摘要:
Sparse matrix-vector multiplication SpMV is of singular importance in sparse linear algebra, which is an important issue in scientific computing and engineering practice. Much effort has been put into accelerating SpMV, and a few parallel solutions have been proposed. This paper focuses on a special type of SpMV, namely sparse quasi-diagonal matrix-vector multiplication SQDMV. The sparse quasi-diagonal matrix is the key to solving many differential equations, and very little research has been done in this field. This paper discusses data structures and algorithms for SQDMV that are efficiently implemented on the compute unified device architecture CUDA platform for the fine-grained parallel architecture of the graphics processing unit GPU. A new diagonal storage format, a hybrid of the diagonal format DLA and the compressed sparse row format CSR HDC will be presented, which overcomes the inefficiency of DLA in storing irregular matrices and the imbalances of CSR in storing non-zero elements. Furthermore, HDC can adjust the storage bandwidth of the diagonal to adapt to different discrete degrees of sparse matrix, so as to get a higher compression ratio than DLA and CSR, and reduce the computational complexity. Our implementation in a GPU shows that the performance of HDC is better than that of other formats, especially for matrices with some discrete points outside the main diagonal. In addition, we combine the different parts of HDC to make a unified kernel to get a better compression ratio and a higher speedup ratio in the GPU. Sparse matrix-vector multiplication SpMV is of singular importance in sparse linear algebra, which is an important issue in scientific computing and engineering practice. Much effort has been put into accelerating SpMV, and a few parallel solutions have been proposed. This paper focuses on a special type of SpMV, namely sparse quasi-diagonal matrix-vector multiplication SQDMV. The sparse quasi-diagonal matrix is the key to solving many differential equations, and very little research has been done in this field. This paper discusses data structures and algorithms for SQDMV that are efficiently implemented on the compute unified device architecture CUDA platform for the fine-grained parallel architecture of the graphics processing unit GPU. A new diagonal storage format, a hybrid of the diagonal format DLA and the compressed sparse row format CSR HDC will be presented, which overcomes the inefficiency of DLA in storing irregular matrices and the imbalances of CSR in storing non-zero elements. Furthermore, HDC can adjust the storage bandwidth of the diagonal to adapt to different discrete degrees of sparse matrix, so as to get a higher compression ratio than DLA and CSR, and reduce the computational complexity. Our implementation in a GPU shows that the performance of HDC is better than that of other formats, especially for matrices with some discrete points outside the main diagonal. In addition, we combine the different parts of HDC to make a unified kernel to get a better compression ratio and a higher speedup ratio in the GPU.
作者机构:
[Tan Guan-zheng; Tan Yue] Cent South Univ, Sch Informat Sci & Engn, Changsha 410083, Peoples R China.;[Tan Yue; Deng Shu-guang] Hunan City Univ, Sch Commun & Elect Engn, Yiyang 413000, Peoples R China.
通讯机构:
[Tan Guan-zheng] C;Cent South Univ, Sch Informat Sci & Engn, Changsha 410083, Peoples R China.
关键词:
particle swarm optimization;chaotic search;integer programming problem;mixed integer programming problem
摘要:
A novel chaotic search method is proposed, and a hybrid algorithm combining particle swarm optimization (PSO) with this new method, called CLSPSO, is put forward to solve 14 integer and mixed integer programming problems. The performances of CLSPSO are compared with those of other five hybrid algorithms combining PSO with chaotic search methods. Experimental results indicate that in terms of robustness and final convergence speed, CLSPSO is better than other five algorithms in solving many of these problems. Furthermore, CLSPSO exhibits good performance in solving two high-dimensional problems, and it finds better solutions than the known ones. A performance index (PI) is introduced to fairly compare the above six algorithms, and the obtained values of (PI) in three cases demonstrate that CLSPSO is superior to all the other five algorithms under the same conditions.
作者机构:
[涂立] School of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan 430074, China;[张应征] Department of Information and Engineering, Hunan Engineering Polytechnic, Changsha 410004, China;[涂立; 贾丽媛; 张弛] School of Information Science and Engineering, Hunan City University, Yiyang 413000, China
通讯机构:
School of Computer Science and Technology, Huazhong University of Science and Technology, China
期刊:
Journal of Chemical and Pharmaceutical Research,2014年6(6):2583-2592 ISSN:0975-7384
通讯作者:
Huang, Xuehua
作者机构:
[Jia, Liyuan; Huang, Xuehua; Li, Tu] School of Information Science and Engineering, Hunan City University, Yiyang, Hunan, China;[Xie, Chuan] Department of Fundamental Medical and Clinial Loboratory, Yiyang Medical College, Yiyang, Hunan, China
作者机构:
[Tao, Wang] College of Information Science and Engineering, Hunan City University, Yiyang, China;[Wenwei, Li; Shiming, He] College of information science and engineering, Hunan University, Changsha, China
摘要:
The temporal evolution of ultrashort pulses before the beam splitting is very useful for filamentation theory. In this article, we experimentally demonstrated the temporal evolution of ultrashort pulses in the different spatial positions during small-scale self-focusing. The simulation results are found to be in good agreement with experiment measurements. (C) 2013 Elsevier Ltd. All rights reserved.
