hint 2: ||part 2 only has unique ranges after processing, no need to worry about counting already-checked combinations||
#advent-of-code
modern terrace
devout turtle
Good morning all,
I'm reading the puzzle and I think I understand what I have to do.
My biggest hurdle in handling the input is to get it worked into some form that I can use the workflow rules nicely, but I'll get that worked out somehow.
What I can't seem to get my head around though, is for the example input, how lines 1 and 3 are evaluated.
After reading the description a couple of times, I'm totally under the impression that every part goes through the workflows in order, I don't quite get WHY or HOW parts 1 and 3 seem to start at workflow 'qqz', instead of 'px' like the rest.
I wouldn't dare to say that there's an error in the puzzle itself, so clearly I'm missing something, but I can't figure out what it is.
Who can help me on that?
hexed axle
Good morning all,
I'm reading the puzzle and I think I understand what I have t...
#aoc-solution-hints is a great place to ask in more depth. Asking here can get messy because of the need to avoid spoilers. If you're comfortable with more spoilers we also have a solutions thread that might have more guidance, at the risk of giving away more
tidal mist
I set an alarm clock for aoc. I'm sure it costs me a few spots to turn it off each day.
lime fulcrum
aoc is too early. I will not stand up at 6 am
lime fulcrum
I always have a bit of problems with wrangling my input with splits and strip. are there better ways to do this?
hidden musk
omg a language that I actually know for once
modern terrace
What's today's language
C#
zealous basalt
C#
yuck
wonder what day will uiua get
<--- doesn't know the mods don't know uiua
zealous basalt
Whats that
cool array lang
lime fulcrum
Why isnt there a pin smh, mods dropping the ball
you know you can just click the overview for spoiler threads and you can directly see the language?
atomic cairn
noble skiff
day 19 may have been my favorite so far
earnest forge
Yeah I liked today's puzzle
noble skiff
the fact that it worked my first try probably biased me π
programming more fun when you don't spend an hour painfully debugging
magic oxide
i haven't done today yet
earnest forge
the fact that it worked my first try probably biased me π
programming more fun...
Me on day 13 with my off-by-one
magic oxide
seems fun tho
noble skiff
Me on day 13 with my off-by-one
i still have one I didn't manage to debug
the one with the broken gears parts or whatever
.#?
earnest forge
.#?
Hot springs
How did you get an off-by-one for hot springs lol surely you either get it right or massively wrong
noble skiff
When implementing the faster approach for part 2, I made some subtle error that only was incorrect for very large examples
idk
it wasn't off by one, to clarify
earnest forge
Ah
noble skiff
I didn't mean "an off by one I didn't manage to debug"
earnest forge
That makes more sense then
noble skiff
well, the underlying error is very likely an off-by-one
that doesn't mean the result is off-by-one
earnest forge
What was your general approach?
noble skiff
||DP||
So far, I didn't have time to fix that, and I also haven't had time to actually do 18; I looked up the formulas everyone's been using because I couldn't stand doing it the same way as day 10 again
And I also want to go back and optimize some days; I have 4 days that are over 1ms
but honestly not sure if any of that will happen
magic oxide
yeah off by one error
||i think i somehow did only 9 steps instead of 10 or something||
dim pivot
I'm terrified for the next 6 days
How many do you think will be very hard puzzles (gold lb cap >= 1 hour)
woven solar
well, we've reached day 19 and it wasn't beacon scanner π₯΄
rapid wolf
I make the bold guess that the ||Chinese Remainder Theorem|| will be relevant tomorrow.
dim pivot
yeah the difficulty seems fairly static this year
dim pivot
I make the bold guess that the ||Chinese Remainder Theorem|| will be relevant to...
well that's why we have sympy
woven solar
in 8 hours, i wake up at day20. it's called "bacon scanner". I go back to sleep.
dim pivot
I hate the beacon ones
I'd be happy with CRT
Or even something like modular inverses/slam shuffle
rapid wolf
I still have zero confidence that the beacon scanner was solvable in the general case
dim pivot
Some other things that can come are reverse engineering or z3able ones
rapid wolf
but whatever julia I scrounged up did work
dim pivot
Or a heuristic puzzle
Last year I only found 1 puzzle nasty (day 16, still don't get it) and 1 more that I couldn't do quickly (day 22, but it was a dumb problem)
rapid wolf
oh interesting, I only got up to day 11 last year.
dim pivot
This year is going better than last year in terms of lb rankings but I'm getting more and more scared every day
rapid wolf
I wonder what I was doing last year that I didn't manage to 50star AOC.
scarlet badger
what is this channel about
scarlet badger
advent of code? what does that even mean
cyan sleet
I still have zero confidence that the beacon scanner was solvable in the general...
the 3d rotation one?
dim pivot
For 22 I had to honestly copy my solution from online
It was worth it though as it got me like 60 points on 25
rapid wolf
advent of code? what does that even mean
It's an event that runs every december where you get one programming puzzle a day
see #aoc-faq for more details
scarlet badger
oo alr
rapid wolf
the 3d rotation one?
Ye, the 12 overlapping beacons per scanner one
dim pivot
Ye, the 12 overlapping beacons per scanner one
I gave up with that on 2021
The ones I didn't do that year were "Amphipods" and that one
My favourite was day 24
The one with reverse engineering
rapid wolf
I just did that one with a constraint solver
cyan sleet
Ye, the 12 overlapping beacons per scanner one
there were some decently strong guarantees, right?
rapid wolf
Yeah, I can't really come up with a counterexample, but intuitively, I don't think the assumptions I made would hold on a malicious case.
but frankly, imagining possibly rotated beacons in 3D is straining my mind as-is
cyan sleet
I liked that problem
it really caught me off guard, took me a bit to see how to approach it in a good way
rapid wolf
I didn't find the clever solution, so I just "bruteforced" it in julia and had the answer in a minute.
lime fulcrum
but frankly, imagining possibly rotated beacons in 3D is straining my mind as-is
beacons?
lime fulcrum
https://adventofcode.com/2021/day/19
... yeah that sounds complicated
I think I know how to start with this one though. lot of vectors for this one
noble skiff
I feel like that one was solvable in the general case
maybe not actually, idk
Actually, yeah. Basically, the idea was, scanners have at least 12 beacons in common. distances between beacons are invariant under translation and rotation of the beacons
so you could pretty rapidly start to establish connections between points
i'm not sure if it's possible to construct something sufficiently adversarial that would cause this to break down. Or even better than distance, a sorted list of 3 deltas.
