Jump to content

Talk:Freivalds' algorithm

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

As good as this article is in explaining the mathematical proof behind Freivald's algorithm It would be good to see more elaborating on the consequences and meaning of this. In general, more non-math prose in the article giving a layman a better chance of coming out of reading the article with at least a semblance of why this is important and what the ramifications are would be nice AdamJacobMuller 19:23, 16 April 2006 (UTC)[reply]

Use of matrix-matrix multiplication for multiple trials?

[edit]

While recently expanding this article I noticed the fact that matrix-matrix multiplication can be used to perform all trials simultaneously. For large n, the matrix of 0/1 vectors would be tall and skinny, and so this would still be fast - while also yielding much better communication bandwidth than the version based on matrix-vector multiplication (due to higher computational intensity). I didn't add this because it's original research, but if a reference for this technique could be located, I think it would add to the article. Dcoetzee 16:22, 10 September 2012 (UTC)[reply]