advent-of-code

[cyan sleet]

someone good at math?

In [20]: def f(a, k):
    ...:     inc = a.pop(0)
    ...:     for i in range(k):
    ...:         a[i] += inc
    ...:     return inc
    ...:

In [21]: a = [1]*100; [f(a, 0) for i in range(10)]
Out[21]: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1]

In [22]: a = [1]*100; [f(a, 1) for i in range(10)]
Out[22]: [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]

In [23]: a = [1]*100; [f(a, 3) for i in range(10)]
Out[23]: [1, 2, 4, 8, 15, 28, 52, 96, 177, 326]

In [24]: a = [1]*100; [f(a, 4) for i in range(10)]
Out[24]: [1, 2, 4, 8, 16, 31, 60, 116, 224, 432]

In [25]: a = [1]*100; [f(a, 5) for i in range(10)]
Out[25]: [1, 2, 4, 8, 16, 32, 63, 124, 244, 480]

In [26]: a = [1]*100; [f(a, 6) for i in range(10)]
Out[26]: [1, 2, 4, 8, 16, 32, 64, 127, 252, 500]

[lusty rover]

Not I

[hidden musk]

O( a lot )

[cyan sleet]

oh, k=2 is fibonacci - 1

so I guess it would grow quickly if we actually had full matches all the time

but in reality we probably don't have that many most of the time

[hidden musk]

hmm. could probably make a fun matrix operator solution thing for shuffling the numbers around

[leaden tartan]

Ok im losing my mind

[lusty rover]

Welcome to the club

[leaden tartan]

I feel I am going insane

I cannot compare 2 numbers

[lusty rover]

Hey, comparison is hard

[leaden tartan]

why does if (num1) in [list of numbers] not work

[royal gull]

!e if (1) in [1,2,3]:
print("Yes it is")

[marsh current]

@royal gull :white_check_mark: Your 3.12 eval job has completed with return code 0.

Yes it is

[royal gull]

but, perhaps you have a list of strings?

[leaden tartan]

Ive tried converting them

Ive tried direct comparing them 1 by 1

Ive tried copying the numbers out of the debug window and comparing those (it works)

[royal gull]

Let's open a thread in #aoc-solution-hints

[jovial jay]

unrelated, but i just discovered the reason why text files ends with a new line even if i didn't want that when doing the advent of code

[low condor]

Just use .splitlines()

[weary kelp]

What's the reason?

[low condor]

Mostly because Unix systems said so - that’s how they define a line

But there’s good historical reason why they did that - which someone else can explain :)

[jovial jay]

I'm not using Python for the AoC. I'm challenging myself by using a domain-specific language geared for image creation/editing to do the AoC.

[jovial jay]

because some programs wouldn't parse text files correctly if it didn't end in new line. I figured some of my code wouldn't work if I had to account for the lack of new line at the end.

[low condor]

Files ending in a new lines make sense IMO

If you concat 2 files together it just works

[bronze vault]

aoc is awesome

[weary kelp]

indeed tis

[bright laurel]

i feel like the purpose of the site is to teach you how to make parsers. the questions themselves are easily solvable if they were more organized

[jovial jay]

Yeah, it's all about parsing

There should be other things imo

[bright laurel]

indeed

[weary kelp]

advent of parsing

[hidden musk]

the early days this is true. theoretically, the problems will get harder until parsing becomes just the easy part

[low condor]

Also only two out of four days has had difficult parsing. The other two were very easy

I’m more a oneight fan :)

[somber trench]

twone 🔛🔝

[noble skiff]

Wait which days had difficult parsing

[peak dock]

d1p2 and d3

[noble skiff]

Feel like everything so far was pretty trivial to parse

What made d3 challenging to the degree it was, was not the parsing

Organizing the data and writing the algorithm over it in a clean way was

Day 1 had the one edge case in the parsing I suppose that got a lot of people (myself included)

I don't think AOC is generally trying to make parsing hard, generally the data is pretty cleanly separated and you can just split() your way through.
The days with recursive parsing are the only ones I think where it's challenging to parse really

[lapis pilot]

that was absolutely nasty today for me

recursion makes me v emotional

[muted aurora]

Timing execution for all 4 days...

real    0m0.024s
user    0m0.011s
sys     0m0.009s

Pretty happy with that so far

[honest nacelle]

Was Day 3 weirdly difficult for this early on? Or did I just miss an easy solution?

Also, how are the resident golfers doing?

[peak dock]

theyve set up their residence in #1180008645384216606

[honest nacelle]

Thank you, my head already hurts

[jovial jay]

how bad can day 5 be? i'm waiting for that

[honest nacelle]

Setting myself a strict 30 min time limit before I have to go to sleep

[sharp panther]

Language Roulette: Day 5
The language is ... Prolog

[honest nacelle]

I don't even know what prolog is...

[sharp panther]

Prolog is a logic programming language, which is a paradigm that's probably vastly different than anything you've seen before

https://en.wikipedia.org/wiki/Prolog gives a rough overview

[honest nacelle]

Maybe I'll make it my goal to go back after the event and do the daily language stuff

@.@ yeah that'll have to wait

[hexed axle]

That's uhh.. you know what? Got a new appreciation for Python's readability from that wiki page

[covert arrow]

glhf

[earnest forge]

GLHF

[shut trellis]

<@&518565788744024082> Good morning! Day 5 is ready to be attempted. View it online now at https://adventofcode.com/2023/day/5. Good luck!

[honest nacelle]

for real, head is spinning at teh quicksort

[jovial jay]

ok, day 5 is confusing

[honest nacelle]

Oh god...

[jovial jay]

you find it confusing too?

[sharp panther]

this explanation is quite long

[honest nacelle]

Yeah, though I'm getting it I think. Now I wonder how to efficiently parse it

[jovial jay]

you're getting it? wow

[sharp panther]

zero part 2s yet, with over 70 part 1s

sounds like part 2 is a doozy

[hexed axle]

It looks like it's really easy to make a time inefficient solution, I bet part two ruins that

[sharp panther]

and now the p1 leaderboard is filled. first time so far this year.

[hexed axle]

with a single 2 star person

[covert arrow]

what is this p2 😭

[jovial jay]

i'm getting p1 and the leaderboard is filled?

[sharp panther]

yep, 08:15 for p1 #100

[honest nacelle]

I need to stop looking at how fast y'all are getting this before I get discouraged lol

[naive stone]

what did i just read

[compact wasp]

why is this so cringe

[wet yoke]

Curious now what pt 2 is though I'm only doing the code during work hours. 😦

[jovial jay]

pt 2 is probably really bad to deal with. i bet.

[wet yoke]

I see a solve in my head for 1, so eh

[earnest forge]

My vscode crashed wtf

[woven solar]

huh, wrong answer.
EDIT: dumb mistake, nevermind

[jovial jay]

i think i'll probably drop off AoC after this

[magic oxide]

wow that's a long problem description

[pine tiger]

oh god wth is p2

[velvet stag]

F

[livid gull]

my pc is frozen shouldve remembered aoc doesnt like small numbers

[honest nacelle]

Keep with it, there's a lot of speed demons out there but take it at your own pace

[woven solar]

oh no, ||sure hope I won't need sympy for this||

[earnest forge]

That's probably a good idea

[hexed axle]

I've only completed day 1 part 1 thus far. I'll pick them up when I get more free time, they're fun puzzles when you take it at your own pace

[earnest forge]

I'm kinda wondering if Z3 is a good fit

[sharp panther]

and it's done!

