advent-of-code

[limber needle]

yeah i was going through some papers that may have touched on day 19's questions

https://ieeexplore.ieee.org/document/1047997

like this one for example

[sleek cave]

i'm diving into the wonderful world of weighted graphs and search algorithms

i'm in love with aoc

[limber needle]

things i'll never use in the future

[low condor]

You never know - maybe one day you'll have a job as a Travelling Salesman

[sleek cave]

lmao

OMG A* IS ABSOLUTELY BRILLIANT

HEURISTICS ARE BRILLIANT

SEARCH ALGOS ARE BRILLIANT

[placid lake]

lmfao

[robust heart]

when this ends, can we keep some links as reference?

Or do we continue in the algos channel?

[minor cave]

This channel will hang around until at least january, but otherwise the algos channel is a good alternative

[sleek cave]

okay, i understand a star in theory

implementation is a whole other problem...

[proud cosmos]

Actually really simple if you understand the theory and think about it a bit

:)

[sleek cave]

yeah, i watched some computerphile vids, now i'll complete a-star on a small graph by hand, then implement in code

[low condor]

can't wait to tell you what my solution was for that day 🙂

[sleek cave]

i'm thinking of 25 nodes, 5 by 5 simulating the risk levels

[low condor]

in the meantime, best of luck 🙂

[sleek cave]

alright, time to get paper programming

[sleek cave]

thanks! i'll definitely look at your sol after :D

[fossil cipher]

I think I have hit my AoC Wall. Saturday's I think I've finally understood the destructions, just haven't attempted it, yet. Today's is just... wow. I don't even know where to start or how to approach it.

I plan to attempt upcoming puzzles, but my 17-day streak for this year is my best ever, so not too shabby 🙂

[low condor]

I had that initial reaction as well. but after a while I did figure something out. It's worth giving it a good think over first, I think

[fossil cipher]

Well, I'm on leave for the next two weeks, so should have plenty of time to try them. Not giving up, but certainly beyond my skill level (in terms of the maths and algos required). There isn't a big call for 3 dimensional beacon detection with inferior scanners in my day to day use of Python. 😄

[hybrid schooner]

||is the score for day 4 part 2 6426? cant check it right now on the website||

[low condor]

everyone has different inputs so no one can check that without running your input

[hybrid schooner]

ah ok thanks so i guess i need to wait 😄

[sleek cave]

my brain is subsisting suboptimally

[pallid quail]

For day 4, I'm struggling with processing the input data. Not sure how to approach this. The first line, I've put in a variable called draws, and have split that string into a list on the commas, so now I have a list of draws - which is fine.

But for the boards - I want to iterate through the lines, and make each board a dict, made up of a list for each row....

[abstract bluff]

can you link your code?

[pallid quail]

I don't have anything actually of any substance. I couldn't even conceive of how to do it. I've only been learning python for about a month - and I think day 4 is where I am tapping out (for now). I went back to my bootcamp that I am doing to actually learn python - and getting back to that cause I feel like right now I'm being held back because there are still some basics that I am lacking.

its not on the list of recommended learning resources, but I bought it before I found this discord. But so far, I've learned a ton from it. This is the course I'm doing - hope its good!

https://www.udemy.com/course/complete-python-bootcamp/

[abstract bluff]

you said you managed to parse the first line, show me how you did that. perhaps we could work out how to do the rest?

[pallid quail]

Sure, one sec.

Let me re-load the project in vscode

this is the function I use to process the input data - I've been using it since the day 1 challenges.

def process_data(filename,datatype='str'):
    with open(filename, 'r') as f:
        if datatype == 'int':
            data = []
            for i in f.read().splitlines():
                data.append(int(i))
            return data
        elif datatype == 'str':
            data = []
            for i in f.read().splitlines():
                data.append(i)
            return data            

Then I ingested the data into a variable called "data"

data = process_data('d4_input.txt')

[abstract bluff]

okay, so you reading by lines, which is a good first step

[pallid quail]

And took the first line of the data, and put it into a list called "draws" by doing this:

draws = data[0].split(",")

[abstract bluff]

okay thats great

now did you make sure to cast that to a list of integers

?

