检索结果-内蒙古大学图书馆

HPC Simulation of Non-Linear Processes in Microsystems Gas-Metal

LOBACHEVSKII JOURNAL OF MATHEMATICS 2020年第8期41卷 1554-1562页

作者： Polyakov, S. V. Podryga, V. O. Kudryashova, T. A. Russian Acad Sci Keldysh Inst Appl Math Moscow 125047 Russia

The present work is devoted to the development and use of supercomputer technology for modeling gas-dynamic processes using molecular dynamics methods. The relevance of the work is associated with the development of nanotechnology for the manufacture of promising nanomaterials and nanocoatings by the method of supersonic cold gasdynamic spraying. The paper proposes to consider gasdynamic processes at the atomic-molecular level. It is proposed to use Newton's equations for particle dynamics as a mathematical model. The numerical implementation of the model is based on the well-known Verlet scheme. The software implementation of the numerical scheme is adapted for parallel computing on modern supercomputers, including those with a hybrid architecture. To illustrate the developed supercomputer technology, the problem of calculating the parameters of the flow of nitrogen in a microchannel with nickel walls is considered. The conducted computational experiments demonstrated the effectiveness of the developed technology.

关键词： high performance computing modeling of nonlinear processes at the gas– metal interface molecular dynamics parallel algorithms domain decomposition technique load balancing

来源：评论

学校读者我要写书评

暂无评论

Fast Scalable Approximate Nearest Neighbor Search for High-dimensional Data 22

Fast Scalable Approximate Nearest Neighbor Search for High-d...

引用

IEEE International Conference on Cluster Computing

作者： Bashyam, K. G. Renga Vadhiyar, Sathish Indian Inst Sci Dept Computat & Data Sci Bengaluru India

ISBN: (纸本)9781728166773

K-Nearest Neighbor (k-NN) search is one of the most commonly used approaches for similarity search. It finds extensive applications in machine learning and data mining. This era of big data warrants efficiently scaling k-NN search algorithms for billion-scale datasets with high dimensionality. In this paper, we propose a solution towards this end where we use vantage point trees for partitioning the dataset across multiple processes and exploit an existing graph-based sequential approximate k-NN search algorithm called HNSW (Hierarchical Navigable Small World) for searching locally within a process. Our hybrid MPI-OpenMP solution employs techniques including exploiting MPI one-sided communication for reducing communication times and partition replication for better load balancing across processes. We demonstrate computation of k-NN for 10,000 queries in the order of seconds using our approach on similar to 8000 cores on a dataset with billion points in an 128-dimensional space. We also show 10X speedup over a completely k-d tree-based solution for the same dataset, thus demonstrating better suitability of our solution for high dimensional datasets. Our solution shows almost linear strong scaling.

关键词： K-NN Search parallel algorithms Load Balancing Vantage Point Tree HNSW

来源：评论

学校读者我要写书评

暂无评论

Vectorising k-Core Decomposition for GPU Acceleration 2020

Vectorising k-Core Decomposition for GPU Acceleration

引用

32nd International Conference on Scientific and Statistical Database Management (SSDBM)

作者： Mehrafsa, Amir Chester, Sean Thomo, Alex Univ Victoria Victoria BC Canada

ISBN: (纸本)9781450388146

k-Core decomposition is a well-studied community detection problem in graph analytics in which each k-core of vertices induces a subgraph where all vertices have degree at least k. The decomposition is expensive to compute on large graphs and efforts to apply massive parallelism have had limited success. This paper presents a vectorisation of the problem that reframes it as a composition of vector primitives on flat, 1d arrays. With such a formulation, we can deploy highly optimised Deep Learning GPU and SIMD frameworks. On a moderate GPU, using PyTorch, we obtain up to 8x improvement over the best parallel state-of-the-art implemented in C++ and running on an expensive 32-core machine. More importantly, our approach represents a novel abstraction showing that redesigning graph operations as a series of vectorised primitives makes highly-parallel analytics both easier and more accessible for developers. We posit that such an approach can vastly accelerate the use of cheap GPU hardware in complex graph analytics.

关键词： graph analytics k-core decomposition parallel algorithms vectorization GPGPU SIMD PyTorch

来源：评论

学校读者我要写书评

暂无评论

Stairway to Abstraction: an Iterative Algorithm for Whisker Detection in Video Frames 11

Stairway to Abstraction: an Iterative Algorithm for Whisker ...

引用

11th Latin American Symposium on Circuits & Systems (LASCAS)

作者： Betting, Jan-Harm L. F. Romano, Vincenzo Bosman, Laurens W. J. Al-Ars, Zaid De Zeeuw, Chris I. Strydis, Christos Erasmus MC Rotterdam Netherlands

ISBN: (纸本)9781728134277

Automated whisker tracking is important for researching active touch in rodents. Earlier efforts to detect whiskers and represent them in a small set of parameters were either not accurate enough to enable tracking over time, or computationally expensive. In this article we propose an algorithm to cluster whisker centerline points, detected through a curvilinear structure algorithm, using the shape of smaller clusters to form bigger clusters of centerline points. After that, a least-squares approach is used to define each whisker by a set of four parameters. We implemented the algorithm in MATLAB in a parallelized fashion, and found that the processing time per frame is reasonable in MATLAB, and is likely to be short when ported to a lower-level language. When tested on a 33,634-frame segment, 89.2% of the whiskers could be represented in an abstract fashion by four parameters with a mean-squares fitting error of lower than 10 pixels, and visual inspection shows that crossing whiskers are detected and parameterized in an accurate way.