[naive stone]

no matter how many times i read this prompt, i don't get it

[hexed axle]

26 minutes on day 5, it's shaping up to be quite the AoC this year

[pine tiger]

lmao

that was hell

[flat sequoia]

man I trolled so hard

I advanced the stage BEFORE calculating things

so I was looking at all the wrong ranges

[pine tiger]

💀

[compact wasp]

💀

[pine tiger]

idk it took me way too long to realize ranges were needed

wonderful code right here tbh

[naive stone]

I don't understand, why in the example does the seed to soil map have 2 lines? why does it have 2 destinations and 2 sources? which is the right one? am i stupid?

[pine tiger]

it's a set of mappings

like "transform this range by an offset to that range"

each 3-set of numbers is that

[naive stone]

i'm still not getting it

[woven solar]

this is like the third time I see an "ugh, it's fucking ||interval intersections|| again" task in AoC, yet for some reason I have yet to make utilities for that. smh, past-reptile

[honest nacelle]

man y'all are nuts

[stable shale]

My brain can't math this late at night

[honest nacelle]

I'mma sleep on it, night y'all and good luck

[stable shale]

Yupppp

[woven solar]

if the seed is from 98 (inclusive) to 98+2 (exclusive), it gets mapped to 50..52. If the seed is from 50..52, it gets mapped to 52..54. otherwise it gets mapped to itself.

[gentle cape]

My day one solution is a mix of a web api, python, and Haskell

[magic oxide]

today looks interesting

looking forward to it

[woven solar]

aoc day5 be like: ||calculate the minimum of the image of a union of sets under a given transform||

[spice tiger]

AHHHHH IM MISSING RANGES

my nums are skipping over way too much

[naive stone]

i'm still so confused. These ranges overlap

[ripe scroll]

how do they even come up with this kind of problems

[wet yoke]

A contrived example: ```
10 5 2
3 1 3

This means a source range starting at 5 of length 2 (so 5 and 6) maps to destination range of the same length starting at 10 (10, 11).
There's also a range starting at 1 length 3 (1,2,3) mapping to destination starting at 3 (3,4,5).
Any numbers not covered by a source range are just 1:1 mappings.

So you would have:

1 -> 3
2 -> 4
3 -> 5
4 -> 4 # not covered, 1:1
5 -> 10
6 -> 11
7 -> 7 # also 1:1

and so on

Essentially you map a seed from the source to the given destination. And then keep doing that through like 6 other maps to end up in a "location".
For all of those starting seeds, find their resultant locations and then find the one that's lowest.

[livid gull]

i've just solved part 1 after using a for loop to add the maps to my map lol, which froze pc obv

[wet yoke]

I don't think that matters.

[woven solar]

they don't, no. range(50,52) and range(52,54) have no common elements. nor do range(98,100) and range(50,98), if you mean the source ones.

[naive stone]

these ranges overlap don't they?

50 98 2
52 50 48

how would that not matter? that would mean that a source has 2 different destinations

[wet yoke]

No, 50 length 48 stops at 97

50 is included as the first

[woven solar]

the length is the third number

[naive stone]

right, ok. So ranges never overlap then, this is just multiple ranges

[woven solar]

yup, source ranges should never overlap. destination ranges... I think it wouldn't be a problem if they overlap (haven't checked if they do in my input)

[wet yoke]

I figure some destinations may overlap? Don't think that part matters.
Sources should not, though

[livid gull]

if destinations overlap thats a problem

right?

2 seeds could go to same spot

[woven solar]

yeah, but I don't see why that'd be a problem for this task

[wet yoke]

As long as there's a direct source -> destination with a single answer, it should be fine

[naive stone]

ok so then the first range only tells you where seeds 98 and 99 go, the second range tells you all seeds from 50 to 97, all other seeds are identical to its source number?

[pine tiger]

lmao my code is so garbage

[wet yoke]

Right.

[pine tiger]

i'm gonna delete it and start again tomorrow

for actually presentable code

[naive stone]

alright, thanks guys

[wet yoke]

Go forth and beat me to it as I won't start for at least 10 hours 🙃

[naive stone]

lol

[pine tiger]

i find it somewhat amusing
that even though i'm 72 on global
i'm #2 on this place

[humble copper]

for the record: ||i hate range splitting||

[woven solar]

i suspect that the fact the destination start number comes first is a chatgpt measure, btw. it's just such a weird order to put them in 🥴

[humble copper]

oh you're sebbylaw?

[pine tiger]

ok but i hardly believe chatgpt could do that parsing thing

[pine tiger]

kevin

[naive stone]

yea that wasn't helping my understanding lol

[humble copper]

am i looking at the wrong leaderboard

[pine tiger]

uh idk

#aoc-bot-commands message

[placid lake]

caching the brute force way is too slowww :))))

sigh

[humble copper]

ah that one

[placid lake]

this was predicted to happen

[woven solar]

smh, you're reinventing the wheel. clearly the correct solution is to put all the data in SQL tables; that way you can take advantage of its highly optimized lookups 😉

[placid lake]

smh

[pine tiger]

i wonder is there a way to get the constant factor down enough

so that brute force works

[placid lake]

that's what i was hoping for

[royal gull]

||cry. I did this. Then looked at my input data||

[placid lake]

okay i guess time to try the other way

[humble copper]

if i wasnt doing importless i'd write a custom lib for problems SPECIFICALLY like this

[covert arrow]

I need to have a custom Range class with unions/intersects/etc. in my utils

[livid gull]

im now implementing that

its now my idea

[earnest forge]

Yeah I should update my custom range from being an iter to being a class

[compact wasp]

man what is eric cooking for future days 😭

this is not day 5

[humble copper]

day 19 is actually just gonna be "here's input for a conversation, write a LLM that can reasonably pass the turing test"

[earnest forge]

LMAO

[sharp panther]

in 2022, the first day to take longer than today's (to fill up the second star leaderboard) was day 15

[wet yoke]

"So it's AI, is it? Let's see you do this!"

Honestly I wouldn't be surprised if they ran a couple of these through GPT early to test whether it knew how to interpret the text...

[humble copper]

or, better:

The elves are feeling lonely on the island. They want to have more elves to talk to.
Regrettably, sending elves up to the island is quite costly at this time of year, and Santa is a bit of a cheapskate.
The elves have turned to making an AI that they can talk to about their elven lives.
Create a Pytorch model given some conversational training data.```

[magic oxide]

maybe odd days are just cursed this year

[candid berry]

oof

[compact wasp]

[brazen bridge]

For someone who isn't math-savvy at all, does anyone have any hints for part 2? I can't seem to figure it out at all

[humble copper]

||is there any way you can make the ranges more manageable?||

[flat sequoia]

Dude I implemented one a few days ago out of pure coincidence but I never implemented range subtraction so it was useless

[compact wasp]

😔

[flat sequoia]

I spent so long trying to figure out how to use it but for some reason I couldn't figure out range subtraction was the thing that was missing

so i ended up just wasting a ton of time lol

[candid berry]

||don't try to work with all the numbers. work with ranges.|| even spoilerier: ||but remember that a range might only partially overlap with a mapping, or might overlap with multiple mappings, so the number of ranges you're analyzing will constantly be growing as you need to split existing ranges up based on what got remapped and what didn't||

[humble copper]

out of curiosity, on average how many people participate in AOC every year?

[sharp panther]

take a look for yourself: https://adventofcode.com/2023/stats (this year), https://adventofcode.com/2022/stats (last year)

[humble copper]

(genuinely slipped my mind that that page exists, ty!)

[placid lake]

yay p2 done

damn ||intersecting ranges||

all too familar from past years

[candid berry]

these problems are feeling a lot less bite sized than usual for this early in

[royal gull]

P1 done after doing it twice. (In sql)

[candid berry]

ooh boy, get ready for part 2

actually, if you can get part 2 working in SQL, I'd be very curious to see it

[covert arrow]

I just went and implemented one on the spot today: ||https://paste.pythondiscord.com/QKUA||

[compact wasp]

i'm convinced it's anti-gpt

no shot eric doesn't know what he's doing here

nor his beta testers

[royal gull]

Yah, I’m. Unsure about this.

