I just have completed a preliminar theoretical study of the important SVD decomposition. Now, I'm moving to numerical calculation of SVD.
I would like to learn directly a parallel algorithm to accomplish the task, or at least an algorithm well-suited for parallelization.
Can you tell me: 1) One (or more) of existing algorithms that match the requirement. 2) Where I can study it from (online refereces and/or books).
Many thanks.
I also worked on a CUDA implementation of parallel SVD algorithms and found this paper that helped me a lot:
Parallel Algorithms for the Singular Value Decomposition
They describe SVD algorithms for both dense and sparse matrices and discuss parallelization possibilities of sparse matrices.