Note on the non-stationary iterative method for the estimation of the optimum overrelaxation parameter

We study the cause for the overestimation of the optimum overrelaxation factor when a non-stationary iterative method of Carré's or Kulsud's type is used. In addition we show that when the overrelaxation factor is substantially changed the estimation process becomes more difficult. The overestimation phenomena are also shown numerically.