Parallel depth first search. Part I. Implementation

This paper presents a parallel formulation of depth-first search which retains the storage efficiency of sequential depth-first search and can be mapped on to anyMIMD architecture. To study its effectiveness it has been implemented to solve the 15-puzzle problem on three commercially available multiprocessors—Sequent Balance 21000, the Intel Hypercube and BBN Butterfly. We have been able to achieve fairly linear speedup on Sequent up to 30 processors (the maximum configuration available) and on the Intel Hypercube andBBN Butterfly up to 128 processors (the maximum configurations available). Many researchers considered the ring architecture to be quite suitable for parallel depth-first search. Our experimental results show that hypercube and sharedmemory architectures are significantly better.At the heart of our parallel formulation is a dynamic work distribution scheme that divides the work between different processors. The effectiveness of the parallel formulation is strongly influenced by the work distribution scheme and architectural features such as presence/absence of shared memory, the diameter of the network, relative speed of the communication network, etc. In a companion paper,(¹) we analyze the effectiveness of different load-balancing schemes and architectures, and also present new improved work distribution schemes.This work was supported by Army Research Office Grant No. DAAG29-84-K-0060 to the Artificial Intelligence Laboratory, and Office of Naval Research Grant N00014-86-K-0763 to the Computer Science Department at the University of Texas ay Austin     

Parallel state-space search for a first solution with consistent linear speedups

Consider the problem of exploring a large state-space for a goal state where although many such states may exist in the state-space, finding any one state satisfying the requirements is sufficient. All the methods known until now for conducting such search in parallel using multiprocessors fail to provide consistent linear speedups over sequential execution. The speedups vary between sublinear to superlinear and from one execution to another. Further, adding more processors may sometimes lead to a slow-down rather than speedup, giving rise to speedup anomalies reported in literature. We present a prioritizing strategy which yields consistent speedups that are close toP withP processors, and that monotonically increase with the additon of processors. This is achieved by keeping the total number of nodes expanded during parallel search very close to that of a sequential search. In addition, the strategy requires substantially smaller memory relative to other methods. The performance of this strategy is demonstrated on a multiprocessor with several state-space search problems.This research has been supported in part by the National Science Foundation under Contract No. CCR-89-02496.     

多核CPU-GPU协同的并行深度优先算法

针对多核CPU和GPU环境下图的深度优先搜索问题,提出多核CPU中实现并行DFS的新算法,通过有效利用内存带宽来提高性能,且当图增大时优势越明显.在此基础上提出一种混合方法,为DFS每一分支动态地选择最佳的实现:顺序执行;两种不同算法的多核执行;GPU执行.混合算法为每种大小的图提供相对更好的性能,且能避免高直径图上的最坏情况.通过比较多CPU和GPU系统,分析底层架构对DFS性能的影响.实验结果表明,一个高端single-socket GPU系统的DFS执行性能相当于一个高端4-socket CPU系统.     

迷宫搜索算法的比较研究

研究面向搜救的应用,将事故环境抽象为一个迷宫,通过仿真实验比较研究了深度优先搜索算法和三种不同启发式函数的A*算法在Perfect迷宫中的应用,并分别将深度优先搜索算法和A*算法用于实际迷宫中进行实现与比较.在实验中,迷宫环境对机器人是未知的,而由于迷宫环境的特殊性——未知的迷宫环境中很少有不会碰撞的路径,从而增加了机器人搜索的难度.通过仿真实验对比了不同启发式函数的A*算法与深度优先搜索算法的性能,最后得出在迷宫搜索中A*算法要优于深度优先搜索算法;同时,在实际迷宫中实现了深度优先搜索算法与A*算法的搜救应用.     

通过八数码问题比较搜索算法的性能

搜索算法的核心在于搜索策略的制定.一般的搜索算法采用无信息指导的搜索策略,如深度优先搜索(DFS)和宽度优先搜索(BFS),还有一些搜索算法采用了启发式信息指导的搜索策略,如A*算法.不同的搜索策略会使得搜索算法的性能有很大的差异.使用以上3种搜索算法实现八教码问题的求解,分析和比较三者所表现出来的性能,同时指出3种搜索算法的特点和应用范围,最后给出分析结论以指导开发和使用更加高效的搜索策略.     

Parallel tabu search algorithm for the hybrid flow shop problem

The paper deals with the parallel variant of the scheduling algorithm dedicated to the hybrid flow shop problem. The problem derives from practice of automated manufacturing lines, e.g. for printed packages. The overall goal is to design a new algorithm which merges the performance of the best known sequential approach with the efficient exploitation of parallel calculation environments. In order to fulfill the above aim, there are two methods proposed in this paper: the original fast method of parallel calculation of the criterion function and the local neighborhood parallel search method embedded in the tabu search approach. The theoretical analysis, as well as the original implementation, with the use of vector processing instructions SSE2 supported by suitable data organization, are presented below. Numerical properties of the proposed algorithm are empirically verified on the multi-core processor.     