I guess I know what to do, I just don’t want to.

[candid berry]

yeah, I get that, but it's a shame if the anti-gpt measures make the whole thing less fun

[compact wasp]

yeah for sure

[brazen bridge]

especially for people who are new to programming/this kind of stuff and just trying to have a nice time for the first couple of days

[earnest forge]

Isn't that a multirange?

[covert arrow]

yes it is

[earnest forge]

Multiranges are neat but I don't think they should be the default for 'range'

[covert arrow]

I initially had a normal range but then changed it later

[compact wasp]

i know someone at lebanese university who told a bunch of people that the first 10 days would be trivial

[candid berry]

yeah. We usually tell people that the first days are pretty easy, and they should try them out. Today, not even part 1 was particularly easy, and part 2 would definitely not be approachable if you didn't already have a good understanding of working with ranges...

[compact wasp]

they're probably traumatized

[lime fulcrum]

Wtf is wrong today

[compact wasp]

lmao

[wet yoke]

If this was old WoW days I'd call this a gear check.

[woven solar]

No, that was day 3.

[lime fulcrum]

Duh

[royal gull]

Ok, im going to bed. I know -how- to approach part 2, but, boy I don’t want to.

[worldly raven]

haha, i feel that

[edgy hearth]

this advent of code is so hard wtf

puzzle questions

I did day 2 p1 wrong and I don't have the will to keep going idk how yall doing all this its gibberishhh

[magic oxide]

another day where i wish python had a bst built in

[broken dock]

(day 5 part 2 spoilers) holy hell that was some ||nasty backtracking||

wasn't expecting to see something like that this early on

[broken dock]

right?? i didn't know the first week was going to be like this

i didn't find days 1 and 3 difficult per se, more so immensely tedious

[atomic cairn]

God damn this is a whole book to solve

[naive stone]

i feel like i need an extremely light hint for part 2. I don't want to be spoiled, but i'm failing to come up with any ideas

[broken dock]

my spoiler is a pretty light hint

the word alone won't give much insight into the approach I came up with

but it at least tells you what general type of approach I went with

[earnest forge]

What on earth were you doing that involved ||backtracking|| lmao

[candid berry]

||work with ranges, not individual numbers||

[naive stone]

hmm

[broken dock]

do you wanna see my part 2?

rofl

[earnest forge]

I do

[broken dock]

i'll dm u

[placid lake]

||backtracking??||

[broken dock]

actually i think i'll hold off

[earnest forge]

https://ptb.discord.com/channels/267624335836053506/1181459209422917683 ?

[broken dock]

i didn't know that existed but still, i'll abstain for the time being

[timber glacier]

~~can someone help #1181484585574486088 ~~

[broken dock]

||i was initially thinking of doing a three-state DP, but turns out the DP isn't necessary||

||normal backtracking will run in time just fine||

[placid lake]

||interesting.. i'm not even sure how that'd work ngl||

[broken dock]

actually i'll post my solution wtv

[placid lake]

||i understood today as intersecting ranges||

[timber glacier]

i'm trying that but I'm just not getting it for some reason

[long jacinth]

D5P1 evaluation taking so long....

i better go shovel snow!

[broken dock]

gl 🫡

[bright laurel]

anyone knows edge cases for day 4, sample input gives right answer, unlike the actual input

[long jacinth]

python taking over 8 gigs

whew

i need more ram, lmao

[ripe scroll]

My laptop hanged lmao

it works though

[somber crow]

is it bad if my solution triggers the OOM killer on and hangs a 32gb machine?

[ripe scroll]

yep my fan sounds like a jet engine rn

let me try an Online interpreter. hope it doesn't DOS it.

[long jacinth]

yeah, also using dicts lol

[zinc kindle]

Ouch. I banged my head against a wall for over an hour in part2. and I figured out a correct approach more or less instantly (I have literally used elements of said approach for my job) it kept being buggy and giving the wrong answer on the sample input, and that's when it didn't enter an infinite loop.
Then I went and had dinner, rewrote part 2 in python, and after one trivial bugfix got the correct answer.

[long jacinth]

ill have to rewrite it later, work time now

[rich delta]

anyone tried using dicts and worked??

[long jacinth]

it works

if you have enough ram

[finite stag]

any bruteforcing works in that matter :D

[rich delta]

8 gb should be sufficient. if not then 💀

[long jacinth]

ive gotten over 12gb before i just killed it

[finite stag]

i had 16 and still crashed my program

[rich delta]

bruhh

backtracking it is

[zinc kindle]

Ugh. Turns out the issue wasn't due to a logic error in the code, it was because when converting the seeds in part 2, I used vec.windows instead of vec.chunks, resulting in my input being read incorrectly. My actual algorithm was fine. Can't belive I didn't notice that.
(equivalent of itertools.pairwise vs itertools.batches)

[rancid creek]

If your answer in python takes longer than a few seconds you might want to consider a different algorithm.

I don't think this is a brute force ish sort of solution. You could spin a long time.

[digital whale]

someone

hel pme

i dont understand day 3 at all

just doesnt make sense

so apparently from what they have said

they say that numbers adjacent to a symbol and diagonally

what do they mean by diagonally do they mean like

lets say the number is in line 1 at the 4th position

the symbol should be in the 3rd or the 5th position in the next line?

and

why is 58 not correct?

it is adjacent to a symbol no?

[bitter osprey]

If there is a symbol on line 41, position 121, a number on line 42, position 120, 121 and 122 would be correct.

[ripe scroll]

xxxx
x58x
xxxx

here 'x' is adjacent to 58 diagonally as well as from sides.

[digital whale]

oh?

then why is 114 not correct?

it is in the 4th position

[ripe scroll]

In the above case 58 isn't part of the adjacent area.

[bitter osprey]

no

[digital whale]

and * is in the 4th position on the second line

[digital whale]

but + is adjacent to 58 right?

[ripe scroll]

No it should be close to it. (touch it)

[digital whale]

[bitter osprey]

467..114

4: index 1
6: index 2
7: index 3
first ".": index 4
and so on

[digital whale]

so in 467

4 is an individual number and so and so?

when we are adding should we add it as

4+6+7

or 467 +....+

[cyan sleet]

467 is one number

[digital whale]

[bitter osprey]

467, the "7" is within the "*" symbol, so then the full number "467" gets caugth

[digital whale]

this is my first time doing advent of code i didnt know it will be this hard

[ripe scroll]

114 and 58 are not the part numbers here

[digital whale]

hmmm

i guess i have an idea for now lets see

lemme sum up the idea

so basically here adjacent part number means it should touch it

not like left hanging in there

and for other numbers

one of the digits in that number

should touch the symbol

[bitter osprey]

atleast one yes

[digital whale]

as in to its side or directly or to the other side

alright

thanks for the help i wish it was worded a bit better is very weird

[lime fulcrum]

so I just read through day 5 again. I'm still not understanding it XD

[digital whale]

[edgy hearth]

I don't understand what day 2 part 2 is asking wth

[digital whale]

even if it was worded better i dont think i can solve it anytime soon

[edgy hearth]

good to know its consistent 🤣

[digital whale]

what exactly you cant?

[edgy hearth]

I don't unerstand the example

is it creating rules for each game or its based off power? which it says in bottom paragraph

[digital whale]

they are simply asking you to find

the max cubes needed

of each color per game

and then

[edgy hearth]

so add all the colors?

[digital whale]

do red*green*blue for each game

and add em up

[edgy hearth]

like game 1 (add red), (add green), (add blue) then sum it ?

[digital whale]

not add red

do you understand whats max cube needed per game?

the minimum amount of cubes you need to have to play all the sets in each game

[edgy hearth]

so max(green), max(blue), max(red), where each respective color is a list of all the color values?