[pallid quail]

Its taking the actual boards, and putting them into a workable format.

I did not, I was going to make them integers when I called them later

by just doing int(draws[someindex])

[abstract bluff]

okay. fair enough

so if I gave you only one board would you then know how to parse it?

22 13 17 11  0
 8  2 23  4 24
21  9 14 16  7
 6 10  3 18  5
 1 12 20 15 19

[pallid quail]

Heh, hadn't gotten that far yet!

But I digured I would have put it in a list, then put that list in a dict, with each board having a key like 'board1': [[board,rows],[board,rows],[board,rows],[board,rows],[board,rows]]

And then, for each draw, going through and checking each board for the draw int

[abstract bluff]

yeah, you got the gist of the idea

[pallid quail]

and thats about as far as I had gotten in terms of figuring out how to deal with it

[abstract bluff]

but focus on parsing a single board first

it's kind of like the first line, except you split by a space instead

you also know each board will have 5 lines

[pallid quail]

My problem was, I don't know how to write my code to split each 5 lines of a board into a a dict key value, then ignore the blank line, and repeat.

[abstract bluff]

do you know how to use a for loop?

[pallid quail]

Oh yeah, I even wrote a few recursive functions to solve day 1-3

[abstract bluff]

board = []
for line in lines:
  row = line.split(" ")
  board.append(row)

[pallid quail]

Thats what I was sort of starting to write - but how do you handle the blank lines?

Would I just write something like

board = []
for line in lines:
  if line == '\n':
    pass
  row = line.split(" ")
  board.append(row)

[abstract bluff]

yes! you're almost there

instead of pass, we use continue

[pallid quail]

Ah right

[abstract bluff]

whenever you hit a newline, you know a new board is starting

[pallid quail]

pass is just if I define a function, but don't have anything in it yet...

[abstract bluff]

so append the old board to another array maybe called all_boards

[pallid quail]

Well, I had created an index variable.

That I was going to add 1 to, each time a new board is started

and use that variable, to add the board to a dict key value

calling it f'board{index}'

[abstract bluff]

a dictionary isn't quite useful here, another array will allow you to index boards the same way

without you having to manage the index variable

[pallid quail]

So, thats what I also meant by sort of being limited by my limited knowledge of python.

I've never used anything like numpy or anything like it - if thats were you're going in terms of using arrays.

[abstract bluff]

no numpy isn't needed at all

[pallid quail]

So when you say array, do you just mean a list

or list of lists?

[abstract bluff]

yeah!

list

all_boards = []
board = []
for line in lines:
  if line == '\n':
    all_boards.append(board)
    board = []
    continue
  row = line.split(" ")
  board.append(row)

this is what I was getting at

[pallid quail]

Okay, yeah - that's sort of where I was going with this. But I guess I was overcomplicating it by trying to use dictionaries.

[abstract bluff]

also make sure you ignore the \n after the first line and append the last board after the for loop.

[pallid quail]

So, I actually just looked at my data after ingesting it

[abstract bluff]

since there is no newline at the end of the file

[pallid quail]

I don't have any \n characters

[abstract bluff]

ahh, yeah splitlines() probs removed them for you

[pallid quail]

my ingest function removes those, yeah

[abstract bluff]

just check if len(line) > 0

since splitlines() will have a empty string for a empty line

[pallid quail]

Actually, no not seeing that either

[abstract bluff]

what are you seeing then?

[pallid quail]

Here is what is in my 'data' variable after ingesting the text file

oh wait, its too big. Gonna throw it into a pastebin

https://paste.pythondiscord.com/ekaxeyiyog.json

[abstract bluff]

yeah you do have a empty line see line 2 of the pastbin

there is a '',

[pallid quail]

Oh, I thought you meant at the end

Gotcha.