You only need each beacon to have a single other beacon, which which it shairs a pair of points, that has a unique sorted list of 3 deltas
cyan sleet
Actually, yeah. Basically, the idea was, scanners have at least 12 beacons in co...
the distance thing is just an optimization since you can reject things quickly (cheap computation and rotation invariant)
you can ||try all rotations and compute differences between all points, the same vector should appear at least 12*11/2 times||
noble skiff
the distance thing is just an optimization since you can reject things quickly (...
yeah, that's true, there aren't that many rotations anyway
I think that year was the last time I fully solved AoC
noble skiff
actually, no, that's a lie
cyan sleet
a factor 24 more work can hurt
noble skiff
I just mean in the grand scheme of things, x24 will not make a feasible solution infeasible
sure
I mean there's a reason I used the optimization π
looks like 2020 is the last time I got all 50; I felt like that year was easier. 2021 and 2022 had some real doozies in the last 5 days
cyan sleet
I think I needed it to do the whole year sub a second π
noble skiff
heh that's roughly my goal this year, first time i've really looked at perf much in aoc
I have 4 days over 1 ms atm
I suppose since I'm still under 1 second for everything, there's a decent chance I don't need to optimize any existing stuff... but still
cyan sleet
yeah, most tasks are pretty trivial, then you get the actually computationally heavy stuff
2021
and 2022
noble skiff
very impressive
I probably won't actually make it; my algorithm chops tend to not be good enough, on average, to get all of the problems without a lot more work than I'm willing to put in (or more copying/getting help, I suppose)
hidden musk
hmm. i've only used c# for work, so now the thought of doing aoc in c# is kinda :(
rancid creek
hmm. i've only used c# for work, so now the thought of doing aoc in c# is kinda ...
Unrestricted C# can be a lot of fun. You can challenge yourself and solve the puzzle as a single LINQ query. π
noble skiff
I tried pretty hard to do that with kotlin (in a past year)
Kotlin has... a lot of stuff that lets you just keep chaining expressions forever
like it has all the typical iterator stuff but then on top of that it has Groupings and things that turn iterators into dicts/Maps or lazy versions thereof
and it has the scope functions which also let you keep chaining stuff
it does get a bit crazy though
hidden musk
huh. more ||range|| shenanigans today
cyan sleet
huh. more ||range|| shenanigans today
indeed
jovial jay
Pretty close to day 20 now.
earnest forge
GLHF everyone
modern terrace
Pretty close to day 20 now.
and by extension the end of aoc
covert arrow
GLHF
covert arrow
and by extension the end of aoc
and by extension christmas
earnest forge
Pfft, it's clearly halloween
modern terrace
Pfft, it's clearly halloween
hmm
have you considered changing from octals to decimals
modern terrace
fair
earnest forge
modern terrace
anyway
my expectations suggest that i can't solve this day
past 19 days were alright
covert arrow
same, we're overdue for a hard puzzle
minor cave
Language Roulette: Day 20
The language is ... C++
modern terrace
it's like those calamity predictions where a calamity happens every 100 years but it's been like 130 or something years
earnest forge
**Language Roulette: Day 20**
The language is ... *C++*
I can't believe you'd pick C++ over a good language like Rust smh
covert arrow
surely rust is in the pool, right? 
earnest forge
I can't imagine they'd put C++ in without also putting Rust in
Also Day 25 is clearly going to be Python
modern terrace
Also Day 25 is clearly going to be Python
merging into normality
earnest forge
Anyway GLHF everyone
covert arrow
GLHF
shut trellisBOT
<@&518565788744024082> Good morning! Day 20 is ready to be attempted. View it online now at https://adventofcode.com/2023/day/20. Good luck!
modern terrace
absolutely wrecked trying to part 1
earnest forge
I made the mistake of trying to be smart instead of brute-forcing lmao (EDIT: FOR P1 DUMMIES)
Cost me loads of time
naive stone
i wish i could understand the problem as quickly as you guys solve it
pine tiger
I made the mistake of trying to be smart instead of brute-forcing lmao (EDIT: FO...
||brute force works for p2?||
modern terrace
||brute force works for p2?||
THERE'S A SMART WAY TO DO PART 2???
earnest forge
THERE'S A SMART WAY TO DO PART 2???
There's definitely a smart way of doing P2 and I probably shouldn't be trying to do it by hand but I'm not smart enough to write code for it
covert arrow
how
modern terrace
i'm bruteforcing rn
covert arrow
same lol
modern terrace
how long does it take
earnest forge
Same
modern terrace
uh oh
earnest forge
Might RIIR to hope for speedup
covert arrow
Join the language roulette and write it in C++
devout tusk
i don't think i even fully understand the state changes in the question today
modern terrace
Join the language roulette and write it in C++
i have no clue how to write in C++
neither rust
i'm officially damned
covert arrow
hmm, ||I think it's doing binary counting||
earnest forge
Yeah I think it probably is but I can't figure out a good way to speed that up with a program
||Tried doing it by hand but that will take ages||
modern terrace
hmm, ||I think it's doing binary counting||
||the hell is that||
devout tusk
I made the mistake of trying to be smart instead of brute-forcing lmao (EDIT: FO...
is p2 the same with like a ||gazillion|| presses?
earnest forge
is p2 the same with like a ||gazillion|| presses?
(no)
((that would have made my half-finished attempt actually useful))
stray flower
fleet fox
π advent of realizing i cant read is back
jovial jay
π advent of realizing i cant read is back
Same. I couldn't figure it out yet. I'm just here because I'm hoping for another fun problem
oblique pasture
Havent Solved Since Day3 Part2 π¦ I Should Do Something
earnest forge
Well those are some bugs I should fix
Python brute-forcer has a 35 minute head-start, who would win
earnest forge
Python brute-forcer has a 35 minute head-start, who would win
OH MY GOD I FORGOT THE RETURN IN THE PYTHON ONE
Even if it finishes it won't help
π€¦
pine tiger
π
slate berry
flat sequoia
If anyone is stuck try working through this simpler example by hand
broadcast -> a
%a -> b, e
%b -> c
%c -> d, e
%d -> e, e
&e -> a, b, output
i think this works??
cyan sleet
||what's updog?||
cyan sleet
aww, my joke doesn't work now
earnest forge
Lol
sharp tapir
hi team ive done upto 9, but have had to hand in a final project for my sql bootcamp last week so ive fallen by the way side. Would someone be able to tell me which days are the easiest from day 9?
velvet basalt
W
lime fulcrum
hi team ive done upto 9, but have had to hand in a final project for my sql boot...
