That paper was front page 2 days ago. It's the reason this paper is front page today. And while it is fantastic work, it has no obvious break-through application, as was discussed: https://news.ycombinator.com/item?id=43004955
Yeah, I posted about that algorithm on Nim Lang forum. Someone used deepseek to create a Nim version [1]. It seems the constant factor overhead is quite large. Well presuming deepseek implemented it correctly. There's obvious easy perf tweaks, but still seems to have high overhead.
1: https://forum.nim-lang.org/t/12679#78170