[abstract bluff]

no every empty line gets converted to a empty string by splitlines()

[pallid quail]

I added:

for row in test:
    if index == 0 and row == '':
        continue

test is just the sample data from the challenge.

[abstract bluff]

how about index?

[pallid quail]

oops

that should be line

forgot to edit that line after I removed the index variable

err row

for row in test:
    if row == '':
        continue
    elif something

Oh wait, that won't work now

[abstract bluff]

you want to just check the row itself not the line number

since after each bingo board there is a empty string

you can look at your data dump again

[pallid quail]

I've got it it up on my other screen

[abstract bluff]

ok now just use the same logic for the board parser you had earlier

[pallid quail]

ok cool - thanks for talking me through it - heh I think I can figure out this part now.

[abstract bluff]

good luck 🙂

[pallid quail]

thanks!

One last question

Is it possible to execute certain actions, just on the first run through a for loop somehow?

without using enumerate

[abstract bluff]

if you just want the first run, then just do it outside the loop right before it

[pallid quail]

Oh

right

I wanted to remove the first blank space by popping it out

I can do that right before the for loop starts

[abstract bluff]

lines = lines[2:] should work

[pallid quail]

Oh right that would work too

Is that more performant than say, pop? A friend of mine who is a professional programmer mentioned something about pop being less than ideal when working with larger data sets.

Or maybe I just misunderstood what he was explaining. or rather, when popping stuff off from the start of a list.

[abstract bluff]

that depends how this pop() method you are referring is implemented

[pallid quail]

on one of my previous solutions, in a recursive function - on each run through I was popping out the item at index 0

[maiden brook]

if you pop a list at the start you have to shift everything one index to the left

[pallid quail]

Yeah - it was in the sonar_sums function here: https://github.com/max-is-loud/adventofcode21/blob/main/day1.py

I was pretty stoked about that - it was my first ever recursive function lol

[abstract bluff]

python uses dynamic arrays for lists i think, so unless you pop() from the end, its O(n)

[pallid quail]

(spoiler for day 1)

Gotcha, so for a list with 2000 items, if I pop one from the left, then it has to do 1999 actions, to move everything over to the left by one?

[abstract bluff]

yeah

however, given that you seem to be just starting out programming, worrying about complexities of builtin functions shouldn't be on your list of concerns

[pallid quail]

haha - thats what I said to my friend.

[abstract bluff]

undoubtely complexity is important, but not "for you right now"

[pallid quail]

Okay, making progress - but I have another little problem.

For numbers that are just 1 digit, there are two spaces

So I end up with: ['', '2', '', '0', '12', '', '3', '', '7']

[abstract bluff]

use the same technique as earlier filter the list for elements with length > 0

[woven sable]

You can use split with no arguments to split by any amount of spaces if you're splitting it

Unless, that's not what you're doing

!e
print("2 3 4".split())

[marsh current]

@woven sable :white_check_mark: Your eval job has completed with return code 0.

['2', '3', '4']

[pallid quail]

oh ok

[abstract bluff]

^ that works too

[pallid quail]

perf

[
  [
    ['22', '13', '17', '11', '0'],
    ['8', '2', '23', '4', '24'],
    ['21', '9', '14', '16', '7'],
    ['6', '10', '3', '18', '5'],
    ['1', '12', '20', '15', '19'],
    ['3', '15', '0', '2', '22'],
    ['9', '18', '13', '17', '5'],
    ['19', '8', '7', '25', '23'],
    ['20', '11', '10', '24', '4'],
    ['14', '21', '16', '12', '6'],
    ['14', '21', '17', '24', '4'],
    ['10', '16', '15', '9', '19'],
    ['18', '8', '23', '26', '20'],
    ['22', '11', '13', '6', '5'],
    ['2', '0', '12', '3', '7']
   ],
   [
    ['22', '13', '17', '11', '0'],
    ['8', '2', '23', '4', '24'],
    ['21', '9', '14', '16', '7'],
    ['6', '10', '3', '18', '5'],
    ['1', '12', '20', '15', '19'],
    ['3', '15', '0', '2', '22'],
    ['9', '18', '13', '17', '5'],
    ['19', '8', '7', '25', '23'],
    ['20', '11', '10', '24', '4'],
    ['14', '21', '16', '12', '6'],
    ['14', '21', '17', '24', '4'],
    ['10', '16', '15', '9', '19'],
    ['18', '8', '23', '26', '20'],
    ['22', '11', '13', '6', '5'],
    ['2', '0', '12', '3', '7']
  ]
]