13 to 16 were pretty simple
covert arrow
my opinion:
10 - p1 is alright, p2 is a bit tricky
11 - pretty straightforward
12 - very hard (arguably the hardest so far)
13 - medium
14 - medium-hard
15 - reading comprehension, otherwise relatively easy
16 - medium-hard
17 - hard
18 - hard
19 - medium-hard
20 - more reading comprehension, p1 is ok but p2 is hard
you can also see the leaderboard fill times (https://adventofcode.com/2023/leaderboard/day/10) to get a sense of how hard a puzzle is
sharp tapir
my opinion:
10 - p1 is alright, p2 is a bit tricky
11 - pretty straightforward
1...
amazing thanks
10 i skipped
reading 11
noble skiff
my opinion:
10 - p1 is alright, p2 is a bit tricky
11 - pretty straightforward
1...
hard agree, 12 part 2 was annoying
i don't think 19 is hard though
covert arrow
on second thought, maybe
noble skiff
I would say 16 is hard; it only becomes reasonable if you know a certain formula that most people didn't think of themselves
a lot of people (self included) grinded out 16 by doing things like tracing paths across and tracking boundary crossing
covert arrow
there's a formula for 16?
noble skiff
there's a formula for 16?
||shoelace + pick's||
covert arrow
16 was the ||mirror|| one, no?
noble skiff
oh my bad
covert arrow
maybe you're talking about 18?
noble skiff
i'm thinking of 10
noble skiff
for 10p2?
covert arrow
yes
noble skiff
even doing ||flood fill|| correctly is pretty obnoxious; there's no guarantee ||the interior is connected||
lime fulcrum
i did it by going row by row. meant ||18 didn't go as well||
noble skiff
So you have to somehow deal with that
covert arrow
even doing ||flood fill|| correctly is pretty obnoxious; there's no guarantee ||...
the way I did it was ||floodfill from every point on one side of the loop||
noble skiff
the way I did it was ||floodfill from every point on one side of the loop||
right , but then the hard task becomes finding all those points
covert arrow
the way I did p1 made it pretty trivial
noble skiff
idk, it tends to be a pain I find
noble skiff
lru?
covert arrow
just lru cache surely
Getting to the point where you can slap on ||memoization|| is the hard part imo
dim pivot
Hm
For me it was quite natural
I found stuff like day 10 much harder
For me the two longest days so far are day 10 and day 17
The rest are much shorter
covert arrow
I guess it depends on how familiar you are with concepts like ||dp||
ripe meteor
The instructions are getting too long π
My smooth brain cant focus long enough to go through them
proven linden
hidden musk
wow C++ is annoying
minor cave
lmfao, I feel like roulette has just resulted in you and fenix really liking some functional languages and being unimpressed with the rest
hidden musk
i've only ever used C++ for cp but this is harder than any problem i've done in c++ lol
but yeah. the functional langs in the beginning were pretty fun
certainly a lot more fun than vb or like
fortran π
still haven't finished that day btw
jovial jay
wow C++ is annoying
i feel you. i only use it in case i want to contribute to open-source softwares that requires it. otherwise, i only go by interpretative languages
minor cave
Language Roulette: Day 21
The language is ... Java
covert arrow
GLHF
earnest forge
GLHF
earnest forge
**Language Roulette: Day 21**
The language is ... Java
Sadists
naive stone
is the pool of languages in the roulette viewable somewhere? i'm curious what's in there
covert arrow
I saw it on the notion somewhere
hexed axle
is the pool of languages in the roulette viewable somewhere? i'm curious what's ...
It makes using fun programming languages all the better
earnest forge
It makes using fun programming languages all the better
So why is Java in the pool then
naive stone
lol
shut trellisBOT
<@&518565788744024082> Good morning! Day 21 is ready to be attempted. View it online now at https://adventofcode.com/2023/day/21. Good luck!
magic oxide
oh
hexed axle
So why is Java in the pool then
Gotta be reminded of the unfun
potent pumice
day 24 aoc should be holyc
earnest forge
part 2 is pain
Agreed
woeful elbow
Umh
modern terrace
how to p2
covert arrow
||the borders might make it easier to calculate how far you can travel||
modern terrace
i think i figured oti out
naive stone
i can't believe you guys finished part 1 so quickly. I thought it'd be easy, but it seems to have a trick to it or something
things get real slow for me after just 12 steps. hell before that even
woven solar
make sure that you ||don't count cells twice, and get rid of the cells reachable at previous step.||
naive stone
hmm, yea good point
woven solar
it's pretty much ||bfs where you replace your population of cells (a set) with a new one each step||
for part2 though, ugh.
i'm currently thinking of ||memoizing what cells are reachable from this cell in 100 steps||, so that I can ||reduce the number of steps required by a factor of 100||. not sure it'll be enough, though.
fleet fox
how to part 2 π
naive stone
alright well now my code finishes quickly, but gives me the wrong answer lol. still, progress i guess
earnest forge
def part_two(): return 0
woven solar
i'm currently thinking of ||memoizing what cells are reachable from this cell in...
oh dear, something's not right with this
earnest forge
(pls dont make me multiply my input and do threading and make my pc cry)
(casual 2 petabytes of ram)
modern terrace
OH
jovial jay
Wow leaderboard ain't filled yet
earnest forge
WHAT HOW
earnest forge
Oh yeah that'll do it
fleet fox
π
fleet fox
help π
covert arrow
π© why
flat mica
fleet fox
WhAT!!!
flat mica
devout tusk
considering there's no spoiler chat, p2 must be insane, ig
fleet fox
considering there's no spoiler chat, p2 must be insane, ig
π u bet
covert arrow
yup
modern terrace
nvm i didn't see.
devout tusk
π u bet
is it like a ||million|| steps?
fleet fox
is it like a ||million|| steps?
||twenty million|| with ||infinite garden plots||
devout tusk
||twenty million|| with ||infinite garden plots||
ah, seems easy to do though, if you did p1 efficiently π
fleet fox
ah, seems easy to do though, if you did p1 efficiently π
guess ima go bruteforce π π
covert arrow
ah, seems easy to do though, if you did p1 efficiently π
there are ||16 million positions after just 5k steps||
woven solar
there's ~n^2 positions within n steps, so you're looking at O(n^3) complexity if you keep track of them
devout tusk
there are ||16 million positions after just 5k steps||
can't be more than ||17161||
covert arrow
?
modern terrace
found a faster bruteforce that goes like 15k per sec
that's still like half an hour
RIIC++ i guess
tall wasp
how does one even start with this π₯²
devout tusk
?
idk wym, solution can't|| exceed that value, actually lesser (considering the amount of #s||
tall wasp
idk wym, solution can't|| exceed that value, actually lesser (considering the am...
its an ||infinite repeating grid so yes it can||
woven solar
idk wym, solution can't|| exceed that value, actually lesser (considering the am...
well, read part2
by their nature all(?) of the cells inside the covered area ||oscillate between being on and off with period 2||. so you can probably reduce a bruteforce solution to only considering ||the outgoing part of the wave, which scales with O(n), and then do some memoization on that||. sounds far too annoying to me, though, so I might ||have breakfast|| instead.
