Papers by Keyword: Parallel Computation

Paper TitlePage

Authors: Zhi Min Chen, De An Zhao
Abstract: The substantial difficulties are encountered as the huge computation of complex three-dimensional finite element model. Network parallel computing is one of the most compelling topics at the forefront in the current field of parallel computing, the size and the speed of structural analysis can be increased by the combination of both, so that large and complex three-dimensional finite element calculation can be carried out smoothly. According to Wushaoling tunnel, three-dimensional finite element model is established and implement under the support of the finite element parallel computing environment of Deep Comp 1800 cluster system based on the analysis of neural network. The realization of 3D geostress analysis is also depend on the artificial neural network computation program ANNBP and MEBAC, which is an interface program connecting ANNBP and ANSYS. The results show that the computation efficiency is highly improved by the Deep Comp 1800cluster system, the distribution of the initial geostress field is compacted significantly by the faults and there is a vertical extrusion characteristic with the tunnel at the fault zone.
2790
Authors: Chang Feng Li
Abstract: Two-phase immiscible displacement in porous media is described by a coupled nonlinear system of an elliptic equation (for the pressure) and a parabolic equation (for the saturation). For the saturation changes much rapidly than the pressure, a more accurate solution (in both time and space) should be illustrated in practical numerica simulaiton for the former unknown. In this paper we present a seven-point central finite difference scheme to simulate the pressure and a characteristic finite difference combinng with domain decomposition method for the saturation equation. This method consists of reduced two-dimensional computation on the subdomain interface boundaries and fully implicit computation parallelly in subdomains. Aparallel algorithm is outlined and an error estimate in discrete norm is derived by introducing new inner products and norms. At the end of this paper, numerical experiments are presented in order to demonstrate theoretical results and the efficiency.
915
Authors: Hong Bo Xu, Nian Min Yao
Abstract: The performance of the spatial range query algorithms based on Brute-Force method, R-tree, VA-file and NB-tree suffers greatly in high-dimensional space. So the reduction of the dimensionality is the key to the spatial range query in high-dimensional space. The paper uses the parallel technique to present a spatial range query parallel algorithm in high-dimensional space. The algorithm transforms d-dimensional spatial range query to the linear space on d slave node processors. The d slave node processors run parallel. The master node processor only need calculate the union of d results which d slave node processors return. The experimental results indicate that its performance is better than that of the spatial range query algorithms based on Brute-Force method, R-tree, VA-file, NB-tree.
2308
Authors: Zhi Yu Chen, Jian Zhong Fu, Hong Yao Shen, Wen Feng Gan
Abstract: Amongst the flourishing Delaunay Triangulation methods, growth algorithm has been widely accepted because of its reputation of being simple and elegant. However, the parallelization of growth algorithm has not been fully exploited. In this work, a novel Growth algorithm of Delaunay Triangulation is proposed. The point cloud is first divided into two parts by a suitable curve and the separated areas are calculated by incremental algorithm. Triangles which cross with the curve are generated by a growth algorithm associated with uniform grid. At the process of merging, these grew triangles are used to detect incorrect triangles of the incremental algorithm areas. Method about generating triangles on curve is elaborated and a simple way to detect interferential triangles is also explained. With above method, triangulation calculation can be parallelized. Unlike the traditional divide-and-conquer method, no flip operation is needed in the proposed methodology. Thus, three dimensional applications are also made possible. A comparative research between tradition incremental algorithm and the proposed method has been conducted. Results show, the algorithm has a higher performance with less computation time.
299
Authors: Jun Feng Lu, Hao Zhang, Wei Fan
Abstract: Hemodialysis is a process to clean the metabolic wastes inside human body. Our previous researches on this topic fundamentally used contemporary numerical methods (such as CFD (computational fluid dynamics), MD (molecular dynamics), etc.) to study the ultra-filtration properties and made several progresses. However, our step-forward MD study on amino-acid leaking problem which natively exists during the hemodialysis process makes the numeral calculations more complicated. To control the heavy computation load during MD simulation, in this paper, a novel parallel domain separation scheme is proposed and its optimization method is further discussed. The unique feature of this scheme is that it introduces fibers as well as "compute shaders" into the parallel computation architecture. And the further application of this parallel scheme could form a new concept of MD simulation.
607
Authors: Yao Yuan Zeng, Zheng Hua Wang, Wen Tao Zhao
Abstract: As for the problem of numerical simulation oflaser propulsion of three dimensions and multi-sub domains, the domain decomposition strategy based on message passing mechanismis applied in this paper to realize parallelization. The cell-centered finite volume scheme is performed to solve Euler equation. A five-step Runge-Kutta scheme of explicit integral model is used for time advancement. The spatial discretization of inviscid fluid is estimated byhigh-order Godunov-type scheme. We test some different examples on a cluster system and the results show the smallest number of speedup is more than 5.19 when the degree of parallelism is 8. In a word, parallel computation is an inevitable choice to achieve the aim of accelerating the study of the mechanism of laser propulsion.
3027
Authors: R.D. Huber, K.E. Simmonds, R.S. Schechter, R.B. Mignogna, P.P. Delsanto
373
Authors: Jing Cun Bi, Cheng Guang Zhang, Yan Fei Liu, Wen Hui Dou
Abstract: Compared with CPU, GPU has a stronger parallel architecture and with specific API, the powerful potential contained in GPU can be released to become GPGPU. High computational capability of GPGPU can be used in scientific computation and data processing. This paper focuses on the research of the high frequency collision crack for normal Windows NTLM-Hash by resorting to CUDA API. The result shows that the computation capability of low-end graphic cards could be almost 100 times that of the normal dual-core CPUs, which could greatly lift the efficiency of encryption cracking.
239
Authors: Yi Ming Lu, Shi Yin Qin
Abstract: In this article, a fast approach of video stabilization is presented based on haerachical estimmation of global motion and multi-thread parallel computation. At first, the structure-texture decomposition is adopted to tackle the negative effect from illumination changes. Then according to pyramid structure of image texture a hierachical model is built to estimate global motion parameters with least square method. Afterwards Gaussian smoothing method is employed to eliminate error accumulation in motion compensation. Meanwhile the multi-thread parallel computation is used to speed up the processing efficiency so as to achieve a real time performance for videos with 25fps.
618
Authors: Yong Sun, Qing Shan Wang, Guan Xiang Wu
Abstract: Neighbor statistic algorithm is a kind of commonly used algorithm, statistical analysis was carried out on certain analysis value through the grid in the window, to reflect the local and zone terrain feature. This paper discussed the method of parallel algorithm based on neighborhood statistics serial algorithm parses and terrain algorithm. Focus on the data partition strategy and aperture effects processing strategy. Experiments show that, the two optimization method proposed can make the parallel program neighborhood grid processing algorithms make full use of parallel computing resources, and then further enhance its parallel performance on the general parallelization.
97
Showing 1 to 10 of 18 Paper Titles