#Day 21 Monkey Math
17 messages · Page 1 of 1 (latest)
oh man so close. would have been top 100 if || i had done the smart thing immediately instead of trying to brut force it. ||
Today kind of looks like what people say prolog would be good at
good old eval and lave
and i agree it's very backprop, but afaik you'd want prolog with a constraint extension
mostly just because my gut instinct is to reach for a constraint solver though it's overkill
i just ended up just writing the obvious evaluator
I did part 1 "properly"
Then I made a working AOT compiled version to try to brute force part 2... and it doesn't work 
Maybe I need to check a wider range than ±2^32
updating my GPU drivers
Still gonna take 140ish hours to check ±2^53 on an RTX 2060
I knew I should've left my desktop on to remote into
Well I got an answer quickly by going up from 0, but the site doesn't accept it.
They wanted the lowest one that gave the right answer
Yay race conditions
||oh lol, you can use binary search for the right answer||
Or you could calculate it directly by || rearranging the expression and solving for humn ||
Yeah that's what I did, recusively