>>12743573I went to a talk like this from Texas A&M a while ago which was on the application of something like this to plasma physics.
Honestly, it seemed mostly like hype but little in the way of substance. As far as I understood it, it was really just a good method of curve-fitting, and as long as nothing too unexpected happened, it would work well. Which is great for efficiency, but the speaker was totally unable to answer the natural questions the audience had about what physical insights this would bring. So since then I've associated these talks with academic meme-speak.
Having said that, this is in CMS, and the method is clearly faster than current ones, and looks like it has some pretty great accuracy. So it's a great achievement in terms of computational efficiency. But I hope people don't get too sucked into this as a 'hot topic' because there is so much low quality crap on this shitting up the journals right now for easy publications.
