Ok I thought we tried that and found graphx based flow was faster due to
some inherent problem structure (graphx can compute K eigenvectors at the
same time)
I will report some stats on row similarities experiments on vector blocked
index row matrix multiply vs current pic flow...
On May 26, 2015
That's a good question; I could imagine it being much more efficient if
kept in a BlockMatrix and using BLAS2 ops.
On Sat, May 23, 2015 at 8:09 PM, Debasish Das
wrote:
> Hi,
>
> What was the motivation to write power iteration clustering using graphx
> and not a vector matrix multiplication over
