摘要
稀疏矩阵的乘法运算可用于解决许多实际的应用问题。提出一种新颖的稀疏矩阵相乘算法,算法实现中将计算单元由单个元素扩展至行向量,避免了矩阵的转置,减少了扫描次数。利用三元组和少量的额外辅助空间实现稀疏矩阵的相乘。实验结果表明了该算法的有效性。
The multiplication of sparse matrix can be used in many applied problems.A novel muhiplying algorithm of sparse matrix is presented here.Using the row vector as computing unit instead of single element,the matrix＇s revolution is avoided,and the scan times are reduced largely.By the list of three-tuple and small additional aided space,the multiplication of sparse matrix is achieved.Experiment results show that the new method is effective.
出处
《计算机工程与应用》
CSCD
北大核心
2009年第19期55-57,共3页
Computer Engineering and Applications
基金
"上海市教育高地--计算机科学与技术建设项目"资助
关键词
稀疏矩阵
三元组
矩阵乘法
行向量
sparse matrix
list of three-tuple
matrix multiplication
row vector
