Mathematical Sciences Computing Research Environments

佐治亚大学计算机科学系建议将其英特尔iPSC/2 D4超立方体交换为英特尔iPSC/860 Model 8超级计算机系统（参见第五部分的规范）。这些设备将专门用于支持数学科学的研究。该设备将用于若干研究项目，其中特别包括：(i) T. Taha博士（计算机科学系）为求解一维和多维非线性偏微分方程的逆散射变换IST数值格式开发并行算法。这些方程出现在许多应用领域，包括非线性光学，信号传播，等离子体物理等。（ii） S. Cater博士和R. Robinson博士（都来自计算机科学部门）在基于GNU c++编译器和库的英特尔iPSC/860超级计算机上开发原型计算环境。目标是为用户提供一个高效的系统，并有效地利用并行计算资源。（iii） Dr. D. Gordon（计算机科学系）开发了一种使用共轭梯度算法求解大型稀疏矩阵的并行算法。（iv） A. Granville博士（数学系），以便更好地理解长度为z的区间内不含素数因子y的整数数量的分布函数。并行筛选不同的区间将大大加快计算速度。(v) j.w. Smith博士（计算机科学系），发展布尔（位）矩阵约简的一般并行方法。

The Department of Computer Science at the University of Georgia proposes to trade-in its intel iPSC/2 D4 hypercube for an intel iPSC/860 Model 8 supercomputer system (see specifications in part V). This equipment will be dedicated to the support of research in mathematical sciences. The equipment will be used for several research projects, including in particular: (i) Dr. T. Taha (Computer Science Department) for developing parallel algorithms for Inverse Scattering Transform IST numerical schemes for solving both one dimensional and multidimensional nonlinear partial differential equations. These equations arise in many areas of application including nonlinear optics, signal propagation, plasma physics, etc. (ii) Dr. S. Cater and Dr. R. Robinson (both from the Computer Science Department) for developing a prototype computing environment on an Intel iPSC/860 supercomputer based on the GNU C++ compiler and libraries. The objective is a system which is productive for the user and effective in the exploitation of parallel computing resources. (iii) Dr. D. Gordon (Computer Science Department) for developing a parallel algorithm for solving large sparse matrices using conjugate gradient algorithm. (iv) Dr. A. Granville (Mathematics Department), for better understanding of the distribution function of the number of integers, in intervals of length z, that are free of prime factors y. Sieving different intervals in parallel will speed up the computations considerably. (v) Dr. J. W. Smith (Computer Science Department), for developing a general parallel method for the reduction of boolean (bit) matrices.