[abstract bluff]

great! just remember to cast them to integers

[pallid quail]

Actually, not perfect.

I ended up with the array above, which is the three sample boards, twice

in two lists within the all boards list.

[abstract bluff]

what is your code?

[pallid quail]

draws = test[0].split(",") # takes the draw numbers on the first line, and splits them into a list.
lines = test[2:] # grabs everything after the draws and the first empty line.

all_boards = []
current_board = []
for row in lines:
    if row == '':
        all_boards.append(current_board)
    else:
        row = row.split()
        current_board.append(row)

[abstract bluff]

you forgot to set the current_board to [] after you append it to all boards

[pallid quail]

lol oops

@abstract bluff One last quick question - when I put the draw numbers into the draws variable above - is there anyway to convert those to ints as part of that variable assignment?

I tried

draws = int(test[0].split(,))

but that didn't work.

[abstract bluff]

draws = [int(c) for c in test[0].split(',')]

[pallid quail]

ahhh list comprehensions.

Didn't think of that

👍

Might write a function for that so I can have it work on any variable name, not just test.

Thanks again!

[abstract bluff]

no problem! any question just ping!

[pallid quail]

Thanks, I appreciate it - I was about to give up haha

[humble copper]

well its almost time

99.9% chance i wait until tomorrow to start

[prisma sundial]

gl

[pallid quail]

Oh the new challenge heh yeah. Man today's challenge makes me think I'm screwed once I get to it if day 4 was a struggle.

[woven sable]

Don't worry, day 4 was kinda difficult for me too

[hidden copper]

<@&518565788744024082> Good morning! Day 20 is ready to be attempted. View it online now at https://adventofcode.com/2021/day/20. Good luck!

[humble copper]

which one was that

[woven sable]

and a few others

[foggy zealot]

third

[pallid quail]

bingo

[humble copper]

yeah that was hell

[wide garnet]

:v I lost hope after not being able to solve 3 puzzles around day 16

[humble copper]

after today's challenge ill be 80% of the way there

(maybe 81.63% if ||d25p2 is a freebie like 2020||)

[pallid quail]

Does it stay up after the fact - like, if I don't get them all by the 25th can I still attempt them later?

[humble copper]

yep

[woven sable]

Yep!

[humble copper]

if youre doing the leaderboard here and you want the role, both parts of all 25 days need to be completed by january (i think)

[pallid quail]

Naw, I haven't bothered with that.

Maybe next year when I've been programming for more than a month

[woven sable]

I'm really excited for the role icon tbb

never had a role icon ever in any server

[pallid quail]

I work at a saas company in client services, and I joined the private leaderboard the software engineering team is using...

they've been super encouraging on slack - but I try not to bug them with questions to much - cause its a work slack and they are busy. A noob in client services is probably not something they want to deal with.

[woven sable]

Well, you can always ask here 😄

[humble copper]

im going strong on 6 years of mostly self taught python and im barely getting by with 2 starring all days, but you might have a different experience

ig it really depends on what you get experience with

[pallid quail]

I'm hoping that I can get to a point where I can join said software engineering team.

[humble copper]

like it you build discord bots for 10 years you might do extremely well on a discord bot building challenge

but who knows on anything else

[humble copper]

confidence is key

[pallid quail]

For sure!

Anyway - its getting late, going to take a break of staring at vscode and check out halo infinite for a bit. I'm on vacay and my kid is in daycare still so I'll have all day tomorrow to mess around with these challenges.