自适应布谷鸟搜索的并行K-means聚类算法*

针对K-means聚类算法受初始类中心影响,聚类结果容易陷入局部最优导致聚类准确率较低的问题,提出了一种基于自适应布谷鸟搜索的K-means 聚类改进算法,并利用MapReduce编程模型实现了改进算法的并行化。通过搭建的Hadoop分布式计算平台对不同样本数据集分别进行10次准确性实验和效率实验,结果表明：(1)聚类的平均准确率在实验所采用的4种UCI标准数据集上,相比原始K-means聚类算法和基于粒子群优化算法改进的K-means聚类算法都有所提高;(2) 聚类的平均运行效率在实验所采用的5种大小递增的随机数据集上,当数据量较大时,显著优于原始K-means串行算法,稍好于粒子群优化算法改进的并行K-means聚类算法。可以得出结论,在大数据情景下,应用该算法的聚类效果较好。     

基于深度优先搜索的快速指纹细化算法改进

对指纹图像的细化算法进行了较深入地研究,分析了两种常用细化算法--快速细化算法和改进的OPTA算法各自的优缺点.针对其中存在的迭代次数多、细化速度慢、图像局部细化不彻底等问题,提取了一种无回溯深度优先搜索的快速指纹细化算法.实验结果表明,该算法在保证对图像完全细化的同时,也具有较快的细化处理速度.     

Parallel breadth-first search algorithms for trees and graphs

Parallel Breadth-First Search (BFS) algorithms for ordered trees and graphs on a shared memory model of a Single Instruction-stream Multiple Data-stream computer are proposed. The parallel BFS algorithm for trees computes the BFS rank of eachnode of an ordered tree consisting of n nodes in time of 0(β log n) when 0(n ^1+1/β) processors are used, β being an integer greater than or equal to 2. The parallel BFS algorithm for graphs produces Breadth-First Spanning Trees (BFSTs) of a directedgraph G having n nodes in time 0(log d.log n) using 0(n ³) processors, where d is the diameter of G If G is a strongly connected graph or a connected undirected graph the BFS algorithm produces n BFSTs, each BFST having a different start node.     

时频分布算法的并行实现

二次时频分布(TFD)是一种常用的非平稳信号分析方法,以最常用的Wigner-Ville分布(WVD)为例,分析了离散时频分布算法实现,针对该算法的复杂性和计算量大的特点,实现了离散信号时频分布的并行算法,通过采用缓冲池和计算结果的压缩传输,同时减小了并行计算中传输数据总量和传输次数,将通信开销从O(N2)降低到O(N),解决了并行时频分布计算中数据传输“瓶颈”问题。     

基于深度优先贪婪搜索的可重构硬件任务划分算法

针对可重构计算硬件任务划分通信成本较小化的问题,提出了一种基于深度优先贪婪搜索划分(DFGSP)算法。首先,从待调度的就绪队列中取出队首任务,在某一硬件面积约束下,按深度优先搜索(DFS)方式扫描一个计算密集型任务转换来的有向无环图(DAG),逐个划入满足要求的节点;然后,一遇到不满足面积要求的任务节点时,就计算当前划分模块间输出边数(可量化为通信成本);最后,跳过当前不满足要求的任务节点,继续搜索该点之后处于就绪状态的节点,当搜索到满足要求的点时,按加入该点后不增加当前划分块间输出边数和尽可能填满可重构运算阵列的原则进行。实验结果表明,与现有的簇划分(CBP)、簇层次敏感两种划分算法相比,提出的算法获得了最小划分模块数和平均跨模块间I/O边数最小的均值,通过实际验证,算法显著地改善了硬件任务的划分效果,而且运行开销没有明显增加。     

G.729中固定码本快速搜索算法研究

虽然G.729中采用的集中搜索和G.729a中采用的深度优先树搜索可以有效减少固定码本搜索复杂度,但固定码本搜索在整个语音编码算法中仍占有较大比重.为了在基本维持语音质量的前提下,减少搜索运算量,研究了几种快速搜索算法,脉冲替代和预选替代一个脉冲搜索算法可以大大减少搜索次数,但语音质量明显下降,因此提出每次替代两个脉冲搜索算法,得到比替代一个脉冲较为完整的搜索,产生较好的语音质量.仿真结果表明,该算法可以大大减少搜索运算量,并且保持了和G.729a深度优先树搜索算法相同的语音质量.     

Parallel and distributed local search in COMET

The availability of commodity multiprocessors and high-speed networks of workstations offer significant opportunities for addressing the increasing computational requirements of optimization applications. To leverage these potential benefits, it is important, however, to make parallel and distributed processing easily accessible to a wide audience of optimization programmers. This paper addresses this challenge by proposing parallel and distributed programming abstractions that keep the distance from sequential local search algorithms as small as possible. The abstractions, including parallel loops, interruptions, thread pools, and shared objects, are compositional and cleanly separate the optimization program and the parallel instructions. They have been evaluated experimentally on a variety of applications, including warehouse location and coloring, for which they provide significant speedups.     

