Matrix inversion algorithm for linear array processor (Q1206194)

The authors describe systolic arrays for calculating the inverse of a matrix by using a Gauss-Jordan algorithm. Their arrays contain \(n\) elements (where \(n\) is the dimension of the matrix), each of which has \(O(n)\) storage locations. One of the arrays performs only serial data transfer, while the other allows parallel data transfer, so that computation and communication can be overlapped. An unusual feature of the paper is its emphasis on the engineering aspects of systolic array design.