并行算法

The Greenfield algorithm has been renewedly designed .It is a fast algorithm for solving linear algebra equation group of coefficient matrix for nubby tri-diagonal symmetrical matrix.

研究了电磁场有限元法解释的串行计算的算法和软件实现、并行计算的算法和软件实现，给出了电磁法有限元法解释计算任务并行的划分和分治，解决了有关并行计算的负载均衡技术，合理地给出了算法的粒度选取。

The elevations add a Z value to the grounds X and Y horizontal coordinates. However, it is always more difficult to get DEM and normally needs more computing time to generate DEM since its complicated algorithm and huge data.

研究中 ，分别从数据并行和算法并行的角度，对 DEM生成算法并行化进行了分析，并在网络分布式机群下进行了数据处理实验，取得了较好的并行处理效果。

The main results of this research are as follows:(1) A parallel algorithm for solving dense matrix linear equations on distributed-memory multi-computer is presented. Convergent velocity of this algorithm is better than parallel Jacobi method"s, and parallelism is better than Gauss-Seidel method"s. According to theoretical analysis, convergent conditions are given. Moreover numerical experiments indicate that results agree with theory.

主要完成了如下研究工作：(1)提出了一种求解系数矩阵为稠密矩阵的线性方程组的适合于MIMD分布式存储的并行迭代算法，该算法设计上兼取了Jacobi算法并行性好的优点和Gauss-Seidel算法收敛速度快的长处，并且在理论上给出了该算法成立的收敛性条件，并通过数值计算验证了此算法结果与理论相符合。

That is to utilize general MPI programming environment, program on the NHBL parallel computational model, analyze and predicate the running cost of program by using the cost function of this model, and analyze the scalability of program by using the time-scale scalability metric. Finally, we analyze the numeric algorithm example---matrix multiplication and the nonnumeric algorithm example---PSRS sorting on Dawning-1000 MPP and NHPCC-cluster NOW system by using the integrative method.

提出了将并行计算模型、算法、编程进行一体化的研究的方法，即利用通用的标准的MPI编程环境，将编程建立在NHBL并行计算模型上，并且利用该并行计算模型的评价函数进行分析、预测程序的运行代价，再利用时间比例的可扩放性分析方法来分析程序的可扩放性性能；最后我们用PSRS排序算法和矩阵乘法这两个典型的数值和非数值的算法，在曙光-1000并行机和NHPCC-Cluster工作站集群上进行了一体化的分析。

Combined genetic algorithms with domain decomposition parallel FEM, designed parallel FEM evolutionary displacement back analysis algorithm, developed corresponding program- GAParal-lelFEM3D(version 1.0)using C++.

把遗传算法和区域分解并行有限元算法结合，设计了进化并行有限元位移反分析算法，采用Visual C++开发了进化并行有限元位移反分析计算软件GAParallelFEM3D(v1.0)。

After the parallelizability analysis, the simulation method is parallelized with parallel programming tool OpenMP.

然后通过对仿真算法流程的并行性分析，以并行编程标准OpenMP为基础对仿真算法进行了并行化。

The new algorithm effectively combines the advantages of ACO and PSO. It adopts the grid method for environment modeling and makes use of the efficiency and succinctness of PSO to obtain the initial distribution of pheromone, reducing the number of iterations and accelerating the convergence. At the same time, by using the parallelizability of ants and distributed parallelized-searching technology, the performance of the algorithm is effectively improved.

该方法有效地结合了粒子群算法和蚁群算法的优点，采用栅格法进行环境建模，利用粒子群算法的快速简洁等特点得到蚁群算法初始信息素分布，以减少迭代次数，加快算法的收敛速度；同时利用蚁群算法之间的可并行性，采用分布式技术实现蚂蚁之间的并行搜索，求解精度高等优点，求精确解。

The domain decomposition parallel iterative algorithm for the 3-D time-dependent neutron transport difference equation is presented in Chapter 5. This algorithm is highly parallelizable and scalable.

第五章研究了基于几何空间区域分解的三维非定态中子输运差分方程的并行迭代算法，该算法达到了多处理机并行的高并行度和可扩展性。

The results show: all processors have better load balance; when the ratio of the execution time of a point addition to that of a point doubling is 3 during computing the scalar multiplication, the parallel efficiency of the proposed algorithm is up to 90%.

结果表明：各并行处理器单元具有较好的负载均衡特性；当执行并行基点判断算法，其标量乘的点加计算时间是点倍数计算时间的三倍时，算法的并行效率可达90%。

At the same time, the paper researches the parallelable of dynamic program, and discusses two kinds of parallel algorithm on SMP computer. After that, the paper give a parallel algorithm on cluster of SMP computers, and discuss how to divide the data getting more effective execution of parallel algorithm.

论文也讨论了动态规划算法的可并行性，研究了两种在SMP计算机上实现的动态规划并行方法，分析了在SMP机群系统中，如何通过分割数据的方法实现机群系统上算法的并行优化。

The split between the two groups can hardly be papered over.

这两个团体间的分歧难以掩饰。

This approach not only encourages a greater number of responses, but minimizes the likelihood of stale groupthink.

这种做法不仅鼓励了更多的反应，而且减少跟风的可能性。