[humble copper]

it may even help to start keeping a github repo of your aoc solutions

[pallid quail]

Already doing that

[humble copper]

same

[pallid quail]

https://github.com/max-is-loud/adventofcode21/

[prisma sundial]

damn not quite lb today

[pearl moon]

The infinite size thing for 20... Is that just to give you surrounding values for image positions on the borders of the image?

[prisma sundial]

in every direction the image is infinitely . (with the exception of course of the provided image)

[pearl moon]

But what's the relevance of that information

[prisma sundial]

it basically tells you that when you do these 3x3 things, the image can grow

[pearl moon]

I'm saying it seems like that is just there to give you something to use for valued on the border

[prisma sundial]

oh yeah i think we're saying the same thing

[pearl moon]

Err...isn't the output image the same size as the input?

[prisma sundial]

i mean they're both infinitely large

[pearl moon]

Oh I guess not

[prisma sundial]

in terms of the interesting region, no

[pearl moon]

Looking at examples...I meant the non infinite dot portion

Ok so it really means you need to apply it to all the pixels which aren't just dots surrounded by all dots...presumably

That would include a few pixels just outside the border of the finite region

[woven oriole]

how is it replicating out of the borders of the initial image

each pixel is decided by 9 pixels right , we never iterate over outside the image (only check neighbours), so how did it turn into light pixels 🤔

[woven oriole]

how are we expected to iterate infinitely lol

we should only check the part given in the input

[somber crow]

thats the puzzle

[woven oriole]

mhm

[somber crow]

||think outside the box, || ||do you need to literally have infinite largeness?||

[late frigate]

rip I did ||+1 instead of -1|| and spent 30 mins debugging

[brisk briar]

infinite largeness? sounds like my mom

[winter vigil]

does anyone have any hints for day 14 part 2?

it is kind of like the lantern fish problem

[woven solar]

Yup, very much like the lantern fish problem, and can be solved in a very similar way - ||keep track of the counts of pairs of adjacent atoms||

[pearl moon]

My last messages explained this I think

[candid berry]

Hey, don't disrespect my girlfriend like that

[brisk briar]

lmao

[pearl moon]

I'm assuming that a position surrounded by all dots will stay a dot. So most of the infinite image isn't going to change. But you have to discern the parts that will change...

[woven solar]

I'm assuming that a position surrounded by all dots will stay a dot.
Only if the first character of the algorithm is ., like in the example... ||but not like in the actual input||.

[pearl moon]

Well in that case almost all of the characters will still be changing to the same thing...either way the finite region will be surrounded by infinity of the same character

[woven solar]

that's true, yeah

[dense shell]

Can anyone help me solve max recursion depth reached in my code

[earnest forge]

Well you've almost certainly done something wrong, but if you want a footgun check out sys.setmaxrecursionlimit

[earnest forge]

That seems to be true of all inputs. If you want a more useful test, ||swap the leading . and trailing # of the algorithm for the example. The answer is 24||

[candid berry]

This really needs to be in spoiler tags. It's giving away a non obvious piece of the puzzle.

[dusk cove]

ok I deleted the message

[livid gull]

did the example help? did you find your error?

[earnest forge]

Eventually

Having something to repeatedly debug against meant I spent less time being timed out

[livid gull]

yeah i normally try the test data for the harder days before my own data, but todays was hard to use

[cunning summit]

hii

[robust heart]

trying to figure out the same

Is that a trench?

[sonic silo]

its a trench

[wooden jewel]

does everyone else have ||# for the replacement of 9*.||? It's driving me up the wall

[sick crag]

yup

||that and . being the last character||

[wooden jewel]

||Well yeah you'd need that if the first character is '#', otherwise you'd get an infinite answer ;)||

[earnest forge]

FYI mods consider this a spoiler; see my spoilered message above for a useful example

[wooden jewel]

yay

Thanks for the test case, that really helped in debugging :)

has anyone written an image encoder for this problem yet?

