网格中执行SLA的并行多群体协作PSO框架

针对广域范围内的计算资源加入粒子群优化(PSO)执行从而提高计算效率并降低计算成本,提出一个网格环境下执行SLA的并行多群体协作PSO框架(PMCF)。给出一个适应网格环境的并行多群体协作PSO(PMCPSO)算法,基于WS-Agreement研究了利用网格技术和PMCPSO算法设计PMCF,把一个元任务调度器用于无缝的资源发现、选取及协商服务质量,通过模拟试验对PMCF进行了评估。     

基于强化学习的多机群网格资源调度模型

在由多个计算机集群构成的多机群网格环境下，为了解决数据并行型计算（DPC）与计算资源的有效匹配问题，提出了一个基于强化学习机制的网格资源调度模型；给出了由多个计算机机群组成的多机群网格、逻辑计算机机群、数据并行型计算和一系列Agent的定义；利用多Agent的协作做竞争机制、基于强化学习的匹配知识库的修正方法，研究了逻辑计算机机群与DPC资源供需之间的有效匹配问题；描述了网格的资源调度模型。理论分析和实践表明，该模型有效地解决了多机群网格环境之下数据并行型计算所需的资源优化使用问题。该模型适合于基于多机群网格的数据并行型计算。     

并行多块结构重叠网格装配算法及应用

针对多块结构重叠网格并行装配的问题,设计了支持初始网格系统细分的多块结构重叠网格框架,并在此框架基础上提出了基于局部洞映射的并行挖洞算法、格心网格下可跨块寻点的并行搜索算法,使之可适应大规模并行数值模拟时的分布式计算环境。此算法被模块化的集成到了自主研发的大规模多块结构网格数值求解器（CCFD-MGMB）中,可支持大规模并行非定常多体分离数值模拟。并行测试结果表明,本文发展的算法具有良好的局部数据结构组织,数据可扩展性强。数值应用模拟结果表明了该算法的有效性及正确性,千核并行非定常数值计算效率（相对于64核）可达58%。     

基于网格的混合微粒群算法解决任务调度问题

网格任务分配是一个NP难问题,结合微粒群优化（Particle Swarm Optimization,PSO）算法,和网格自身的特性,提出了基于网格的混合微粒群算法。算法对问题的解空间进行变换、重定义,使之更加符合PSO算法的求解环境,实现了网格资源的优化分配。与离散微粒群（DPSO）算法和遗传算法进行了仿真比较,结果表明,新的PSO算法具有较好的性能。     

一个数据并行型计算的网格资源的管理模型

在由多计算机集群构成的数据网格环境下，挖掘网格计算节点的空余资源来支持数据并行型计算（Data Parallel Computing，DPC），提出了一个基于分类、统计机制的数据网格管理模型。根据不同时间的网格资源的空余、各类DPC以及逻辑计算机机群，研究了支持DPC的网格资源管理模型。实验表明，该模型有效地解决了网格环境下数据并行型计算所需的空余资源优化使用问题。     

一种分布式网格计算框架以及在大规模分子动力学模拟中的应用

分子动力学（molecular dynamics）模拟蛋白质等大分子内原子间的相互作用,蛋白质折叠所需的时间通常在微秒（10^-6s）量级，而进行模拟的时间步长在飞秒（10^-15s）量级，并且每步需要计算大量的相互作用（O（n^2），n为原子数），以致于无法模拟足够长时间的折叠过程．现今在满足精确度的需求下没有更好的模拟算法．最近，生物学家研究了一种分布式的动力学方法，使得可以利用分布在Internet上的计算机进行并行模批成为可能,本文的目标是设计并实现在分布式P2P和网格计算环境等多种异构计算资源下进行动力学模拟的可靠框架，以便更大限度地利用计算资源，加快计算过程．我们基于Java和web service技术，已经实现了对应用透明的计算框架，并已将它扩展到我们的网格计算环境,实验表明分子动力学模拟程序在该框架下运行良好．     

基于网格的并行FFT计算研究

快速傅里叶变换（FFT）在科学和工程领域有着广泛的应用。在网格环境下进行并行FFT计算可以提高运算速度,促进FFT的应用。在介绍了网格计算发展状况的基础上,详细阐述了基于网格的分布式并行计算。实验以FFT算法为背景,在Globus Toolkit 4平台下实现了并行FFT计算,并对实验数据作了分析,说明了基于网格的并行FFT计算的可行性。最后指出网格资源调度对并行计算的重要性。     

基于经济模型的资源代理

网格计算是为解决大规模资源密集型问题而提出的新一代计算平台,是当前并行和分布处理技术的一个发展方向,而资源管理是计算网格的关键技术之一。对各种各样可利用资源的整合和管理是网格应用的基础,而资源的分布性、动态性、异构性、自治性和需要协调一致性使得网格资源的管理调度成为一个棘手的问题。目前基于市场的经济资源管理和调度算法非常适合计算网格中的资源管理问题,但有调度价格不能更改、负载平衡等问题。文中提出了“网格环境下基于经济模型的资源代理”,依靠多维QoS指导的调度策略和经济模型的启发式调节资源价格,改进和优化计算网格资源的分配。     

