Re:Exact numeric nth derivatives
by 30thElement on Thursday October 24, 2013 @02:45PM
(#6607091)
If I remember correctly, the most efficient algorithms have a ridiculously large constant factor, to the point where you don't have enough RAM to even store the matrix, let alone do anything with it.
Re:Exact numeric nth derivatives
by rcthompson on Thursday October 24, 2013 @05:14PM
(#6608020)
That's not what GP is talking about. Those algorithms are the sub-cubic ones, the ones that are O(x^2.something). GP is talking about an algorithm that is still O(n^3), but is not just a straight translation of the matrix multiplication formula into code, but rather does things in a way that is more friendly to the CPU cache, resulting in significant speed gains.
It's not the naive O(N^3) algorithm though - it's a block multiplication algorithm which generally gives faster results than the naive algorithm, because it's able to exploit locality so that you don't need to shift data into and out of the CPU cache all the time.