[woven oriole]

is that really an image

[jade root]

aa my solution for day 20 works on example but not on full input :((

[frigid mural]

yall ever find it satisfying when 2 consecutive lines have the same length

[wooden jewel]

welcome to the club

[woven oriole]

||the algorithm of puzzle input is not the same||

[jade root]

ik

[woven oriole]

||9 dots can result in a # if 0th index is a #||

[jade root]

E

[woven oriole]

||and the # formed can once again be changed to . if the last char of algo is a . lol||

[jade root]

my brain lahfdsjfhasdfa

[woven oriole]

relax

[jade root]

||the image is infinite but 9 .s can form a # so the answer is infinity aaaaa||

[frigid mural]

||no cos it happens twice so all the # get turned back||

[woven solar]

||on step 1, yes. But you're not asked about step 1.||

[jade root]

o

[woven oriole]

||on step 2 it will again become dots||

[jade root]

omg yay

oh no part 2

YAY

finally ;u;

[frigid mural]

that was a very quick rollercoaster of emotions, omg yay to oh no to YAY in 2mins flat lol

[jade root]

yes lol

i was so worried it wouldn't work for part 2

[glad tartan]

the example after 50 enhancements

[cosmic herald]

Which should I use... file.read() or file.readlines() for reading an entire file?

[wooden jewel]

You can just do for line in file and get every line in the file without reading it into memory all at once

[cosmic herald]

True... keep forgetting about that...

Thanks!

[frigid mural]

oh lol i wasnt sure if this S-tier fudge i was pulling would work but it does

[brave wren]

The example runs exactly as expected but the actual input isnt correct :c

[glad tartan]

there is a catch in the input :). what happens to the infinite background after the first enhancement?

[brave wren]

oh god

[sleek lark]

Any tip for day 17 ? Max x speed is pretty easy to calculate but y is a bit tricky.

[proud cosmos]

Consider what happens for a y speed ignoring x

Then attempt tofigure out how high y speed could possibly be

[sleek lark]

But thinking about it, any Y height is possible

[woven solar]

you can use math to make part 1 a one-liner and part2 fairly straightforward, but it's actually also possible to just bruteforce all possible vx and vy values.

[proud cosmos]

Is it really?

Think some more

[sleek lark]

Even at high speed it is very possible for the probe to hit the target, so unless I can only constrain Y with X, please be a little less condescending

[proud cosmos]

I wasn't condescending I was trying to point you to the fact there is in fact an upper limit for y

Especially since the example even admits only y=9 works and nothing more in their case

I figured out what the max limit is by looking only at y and how it behaves for different y values

Guess the tip wasn't clear enough though.

[sleek lark]

But it is only because the exemple is built like this. The input could be not as much constrained.

[mossy basin]

you can solve for max y by considering the ||velocity of the drone on the way back down as it crosses the x-axis||

[sleek lark]

Thank you, you nudged me enough so that I have a better understanding of the problem.

[humble copper]

doing this suoer early on a second machine which i just remembered doesnt have vscode installed

i have to write everything in nano

not fun

[brave wren]

todays solution was definitively not my fastest but at least it works

[mossy stream]

That's not the right answer; your answer is too low. Curiously, it's the right answer for someone else; you might be logged in to the wrong account or just unlucky.
that's pretty KEKW

[devout tusk]

Had the same happen to me today

[mossy stream]

oof da heck is day 19

[velvet stag]

the skip day

[mossy stream]

Well I have an idea on how to do this, but I feel like it will take ages to complete

why did they make the direction change smh

[humble copper]

im somehow getting the example right but the input wrong
for d20p1

[mossy stream]

I just read the prompt of day 20

[sleek cave]

oh god this is where stuff goes downhill really quick

I'm skipping d19 for now, gonna attempt d15 and d20 today

[sleek cave]

so enhanced

[mossy basin]

i think 19 was more challenging than todays

[sleek cave]

all the days starting from 18 are challenging 🎄

[peak dock]

19 has been the hardest so far imo

[sleek cave]

I don't even want to read the prompt

[humble copper]

and yet im still getting it wrong ☹️

[sleek cave]

some guy said 19 was comparable to sea monster, oh no...

