Matrix multiplication in Hadoop
I
tried to build a code for the page ranking algorithm, the main complexity is to efficiently solve matrix multiplication, but I don’t understand how this task is performed, I read some papers on it, but this is beyond my scope. I don’t understand the concept he applied.
So, can you give me a concept behind the matrix multiplication mapper and reducer function? Thanks in advance.
I read thislink
Solution
The idea is that you can use a Strassen algorithm like decomposes matrix multiplication into subproblems, and then sends those subproblems to a bunch of different computers. Once these subproblems are completed, it is also possible to deal with adding different subproblems together into the matrix itself. The key to using Mapreduce is that all subproblems can basically be computed in parallel, and that’s it… The purpose of MapReduce.