Matrix Multiplication Using Incoherent Optical Techniques

Abstract

The use of incoherent electrooptical analog methods for performing matrix-vector multiplication has been investigated mathematically. A technique for encoding the matrix information on a two-dimensional binary optical transparency by means of an area modulation scheme is described. The one-dimensional discrete finite Fourier transform, viewed from the standpoint of matrix-vector multiplication, has been performed experimentally to demonstrate feasibility. Matrix and vector array sizes employed were 33 x 33 and 33 x 1, respectively. The average value of the correlation coefficients between theoretically derived andexperimental data was found to be 0.95.