Algorithms for matrix estimation (Q1287259)

Matrix estimation means the restoration of an exact value of a matrix using the measurements of a certain (usually linear) function of the matrix. The measurements are assumed to be either accurate or corrupted by random noise. The authors investigate algorithms for matrix estimation in the presence of random disturbances for various observation schemes. Theorems on convergence of recursive algorithms are proved and the results of numerical simulation are presented. The incorporation of artificial noise into observations is considered whereby the restrictions imposed on the measurement points can be weakened. An application of matrix least-squares estimation to the problem of searching for a root of a function of several variables is proposed. It is astonishing that almost no special attention has been paid to the problems of matrix estimation, although these issues were actually discussed in the literature.