[digital whale]

per game yes

you could do it that way as well

[edgy hearth]

given Game 4: 1 green, 3 red, 6 blue; 3 green, 6 red; 3 green, 15 blue, 14 red
Green = 3, Blue = 15, red = 14

[digital whale]

yes

and also keep that

first game constraints in mind

god day 3 keeps going worse and worse now

[lime fulcrum]

its just quite abstract and I'm quite tired today. I need to sit down and think solely about how it works for 15 min

[rocky spire]

Code passes the example inputs & actual input part 1 but not part 2:
oh no

[modern terrace]

day 5 is ||dp||?

[rocky spire]

I wouldn't say so

[modern terrace]

well part 2 seems like it

[long jacinth]

do i wait for part 2 to take howevver long it wishes or do i come up with something smarter

i have work to do instead, can just let it run meanwhile lol

[rocky spire]

Well it doesn't seem that way to me

[modern terrace]

the way i did it feels like so

i mean

[rocky spire]

Numbers are in the 10^9 range, I doubt any naive solution could finish in a reasonable amount of time

[long jacinth]

i have time 👀

[modern terrace]

coming up with something smarter would surely take much less time than bruteforcing, no?

[obsidian swallow]

any hints on where to start on 5?

[rocky spire]

No I mean it may run for like 10s or 100s of years

[long jacinth]

i know what must be done but i dont know if i have the strength to do it

[rocky spire]

I'm probably overestimating, but I just mean it'll take looooong

[long jacinth]

ill let the first range run, see if it gets solved, if it doesnt within an hour ill stop it and rework it after work

although im doing it in pure python

so its likely to take 100s of years

[obsidian swallow]

[obsidian swallow]

part5?

[nocturne blaze]

day5 part2

[silver karma]

Is someone able to nudge me in the right direction in #aoc-solution-hints?

[obsidian swallow]

wth

[obsidian swallow]

any suggestions other than brute force? im stuck in part1? where should i start looking

[nocturne blaze]

okay in #aoc-solution-hints

[long jacinth]

alright i stopped it after 20 minutes

ill just do the smart thing and rewrite a thing

[slate berry]

I have risen

[slate berry]

wtf strong

[nocturne blaze]

took me a while to code bug free

[edgy hearth]

WHY D2P2 EASIER THAN P1

[slate berry]

It's not

oh wait d2 nvm

[rancid creek]

Am I overthinking it, or is the last part of this a red herring:

Every type of seed, soil, fertilizer and so on is identified with a number, but numbers are reused by each category - that is, soil 123 and fertilizer 123 aren't necessarily related to each other.
As long as you apply them in the correct order.

[spice hemlock]

Wow

[rocky spire]

Can't debug for the life of me
I guess I'll try sacrificing performance

[modern terrace]

"aren't necessarily" sort of implies a situation of "may or may not be"

[halcyon pulsar]

Mm part two seems tricky

What is the longest it would take the top speedrunners to solve the hardest problem?

[modern terrace]

20 something minutes

[halcyon pulsar]

It took pretty much everyone in this leaderboard 20 minutes to do part 2 today

Would you say part 2 is difficult?

[modern terrace]

yep

[covert arrow]

wayyy too difficult for day 5

[modern terrace]

took me a few minutes to figure out traversing in reverse doesn't work the way i thought it would

[digital whale]

which day yall in?

[cyan sleet]

doing this in prolog sounds fun

I'll probably make a program to generate prolog code though 🥴

[digital whale]

hello people

can someone run through my day 3 text file and let me know what the answer is, i seem to get 531267 for my text file but its not the right answer apparently

i want to know where i missed

[thin blaze]

Part 2 is killing my brain.

[digital whale]

which day?

[thin blaze]

Day 5.

I managed to manually split ranges and succeed.

But my brain hurts.

[earnest forge]

Same here lool

[digital whale]

;-;

my brain hurts too

[silver karma]

Most people are in day 5 as that's the most recent one

[digital whale]

i see

[silver karma]

I can tommorow if you still are having issues, the website should tell you if the number is higher or lower so I'm not sure how knowing the exact number it's meant to be is any help :P

[digital whale]

im closer

my answer is off by -1000 well i did check with someone on whats the answer, not because i want to enter the answer

i want to know where im going wrong

but id love to get help yea

[silver karma]

Are you wanting a full explanation of what to do or a nudge in the right direction?

[digital whale]

a nudge, ik how the problem works

but it seems something is wrong with my code

and.. im not so familiar with libraries

people use a lot of libraries to get their work done simply

im doing it full vanilla way with basic functions

[digital whale]

sent a pastebin link in your dms, if you want to take a look at the code

[sleek cave]

today's problem looks fun

[hidden musk]

giving prolog today is evil ngl

[lusty rover]

Why?

[hidden musk]

i know exactly what i want to do, and i have no idea how to express it

the previous 3 days were "fine" compared to this 😩

[median juniper]

i accidentally allocated too many memory while solving AoC

[lusty rover]

Counted out the numbers I see

[hidden musk]

ok actually this might be fun to do

[median juniper]

second part is not that hard i think
||i can implement naive solution and just wait for several hours for it to complete||

[sinful pewter]

Day03 just kicks buttocks. And day05 is even harder. What's happening?

[vestal breach]

years

[magic oxide]

possibly a countermeasure against AI?

i'm up for the challenge tho

[sinful pewter]

Yeah, I am on day03 and churning out code like crazy. Exponential stretch in brain muscles

[median juniper]

whoops, i did it again

i need roughly 100gb of ram

[lusty rover]

Ask Eivl to borrow some

[hidden musk]

clausal logic is based

[zenith elbow]

hah if 64 bits are used then more memory is needed than my whole disk space itself

[hidden musk]