期刊:
Lecture Notes in Electrical Engineering,2014年163:1211-1216 ISSN:1876-1100
通讯作者:
Tu, Li(tulip1907@163.com)
作者机构:
[Li Tu; Chi Zhang] Department of Computer Science, Hunan City University, Yiyang, Hunan 413000, China
会议名称:
2012年控制与信息国际会议
会议时间:
2012-09-21
会议地点:
中国重庆
摘要:
In this paper an improved Hopfield neural network algorithm was given,the applicable conditions of Hopfield neural network models of memory storage is greater than the storage capacity of the perfect
期刊:
Lecture Notes in Electrical Engineering,2014年163:1941-1946 ISSN:1876-1100
通讯作者:
Tu, Li(tulip1907@163.com)
作者机构:
[Li Tu; Jianjun Wu] Center of Network, Hunan City University, Yiyang, Hunan 413000, China;[Li Tu] Department of Computer Science, Hunan City University, Yiyang, Hunan 413000, China
会议名称:
2012年控制与信息国际会议
会议时间:
2012-09-21
会议地点:
中国重庆
摘要:
This paper mainly studies the scrambling technology of digital image;discussing the image hashing algorithm based on Joseph Traverse. This paper explains the image encryption process based on Josephus
摘要:
In this letter, we propose a method for measuring the pulsewidths at different spatial positions of ultrashort laser pulses. By measuring the pulsewidths varying with spatial positions of chirped and chirped-free femtosecond laser pulses, it is found that pulsewidths at edge positions are longer than that of central positions due to the effect of residual spatial chirp. Then, we measure the temporal evolutions of pulse at the strongest spatial modulation position after small-scale self-focusing and our results show that the pulsewidths become narrower with increasing spatial contrast due to spatiotemporal coupling effect. We find that the method is reliable and feasible.
期刊:
Journal of Chemical and Pharmaceutical Research,2014年6(3):1312-1319 ISSN:0975-7384
通讯作者:
Zhang, C.
作者机构:
[Jia, Liyuan; Zhang, Chi; Li, Tu] School of Information Science and Engineering, Hunan City University, Yiyang, Hunan, China;[Xie, Chuan] Department of Fundamental Medical and Clinical Laboratory, Yiyang Medical College, Yiyang, Hunan, China
摘要:
In recent years, the chaos based cryptographic algorithms have suggested some new and efficient ways to develop secure image encryption techniques. In this paper, we propose a novel method to encrypt a gray image based on Arnold transform and Kent mapping. It is a digital medical image encryption algorithm, dual scrambling that based on chaos of pixel position and pixel values. The results of experimental and the statistical analysis show that the histogram of the encryped image is fairly uniform and significantly different from the histogram of the original image. The algorithm improves the traditional image encryption algorithm of image scrambling based on chaos, and it is valid and has good performance.
期刊:
Journal of Chemical and Pharmaceutical Research,2014年6(7):2073-2082 ISSN:0975-7384
通讯作者:
Tu, L.
作者机构:
[Jia, Liyuan; Tu, L.] School of Information Science and Engineering, Hunan City University, Yiyang, Hunan, China;[Huang, Xuehua; Zhang, Yingzheng] Department of Information and Engineering, Hunan Engineering Polytechnic, Changsha, Hunan, China
期刊:
Journal of Networks,2014年9(8):2121-2128 ISSN:1796-2056
通讯作者:
Ouyang, A.(ouyangaijia@163.com)
作者机构:
[Pan, Guo] College of Information Science and Engineering, Hunan University, Changsha 410131, China;[Ouyang, Aijia] School of Information Science and Engineering, Hunan City University, Yiyang, Hunan 413000, China;[Ouyang, Aijia] College of Computer, Hunan Science and Technology economy trade vocation college, Hengyang 421001, Hunan, China;[Pan, Guo] Logistics Information Dept, Hunan Vocational College of Modern Logistics, Changsha 410082, China
期刊:
Lecture Notes in Electrical Engineering,2014年272(1):807-812 ISSN:1876-1100
通讯作者:
Liu, H.(338945249@qq.com)
作者机构:
[Liu, Hean] Hunan University of Science and Technology Xiangtan, 411201 Hunan Province, China;[Zhou, Minghui] Network Information Center, Hunan City University, 413000 Yiyang, China
期刊:
Lecture Notes in Electrical Engineering,2014年272(1):793-800 ISSN:1876-1100
通讯作者:
Liu, H.(heanliu220@163.com)
作者机构:
[Shen, Hao] Institute of Computer and Communication, Hunan University of Technology, Hunan, 412007 Zhuzhou, China;[Liu, Hean] School of Information Science and Engineering, Hunan City University, Hunan, 413000 Yiyang, China
作者机构:
[曲咏哲; 蒋冬初; 李玉山; 闫旭] School of Electronic Engineering, Xidian University, Xi'an 710071, China;[蒋冬初] School of Communication and Electronic Engineering, Hunan City University, Yiyang 413000, China
通讯机构:
School of Electronic Engineering, Xidian University, China
期刊:
International Journal of Applied Mathematics & Statistics,2013年51(21):484-493 ISSN:0973-1377
作者机构:
[Xiao, Weichu] College of Communication and Electronic Engineering, Hunan City University, Yiyang 413002, Hunan, China;[Chen, Weihong] College of Information Science and Engineering, Hunan City University, Yiyang 413002 Hunan, China
关键词:
Benchmark information diffusion;LBP model;Remote sensing image;Support vector machine