tall wasp
by their nature all(?) of the cells inside the covered area ||oscillate between ...
i love how ||have breakfast|| is a spoiler
fleet fox
i was about to say that lmfao
tall wasp
gave me a good chuckle in between crying about the puzzle
earnest forge
I have an idea of a reasonable algorithm but there are so many edge cases
tall wasp
i have no reasonable ideas
devout tusk
its an ||infinite repeating grid so yes it can||
Ah
dim jewel
real
mossy basin
i want yall to know that i initially tried to solve part one by taking matrix powers and numpy complained about initializing a 1.5gb array
dim jewel
@covert arrow what is the golf right now
tall wasp
today might be my first day i dont finish before work
which means ill have to finish it during work instead /s
devout tusk
<@726368249666732074> what is the golf right now
Is there even a golf π
mossy basin
my input looks like a diamond
fleet fox
leaderboard still isnt filled
dim jewel
Is there even a golf π
im hoping he does
hes called golfing enthusiast after all
thought hed be good at it (jk
tropic wind
npc
dim jewel
real
tall wasp
is this the hardest aoc problem ever
i have no idea where to even start π
dim pivot
Yeah it's really hard
minor cave
I think an int code year problem is still classed as the hardest based on lb fill time
tall wasp
my part 1 code also just has zero scalability
tall wasp
my part 1 code also just has zero scalability
its just ||str.replace|| which is just gonna die on any larger scale
minor cave
ah yeah, 2019 d18p2 had a nearly 2 hour lb fill time
dim pivot
Well then 2019 day 22 would be harder
tall wasp
I don't really want to help until the lb is filled tho
yeah no thats fair
dim pivot
But I didn't find it very hard (as I am mainly a mathematician)
devout tusk
thought hed be good at it (jk
I'm a car enthusiast. But I hate driving
dim pivot
Compared to something like this year's day 17 - that was an implementation hell for me
hexed axle
ah yeah, 2019 d18p2 had a nearly 2 hour lb fill time
I've been wondering how much of the lb fill time is tied to more folks participating
marble summit
hello
fleet fox
minor cave
I've been wondering how much of the lb fill time is tied to more folks participa...
It probably pays a non-trivial part in it tbh. We have a lot of consistent leaderboarders now that probably didn't know about this 4 years ago
fleet fox
im gonna lose it
marble summit
can someone help me with day 5 pt 2 pls
i hard coded it and i pressed run like 10 minutes ago and nothing has popped up so i think im cooked
jovial jay
Lb still ain't filled? Wow, just wow.
jovial jay
i hard coded it and i pressed run like 10 minutes ago and nothing has popped up ...
You can look at the day 5 thread
hexed axle
80 minutes, oof that was a long one today
jovial jay
It's filled now?
hexed axle
Yup, just got there
jovial jay
I need to try this after I sleep
modern terrace
C++ bruteforcer started
woven solar
now that I think of it, it should be possible to ||essentially factorize the task into movements within a board and between boards||, by ||precalculating possible transition periods between boards||.
modern terrace
wait a minute i'm not thinking right with this am i
i should be calculating how far the thing should reach
.-.
aanyway
covert arrow
now that I think of it, it should be possible to ||essentially factorize the tas...
that's what i've been trying to do for the past hour π©
modern terrace
hoping one of my bruteforce approaches come back with good results
π¬
wuh oh
OH THAT'S THE C++ BRUTEFORCER
???
earnest forge
Oh my god ||there aren't any obstacles in the same row as your start square||
This makes it so much easier
modern terrace
i don't think my C++ bruteforce works
lucid narwhal
p2 how?
modern terrace
grr
covert arrow
This makes it so much easier
why isn't the example input like that π
earnest forge
why?
||add or sub multiples of grid size||, ||track your solution in terms of grids while you have enough area||
modern terrace
so it seems like there are 15604 reachable squares out of 17161 in my input
what do i do with that info
covert arrow
oh, it's my cake day today
modern terrace
Oh my god ||there aren't any obstacles in the same row as your start square||
so ||horizontally the max reach is just Β±26501365 units from the original||?
how do i approach this-
woven solar
...okay, got it. I'm not sure I enjoy how many aoc days I solved by ||spotting a pattern in data without understanding what causes it||
tall wasp
alright, im off to work - maybe later today
earnest forge
My solution has been running for 18 minutes and I don't see why it's so slow
It's stuck in the 'explore trivially-explorable grids' step
oh that's still like 41 billion things though lol
devout tusk
<@726368249666732074> what is the golf right now
today's golf : ||#1180008645384216606 message||
thin blaze
I hate that day.
magic oxide
how difficult is yesterday and today?
uni offers just came out and I'm swamped with work, so I need a rough estimate of how much time I should dedicate
thin blaze
how difficult is yesterday and today?
Yesterday part 1 is simple enough, part 2 requires a bit of maths.
thin blaze
how difficult is yesterday and today?
Today part 1 is doable with graph computation.
Part 2 is just pure hell.
This is the first day I failed though.
woven solar
yesterday wasn't very hard (hint: it's one of the "||probably near-impossible in the general case but all the inputs are such that it's pretty easy if you look at them||" days that are very common this year). today is... objectively harder given the leaderboard fill speed, but I solved it by ||finding a pattern in the answers||.
thin blaze
I still didn't understand most of people's solutions for Day 21/2.
wispy folio
For Day 18 (part 1), do we have to mind the subtleties of day 10 (cf. screenshot)
magic oxide
I believe so
hollow wharf
I=inside
O=outside
wispy folio
so same as day 10 damn
pine tiger
wait is there prize for pydis top finisher(s)
i might be hallucinating but i remember a discord role or something
minor cave
the Completionist role will return, yeah
simple raven
I can't believe the 2015 AoC had you mine bitcoin for the authors π€¦
pine tiger
any suggestions for what i should do a year of aoc in casually
i've tried rust, go, and haskell (unsuccessfully)
hearty harness
anyone keep a note of what algorithms or data structures worked best over the years of each day?
noble skiff
idk about "best" - do you mean most commonly?
hearty harness
idk about "best" - do you mean most commonly?
yeah I think the most common algos and data structures is what I was going for...
hidden musk
any suggestions for what i should do a year of aoc in casually
ocaml π₯°
noble skiff
yeah I think the most common algos and data structures is what I was going for.....
i think all the bread and butter of CS, mostly
arrays, hash tables, sets, linear search, binary search, graph traversal, Djikstra's
hearty harness
yeah, this is where 'self-taught' and not seeing those in application at least for me hurts.
