一种因子化的稀疏矩阵转置算法

A Transposition Algorithm for the Factorizable Sparse Matrix

在定义了二元组结构的基础上,对稀疏矩阵中非零元素的行、列坐标值进行因子化,以此实现了一种新的稀疏矩阵压缩存储结构.并在应用该压缩存储结构的基础上,提出了一种稀疏矩阵转置操作的改进算法.该算法较传统算法而言,在时间复杂度不变的前提下将空间复杂度降低了30%.

Based on the defined 2-taple vertex,a new compression method is realized to store the sparse matrix using the coordinate factor ranks of the matrix non-zero values.A new transposition algorithm over this matrix data structure is proposed.This new developed algorithm reduces the space complexity by 30% with the even time complexity compared with those traditional algorithms.