$ sh run.sh
hello world
``` only took an hour 😩

[sinful pewter]

Need some help here in day03
#aoc-solution-hints message

[hidden musk]

@cyan sleet what if i ||preprocess the input into facts and rules|| 😳

[median juniper]

it is possible to fit it into 1gb ||using bit arrays instead of sets||

[cyan sleet]

that's my idea, generate code

[hidden musk]

i found some resource from 2002

[cyan sleet]

like, I have a decent plan, I can't really try it until after work though

[hidden musk]

this is similar to another from last year i think

[jade atlas]

is there any good data structure / algorithms to break down / merge n-dimensional ranges?

i hate the code I just wrote for the range method and I don't know how I could improve

[onyx flicker]

i'm not gonna lie

that's hella reading for the problem today

[noble skiff]

IMHO day 5 part 2 is a perfect example of a problem where you derive benefit from using a fast language/are penalized for using python

just using my part 1 solution to brute force part 2 takes about a minute. Not great, but okay.
in python it would probably take over an hour

sometimes constant factors matter

[onyx flicker]

i'm curious now
ig i'll have to read

[hollow wharf]

#aoc-solution-hints message
any1 can help me with this 😅 ?

[hidden musk]

this task is actually super suited to prolog lol

is it cheating if i preprocess the input in not prolog

[cyan sleet]

I do not want to try to use prolog to generate rules based on a file, at least not on a first iteration 🥴

[simple raven]

isn't python's native range() function super optimised for these kinds of problems? like finding whether 654065230189406216 is in range( 65103125198516021, 908798490546516216541940951621, 65462) is pretty much instant

[cyan sleet]

I'll probably do python to generate prolog code

[cyan sleet]

it's an O(1) check yeah

[simple raven]

bless the genius who wrote that thing

[terse comet]

kind of, but not as much for the

break down / merge n-dimensional
part

[hidden musk]

yeah we're gonna be python-codegenning 😩

[cyan sleet]

I mean, it's not hard

[terse comet]

isn't that just ```py
start <= number < end
and (number - start) % step == 0

[simple raven]

hm... maybe it is... good point

[cyan sleet]

(beware of negative steps and decreasing ranges)

that's the only real annoyance with range

if step is positive, then things are nice and simple

negative is also simple, but a special case

[jade atlas]

I was talking about merging ranges and creating new ones.

[terse comet]

true

[bronze vault]

That moment when the test case works but the puzzle output is incorrect

[simple raven]

I'm about to start working on part1 of day5 so maybe I just haven't found the problem yet, but I think ||I can just use if in range() to find out if I need to translate and if I do, calculate the translation instead of building the whole map||

[terse comet]

part 1 is easy
part 2 is a numerical hell that requires knowing how to work with ranges effectively (spoiler: ||ranges of seeds with total length of 10e10 or so||)

[simple raven]

ah, so just like yesterday, take a breather in part1 and then get shut down in part2

[terse comet]

yesterday's part 2 was really not that bad?

did you ||use recursion instead of just keeping track with a dict/hashmap|| or something

[bronze vault]

lmao i dont even know how im supposed to start debugging this

[simple raven]

yes 😄

[rocky spire]

Depends on how you did it
My solution in part 1 allowed me to seamlessly transition into part 2 on day 4

[simple raven]

in my defense I wanted to practice it so I spent a long time trying to figure out how to write the function

[terse comet]

for some days (in particular day 3), I find that handcrafting inputs equivalent to the example which you think that that may cover edge cases not present in the example can help

[rocky spire]

The way you debug this is to rewrite the entire thing and hope it works the second time around
At least that's what I did and it worked

[simple raven]

tbf now that you mention it, it would make sense that this would just be a ||memoize|| solution... oh well

[bronze vault]

hold on lol I was reading the wrong input

ok i got it 🤦‍♂️

i made a mental note to change it when i get to the real input and then forgot

[jade pecan]

Guys I need help in dynamic programming.Executor June15 converts a number on the screen. The executor has two commands which are assigned numbers:

1. Add 1
2. Multiply by 2

The first command increases the number on the screen by 1, the second multiplies it by 2. The program for the June15 executor is a sequence of commands. How many programs exist for which, given an initial number 2, the result is the number 34 and yet the computation path contains the number 10 and does not contain the number 28?

[terse comet]

ignoring the parsing and matches/scoring logic, I did something like || ```py

part 1

sum(Game.parse_line(line).get_score() for line in input.splitlines())

part 2

counts = {}
for line in input.splitlines():
game = Game.parse_line(line)
card_count = counts.setdefault(game.id, 1)
matches_count = game.get_matches_count();
for next_id in range(game.id+1, game.id+count+1):
counts[next_id] = counts.get(next_id, 1) + card_count
sum(counts.values())

[rocky spire]

You should ask this in #1035199133436354600

[onyx flicker]

i did 0 reading of this channel. but isn't the first example for day 5 wrong? (partially)
shouldn't this not have 99

wait nvm im dumb
idk how to read

[hollow wharf]

it should

[onyx flicker]

yeye i misread and thought that that range was source, not destination

[dense osprey]

the thought of doing aoc was way more appealing than actually doing it

i feel like last year was way more fun or maybe i just dont like doing problems anymore

[digital whale]

this is my first aoc

i actually thought it was going to be fun but oh boy its a sweaty challenge ngl

for beginners ig

[naive stone]

have you done all the days so far? I agree (i think most do), this year has gotten quite difficult a lot sooner than previous years

[solid gust]

This year has definitely been harder than other years— honestly kinda demotivated me to do it.

[digital whale]

ive only finished till day 3 part 1

now making my brain into a mush for this part 2

part 1 ate up almost half of my day

[naive stone]

that's quite good i would say, well done

[digital whale]

i thought ill encounter stuff like this

around day 15 or 16 lol

well nothing can be done

[naive stone]

lol yea, usually that is when you'd encounter it

[digital whale]

chose the wrong year this time ig shouldve joined last time

i thought i could binge till day 5 today but alas

had to miss out on the chance yesterday thanks to a storm over here

[dense osprey]

i mean it isnt really difficult but reading input is exhausting and they are just implementation heavy (but ig that was the same last year for the first few problems too)

[digital whale]

waw pro

[dense osprey]

oh no sorry

i meant like

[digital whale]

i do agree the input is a lot

[dense osprey]

the reason it is difficult isnt because there is some algorithm needed which people may not know

[digital whale]

like i spent way too much time not being able to find the bug cuz its not on the syntax

i was going through the input file for so long

until i found it luckily

[dense osprey]

but for example day 3 was just writing many cases

maybe my implementation was just suboptimal (probably the reason)

[digital whale]

no i did write a lot of cases too

some people just wrote very very simple code that i wish i could spent time analysing bit by bit

but there is no time when day 6 is already here

[dense osprey]

yea some people try to use the least amount of chars possible

havent done day 5 yet

fell asleep yesterday

solved it today

[vestal token]

do you think that aoc is hard this year so that people cant just use AI to get the whole solution? isn't chatgpt against the rules anyway?

[digital whale]

yea its hard due to that ig

[dense osprey]

idk how using chatgpt would be fun

but i dont think that it is against the rules as long as you dont use it for the main leaderboard

[vestal token]

if the questions were really hard and they did allow the use of AI, i think for me it would still remain fun as there's more high-level work and the AI can code various functions

[dense osprey]

and it is impossible to detect it i think

[digital whale]

but because they made it hard so we wont use chatgpt

it doesnt feel fun, its more exhausting

:(

[dense osprey]

i dont think that that is the reason

[digital whale]

oh

then?

[dense osprey]

i dont know the reason

i dont even know if they tried to make it extra difficult this year

[digital whale]

i cant imagine what day 25 is going to be

[vestal token]

can i use python for day 5 part 2 or will it take too long to run?

[digital whale]

like you can see such a big drop from day 2 to day 3

which day saki?

if day 5 idk

[weary kelp]

Easy, cause it's christmas day

[digital whale]

i hope so

[naive stone]

well, easy compared to whatever nightmare the previous week worth of puzzles has been anyway lol

