Specifically, (n log n)^{1 - 10^{-13}}). There are likely no practical industrial problems of a size that would benefit from that specific reduction, but just breaking the nlogn barrier suggests that there may be more and better fruit here in the future.
Yeah, I am kind of freaking out at some of these. I called a math friend to bring me down to Earth and he is freaking out even harder.
FFT BELOW NLOGN
Specifically, (n log n)^{1 - 10^{-13}}). There are likely no practical industrial problems of a size that would benefit from that specific reduction, but just breaking the nlogn barrier suggests that there may be more and better fruit here in the future.
SUBSET SUM AT 0.49 WTF
matmul <= 2.25 is no big surprise tbh. unique games and mul < n log n are much bigger.
Is anyone verifying these? And then, as a next step, how can they get a voice?
What the hell, indeed.