Parallel permutation generation on linear array

Given n items, a parallel algorithm for generating all the n! permutations is presented. The computational model used is a linear array which consists of n identical processing elements with a simple structure. One permutation is produced at each other time step. The elapsed time to produce a permutation is independent of the integer n. The basic idea used is the iterative method and the exchange of two consecutive components in an existing permutation. The design procedures of this algorithm are considered in detail. The ranking and unranking functions of the required permutations are also discussed.     

Cuckoo search with varied scaling factor

Cuckoo search (CS), inspired by the obligate brood parasitic behavior of some cuckoo species, iteratively uses Lévy flights random walk (LFRW) and biased/selective random walk (BSRW) to search for new solutions. In this study, we seek a simple strategy to set the scaling factor in LFRW, which can vary the scaling factor to achieve better performance. However, choosing the best scaling factor for each problem is intractable. Thus, we propose a varied scaling factor (VSF) strategy that samples a value from the range [0,1] uniformly at random for each iteration. In addition, we integrate the VSF strategy into several advanced CS variants. Extensive experiments are conducted on three groups of benchmark functions including 18 common test functions, 25 functions proposed in CEC 2005, and 28 functions introduced in CEC 2013. Experimental results demonstrate the effectiveness of the VSF strategy.     

基于矢量量化和深度优先搜索的轨迹分布模式学习算法

在对轨迹流矢量进行量化编码的基础上,提出了一种基于深度优先搜索的轨迹分布模式提取算法,生成了能够描述轨迹分布的序列模式图,并给出了与之相应的异常检测和行为预测方法。对不同场景的可见光和红外序列图像的实验表明,本文方法不仅能够学习轨迹中流矢量的分布,而且能够反映它们之间的时序关系,可以应用于室外复杂场景的目标异常行为检测。     

基于旋转学习策略的共生生物搜索算法

为提高共生生物搜索算法(Symbiotic Organisms Search, SOS)的性能,提出一种基于旋转学习策略的共生生物搜索算法(Symbiotic Organisms Search Using Rotation-Based Learning, RSOS)。该算法将串行个体更新方式改为并行种群更新方式,提高算法收敛速度;引入遍历保优的旋转学习策略,代替寄生机制的盲目随机搜索,增大保留新个体的概率,补充种群多样性,提高算法跳出局部最优的能力。对于8个标准测试函数仿真表明,RSOS算法较基本SOS算法在收敛速度、收敛精度及稳定性上得到了明显提升。     

混合策略改进的乌鸦搜索算法

为了改善乌鸦搜索算法(crow search algorithm,CSA)收敛速度慢、收敛精度不足的问题,提出一种混合策略改进的乌鸦搜索算法(MSCSA).首先在算法运行前期引入tent序列扰动的自适应权重系数,提高算法收敛速度;其次在算法后期引入混合黄金正弦与飞蛾扑火算子,避免算法后期陷入局部最优值;最后通过改进算法的发现概率AP,增加算法的随机性从而提高算法的收敛精度.通过在九个基准函数上对比测试,确定迭代系数的取值,通过Wilcoxon秩和测试验证算法性能.实验结果证明,所提出的MSCSA的性能更为优秀.     

基于引力搜索的布尔函数生成算法

布尔函数是在密码学、纠错编码和扩频通信等领域有着广泛应用的密码函数,寻找性能优良的布尔函数一直是密码学领域的重要问题之一。基于引力搜索算法设计了一种搜索布尔函数的新算法。该算法模仿万有引力定律,以n维空间中的质量点表示布尔函数,以布尔函数的密码特性作为目标适应度函数进行搜索。实验结果表明,算法使用新设计的目标适应度函数可以直接生成具有1阶弹性、1阶扩散准则和高非线性度、高代数次数以及低自相关指标等多种密码学指标的平衡布尔函数,并且进一步给出了直接生成2输出平衡布尔函数的计算机搜索算法。     

A beam search approach for solving type II robotic parallel assembly line balancing problem

In a robotic assembly line, a series of stations are arranged along a conveyor belt and a robot performs on tasks at each station. Parallel assembly lines can provide improving line balance, productivity and so on. Combining robotic and parallel assembly lines ensure increasing flexibility of system, capacity and decreasing breakdown sensitivity. Although aforementioned benefits, balancing of robotic parallel assembly lines is lacking – to the best knowledge of the authors- in the literature. Therefore, a mathematical model is proposed to define/solve the problem and also iterative beam search (IBS), best search method based on IBS (BIBS) and cutting BIBS (CBIBS) algorithms are presented to solve the large-size problem due to the complexity of the problem. The algorithm also tested on the generated benchmark problems for robotic parallel assembly line balancing problem. The superior performances of the proposed algorithms are verified by using a statistical test. The results show that the algorithms are very competitive and promising tool for further researches in the literature.