[compact wasp]

of course they did

[naive stone]

idk if the difficulty jump is enough to consider it objectively extra difficult than previous years

i've been stumped by early days before

[compact wasp]

look at leaderboard times though

[magic oxide]

yeah this year's odd days have definitely been a step up from previous years

day 5 was really fun tho at least for me

[naive stone]

i mean, 2019 was an extremely difficult year

which would have also probably been very difficult if not impossible for LLMs

[magic oxide]

I wonder how many people will quit day 5 this year

[naive stone]

they seem insanely impressive to me as always

[compact wasp]

ty

[naive stone]

day 5? day 3 has already been the killer for people

[compact wasp]

i know someone who said stuff before day 10 would be easy, and recommended beginners to try aoc up to that point

his credibility is in shambles

[naive stone]

lmao

[magic oxide]

I used to say that as well

[naive stone]

i've gotten countless people into aoc and i kind of just avoid talking about the difficulty lol. I'm very vague about it

[compact wasp]

good strat

[naive stone]

well enough stalling, i have to go back to inventing a way to solve part 2 today somehow

[magic oxide]

that's a good strat for a lot of things now that I think about

[hidden musk]

day 7 in some past years used some dfs iirc

shiny gold bags

[minor cave]

shiny gold bags was an iconic puzzle tbh

[hidden musk]

so true

[naive stone]

i'd love a path finding problem right about now. I can write bfs in my sleep

idk what to do with this nightmare of a day 5

[hidden musk]

||isn't that today though||

[magic oxide]

||you did bfs today?||

[naive stone]

i hate how tempting it is to click spoiler tags

[magic oxide]

resist the temptation

[minor cave]

||pop||

[simple raven]

I also told a couple friends who have never done any CS-style programming (they are scientists) about AoC and I don't think they have done day1 yet 😦

[minor cave]

I have a friend who's been doing cs/software stuff since foreveeeer and yet has never done AoC. He loves a challenge so I've been trying to get him into this year because of the weird sort-of spike in difficulty

[hidden musk]

no, but it's possible I think. there's no such thing in prolog though

[compact wasp]

tell him to do codeforces 😈

[simple raven]

I think if you get into this with a solid CS understanding you will enjoy using your toolbox to solve these problems, for the rest of us we are forced to learn new things 😄

[hidden musk]

yeah, so far it's been pretty much fundamentals

except some people have to relearn all the fundamentals in a new lang 😩

today might be the first day I don't complete it before the next opens

[compact wasp]

😔

[simple raven]

I can't even read the language Nahita's using, crazy stuff

[naive stone]

i do wish i could complete an entire year without a single >24h sullying my times. That is unlikely though

[weary kelp]

What year was shiny gold bags?

[tepid cairn]

school’s been keeping me busy 😩 I have to grind the days I’ve missed on the weekend

[hidden musk]

2020?

[compact wasp]

me last year lmao

only 24h was day 1

[simple raven]

have you solved everything else?

[compact wasp]

yeah

[simple raven]

nice

[compact wasp]

i was just not worried about leaderboards

so i didn't start as soon as they released

[naive stone]

haha nice

[magic oxide]

I'm honestly somewhat glad I can't make time for aoc right when it drops tbh
not having the pressure of trying to leaderboard is nice

[compact wasp]

i just need to get to like 700 pts to guarantee i stay within top 100

after that i might stop trying

[magic oxide]

and I'm not good enough to leaderboard anyways, so it's a good excuse to not feel sad

[naive stone]

trying for global leaderboard? or in general?

[compact wasp]

for global

i.e. might just do it the next day

[magic oxide]

devote everything to the golf instead

[compact wasp]

i know i already turned off my brain by the time aoc starts

so i can only leaderboard on the easy ones anyways

[naive stone]

@compact wasp did you know exactly how to solve part 2 of today instantly?

[compact wasp]

yeah i knew i just couldn't be bothered to do it quickly

so i took a shower before trying it lmao

[naive stone]

lol

[magic oxide]

did you brute force p2?

[compact wasp]

no

[magic oxide]

oh did you not leaderboard p2 today

[compact wasp]

i didn't

[magic oxide]

ah

[compact wasp]

as i said i took a shower in between

so my time ended up being 50 mins or something

[magic oxide]

given your skill level I wouldn't be surprised if you could afford a shower

[compact wasp]

i probably could've leaderboarded if i didn't shower

given how slow others were

[weary kelp]

I'm looking forward to the dijkstra problem

[dense osprey]

ok can you do fft

(i cant)

[naive stone]

haha no

[dense osprey]

send him a 3600 problem

[dense osprey]

its just implementation problems

[woven sable]

i'm worried about part 2 after hearing about it

ky@solutions too slow for part 1 LMAO

my*

i know what to do tho i'll do it when i have time

[dense osprey]

you could do bin search today for finding intervals but it doesnt really make sense to do due to the small input size

[naive stone]

honestly writing bfs from scratch is so much easier than people think. I would find it easy to explain even to a beginner. I feel like people hear the word dijkstra and get scared, but the overall concept is very simple

networkx is overkill

[dense osprey]

yea i mean it would be weird if it was hard to explain to a beginner

since it is an algorithm which you learn as a beginner

[compact wasp]

if bfs isn't a beginner topic then idk what is

it's the kind of stuff that anyone could independently come up with

[naive stone]

ok i think you guys are going too far now lol

[dense osprey]

isnt it the first or second graph algorithm you learn

[naive stone]

you've forgotten what it's like to be a beginner

[compact wasp]

if you ask a beginner how to go to every node in a graph then bfs or dfs is something they'd be able to come up with

[naive stone]

maybe in a theoretical sense. I don't think they'd be able to code it

[compact wasp]

ig

[naive stone]

intuitively yes, flood fill is something anyone can imagine

it's not necessarily obvious how to make a computer do it

[cyan sleet]

and Dijkstra is just BFS with a priority queue rather than a queue

(and DFS if a stack)

[simple bison]

As someone who's never had formal CS learning, I'm pretty sure I did work out floodfill/dfs myself, maybe bfs. I'm sure they were inefficient and I probably never thought about which one I was using, but I think they were recognisable. Dijkstra and A* I definitely only discovered through reading (because of aoc, actually).

[dense osprey]

yea but i dont think that you can come up with dijkstra by urself that easy

[slate berry]

hi : )

[halcyon pulsar]

What day would you guys expect the part 2 of this problem to be from?

[minor cave]

it always takes me a bit to code up floodfill logic-wise despite needing to do it every year and understanding how it works intuitively

[naive stone]

haha yea, that's totally fair, i think people are getting a little excited by how good they are at it. I like everyone's confidence though

[noble skiff]

I think flood fill is harder for a beginner than bfs

Because before you simply don't think of reframing a grid as a graph

I remember a problem where I failed to come up with flood fill from scratch in high school

Once you spend some time studying graphs you start reframing other things as graphs

At the time though I didn't even know what a graph was

[dense osprey]

hi

[dense osprey]

yea just cause its easy to implement the "default version" of an algorithm, doesnt mean that problems involving that algorithm have to be easy

idk why but putting these logical stataements in equations feels wrong

i always feel like they might cause some problems

even thou they dont

[simple bison]

Unless you're in #1180008645384216606, I would say it's generally bad readability. But yeah, works fine.

[compact wasp]

it really does depend though

[dense osprey]

but writing an if statement for this is also weird

i usually try to avoid this

[sharp crown]

Could somebody explain the meaning of todays challenge to me? I have different maps with different mappings. I choose the mapping that fits my seed number? But all the submaps within one map are all equal in relevancy?

[scarlet wasp]

The maps are a chain, from seed -> soil -> whatever -> something else -> ... -> location

[dense osprey]

oh

ill have to rewrite everything for part 2

[lime finch]

Always 😭

[slate berry]

nice

[slate berry]

honestly it doesn't matter if u really check

the type of map doesn't matter

just keep changing value for every key seed while u go through all ranges

minimum value of a key value pair is answer

[scarlet wasp]

the fact that they are a chain really matters 🙂

[slate berry]

yeah true I mean if it was jumbled

but for same order it doesn't

[quartz badger]