基于自适应线程束的GPU并行粒子群优化算法

基于统一计算设备架构（CUDA）对图形处理器（GPU）下的并行粒子群优化（PSO）算法作改进研究。根据CUDA的硬件体系结构特点,可知Block是串行执行的,线程束（Warp）才是流多处理器（SM）调度和执行的基本单位。为了充分利用Block中线程的并行性,提出基于自适应线程束的GPU并行PSO算法：将粒子的维度和线程相对应;利用GPU的Warp级并行,根据维度的不同自适应地将每个粒子与一个或多个Warp相对应;自适应地将一个或多个粒子与每个Block相对应。与已有的粗粒度并行方法（将每个粒子和线程相对应）以及细粒度并行方法（将每个粒子和Block相对应）进行了对比分析,实验结果表明,所提出的并行方法相对前两种并行方法,CPU加速比最多提高了40。     

一种带数据整数规划的新型并行自平衡PSO算法

为有效解决标准粒子群(PSO)算法在进化后期缺乏多样性且精度不高的问题,利用多核系统及实际高校地理数据,给出一种高校数据的整数规划方法及并行自平衡PSO算法模型来并行求解高校路网问题,同时体现算法性能。将自平衡机制采用多核系统并行处理方式生成相互独立的子群体,每个子群体间并行求解,最终生成主群体最优路径即高校路网。在Visual Studio2005.NET环境下用C++编程实现仿真。实验结果表明,此算法从求解精度及计算时间两个重要方面综合改善了算法性能。     

云计算环境下基于可靠性感知的任务调度算法

针对云计算环境下并行任务易受资源失效的影响而无法完成,且动态提供云资源可靠性较低的问题,首先,引入失效恢复机制,由于在失效可恢复情况下资源失效规律动态变化,使用两参数Weibull分布对不同时段资源节点和通信链路失效规律的局部特征进行描述;然后,根据并行任务之间存在的各类交互关系分析,提出了一种基于变参数失效规则的资源可靠性评估模型;最后,将该模型并入粒子群算法得到基于可靠性感知的自适应惯性权重粒子群资源调度算法R PSO,从而在计算适应度时充分考虑备选资源的可靠程度。仿真实验结果表明,当选择了合适的失效恢复参数时,提出的R PSO算法能够大幅度提高云服务可靠性,且只会增加少量的额外失效恢复开销。     

MQPSO: 一种具有多群体与多阶段的QPSO算法*

提出了一种改进的QPSO(Quantum-behaved Particle Swarm Optimization)算法,即一种具有多群体与多阶段的具有量子行为的粒子群优化算法.在该算法中,粒子被分为多个群体,利用多个阶段进行全局搜索,这样可以有效地避免粒子群早熟,提高了算法的全局收敛性能.对几个重要测试函数的测试结果证明,MQPSO算法的收敛性能优于标准粒子群算法(Standard Particle Swarm Optimization, SPSO)以及QPSO算法.     

The Triangular Pyramid Scheduling Model and algorithm for PDES in Grid

Grid is a perfect environment for the large scale Parallel Discrete Event Simulation (PDES), because its distribution and collaboration features match the PDES applications well. The PDES tasks or applications are modeled as a Directed Acyclic Graph (DAG), in which the simulation resources consist of three critical factors, simulation hosting machine (SHM), simulation service (SS) and simulation data (SD) in Grid environment. By solving the model we attempt to obtain an optimized triangular matching of the simulation resources on Grid, so that it can support the PDES activities better. We name the algorithm of solving the model Triangular Pyramid Scheduling (TPS). The PDES DAG is divided into three basic graph structures: Sequential structure, Fork structure, and Join structure. The TPS algorithm is developed based on these graph structures. The simulation results show that TPS algorithm can reduce the makespan and congestion, improve the simulation efficiency, and increase the resource utilization efficiency, compared to the existing algorithms.     

An algorithmic framework of discrete particle swarm optimization

Particle swarm optimization (PSO) was originally developed for continuous problem. To apply PSO to a discrete problem, the standard arithmetic operators of PSO are required to be redefined over discrete space. In this paper, a concept of distance over discrete solution space is introduced. Under this notion of distance, the PSO operators are redefined. After reinterpreting the composition of velocity of a particle, a general framework of discrete PSO algorithm is proposed. As a case study, we illustrate the application of the proposed discrete PSO algorithm to number partitioning problem (NPP) step by step. Preliminary computational experience is also presented. The successful application shows that the proposed algorithmic framework is feasible.     