关键词： Biology computing Clustering methods High performance computing Image processing MATLAB parallel algorithms parallel processing

来源：评论

学校读者我要写书评

暂无评论

parallel Three-Dimensional Distance Transform for Railway Alignment Optimization Using OpenMP

引用

JOURNAL OF TRANSPORTATION ENGINEERING PART A-SYSTEMS 2020年第5期146卷

作者： Song, Taoran Pu, Hao Schonfeld, Paul Li, Wei Zhang, Hong Ren, Yuhan Wang, Jie Hu, Jianping Peng, Xianbao Cent South Univ Sch Civil Engn Natl Engn Lab High Speed Railway Construct Changsha 410075 Peoples R China Univ Maryland Dept Civil & Environm Engn College Pk MD 20742 USA China Railway First Survey & Design Inst Grp Co L State Key Lab Rail Transit Engn Informatizat Xian 710043 Peoples R China China Railway Eryuan Engn Grp Co Ltd 3 Tongjin Rd Chengdu 610031 Peoples R China China Railway Siyuan Survey & Design Grp Co Ltd 745 Heping Ave Wuhan 430063 Peoples R China

Railway alignment optimization is a large-scale and time-consuming civil engineering problem. To solve it, a three-dimensional distance transform (3D-DT) algorithm, which is a variant of the three-dimensional Euclidean distance transform (3D-EDT), was previously designed. However, that algorithm is quite computationally intensive. In addition, the 3D-DT is inherently sequential, and it is thus challenging to parallelize. Thus, this study focuses on improving the sequential 3D-DT by transforming it into a parallel one. First, existing representative parallel EDT methods are reviewed and assessed. Then the railway alignment optimization model and the sequential 3D-DT are described. After that, critical execution properties of the 3D-DT that significantly influence its parallelization are explored in depth. Lastly, a novel so-called parallel linkage method is presented. This parallel implementation, which is developed using the OpenMP library, is highly effective and scalable by fully exploiting the parallelism of the algorithm. Using this parallel 3D-DT method, a large-scale, real-world railway case is tested and analyzed in detail. The outcomes verify that the proposed parallel method can accelerate the optimization process significantly without reducing the quality of computation results.

关键词： Alignment optimization parallel algorithms Three-dimensional distance transform OpenMP

来源：评论

学校读者我要写书评

暂无评论

Multiscale Simulation of Gas Cleaning Processes

引用

Mathematical Models and Computer Simulations 2020年第3期12卷 302-315页

作者： Polyakov, S.V. Karamzin, Yu. N. Kudryashova, T.A. Podryga, V.O. Puzyrkov, D.V. Tarasov, N.I. Keldysh Institute of Applied Mathematics Russian Academy of Sciences Moscow 125047 Russian Federation Moscow Automobile and Road Construction State Technical University Moscow 125319 Russian Federation

Abstract—: This paper considers modeling the processes of cleaning air from finely dispersed solid contaminants clustered in the form of nanoparticles. The purification technology chosen for the study involves the use of a system consisting of nanofilters and sorbents. Both cleaning methods used in it are currently in high demand and are often combined in appropriate devices. The first cleaning method using nanofilters ensures a high purification quality. However, this method is expensive as it requires frequent replacement of the filter elements (membranes) and the disposal of these elements. The second method of cleaning with sorbents gives a relatively low quality of cleaning but enables multiply repeated a purification procedure after washing the sorbent with special liquids. The optimization of air cleaning devices using nanofilters and sorbents requires a detailed investigation of the processes occurring in the cleaning system. The proposed study addresses part of the problem associated with the passage of an air flow containing solid contaminant nanoparticles through a layer of granular sorbent. For this purpose, a multiscale mathematical model, a numerical algorithm, and a parallel implementation of the model on a macroscopic scale have been developed. The novelty of the approach consists in the use of a quasigasdynamic model for describing the flow in the absorbing layer and in the proposed multiscale formulation of the problem. The preliminary calculations based on the macromodel showed the efficiency of the proposed approach. © 2020, Pleiades Publishing, Ltd.

