Suppose that we want to use the map-reduce framework of Section 20.2 to compute one iteration of the Page Rank computation. That is, we are given data that represents the transition matrix of the Web and the current estimate of the Page Rank for each page, and we want to compute the next estimate by multiplying the old estimate by the matrix of the Web. Suppose it is possible to break the data into chunks that correspond to sets of pages - that is, the Page Rank estimates for those pages and the columns of the matrix for the same pages. Design map and reduce functions that implement the iteration, so that the computation can be partitioned onto any number of processors.