noble skiff
yeah, it's definitely much easier with a freshman DSA course under your belt
hearty harness
once we get into the pathfinding, searching etc. I start to lose it and need to start watching and reading what others are applying and I was trying to get an idea of what I should be reviewing.
noble skiff
i'd suggest just doing a few DSA classes on coursera
if what you want is to learn this stuff
hearty harness
yeah, I would like to expand my knowledge in those areas for sure. Then again, in my day to day I hardly ever encounter these types of problems or I am ignorant to the solution and shape of the problem that fit together. I will check out some of the DSA documentation.
hidden musk
just use the oop parts of ocaml :D
pine tiger
does it have for loops
pine tiger
i mean last time i tried haskell and i couldn't get past day 7 lol
granted it was 2019 but still
ripe meteor
+1 for ocaml or f#
pine tiger
no but seriously having worked with imperative for so long
i think i'm too dumb for functional lmfao
ripe meteor
both langs give you an "out" if you want to mutate stuff or iterate imperative style
agile silo
|| ```py
def sum_distances(gxs, rows, cols):
total=0
for (ga_x, ga_y), (gb_x,gb_y) in gxs:
dist=0
if ga_x < gb_x:
dif = gb_x - ga_x
for i in range(ga_x+1, gb_x):
if i in cols:
dif+=1
dist+=dif
else:
dif = ga_x - gb_x
for i in range(gb_x+1, ga_x):
if i in cols:
dif+=1
dist+=dif
if ga_y < gb_y:
dif = gb_y - ga_y
for i in range(ga_y+1, gb_y):
if i in rows:
dif+=1
dist+=dif
else:
dif = ga_y - gb_y
for i in range(gb_y+1, ga_y):
if i in rows:
dif+=1
dist+=dif
total+=dist
return total
i dont get what's wrong in my logic
i am getting in day11 part1 less than i should
I ||count empty cols x2 and same with rows||
simple raven
|| ```py
def sum_distances(gxs, rows, cols):
total=0
for (ga_x, ga_y), (...
if you want help, use #aoc-solution-hints where you could even open a new thread just for yourself and when you do, post the whole code so people can run it on the test input
magic oxide
idk how i feel about day 20
agile silo
if you want help, use <#1178802982981214278> where you could even open a new thr...
oh, okay sorry
simple raven
π
magic oxide
definitely a go to bed moment
have some ideas but i'm gonna need some sleep to solve this
simple raven
gn
agile silo
i was confusing x with y's
pine tiger
that's why i always use rows and columns to represent my points instead
ivory vigil
dim pivot
I always use grid[y][x]
stark flume
its even more confusing when game libraries invert y...
rancid creek
when 99% of the time in math you go (x, y)
ues dictionaries then. π
gardens = { (x,y): cell for y,row in enumerate(aocdata) for x,cell in enumerate(row) }
by hit you mean "operate faster" then sure. π
also boundary checks are a lot easier.
rancid creek
especially useful for any sort of problem where you want to exclude content or find a specific point.
gardens = { (i,j): cell for j,row in enumerate(aocdata) for i,cell in enumerate(row) if cell !='#'}
start = next((i,j) for i,j in gardens if gardens[i,j] == 'S')
i dont know what to do
start over from what you remember, or grab someone else's code and feel guilt free about it.
which day were you working on?
earnest forge
GLHF everyone
sharp panther
Language Roulette: Day 22
The language is ... Julia
sharp panther
**Language Roulette: Day 22**
The language is ... *Julia*
covert arrow
GLHF
shut trellisBOT
<@&518565788744024082> Good morning! Day 22 is ready to be attempted. View it online now at https://adventofcode.com/2023/day/22. Good luck!
covert arrow
my code isnt working for some reason π¦
fleet fox
im gonna lose it
sharp panther
:ooo congratulations!
earnest forge
I'm losing it, why am I getting the wrong answer on the real input
What edge case am I missing lol
flat sequoia
MY PLEASURE RECEPTORS
Congrats
earnest forge
Are you ||sorting the bricks||
Yes
flat sequoia
Oop
earnest forge
You can see my code in the spoiler channel
pine tiger
haha i didn't even ||sort the bricks||
my solution was the worst you could ever think of
modern terrace
i sorted the bricks
what to do
uh
why part 2 wrong
why part 2 wrong ;-;
work for test input
not for real
:(
flat sequoia
haha i didn't even ||sort the bricks||
Wtffff
Man I failed so hard today
First I failed at reading
And then I failed at writing
modern terrace
my part 2 no workkk
pine tiger
how are you doing aoc on a plane
flat sequoia
pine tiger
what wifi
flat sequoia
I canβt believe I paid for the expensive ass $40 plane wifi only to get 6 points
Itβs usually like 8 bucks
But on this flight it was 40 for some reason
modern terrace
how do part 2
pine tiger
imagine "failing" and still getting t100
modern terrace
I canβt believe I paid for the expensive ass $40 plane wifi only to get 6 points
top 100 is NOT a failure
coming from an average aoc participant
modern terrace
anyways why does my part 2 not work
flat sequoia
Like yesterday and the day before I genuinely couldnβt solve part 2 in time at least
pine tiger
You did so well!!!
it was a fluke, really
modern terrace
it ||accumulates the bricks that can fall and subtracts the bricks that can't fall from it||
flat sequoia
it ||accumulates the bricks that can fall and subtracts the bricks that can't fa...
R u checking falling in order
And removing fallingones before checking next ones
modern terrace
it ||accumulates the bricks that can fall and subtracts the bricks that can't fa...
||```py
sup is a list of the bricks that a certain brick supports
cache = [*map(len, sup)]
newcache = cache[:]
nmin = [0]*len(cache)
for j in range(len(sup)-1, -1, -1):
x = sup[j]
for i in x:
if nsups[i] == 1:
cache[j] += cache[i]
else:
nmin[j] += 1
p2 = sum(cache) - sum(nmin)
Oh wow you did this completely differently from me
pine tiger
yk speaking of plane aoc
i recall doing a day on bart
didn't think i even t1000'd that one
and it was an impl day asw
earnest forge
I'm really not seeing my bug, wtf
modern terrace
I finally did it
can u help wit my code.. pls
earnest forge
Same here LMAO
covert arrow
Is it in #aoc-solution-hints ?
modern terrace
k i'll post it there wait
earnest forge
I'm the only one to post in the spoiler channel so far and it doesn't even work
pine tiger
what time zone ru in
pine tiger
what region of the world is that
covert arrow
china
covert arrow
I'm also in shanghai 
pine tiger
massive
flat sequoia
i recall doing a day on bart
Yooo youβre from the bay?
Lowkey train is more rocky than plane
Head would hurt
Also it goes underground and u lose internet anguishe
Ayy I have landed sfo
devout tusk
I'm also in shanghai <:ducky_concerned:1178032077514477629>
since when did Discord start working in China? π€
You on VPN?
covert arrow
shhhhhh
covert arrow
since when did Discord start working in China? π€
You on VPN?
