dsimcha Wrote: > == Quote from Kevin Bealer (kevinbea...@gmail.com)'s article > > (Non-software) people doing routine tasks often come up with better > > algorithms > intuitively than my intuition expects them to. > > I think a lot of people would do even better than insertion with a deck of > > poker > cards -- they might group cards by either suit or rank (or rank groups) (e.g. > "Hmm, I'll make three piles of 1-5, 6-10, and J-A"), then order the "buckets", > then stick these ordered sets back together. If you think about it, this is > a lot > like a radix sort or a multi-pivot cousin of the quick sort. > > You mean a bucket sort? http://en.wikipedia.org/wiki/Bucket_sort
More or less, though I think human beings use algorithms in a more artistic way than sticking to any one algorithm. I'm curious if the multi-pivot quicksort (I think everyone gets what I mean by this? Divide by more than one pivot on each pass? I can give details if you like ...) has been tried out much. It seems like it must have been, but it also seems like something that would have cache-awareness advantages that would not show up in the simplified comparison-counting way of thinking about efficiency. Kevin