An original approximate method for estimating the invariant probability distribution of a large class of multi-dimensional nonlinear stochastic oscillators

This paper proposes an original method for obtaining analytical approximations of the invariant probability density function of multi-dimensional Hamiltonian dissipative dynamic systems under Gaussian white noise excitations, with linear non-conservative parts and nonlinear conservative parts. The method is based on an exact result and a heuristic argument. Its pertinence is attested by numerical tests.     

Three-dimensional MHD high-resolution computations with CWENO employing adaptive mesh refinement

Until recently, numerical simulations of discontinuities in highly super-Alfvénic plasmas have been severely limited by comparatively crude resolution and accuracy. Significant progress in the numerical simulation of such plasmas was achieved with the recently implemented Central Weighted Essentially Non-Oscillatory (CWENO) scheme. Combining this technique with that of adaptive mesh refinement (AMR), we have developed a third-order numerical scheme, which is able to efficiently capture strong gradients on spatial scales being small compared to the overall scale of the plasma system considered. Here, we first describe important algorithmic aspects of the scheme as well as the physics included in it. Second, we present the results of various performance tests. And, third, we illustrate its application to ‘real world problems’ using the example of the dynamics of a Sedov-type explosion.     

Information systems use as strategy practice: A multi-dimensional view of strategic information system implementation and use

Information systems (IS) are strategic in so far as they are used to realize strategic intent. Yet, while much has been said about aligning IS functionality with the strategic intent and how to organizationally implement strategically aligned systems, less is known of how to successfully implement strategic change associated with system use – a truly critical challenge within strategic IS implementation. Drawing on a strategy-as-practice perspective we address this gap by developing a multi-dimensional view of IS strategy, conceptualizing three key challenges in the IS strategy process, to explain how and why a paper mill, despite successfully implementing a strategic production management system, failed to produce intended strategic change. We call this outcome strategy blindness: organizational incapability to realize the strategic intent of implemented, available system capabilities. Using a longitudinal case study we investigate how cognitive rigidity of key actors and fixed, interrelated practices shaped the implementation of the new production system. We also identify core components and dynamics that constitute a richer multi-dimensional view of the IS strategy implementation (alignment) process. In particular, we identify three salient factors that contribute to strategy blindness – mistranslation of intent, flexibility of the IT artifact and cognitive entrenchment – and discuss how they affect strategic implementation processes. We conclude by discussing implications of our findings for IS strategy theory and practice, especially the contribution of strategy-as-practice to this stream of research.     

面向批量处理的大数据检索过滤模型研究

大数据作为新的战略资源,在信息领域发挥着重要作用。大数据的检索规模往往达到十亿甚至百亿级,导致传统的查询机制效率低下成为常态。因此,提高大数据的查询效率、降低查询负担成为大数据研究的重要方面。为 此提出了一种面向批量处理的大数据检索过滤模型IMFM,介绍了其核心思想及工作原理,论证了IMFM对于多维查询的支持,并给出了IMFM的部署策略。在大数据索引结构中的适当位置部署该模型,在检索请求通过节点时对检索请求进行快速过滤,避免无关请求对节点下方索引结构的操作,从而降低检索对性能的消耗。实验证明,在大数据批量处理环境下,该模型可以有效缩短大数据一维和多维查询的路径长度,提高检索效率,大幅减轻大数据存储和处理平台的负担。     

基于人工免疫的多维关联规则挖掘及其应用研究

关联规则挖掘是一个重要的数据挖掘问题.目前,关于单维关联规则的成果已经比较成熟,但是对于多维关联规则问题因为存在显著的组合爆炸问题,至今尚未完美解决.提出了一种基于人工免疫的多维关联规则挖掘算法.算法充分利用了人工免疫的记忆特性,把挖掘的关联规则存入记忆库,加快了多维关联规则的挖掘速度.结果表明,该算法应用于煤与瓦斯突出预测中,具有较好的鲁棒性,能快速、有效地进行全局优化搜索,在多维关联规则的挖掘中具有可行性和高效性.     

基于多维金字塔表达和AdaBoost的高分辨率SAR图像城区场景分类算法

提出了多维金字塔表达算法, 并使用基于多维金字塔表达的AdaBoost实现了高分辨率合成孔径雷达(Synthetic aperture radar, SAR)图像的城区场景分类. 多维金字塔表达算法首先在局部特征的各维计算金字塔表达矢量, 再将所有的金字塔表达矢量连接起来构成多维金字塔表达矢量. 多维金字塔表达算法克服了金字塔表达算法在处理高维局部特征时, 遇到的输出金字塔表达矢量的区分力受计算效率制约的问题. 本文分别在一个TerraSAR-X图像库和一张大幅TerraSAR-X图像上比较基于金字塔表达的AdaBoost和基于多维金字塔表达的AdaBoost的分类性能. 实验结果表明, 与前者相比, 后者显著提高了计算效率同时保证了分类精度.     

Fast,effective heuristics for the 0–1 multi-dimensional knapsack problem

The objective of the multi-dimensional knapsack problem (MKP) is to find a subset of items with maximum value that satisfies a number of knapsack constraints. Solution methods for MKP, both heuristic and exact, have been researched for several decades. This paper introduces several fast and effective heuristics for MKP that are based on solving the LP relaxation of the problem. Improving procedures are proposed to strengthen the results of these heuristics. Additionally, the heuristics are run with appropriate deterministic or randomly generated constraints imposed on the linear relaxation that allow generating a number of good solutions. All algorithms are tested experimentally on a widely used set of benchmark problem instances to show that they compare favourably with the best-performing heuristics available in the literature.     

Multiagent System Simulations of Signal Averaging in English Auctions with Two-Dimensional Value Signals

Computational experiments with a multiagent system show that bidders use signal averaging to avoid the winner’s curse in English auctions. The results vary with the percent of common value in a two-dimensional value signal, information levels, uncertainty, and the number of bidders. The complexity introduced by the combinations of these factors affects the bidding strategies and auction outcomes in interesting ways—usually nonlinearly and sometimes non-monotonically. Of main concern to a seller is the effect of these factors on revenue. I find that revenue increases with the percent of common value in the two-dimensional value signal, decreases with uncertainty, and increases with the number of bidders. There is very little impact of information level on revenue when values are pure private and pure common. However, for the intermediate cases of two-dimensional value signals, revenue decreases with increased information.      

太阳辐射计检定用多维工作台的研制

针对太阳辐射计量仪表进行检定的需要,提出了一种具有四个自由度功能动作的多维检定工作台。工作台在与太阳模拟器联合,可同时对两块待栓仪表的参数误差进行检定。在对太阳高度角和方位角的研究基础上,阐述了工作台旋转机构、升降机构、俯仰机构和转臂机构的设计过程,并对工作台的控制系统进行了说明。测试结果表明多维工作台各级转动误差小于0．05°,由倾角仪和步进电机组成的台面水平控制误差在±10’以内,满足设计指标要求。     

分布式多维标度定位算法的研究

针对经典MDS-MAP定位算法在定位精度和算法复杂度方面的不足,提出一种分布式多维标度定位算法。改进后的算法加入了分簇的思想,将大规模网络分成多个具有簇首的局部网络。局部定位时,引入Hop-Euclidean算法,计算簇内节点间距离,再用局部网络融合算法将局部相对坐标图合并成全局相对坐标图。仿真分析表明,提出的算法在各向同性和各向异性网络中都有很好的定位精度,而且在定位精度提高的情况下可用于不规则网络,有利于网络的扩展,更适用于大规模密集型网络。