Hi how do i put the puzzle input in the code?

[pine tiger]

some read from stdin

some read fromi a file

doesn't really amtter

[slate berry]

which language?

[quartz badger]

Python

[slate berry]

just refer some code from #aoc-solution-hints

[quartz badger]

ty

[simple raven]

I ctrl+A, ctrl+C, ctrl+V the input from the website 🤷‍♂️

but there is a python module to get the inputs automatically

[quartz badger]

What is the module ?

[simple raven]

https://pypi.org/project/aoc-lube/

[quartz badger]

Ty

[dense osprey]

or you request them yourself

[simple raven]

or that

[dense osprey]

but i think every account has different inputs / not every but there are different inputs

[simple raven]

afaik the inputs are generated for each account separately but who knows

[earnest forge]

IIRC there's a pool of 50 or so inputs for most questions (might actually be as low as 20)

[simple raven]

ah okay

[earnest forge]

Eric has previously stated that some generators take 'several hours' to come up with satisfactory inputs

[simple raven]

wow, then that makes sense

[slate berry]

yeah there are different inputs

[slate berry]

is eric on this server?

[naive stone]

i don't think so

i've definitely seen one or two of his mods here before though

[simple raven]

yay! finally done with part1 😄

I see the issue with part2 now...

[dense osprey]

yea

ill try once im home again

[woven sable]

part 2 seems super complicated

[strange zinc]

I'm still wrapping my head around it :}

[cyan sleet]

progol tiem

[simple raven]

coding part2 is making my brain melt

[sour trout]

How much python experience is required to be medium at aoc problems?

They include a little bit of data structures so you need some experience

[scarlet wasp]

What "python experience" is can vary a lot. If the bulk of your python experience is from doing code challenges, you don't need too much (whatever that means). But at the same time, you might have been coding for years and still struggle, because you've been solving very different types of problems. Soooo kindof an impossible question I think

[sour trout]

Yeah

[simple raven]

as someone with no formal CS education but having done a fair few easy leetcode problems, I would say you want to be comfortable with medium leetcode problems for AoC + a tiny bit of practical experience writing bigger code than just 1 function that solves a thing (happy to be corrected if people think otherwise)

[sinful pewter]

Should I brush up on my algos to figure out the puzzles? I can't figure out the puzzles like I used to last year. It's getting more and more complicated and unintuitive.

[sour trout]

My current language is Lua, but im learning python aswell so thats why i joined this, also ive been doing a lot of stuff except solving problems and im actually good in Lua but the problem is that i cant solve aoc problems that well 😔

I struggle too much

[odd wigeon]

I miss the days where I could solve the day 1 aoc with a one liner in the browser console.

[sinful pewter]

I have written software professionally for a year, this year problems are hard

[sour trout]

people struggled a lot today with python's interpreter speed

I dont know why python isnt compiled but is interpreted

[odd wigeon]

python is compiled

[simple raven]

I have written code for my research but I'm using a very different set of skills for that compared to AoC/leetcode, so I can relate

[sinful pewter]

Python is compiled to bytecode

[sour trout]

It is?

oh then im mistaken

[odd wigeon]

you just don't have to compile it manually before running your script

[sour trout]

I thought it was interpreter i remember once seeing this

[simple raven]

if you write a good solution then it's fast, you just shouldn't bruteforce it

[odd wigeon]

it compiles to bytecode.

[sour trout]

It is compiled to byte code then it is executed on the py VM

[odd wigeon]

just like java

[sour trout]

Java is starting to not get as popular nowadays

people rather use js with a lot of frameworks

[odd wigeon]

Because that's what their frontend is written in

[sinful pewter]

But the algos man. They are unintuitive.

[sour trout]

Well you can do the same on backend

Nodejs & js

[odd wigeon]

which is why people use it

why learn javascript, python, java, C, and SQL when you only need to learn javascript?

[sour trout]

foreal

but for ios i dont know the situation there but i mostly see swift

[odd wigeon]

you can use react-native on ios

[sour trout]

Yeah

[sour trout]

to be honest, sometimes you'll need SQL it depends if youre fullstack or work on the backend

[sinful pewter]

I feel JS is messy. Like it is a language for people in a hurry

[sour trout]

i agree i feel the same

[odd wigeon]

The guy that designed it did it in a weekend

[simple raven]

day4 part2 was my first major roadblock, figuring out the recursion for it took me hours 😦

[sinful pewter]

Yeah Brenden Eich, I think

[sour trout]

console.log(typeof(null)) - > object

[simple raven]

nice

[odd wigeon]

typeof isn't a function, it's a keyword

typeof null

[sinful pewter]

I still have to attempt day 4. I am simultaneously working on a paper which is due on the 20th of this month

[sinful pewter]

The famous piece of shittiest code ever

[simple raven]

oh no, I would say just get the paper done 😄 what's the topic?

[sour trout]

python is an awesome language

thats the only i wanna learn

Nothing else needed

[sinful pewter]

Pausing and Jumping in the context of learning how to program
An analysis of how CS1 students temporally distribute their work
on assignments

[sour trout]

Is there a lot of theory at CS? mostly math?

[sinful pewter]

We came up with a new metric for defining the movement in a file of a student solving problems

[sour trout]

i never went to college but i wish to

[sinful pewter]

Oh the irony 🤣

You guys get it... AoC and learning how to solve problems 😄

[simple raven]

yup

very topical

[sinful pewter]

Yup

I am done with the major analysis. My professor wants cool plots

[simple raven]

I am also working as a teaching assistant and sooo many master's students could really do with learning the basics of programming

[sinful pewter]

Nice, I am applying for my PhD. I have got the initial approval. Waiting for LORs and TOEFL

GRE is optional so skipping it

This is going off topic

Back to Python and that shitty mess JS

[sinful pewter]

Programming is art

[simple raven]

good luck!

[simple raven]

yeah I'm still coding it up 😦

[sinful pewter]

Thank you

[sinful pewter]

Is there an absolute need to use a specific algo?

[simple raven]

which task?

[sinful pewter]

Day 5 part 1

[simple raven]

I didn't use anything I would call an algorithm, trickiest bit was translating from source to destination without creating the whole map, but the logic isn't too bad

[sinful pewter]

Got it

So kind of like pattern recognition without the regex, right?

[simple raven]

I haven't used regex so far 😅 I think if I say more I will start spoiling, but I can do that if you want

[sinful pewter]

Please, a few hints won't spoil. This one is initmidating as it is.

[simple raven]

||for part1, instead of calculating the whole map, think about how much the destination shifts the source. The range length is filled one by one, so you don't have to worry about weird jumps in the numbers, if seed is 79, it gets changed into 81 using [52 50 48], how do you calculate the shift?||

[sinful pewter]

Oh man that's actually really easy. ||We shift data all the time in Pandas|| Thank you.

[atomic cairn]

Now the problem is part 2 lol

[simple raven]

||Pandas <3||

[dense osprey]

i just simulated each seed seperatly

there werent that many seeds

but part 2 makes my solution for part 1 useless

[atomic cairn]

Not yet my guy not yet😂

[simple raven]

well this is why I suggested a different solution 😄

[dense osprey]

thats why i said that that solution is useless for part 2

[atomic cairn]

Im clueless on what to do with part 2

[simple raven]

I wrote a solution I'm coding up now in #aoc-solution-hints , it's the latest big block of spoiler if oyu're interested

[dense osprey]

i dont want to be not able to solve day 5

[hazy ermine]

is there some sort of optimisation for part 2?

[dense osprey]

id be kinda embarassed (because id hate myself)

[hazy ermine]

i bruteforced and got the answer in under 4m30s

used Golang tho

[simple raven]

don't ||bruteforce||

[dense osprey]

different algorithm

solution isnt supposed to be bruteforce

[simple raven]

modern computers are powerful O.o

[hazy ermine]

||range query||?

[dense osprey]

thats pretty difficult for day 5

[hazy ermine]

whats the lowest time you all got for part 2?

[scarlet wasp]

My python solution runs in like 3 ms, doing both parts

[hazy ermine]

💀

[dense osprey]

when is usaco december contest

?

[cyan sleet]

@hidden musk how is the pain?

[hazy ermine]

@simple raven is there more to it than just ||removing duplication ranges||?

[simple raven]

this is not something I do so I don't know what approach you have in mind

[terse comet]

completely checking every single seed?

