Numerical method in which the n-th approximation of the solution is obtained on the basis on the (n-1) previous approximations.