2023 Day 06

[near cloud]

It is Day 6: Wait For It
Solve it here: https://adventofcode.com/2023/day/6
View leaderboards here: https://adventofcode.com/2023/leaderboard/private/view/419899?order=stars

[undone ravine]

This seems easy?

[smoky carbon]

today's seems more like a mathematics puzzle than a coding one

[smoky carbon]

brave to say that before seeing part 2 😉

[undone ravine]

fair lmao

[undone ravine]

part 2 was also easy

arguably easier

[icy moat]

isnt its a bit easy for AoC?

[smoky carbon]

did this one in Excel, just because

[undone ravine]

https://github.com/monbrey/aoc2023/tree/main/day-6 ||I suspect there's technically a faster way than starting at 0 of calculating this correctly, but it runs fast enough anyway||

[smoky carbon]

||the elegant solution would be to solve the quadratic||

[undone ravine]

oh of course

||waits, whats squared here?||

||nvm I understand||

I havent been in school for 15 years cut me some slack

[icy moat]

its enough to work

speed is unnecessary

[undone ravine]

lmao

[dry echo]

part 2 is somehow flying over my head

You could hold the button anywhere from 14 to 71516 milliseconds and beat the record, a total of 71503 ways!
and asks
How many ways can you beat the record in this one much longer race?
what?

[undone ravine]

||Yeah so, instead of doing the multiplication thing from part 1, literally just say the number of ways||

||a "way" to win is just a number of ms you can hold it which will result in a record break||

[south valley]

First time I found part 2 easier than part 1

[undone ravine]

Yeah haha

[coarse imp]

it is, if you don't dumbo p1 like i did

because then p2 is not happening apparently it is, just takes a few secs

[south valley]

[dry echo]

I suppose I dumbo'd part 1 as well then, because p2 results in the same as the number already mentioned in p2

[coarse imp]

|| iterating possible races might not be the way to go ||

[smoky carbon]

||floor((sqrt(time ** 2 - (4 * distance)) + time) / 2) * 2 - time + 1 in case anyone was wondering...||

undoubtedly slower than brute-force iteration for the size of numbers in part 1

but likely quicker for the part 2 calculation

[coarse imp]

|| oh... i also could've waited 3 more secs ||
|| naive p1 does actually work for p2, it's not as terrible as it looked ||

[ionic grove]

AoC playing games, yesterday's was so hard and this one's so easy

[agile sphinx]

|| I‘m being proud of everyone not dumbo‘ing it ||

[undone ravine]

||Mine was half dumbo, I just iterated to find the first winning race, but recognised that the last winning race is that time subtracted from the max time||

||Commutative multiplication||

[topaz walrus]

||i did newton's method|| https://github.com/brownie-in-motion/advent-of-code/blob/main/2023/source/bin/day_06.ml

mostly because ||i don't know how floats work|| :p

[smoky carbon]

what is that, MATLAB?

[topaz walrus]

ocaml

||x' = x - (- x * x + t * x - (d + 1)) / (- 2 * x + t)||

||repeat this and it converges very fast||

[smoky carbon]

how many iterations for part 2?

[topaz walrus]

10

||yes five for each root||

[smoky carbon]

||I personally just found one solution and then reflected it about x=(time/2)||

I like the method though, very elegant

[topaz walrus]

oh that's crazy didnt realize u could do that

[agile sphinx]

|| that’s how quadratic functions look like, they are symmetrical ||

[smoky carbon]

||yeah, if you think of the graph, it's always 0 at x=0 and x=time, and the maximum is always halfway between the two||

[topaz walrus]

||ah yes because after u shift it vertically it doesn't change the center||

i feel like you can get it to like ~2 iterations ||using the convexity of parabolas||

[astral geode]

todays was refreshing

[astral geode]

wow the perl devel profiler is really nice

[ocean crater]

Few seconds? ||What magnitude is your input? Mine is around 60mil and solves in 71ms using js||

[ocean crater]

Little faster once the jit warmed up (and I wrote a slightly optimized version)

[coarse imp]

with the naive approach

meant to say: no optimization necessary, even if you took the worst possible solution || iterating through every race and calculating if it is won or not ||

[undone ravine]

that would be shockingly bad to completely ignore ||commutative multiplication|| at least

[coarse imp]

nah, it actually works okay

i expected it to not, which is why i didn't even try it, but after having a better solution (shockingly close to yours, actually) i tried it again

[undone ravine]

I mean yeah iterating every race definitely would work, it's just a waste of time lol

[coarse imp]

i absolutely expected it to not work
that's what he does, usually
naive gets you to 1, but p2 is off limits

[civic galleon]

My solution for p2 was ||brute force and it took only 60ms without logging||