基于带回溯机制冒泡排序法的矩阵最佳分块研究

The Analyzing About the Optimal Solution of Blocked Matrix-Based on Bubble Sort Method With Backtracking Mechanism

为了解决大型、超大型矩阵计算中处理速度较慢的问题,采用代数学、算法学相结合的方法,对缺乏实际分块标准与方法的矩阵分块问题进行探讨。通过引入带回溯机制的冒泡排序法,有效地建立了矩阵最佳分块方法。在此基础上,将最佳分块方法从理论上应用于矩阵乘积运算中,从而为降低复杂矩阵运算复杂度提出了切实可行的明确的改善方法。

In order to solve the problem of large, super large matrix calculation of slower processing speed problem, using algebra, arithmetic method, the lack of actual block standard and method of block matrix problems. By introducing a backtracking mechanism of bubble sort method, effectively established the optimal partition method. On the basis of this, the optimal partitioning method from the theory is applied to the matrix multiplication, so as to reduce the computing complexity of complex matrix and proposes some practical and feasible clear improvement method.