[mossy basin]

it's comparable, but sea monster was a lot more code for me

[sleek cave]

ah, I see

for me, I'm already glad I solved 18 lol

[wooden jewel]

today is the third day that day n has more solutions than day n-1

[sleek cave]

after 5.5 hours of intense programming

[mossy stream]

what the fuck is 19

like I can't find any solution that would run in less than two hours

[sleek cave]

big oof

[mossy basin]

i think my solution only takes a couple of seconds

[mossy stream]

your solution is just numpy on top of numpy on top of numpy smh

[mossy basin]

yes

[wooden jewel]

as is tradition

[sleek cave]

||I'm learning a* on the spot to solve day 15, I'm not going to learn its implementation, though, since I need to figure that out myself in order to have completed the puzzle lol||

I imagine how long it would have taken if that algorithm hadn't been used

[mossy basin]

i'm sure the wiki has pseudo code

[sleek cave]

no pseudocode for me bc I like pain :D

[mossy basin]

||a* doesn't actually save much time over dijkstra for this problem||

[wooden jewel]

||You don't need a proper search algorithm, mine runs in 2 seconds with a fairly dumb search||

[mossy stream]

wtf some people have some >2s pure python solutions

[woven sable]

Am I the only one having trouble with Day 20?
||what do you do if there aren't 9 pixels surrounding the current one?||

[sleek cave]

really? I implemented ||a heuristic for the distance to the end coords||

[woven sable]

||yeah same, heuristic was Manhattan distance and G-cost was the risk factor||

[peak dock]

its an infinite grid :P

[mossy basin]

||people that implemented manhattan distance as a heuristic only saved a couple of cells searched by the end||

[woven sable]

oh, so it's just . if it isn't there?

[peak dock]

for step 1 yeah

[woven sable]

or rather, replace . with all the ones that aren't there

got it

[crisp coyote]

Don't feel you need to rediscover the foundations of computer science to have completed the tasks properly :P Although implementing it from a description can be good to learn it and can be a fun challenge

[sleek cave]

oh, interesting
still doing it cuz why not, learning = good

[sleek cave]

good point, thanks! but yeah, I'll still be giving myself a little challenge, because this is also such a great opportunity to learn about ||priority queues and heaps|| and actually using them in a project

[crisp coyote]

Yeah I did some learning myself on that one, in my initial solution I just used ||.sort|| lol (||instead of a heap||) but then looked into better ways to do it afterwards

[woven oriole]

||its the direct implementation of a* lmfao||

[sleek cave]

that's what I was thinking at first!! but then I was like "hm, shouldn't there be an efficient python library for this" and there was

[woven sable]

I wouldn't have used ||heapq|| if I didn't learn about it in a Beazley talk a few days before

[woven oriole]

||PriorityQueue|| moment

[sleek cave]

lol nice, I just searched up ||"priority queue python"||

[woven oriole]

||first website would have been geekforgeeks heapq /rawguess||

[sleek cave]

nice

[humble copper]

||i tried this and it gave me the right answer for the sample but the wrong one for the input||

[low condor]

Are you talking about today?

[humble copper]

yep

[low condor]

I'm not going to look back over all the spoilers so maybe this has been said already

||but there's a specific trick with the real input that makes it different to the test input in a particular way ||

^ this tripped me up as well

[humble copper]

oh

yep i see it

[woven sable]

Wait, what is it?

[low condor]

did you solve today's puzzle?

[woven sable]

Oh not yet

[low condor]

best to attempt it first 🙂

[woven sable]

ah true, also quick question that I'm gonna ask in #advent-of-code-spoilers-archive because i don't know how to spoiler a code block:

[muted aurora]

Do you mean <2s? Otherwise I'm not really surprised

[hidden musk]

<>2

[mossy basin]

❤️

[muted aurora]

I swear there was some language where you used this instead of ==

[rapid wolf]

there are several

[low condor]

isn't it used instead of != ?

