ABSTRACT
One possible iteration procedure is the following. Let x(k) be the kth approximate solution to equation (1). An improved solution is found from
where ? is a scalar convergence factor that is chosen to ensure that the solution will converge. Obviously, if ?=0, the convergence will be infinitely slow. However, if ? is too large, x(k) will either oscillate around the desired solution or diverge. We shall
see that, if this process converges, it converges to (AtA)- 1Ay. This will not be very useful in practice, but it illustrates the method-we will derive a regularized method presently.