[hazy ermine]

yes

[terse comet]

wow

[hidden musk]

I'm almost there I think for part 1

[simple raven]

I was amazed as well, but I guess 2 billion checks + valid seeds * number of translations could be doable?

[hidden musk]

I saw you got part 1

[hazy ermine]

it wasnt in Python btw (was in Golang)

[terse comet]

even then, that's billions of operations if not trillions

[cyan sleet]

part 1 wasn't too bad with some codegen

part 2 is...worse

[rapid wolf]

Yeah, it is very bruteforcable in not-python. 60s in C a friend got

[hidden musk]

I can't figure out how to do the default mapping

[cyan sleet]

(if else)

[hidden musk]

without basically solving it in codegen

huh. that exists???

[cyan sleet]

( pred1 -> if_true; if_false )

and you can chain it

( pred1 -> if_true1
; pred2 -> if_true2
; if_false )

[hidden musk]

wack

[cyan sleet]

it's not how would prefer to write it, but it works

I'm legit confused about some stuff though

[hidden musk]

oh that's nice

same

[cyan sleet]

how is this Arguments are not sufficiently instantiated

blah(R) :- between(1, 99, X), 0 < R, R < X.

:- blah(X), writeln(X).

[dense osprey]

oh solution for part 2 < 3e8

i dont know if i really want to avoid a naive solution

[dense osprey]

nvm thats false

[rapid wolf]

try :- use_module(library(clpfd)). and #> instead, it might help. The basic problem is that > only works when it was 2 numbers, not when it has a number and a variable. Prolog is full logical impurities like this.

[hidden musk]

huh

[cyan sleet]

huh

yeah that made it do something

[rapid wolf]

oh yeah, you'll get that fancy range answer then

I think clpfd will work quite well for this problem

but I don't quite have the energy to do a prolog solve rn

[cyan sleet]

I...need to actually express things in terms of ranges first 🥴

[woven sable]

i feel like rust would be rlly. good for today

why have the first days been so hard recently 😭

ion remember day 5 being this hard

chat gpt prevention?

[rapid wolf]

If my vague brain is right, I think prolog can crush this problem effortlessly.

[dense osprey]

i hope next problems wont get way more difficult

[cyan sleet]

(if one knows prolog)

[dense osprey]

isnt chatgpt allowed aslongt as you dont use it to get on the leaderboard

[rapid wolf]

indeed

[dense osprey]

and i dont think that the chatgpt usage would hurt anyone

[safe bough]

hey

[hazy ermine]

well theres no way to tell if someone on the leaderboard used it

[dense osprey]

you just sabotage yourself imo

[hazy ermine]

yes

[dense osprey]

yea obviously

[minor cave]

It's allowed but in a disappointed dad way

[hazy ermine]

its pointless to use gpt here lol

[safe bough]

is this the appropriate place to ask for help in day2 part 2?

[dense osprey]

is chatgpt even good at solving problems like these

[woven sable]

i would ask in #aoc-solution-hints but here's fine too, you'll j have to add spoilers i think

or #1047673173447020564

[dense osprey]

i remember it fail9ing when asked to solve problems like these

[minor cave]

#aoc-solution-hints is what you're after. We're more lax about spoiler discussions and you can make a thread in that channel

[terse comet]

it was pretty alright for part 1, but unless you want to brute force, I cannot recommend it for part 2
unless it has some ultra fancy range class I am not aware of

[safe bough]

alright ty

[woven sable]

kotlin was pretty nice for today's challenge mostly bc it has nice implementations for ranges

like|| .intersect & .subtract||

[cyan sleet]

@rapid wolf
could you explain what I'm being dumb about?

seed(between(1, 10)).
soil(between(L, R)) :- seed(between(LL, RR)), LL #< L, R #< RR.

:- soil(X), writeln(X).
:- halt.

Output: between(_7734,_7960)

[woven sable]

basically j set operations

[cyan sleet]

this makes no sense

[noble skiff]

yeah kotlin is a nice aoc language IMHO

[woven sable]

yeah i've had zero struggles w it so far

i j have to finish part 2 when i get home 💀 and hope the solution im thinking of isn't slow

[noble skiff]

yeah. I mean, it's as easy or easier to write than python, and still a lot faster.
not as fast as Rust but you dont' have to deal with borrow checker issues

[woven sable]

also all the predicate functions or wtv they're called

my solutions have been so clean

[dense osprey]

how is youtube checking if i have an adblocker legal

[terse comet]

that's out of topic for this channel

[rapid wolf]

basically, clpfd is creating a variable that is constrained by #<, but that constraint will only manifest at a time when it can be decided - that is, when you put a number to that variable, I would suggest reading up on label for an automated way to do that.

[hidden musk]

that's my feeling

unfortunately, I'm not very good at prolog

[rapid wolf]

which is definitely not confusing at all. God I hate how prolog immediately forces you into clpfd if you want sane math

[cyan sleet]

oh _<number> are placeholders?

variables/whatever

[rapid wolf]

ye

[cyan sleet]

ah

so if I added a constraint that would imply a maximum or something it might work?

I got used that blah(X) got me some solution X

but I guess this doesn't

[rapid wolf]

there are sort of two steps to using clpfd -> you construct your constraints, and get a list of your constrained variables. Then you call label/labeling, which assigns actual values and solves those constraints

you can also ask them to minimize/maximize certain variables

Yeah, variables which have a constraint on them will not get assigned anything

[safe bough]

so i created a thread, now do i ask people in other chats to help me or do i just gotta wait

[minor cave]

you can ask here for help, but people will also check out that channel. I have some time now so let me take a look

[wet yoke]

I must admit, my solution for part 2 today was in no way clever: I just generated all the seeds and brute forced it.
But on the bright side, Rust in release mode got it done in about a minute.

[simple raven]

I am deep in the trenches trying to get part 2 done without bruteforcing and I just realised I backed myself into a corner and have to rewrite half of my logic... fml

[noble skiff]

if you want, you can probably make a one line change or so and get it to run in < 10 seconds w/ rayon, depending on your computer

[hidden musk]

I'm just about done with part 1 😔

I mapped the things and now I just need to find the min of all my query results

which is apparently really hard

[hazy ermine]

wait what? I thought u found the optimal solution

[simple raven]

I got the logic, not the code

[hazy ermine]

ah right

[lime fulcrum]

apply to all 20 and do min?

[rapid wolf]

Ye, if you aren't using label or sth, you need a findall style thing.

[hidden musk]

that's what I surmised

honestly the language seems really cool

[rapid wolf]

It does, I just always get sad when I have to do anything that it doesn't do amazingly well.

[hazy ermine]

u mean Python? yes imo u can really just convert your logic into code without much translation

[hidden musk]

I think it does today really well

I mean, part 1

idk about part 2

[hazy ermine]

nvm... i realised youre talking about some esoteric lang lol

[rapid wolf]

I am hoping the builtin datalog of flix.dev will perhaps give me a similar experience while also giving me just plain old FP for when I don't want DFS

[hidden musk]

prolog 🤩

[lime fulcrum]

not being able to overwrite variables sounds like pain

[cyan sleet]

so apparently the solver easily gets unhappy whenever anything is unbounded

[hidden musk]

not as big of a problem as you might think @lime fulcrum

[cyan sleet]

that's the least of the worries

[hidden musk]

with functional langs that's the norm really. just takes a different kind of thinking

[lime fulcrum]

yeah well first I have to get a bit better in python

[gentle cape]

With functional langs it’s so hard to debug

Because I can’t just add a print call somewhere

[hidden musk]

depends on the language

[gentle cape]

Haskell

I’m fighting the urge to switch to lisp

[pine ravine]

this program just ate 24gb

[hidden musk]

sure. but just break the function into parts and debug

[cyan sleet]

@rapid wolf any hint how to even start to think about modeling the interval logic?

I feel like whatever I try I run myself into a corner

[pine ravine]

how many bytes is python using to store them

[cyan sleet]

what is them?

[leaden tartan]

Had to goto bed before I finishe p2 of yesterdays. Is todays brutal?

I need to sleep at some point

[minor cave]

It's not pretty for a day 5. Like 25 minutes for part 2 global to fill up

[hidden musk]

kat are you prologueing with us

[lime fulcrum]

todays is a lot worse. ( in my opinion)

[hidden musk]

#aoc-solution-hints 👀

[minor cave]

fuuuuck no, I am quite happy sticking with Python

[leaden tartan]

Awesome

[lime fulcrum]

i mean part one is okay. two is a lot harder

[rapid wolf]

I would probably solve it as you would in a normal languages, short of figuring out how to get clpfd to manipulate the ranges, I don't think there is much the prolog search will do for you.