Hi @galencolin, could you do lagrange interpolation, especially power sums. (Basically unconventional concepts, not present on cp-algorithms)
Reference questions : JETPACK and GCDMASK
I know editorials exist, but they are not clear to me.
Also, if you’re going as far as optimizing addition and subtraction, why not go for ints and (1LL * a * b) for multiplication. I have had significant speedups with this (mostly in NTT questions with hard time contraints).