panel: Open Agent Benchmark Initiative for Parallel and Distributed Benchmarking (Chair: Paul Richmond)
    • Round Table (Chair: Vittorio Scarano)