Abstract

Linear algebra problem solving constitutes a major part of all scientific and engineering problems. One of the basic operations in linear algebra is computing the inverse of the matrix that can be used in solving systems of linear equations among others. The proposed processor is based on a parallel optical vector-matrix multiplier. The operation is an iterative one producing an approximate matrix inverse. The computing complexity of this algorithm for computing the inverse of an N × N matrix is O(N), while the complexity using a sequential electronic computer is O(N³). Reducing complexity results in less computing time, which is very appealing for solving large systems of equations like those of computational fluid dynamics problems.

© 1991 Optical Society of America