关键词： high performance computing multiscale simulation numerical schemes on unstructured grids parallel algorithms processes of gas cleaning

来源：评论

学校读者我要写书评

暂无评论

DNA Sequence Alignment Using Dynamic Programming 3rd

DNA Sequence Alignment Using Dynamic Programming

引用

3rd International Conference on Intelligent Computing and Communication (ICICC)

作者： Singh, Niharika Rajput, Gaurav Dixit, Yash Sehgal, Aastha Univ Petr & Energy Studies Sch CS Dehra Dun Uttarakhand India GL Bajaj Inst Technol & Management Dept CS&E Greater Noida Utter Pradesh India UPES Sch Comp Sci Dept Informat Dehra Dun Uttarakhand India

ISBN: (纸本)9789811510847;9789811510830

In today's time, Molecular Science is increasingly dependent on Software Engineering Calculations in the department of Research and Development. Aligning sequences of DNA, RNA is becoming a major part of present-day natural sciences. A hereditary database holds a large amount of unprocessed data that resides very crucial information. A single pair of chromosomes contains approximately 3 billion DNA base pairs. To look through this information, retrieve the connections, and subconnections in it is a very slow procedure. Therefore, scientists are looking forward toward computer science algorithms for faster retrieval of information. This paper is focused upon using a parallel programming algorithm than the previous alignment algorithms.

关键词： DNA Genetics parallel algorithms Thread

来源：评论

学校读者我要写书评

暂无评论

Work-Efficient Batch-Incremental Minimum Spanning Trees with Applications to the Sliding-Window Model 20

Work-Efficient Batch-Incremental Minimum Spanning Trees with...

引用

32nd ACM Symposium on parallelism in algorithms and Architectures (SPAA)

作者： Anderson, Daniel Blelloch, Guy E. Tangwongsan, Kanat Carnegie Mellon Univ Pittsburgh PA 15213 USA Mahidol Univ Int Coll Bangkok Thailand

ISBN: (纸本)9781450369350

algorithms for dynamically maintaining minimum spanning trees (MSTs) have received much attention in both the parallel and sequential settings. While previous work has given optimal algorithms for dense graphs, all existing parallel batch-dynamic algorithms perform polynomial work per update in the worst case for sparse graphs. In this paper, we present the first work-efficient parallel batch-dynamic algorithm for incremental MST, which can insert l edges in O(l lg(1 + n/l)) work in expectation and O(polylog(n)) span w.h.p. The key ingredient of our algorithm is an algorithm for constructing a compressed path tree of an edge-weighted tree, which is a smaller tree that contains all pairwise heaviest edges between a given set of marked vertices. Using our batch-incremental MST algorithm, we demonstrate a range of applications that become efficiently solvable in parallel in the sliding-window model, such as graph connectivity, approximate MSTs, testing bipartiteness, k-certificates, cycle-freeness, and maintaining sparsifiers.

关键词： dynamic algorithms minimum spanning trees dynamic graphs parallel algorithms incremental algorithms

来源：评论

学校读者我要写书评

暂无评论

Distributed Graph Analytics 16th

Distributed Graph Analytics

引用

16th International Conference on Distributed Computing and Internet Technology (ICDCIT)

作者： Srikant, Y. N. Indian Inst Sci Bengaluru 560012 India

ISBN: (纸本)9783030369873;9783030369866

Graph Analytics is important in different domains: social networks, computer networks, and computational biology to name a few. This paper describes the challenges involved in programming the underlying graph algorithms for graph analytics for distributed systems with CPU, GPU, and multi-GPU machines and how to deal with them. It emphasizes how language abstractions and good compilation can ease programming graph analytics on such platforms without sacrificing implementation efficiency.

关键词： Graph analytics Social networks parallel algorithms Multi-core processors GPU computation Graph frameworks Graph processing languages

来源：评论

学校读者我要写书评

暂无评论

A Submatrix-Based Method for Approximate Matrix Function Evaluation in the Quantum Chemistry Code CP2K

A Submatrix-Based Method for Approximate Matrix Function Eva...

引用

International Conference on High Performance Computing, Networking, Storage and Analysis (SC)

作者： Lass, Michael Schade, Robert Kuhnet, Thomas D. Plessl, Christian Paderborn Univ Paderborn Ctr Parallel Comp Warburger Str 100 D-33098 Paderborn Germany Paderborn Univ Dept Comp Sci Warburger Str 100 D-33098 Paderborn Germany Paderborn Univ Dept Chem Warburger Str 100 D-33098 Paderborn Germany

ISBN: (纸本)9781728199986

Electronic structure calculations based on density-functional theory (DFT) represent a significant part of today's HPC workloads and pose high demands on high-performance computing resources. To perform these quantum-mechanical DFT calculations on complex large-scale systems, so-called linear scaling methods instead of conventional cubic scaling methods are required. In this work, we take up the idea of the submatrix method and apply it to the DFT computations in the software package CP2K. For that purpose, we transform the underlying numeric operations on distributed, large, sparse matrices into computations on local, much smaller and nearly dense matrices. This allows us to exploit the full floating-point performance of modern CPUs and to make use of dedicated accelerator hardware, where performance has been limited by memory bandwidth before. We demonstrate both functionality and performance of our implementation and show how it can he accelerated with GPM and FPGAs.

关键词： Accelerator architectures Approximate computing Approximation algorithms Chemistry Density functional theory Linear algebra Open source software parallel algorithms Reconfigurable architectures Scientific computing Solid-state physics

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：