I find sorting networks to be fascinating. Beyond 8 inputs, there isn’t a pattern to follow to construct the optimal sorting network, and there are only “best

I find sorting networks to be fascinating. Beyond 8 inputs, there isn’t a pattern to follow to construct the optimal sorting network, and there are only “best known” sorting networks (but not proven to be optimal) for networks beyond 11 inputs. Unintuitively, the optimal sorting network for, let’s say, 16 inputs is not just an iteration on the optimal sorting network for 15 inputs. So you can’t just build upon what was optimal for a smaller sorting network to get what is optimal for a larger sorting network. I’ve spent more time than I probably should have several years ago experimenting with various genetic algorithms to construct sorting networks for 14, 15, and 16 inputs. It was a fun challenge and learning experience though. Here is a little utility I wrote in Python to check whether a comparison network is a sorting network and can generate diagrams for them in svg format. https://github.com/brianpursley/sorting-network https://github.com/brianpursley/sorting-network