ABSTRACT
In addition, remember that we assumed that we could always start with the approximation x(0)=0. It should be clear, however, that we could have started with some other initial guess. In this case, the result would depend to some extent on what that initial guess is. If the dependence is too large, this dependence could be detrimental because it would bias the results. On the other hand, suppose that we had a sequence of measurements {y i} (i.e., a sequence of measurements of the vector y) and had reason to believe that consecutive sets of measurements are correlated to
some degree. Then using as the initial approximation xi+1 (0) might have some
benefits. Exercise 11-3: Can you analyze this situation, where you allow each initial
approximation to depend on the last result and the iterative solution is purposely allowed to depend on x(0)? Might there be a problem with parasitic solutions that arise because each solution depends on the preceding one?