HPC on the Grid: The Theophys Experience

The Grid Virtual Organization (VO) “Theophys”, associated to the INFN (Istituto Nazionale di Fisica Nucleare), is a theoretical physics community with various computational demands, spreading from serial, SMP, MPI and hybrid jobs. That has led, in the past 20 years, towards the use of the Grid infrastructure for serial jobs, while the execution of multi-threaded, MPI and hybrid jobs has been performed in several small-medium size clusters installed in different sites, with access through standard local submission methods. This work analyzes the support for parallel jobs in the scientific Grid middlewares, then describes how the community unified the management of most of its computational need (serial and parallel ones) using the Grid through the development of a specific project which integrates serial e parallel resources in a common Grid based framework. A centralized national cluster is deployed inside this framework, providing “Wholenodes” reservations, CPU affinity, and other new features supporting our High Performance Computing (HPC) applications in the Grid environment. Examples of the cluster performance for relevant parallel applications in theoretical physics are reported, focusing on the different kinds of parallel jobs that can be served by the new features introduced in the Grid.     

网格环境下费用约束的科学工作流可靠调度算法

网格基础设施是目前科学工作流应用规划、部署和执行的主要支撑环境.然而由于网格资源的自治、动态及异构性,如何在保障用户QoS约束下有效调度科学工作流是一个研究热点.针对费用约束下的科学工作流调度问题,为了提高其执行的可靠性,本文使用随机服务模型描述资源节点的动态服务能力并考虑本地任务负载对资源执行性能的影响,给出一种资源可靠性的评估方法,在此基础上提出一种费用约束下的科学工作流可靠调度算法RSASW.仿真实验结果表明RSASW算法相对于GAIN3,GreedyTime-CD及PFAS算法,对工作流的执行具有很好的可靠性保障.     

多量子粒子群协同优化算法研究

针对标准粒子群优化（PSO）算法及其改进算法存在的局部收敛与收敛速度问题,提出了一种多量子粒子群协同优化（QPSCO）方法。该算法采用双层的多粒子群协同优化结构：用多个量子粒子群在底层独立地搜索解空间,同时引入参数变异策略,以扩大搜索范围;上层用1个量子粒子群追逐当前全局最优解,并对飞离搜索区域粒子的位置用新位置取代,以加快算法收敛。在此基础上,将该算法应用于实际控制系统低阶时滞对象的PID控制器设计中。仿真结果表明,QPSCO是一种有效的参数优化算法,与标准PSO、QPSO等算法相比具有更好的全局收敛性能。     

Enhancing Availability of Grid Computational Services to Ubiquitous Computing Applications

The Grid is an integrated infrastructure that can play the dual roles of a coordinated resource consumer as well as a donator in distributed computing environments. The enormous growth in the use of mobile and embedded devices in ubiquitous computing environment and their interaction with human beings produces a huge amount of data that need to be processed efficiently anytime anywhere. However, such devices often have limited resources in terms of CPU, storage, battery power, and communication bandwidth. Thus, there is a need to transfer ubiquitous computing application services to more powerful computational resources. In this paper, we investigate the use of the Grid as a candidate for provisioning computational services to applications in ubiquitous computing environments. In particular, we present a competitive model that describes the possible interaction between the competing resources in the Grid Infrastructure as service providers and ubiquitous applications as subscribers. The competition takes place in terms of quality of service (QoS) and cost offered by different Grid Service Providers (GSPs). We also investigate the job allocation of different GSPs by exploiting the noncooperativeness among the strategies. We present the equilibrium behavior of our model facing global competition under stochastic demand and estimate guaranteed QoS assurance level by efficiently satisfying the requirement of ubiquitous application. We have also performed extensive experiments over Distributed Parallel Computing Cluster (DPCC) and studied overall job execution performance of different GSPs under a wide range of QoS parameters using different strategies. Our model and performance evaluation results can serve as a valuable reference for designing appropriate strategies in a practical grid environment.     

Swarm Intelligence Approaches for Grid Load Balancing

With the rapid growth of data and computational needs, distributed systems and computational Grids are gaining more and more attention. The huge amount of computations a Grid can fulfill in a specific amount of time cannot be performed by the best supercomputers. However, Grid performance can still be improved by making sure all the resources available in the Grid are utilized optimally using a good load balancing algorithm. This research proposes two new distributed swarm intelligence inspired load balancing algorithms. One algorithm is based on ant colony optimization and the other algorithm is based on particle swarm optimization. A simulation of the proposed approaches using a Grid simulation toolkit (GridSim) is conducted. The performance of the algorithms are evaluated using performance criteria such as makespan and load balancing level. A comparison of our proposed approaches with a classical approach called State Broadcast Algorithm and two random approaches is provided. Experimental results show the proposed algorithms perform very well in a Grid environment. Especially the application of particle swarm optimization, can yield better performance results in many scenarios than the ant colony approach.