Parallel Jacobi-like algorithms are presented for computing a singular-value decomposition of an $mxn$ matrix $(m \geq n)$ and an eigenvalue decomposition of an $n x n$ symmetric matrix. A linear array of $O(n)$ processors is proposed for the singular-value problem and the associated algorithm requires time $O(mnS)$, where $S$ is the number of sweeps (typically $S \leq 10)$. A square array of $O(n^{2})$ processors with nearest-neighbor communication is proposed for the eigenvalue problem; the associated algorithm requires time $O(nS)$. Key Words And Phrases: Multiprocessor arrays, systolic arrays, singular-value decomposition, eigenvalue decomposition, real symmetric matrices, Hestenes method, Jacobi method, VLSI, real-time computation, parallel algorithms.
展开▼
机译:提出了类似Jacobi的并行算法,用于计算$ mxn $矩阵$(m \ geq n)$的奇异值分解和$ n x n $对称矩阵的特征值分解。提出了一个$ O(n)$处理器的线性数组用于奇异值问题,并且相关算法需要时间$ O(mnS)$,其中$ S $是扫描次数(通常为$ S \ leq 10)$ 。针对特征值问题,提出了具有最近邻通信的$ O(n ^ {2})$处理器的方阵;相关算法需要时间$ O(nS)$。关键词和短语:多处理器数组,脉动数组,奇异值分解,特征值分解,实对称矩阵,Hestenes方法,Jacobi方法,VLSI,实时计算,并行算法。
展开▼