Dynamic behavior of the whitening process

We present an insightful view of whitening processes (or orthogonalization processes), which occur frequently in signal processing and neural network applications. For such processes, we show the dependence of the limiting solution on the initial matrix and the precise convergence rate. Simulation results corroborate our theoretical analysis.