Oh....Sorry, It's really difficult to explain it in English. : ( I'll add it to wiki.
On Tue, Oct 21, 2008 at 10:36 AM, Edward J. Yoon <[EMAIL PROTECTED]> wrote: > For example, > We have a 4 * 4 Matrix a : > > [a, b, c, d] > [e, f, g, h] > [i, j, k, l] > [m, n, o, p]] > > Then, we can store an 2 * 2 block matrices to file. > > block(0, 0) is a 2 * 2 sub-matrix as below. > > [a, b] > [e, f] > > Key : block(0, 0), Value : submatrix [a, b], [e, f] > Key : block(0, 1), Value : submatrix [c, d], [g, h] > Key : block(1, 0), Value : submatrix [i, j], [m, n] > Key : block(1, 1), Value : submatrix [k, l], [o, p] > > ..... > > Also, If we can access sub-matrix from Hbase table, blocking can be > pre-computed. > > On Tue, Oct 21, 2008 at 10:18 AM, Samuel Guo <[EMAIL PROTECTED]> wrote: >> should the matrices that do the multiplication have the same blockID layout? >> >> And How to divide a matrix into BlockIDs to make the parallel computation >> most efficiently? >> >> 2008/10/21 Edward J. Yoon <[EMAIL PROTECTED]> >> >>> Can anyone review this ?? >>> >>> http://blog.udanax.org/2008/10/parallel-matrix-multiply-on-hadoop.html >>> >>> -- >>> Best regards, Edward J. Yoon >>> [EMAIL PROTECTED] >>> http://blog.udanax.org >>> >> > > > > -- > Best regards, Edward J. Yoon > [EMAIL PROTECTED] > http://blog.udanax.org > -- Best regards, Edward J. Yoon [EMAIL PROTECTED] http://blog.udanax.org