Yes
hollow wharf
Aoc's still going on?
molten fulcrum
Aoc's still going on?
Yeah. Last day is the 25th
real flare
Hello
lime fulcrum
hi
magic oxide
i am soooo glad i have a display_visited function leftover from last year's aoc
so nice
also nice to see ||my guess about day 21's assumption last night in bed was correct||
ripe meteor
21 part 2 was terrible and i kinda cheated but i dont care tbh
lime fulcrum
21 part 2 was terrible and i kinda cheated but i dont care tbh
define cheating?
ripe meteor
I looked at solutions and hints
I never would have gotten it by myself
It didnt even occur to me to look at my input
Dw im not in the leaderboard
dim pivot
3 more days left....
lime fulcrum
I wouldn't say that is cheating. If I am completel stumped then I do that as well
dim pivot
How many do you think will be "hard"
I don't think there will be any more 3d processing thankfully
lime fulcrum
beacons?
sharp panther
usually i think difficulty peaks around 18-22?
sharp panther
yesterday was a pretty fast day, global lb finished in less than a half hour
tall wasp
yesterday was a pretty fast day, global lb finished in less than a half hour
yesterday? fast??
sharp panther
compared to 21? yeah
lime fulcrum
oh no we havent had beacons yet
because I heard some horror stories about that one
sharp panther
oh no you're right, 22 is today
lime fulcrum
still on 19 XD
sharp panther
but it was before i slept and thus i'm calling it yesterday
tall wasp
because I heard some horror stories about that one
still havent solved 2021 beacons
2022 beacons was a little more doable
tall wasp
but it was before i slept and thus i'm calling it yesterday
ah fair, makes sense
lime fulcrum
but is a very cool problem, hard yes but very cool
tall wasp
yup
noble skiff
I feel like the trend of needing to "explore" the input and discovering its specific properties to solve it continues
day 20 and 21 were again super blatant in that regard
noble skiff
it could just be my memory of course but I definitely do not remember ever feeling this way, to such a degree, in past years
magic oxide
yesterday reminded me of a 2021 day 24
noble skiff
like, I remember in past years there were things where you didn't need to use LCM because a bunch of numbers were co-prime... but LCM is like grade school arithmetic anyway
we've already have like, at least 3 days where the "full" solution would be crazy, but when you realize the specific properties of the input it's actually quite easy
so the main challenge is actually just verifying those properties
For day 20 part 2, as soon as I read the vaguest hint about it, I immediately understood where it was going, I made all the assumptions, and it worked
I'd be curious if this is an intentional change by the creator if they do not see it that way
rancid creek
2022 beacons was a little more doable
There were 2 beacons puzzles? There was 2021/19 which was fun. rotating groups of beacons and matching them up. What was the other one?
pine tiger
Yooo youβre from the bay?
msj co 23!!
mossy basin
i remember using cv2 for beacons
tall wasp
There were 2 beacons puzzles? There was 2021/19 which was fun. rotating groups o...
not at all the same kind of puzzle but I remember when I read this puzzle title I got war flashbacks to 2021/19
https://adventofcode.com/2022/day/15
rancid creek
not at all the same kind of puzzle but I remember when I read this puzzle title ...
Ah yeah, more beacons! I'd completely forgotten about that one.
rancid creek
oh it was 2021 day 19, used cv2.estimateAffine3D
cheats. π
but cleaner than this mess:
||
psub=lambda p1,p2: (p1[0]-p2[0],p1[1]-p2[1],p1[2]-p2[2])
padd=lambda p1,p2: (p1[0]+p2[0],p1[1]+p2[1],p1[2]+p2[2])
rotate=lambda s,a,b,c,i,j,k: (a*s[i],b*s[j],c*s[k])
rotations=lambda scan: {r: {rotate(s,*r) for s in scan} for r in orientations}
mhd=lambda a,b: abs(a[0]-b[0])+abs(a[1]-b[1])+abs(a[2]-b[2])
||
mossy basin
yeah, i like when i can apply my esoteric cv2 knowledge
i had a fun day 5 solution
for 2021, where you had to do lines, but i got cv2 to do it
tall wasp
Ah yeah, more beacons! I'd completely forgotten about that one.
haha yup i had to find the day as well but I remembered the puzzle was there
rancid creek
The only time knowledge of another library really helped me in aoc was the time I had just helped someone with a CRT question on learnpython and the next day was a CRT question on aoc.
'part 2:', sympy.ntheory.modular.crt(m,r)[0]
my utilities has crt in it now
rancid creek
my utilities has crt in it now
I really should start using your lib, or write my own, instead of reinventing the wheel every year.
mossy basin
but i did actually take number theory once upon a time and the only thing i remember is crt and quadratic reciprocity
i think we spent a week proving quadratic reciprocity like 4 ways
rancid creek
but i did actually take number theory once upon a time and the only thing i reme...
jeeze, I'm lucky I remember pigeon hole.
like almost all of the math I learned way back when has faded to fuzzy memory. as long as I can remember something loosely exists I can google it.
mossy basin
wait until we need to solve some quadratic equations modulo something and i'll flex on all these fast guys
or i won't because they know already
the math i use now is too specialized to do me any good in aoc
noble skiff
I remember a decent amount of math but none of it is relevant lol
it's all continuous stuff; tons of courses on calculus, analysis, etc
pine tiger
nothing discrete?
noble skiff
formally, no, except for some counting (i.e. in probability/physics classes)
it's mostly just pure math and CS people who do significant amounts of discrete math
outside of that most fields use a lot more continuous math
dim pivot
I hope there are more maths days but I know it's not coming
We already had quite a few earlier (18,20,21)
ripe meteor
The channel name clearly states advent of code
rancid creek
Waiting for something to hit my speciality and then realizing I might not have one it can hit. π
earnest forge
π€
woven sable
cheats. π
but cleaner than this mess:
||
```py
psub=lambda p1,p2: (p1[0]-p2[0],...
omg i remmeber this day
i hard coded every possible orientation
and spent 6 hours figuring out why my code wouldnt work
then i j went to sleep and solved it in like 30 min the morning of
rancid creek
then i j went to sleep and solved it in like 30 min the morning of
sleep is either underrated or under-utilized. π
rancid creek
i hard coded every possible orientation
I hard coded them at first, and then realized I could generate them, and then went back to the hard coded version anyway.
dim pivot
My fav days of all time are probably day 24 2021 (the one with reverse engineering) or day 22 2019 (the one with modlar inverse and geometric series)
pine tiger
yk i'm having kind of a hard time doing day 22 efficiently here
lime fulcrum
can someone help me debug something for 19 2? i'm going crazy here
zinc walrus
Hi, can somebody answer me a question for d14-1, please? First riddle is that I only pull on the north side and all round stones need to go as far up as possible?
pine tiger
Hi, can somebody answer me a question for d14-1, please? First riddle is that I ...
yup
zinc kindle
I'm going through old AoC problems and I'd like some clarification for 2018 day20 part2: ||does the number of rooms include rooms along a path, or just rooms at intersections/dead ends? e.g. if input was "N"*2000, would answer be 1001 or 1?||
zinc walrus
yup
Thank you. I just finished D14 part 1.
