There has been quite a flutter recently in the iterative world about an algorithm that speeds up the classical Jacobi method by as much as 200 times using a scheduled relaxation method where a combination of over and under relaxation is employed. The results as shown in this paper are remarkable and are indicative of a resuscitation of Jacobi method as a method for practical purposes.
However, when I implemented it I cannot get it to work. I was wondering if anybody else in this community has tried implementing this method and if so was the attempt successful. It would be interesting to hear if there is anything special that needs to be taken care of when implementing this algorithm.