I have been studying the iterative methods recently. For classical iterative methods solving $Ax=b$, I have seen that the most simplest iteration method is the so-called "Richardson iteration". But I am a little confused that the following names of the "Richardson methods" as follows:

  1. Richardson iteration
  2. extrapolate Richardson iteration

In addition, I have found there are 3 versions of the Richardson: $$ x_{k+1} = x_k+r_k; $$ $$ x_{k+1} = x_k +\omega r_k; $$ $$ x_{k+1} = x_k+\omega_k r_k; $$ where $r_k = b-Ax_k$ is k-th residual vector.

My question is: in different papers, one call different names of the Richarson iteration. For example, someone call the first version of the above 3 the Richardson. Someone call the second the Richardson. And someone call the last version the Richardson iteration. I want to get that which one is the real Richardson iteration, and what is the difference among the 3 iterative versions? In general, how should we call the 3 versions? any suggestions are welcome.


The first version only converges if $\|I-A\|<1$.

The second version converges if $\|I-\omega A\|<1$, so the parameter $\omega$ allows you to use the iteration on a broader class of matrices.

The third version modifies $\omega$ at each step to minimize the $\|r_{k+1}\|_2$ with respect to $\omega_k$. The optimal value is given by $$\omega_k = \frac{r_k^TAr_k}{\|Ar_k\|^2_2}.$$ See here for details.

I have heard 2 and 3 both called "modified Richardson iteration" but also just Richardson iteration

  • $\begingroup$ Thanks for your clear reply, get it. I agree with you that the 2 and 3 are extrapolative version or called modified/damped/ weighted Richardson iteration. The original one must be the first version without parameters. $\endgroup$
    – sunshine
    Dec 5 '19 at 6:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.