Performance Of The Faddeeva Algorithm On Optical Processors

Abstract

An optical associative processor realizing the versatile Faddeeva algorithm can be used for several computations such as matrix multiplications, matrix decompositions or inversions, and singular value decompositions just by changing the structure of the data stored. Thus such an optical processor is programmable and efficient. In this paper the realization of this algorithm on various optical structures and their performance are described. For unstable data structures it is necessary to use a preprocessed Faddeeva algorithm on the optical processors for accurate results. The procedure of preprocessing and the performance of the preprocessed Faddeeva algorithm are also described.