My rolling algo is a bit stupid, but what I noticed: when I replace the string values for numbers, I can actually exploit some calculations. But I have to go still through the matrix multiple times. Hmpf
zinc kindle
Alsp, part1 of 2018 day20 was a pain. wrote algorithm that looked plausibly correct, gave too high answer. Wrote a more precise algorithm, and it turns out that no, my original algorithm was correct, I just forgot to .strip() the trailing whitespace in the input.
cyan sleet
so 3 days left, predicting ||rust||, ||typescript|| and ||python|| for the roulette
zinc walrus
Is the answer for each puzzle btw coupled to the account? Or does everybody get the same solution for each day's parts?
minor cave
Is the answer for each puzzle btw coupled to the account? Or does everybody get ...
There are several different inputs, but not enough to be unique for each account
pine tiger
I'm going through old AoC problems and I'd like some clarification for 2018 day2...
doesn't it talk about the # of doors?
earnest forge
Today had better be nice and easy because I didn't get much sleep
covert arrow
GLHF
hexed axle
We're here, the roulette ball just takes a bit of time to land
hexed axle
It was the easiest way I could think of to make Advent of code a bit more analog
earnest forge
GLHF everyone, 5s!
shut trellisBOT
<@&518565788744024082> Good morning! Day 23 is ready to be attempted. View it online now at https://adventofcode.com/2023/day/23. Good luck!
modern terrace
pathfinding :(
covert arrow
^
hexed axle
Language Roulette: Day 23
The language is ... JavaScript/TypeScript
cyan sleet
1 prediction down
hexed axle
Your prediction power is impressive, use it wisely
pine tiger
YOOOOO
earnest forge
Nicely done
I can't quite get ||my heuristic for the path that will be the longest|| right
pine tiger
I can't quite get ||my heuristic for the path that will be the longest|| right
||heuristics aren't necessary||
earnest forge
Bruh
covert arrow
how long would a ||brute force for p2|| take?
earnest forge
More than 17 minutes
covert arrow
hmm
woven solar
today I decided to treat myself and write a nice solution with ||networkx||
part2: now solve an NP-complete problem
earnest forge
Lmao
covert arrow
Maybe you could ||collapse some of the paths|| and it would run faster?
pine tiger
Maybe you could ||collapse some of the paths|| and it would run faster?
||that is precisley what i did!||
earnest forge
I see how the input is structured but honestly I don't see how that would speed up my searcher
earnest forge
||that is precisley what i did!||
Why does this work lol
tall wasp
im getting wrong answer for some reason π©
pine tiger
Why does this work lol
lucid narwhal
i got just 75 π
flat sequoia
OH SHIT
No wonder
BRO WHAT
I thought I fixed the issue
but I'm still getting 175
oh
I just didn't fix it lmaooo
flat sequoia
||that is precisley what i did!||
How long does your solution run in lmao
Should I wait mine out
covert arrow
π₯³
pine tiger
How long does your solution run in lmao
<30 seconds
modern terrace
how do i do part 2
woven solar
pain and suffering
modern terrace
oh
woven solar
i ||simplified the graph by collapsing corridors|| and it still took me >10 minutes because networkx is slow as fuck. probably a handwritten bruteforce would be much better.
woven solar
it's such a shame that the premiere graph library for python with all the fancy algorithms is also pure-python.
covert arrow
wdym by that
||if b is only connected to a and c then you can just have a path between a and c||
modern terrace
||if b is only connected to a and c then you can just have a path between a and ...
i have no idea how to do that with my current code but i'll go for it nonetheless
covert arrow
I'd recommend ||converting the grid into a neighbor dict||
woven solar
(AFAIK there's no solutions other than bruteforce with various optimizations because this problem is NP-hard in general (though not sure if it's NP-hard on a grid, probably yes). some people used ||heuristics||, which makes it much faster but is hard to guess at.)
pine tiger
(AFAIK there's no solutions other than bruteforce with various optimizations bec...
grid -> planar graph -> i don't see anything about that which makes the problem less np complete
onyx knot
modern terrace
I'd recommend ||converting the grid into a neighbor dict||
ok i made a solution that works for the TI, fails for RI
modern terrace
do i have to bruteforce this
dim pivot
I don't see why collapsing corridoors is so effective but it worked for me...
Yes it's NP hard.
I'm really bad at such problems - I like knowing what I have to do
for part 1 I overkilled it using topological sorting so was not that fast
magic oxide
im so confused on day 22
my solution worked for p1, doesn't work for p2
and i just found a bug that means it shouldn't have worked for p1, so i just got lucky?
and it still doesn't work
just figured it out
somehow my two bugs just cancelled each other out and gave me the right p1?
noble skiff
Strange things can happen
magic oxide
truly wild
feral hazel
won't be able to make it today or tomorrow but it was fun while it lasted
see you all next year
(or in june if we're doing roc :)
feral hazel
what's p2 of day 23 btw?
||brute force :D||
pine tiger
any way for me to uh
have a type annotation that specifies
"you can index this and it should give a string as a result"
rapid wolf
"you can index this and it should give a string as a result"
typing.Sequence[str]
rapid wolf
I don't believe it is, it has several crucial things that don't have an alternative
feral hazel
yeah
though iirc now the collections from collections.abc are preferred
eg collections.abc.Sequence
pine tiger
that name is so much longer tho π
feral hazel
from collections.abc import Sequence :D
hollow wharf
import collections.abc as typing
typing.Sequence[str]
``` gThe ironic thing is that by making list and dict more convenient built in type annotations
They discouraged indirectly using Sequence and Mapping
Which should be used a lot more often
hidden musk
so true
sleek cave
I SOLVED DAY 17 LETS GOOOOO
AFTER FIVE HOURS
and not understanding the solution
i made some sense of it
still need to figure out why this ||modified version of dijkstra|| works lmao
elfin sun
why are the days scrambled on the website
pure dove
would anyone here be kind enough to help me out with day 21 part 2 π my thing works for all the test cases provided but not the right answer for my input and ive been at this for 2 days now and idk whats wrong
pine tiger
would anyone here be kind enough to help me out with day 21 part 2 π my thing w...
