A note on the nature of optimality in the discrete Kalman filter

A derivation is presented in which it is shown that the optimal gain sequence of the discrete Kalman filter minimizes not only the trace of the estimation covariance matrix, but any linear combination of the main diagonal elements of that matrix.