wanna tell me your approach/code in the solutions thread?
no guarantees i can help you out but i'll try
pure dove
wanna tell me your approach/code in the solutions thread?
no guarantees i can he...
wait do I ask this in the spoilers channel or the solution/hints channel
pine tiger
idk i'm more acccustomed to speaking there so
cyan sleet
yeah
though iirc now the collections from collections.abc are preferred
eg colle...
collections like collections.abc.Callable π₯΄
!d collections.abc.Callable
marsh currentBOT
sleek cave
day 23 ... definitely no time to solve this today, just caught up with day 17 as winter break is finally here
but first thoughts give me ||longest path in a DAG|| which should just convert to ||shortest path in the negated graph||? i think there might be something fishy with the slopes, though. and since ||the graph is unweighted|| there might be a much easier way
cyan sleet
day 23 ... definitely no time to solve this today, just caught up with day 17 as...
longest path in a graph doesn't translate to a nice shortest path problem
in fact, it's an NP-hard problem https://en.wikipedia.org/wiki/Longest_path_problem
however in part 1 ||the directedness of parts of the graph makes things easier||
and even in part 2 ||you can reduce the problem to a pretty small graph where attacking the NP-hard problem is feasible||
sweet star
why does your laptop look wet
pine tiger
think it's just the texturing
modern terrace
in fact, it's an NP-hard problem https://en.wikipedia.org/wiki/Longest_path_prob...
does NP hard translate to "it needs to be bruteforced"
modern terrace
it means "you can't solve this polynomial time"
i have no clue what that means
pine tiger
do you know what big o notation is
pine tiger
we don't know that
i'm willing to bet money p != np
cyan sleet
do you know what a polynomial is?
magic oxide
you need subscripts on those a's
magic oxide
def polynomial(x):
coefficients: list[number]
return sum(c * x ** n for n, c in enumerate(coefficients))
essentially
cyan sleet
so there is the class of problems called NP where we can verify a solution in O(polynomial(n))
modern terrace
like Ξ£_i=1^n aα΅’xα΅’^i?
magic oxide
i=0
cyan sleet
so there is the class of problems called NP where we can verify a solution in O(...
and the class P where the problem is also solvable in O(poly(n))
P is a subset of NP
NP-hard means "this problem is as hard as any other problem in NP"
and longest path is one of those NP-hard problems
as hard as a bunch of other problems we also don't know how to solve efficiently
modern terrace
so basically it's just problems that take the most time to solve?
cyan sleet
NP-complete problems are problems where we know how to verify an answer efficiently, but we currently don't know if we can also solve them efficiently
cyan sleet
the open P=NP problem is asking whether all problems that can be verified efficiently can also be solved efficiently
pine tiger
the open P=NP problem is asking whether all problems that can be verified effici...
personally, i'm 99.999% sure p != np
but maybe one day the next turing will come along and prove it is
mossy basin
will prove that it's undecidable
pine tiger
π
cyan sleet
annoyingly p != np is the annoying case to prove
pine tiger
π
cyan sleet
non-deterministic polynomial
which is about non-deterministic computing
mossy basin
speaking of nondeterministic, i was think of trying non-deterministic solutions to yesterday
like implementing https://en.wikipedia.org/wiki/Ant_colony_optimization_algorithms on the graph
modern terrace
annoyingly p != np is the annoying case to prove
does comparing P with NP using > or < make sense
cyan sleet
which is about non-deterministic computing
i.e. imagine some non-deterministic program, is there any possible outcome where it solves the problem in poly-time?
P is a subset of NP
so the subset symbol would be relevant
which is kinda β€
magic oxide
if you can solve a problem in polynomial time you can verify it in at least polynomial time by just solving it, so P <= NP
modern terrace
is there such a case that NP would take more time than P
or is it always that the time taken by NP is equal to and potentially less than P
cyan sleet
if you can solve a problem in polynomial time you can verify it in at least poly...
but with the set relation rather than an inequality
cyan sleet
is there such a case that NP would take more time than P
I mean yeah there are plenty of specific problem instances where a "harder" problem can be solved fast, while an "easy" problem can take long time
like solving an NP-hard problem for tiny inputs, while solving a P problem for hugr inputs
magic oxide
day 23 is gonna be interesting to see the solutions for
i'm doing it now and i've become painfully aware how little graph theory i actually know
cyan sleet
asymptotics don't say much about absolute running times
cyan sleet
day 23 is gonna be interesting to see the solutions for
i'm doing it now and i'v...
you don't actually need that much of it
for sure the stuff you need you already know
magic oxide
rn i'm trying to conceptualize a ||longest path algorithm on the graph||
||does dp work?||
cyan sleet
what's the dumbest idea you can come up with?
earnest forge
(and can you implement it well)
((mine took 12 minutes but someone managed to get 23 seconds))
cyan sleet
mine runs in 0.6s, but that's after a bunch of improvements
cyan sleet
my initial thing took like 2 minutes
sleek cave
in fact, it's an NP-hard problem https://en.wikipedia.org/wiki/Longest_path_prob...
right, but our input is ||acyclic||
earnest forge
||not for p2||
modern terrace
is day 25 always easy
sleek cave
right, but our input is ||acyclic||
and since ||you can't backtrack|| it might as well be ||directed||
i don't know
earnest forge
is day 25 always easy
Yeah, usually
sleek cave
these are my first thoughts lol
cyan sleet
is day 25 always easy
it's usually easier yeah
sleek cave
||not for p2||
not reading this spoiler bc i don't know to what extent it gives away day 23 lol :P
cyan sleet
it's usually easier yeah
and ||only 1 part||
magic oxide
my p1 gives 93 for the test input 
cyan sleet
not reading this spoiler bc i don't know to what extent it gives away day 23 lol...
it gives away what the twist is
earnest forge
not reading this spoiler bc i don't know to what extent it gives away day 23 lol...
Not very much (IMO), but it is in reference to part 2
cyan sleet
my p1 gives 93 for the test input <:PensiveCat:783736784168681493>
missed counting the first or last step?
magic oxide
an error in graph construction i think
so yeah
for the test input, it takes 15 steps to reach the first fork from the start right?
cyan sleet
an error in graph construction i think
oh did you ||immediately reduce the graph to the minimal thing||?
or just doing the grid wrong?
magic oxide
oh did you ||immediately reduce the graph to the minimal thing||?
||yeah||
oh i just wrote 0 instead of 1
earnest forge
Lmao
cyan sleet
lol
I did the reduction only after badly solved p1
but I also did the slope logic
it gets pretty intricate
cyan sleet
which part fails? the "immediately" or the "right answer"? π
earnest forge
Sounds like ||you're pruning too aggressively||