[rapid wolf]

pascal and python 2 come to mind

[low condor]

might be wrong on that

[rapid wolf]

oh yeah, !=

[hidden musk]

and in some it's the total comparison op

[muted aurora]

yeah mb

[muted aurora]

wait python2 had <> ?!

that's wild

[rapid wolf]

indeed

[low condor]

Python 2 had all kinds of weird stuff

you could use backticks instead of repr() I believe

[muted aurora]

I...

[low condor]

And then there was input() which by default ran eval() on anything you typed into it

[muted aurora]

yeah I'm aware of the input() and print() stuff, and how decoding/encoding strings changed

[low condor]

My original Python tutor told me that raw_input() was for strings and input() was for ints, which was sort of correct but in a bad way

[muted aurora]

not really sure of what else was really different, I've just been Python3

[mossy stream]

!e

from __future__ import barry_as_FLUFL

print(True <> False)

[marsh current]

@mossy stream :x: Your eval job has completed with return code 1.

001 |   File "<string>", line 3
002 |     print(True <> False)
003 |                ^^
004 | SyntaxError: invalid syntax

[mossy stream]

aw

[muted aurora]

We need a from __past__ import ...

[mossy stream]

Python 3.8.6 (default, Apr 14 2021, 17:21:16) 
[GCC 10.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from __future__ import barry_as_FLUFL
>>> print(True <> False)
True

(it got removed with the PEG parser)

[sonic silo]

<> good

[muted aurora]

still works for me on 3.10?

[mossy stream]

interesting

[muted aurora]

is the PEG one 3.11?

[mossy stream]

should be 3.10

[muted aurora]

Python 3.10.0 (v3.10.0:b494f5935c, Oct  4 2021, 14:59:20) [Clang 12.0.5 (clang-1205.0.22.11)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> from __future__ import barry_as_FLUFL
>>> 3<>4
True

[mossy stream]

!e

import sys
print(sys.version)```

[marsh current]

@mossy stream :white_check_mark: Your eval job has completed with return code 0.

3.10.1 (main, Dec  8 2021, 04:14:05) [GCC 8.3.0]

[mossy stream]

hmm it works for me too

[muted aurora]

maybe 3.10.0 vs 3.10.1?

[mossy stream]

testing that

[muted aurora]

BTW if you're interested in my <2s solution in Python for day 19 I'll ping you on the spoilers channel

[mossy stream]

haven't finished day 19 just yet

:3

[muted aurora]

ah well, ignore me then

[mossy stream]

also

akarys@mojito ~/P/aoc-2021 (main)> PATH="$PYENV_ROOT/versions/3.10.0/bin:$PATH" python
Python 3.10.0 (default, Dec 20 2021, 16:45:22) [GCC 11.1.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from __future__ import barry_as_FLUFL
>>> print(True <> False)
True

so I have no idea what causes this

[muted aurora]

turns out it only works on the REPL

[woven sable]

That sounds weird that it would only work on the REPL

oh well

[woven oriole]

thats why there was raw_input iirc

[low condor]

Yup yup

But the naming is terrible

[woven oriole]

yeah

[wooden jewel]

yep

[mossy stream]

Why did he have to add random rotation to day 19, I swear

[wooden jewel]

I've heard of people learning py3 who are told to use eval to convert strings (from input) to int

like blegh

[low condor]

Oof

[woven oriole]

schools

[proud cosmos]

it works™️

[low condor]

The worse one I saw

[bright haven]

there is plenty of bad advice out there.

[woven oriole]

i myself have found it in a high school book

😔

[low condor]

Was where there was a library that implemented goto

As an actual keyword

[proud cosmos]

Now that's the real programming

I hope it also implemented labels

[woven oriole]

also the funny thing is that they never explained about eval in that book lmfao

[low condor]

There's also a book on Amazon that, because Python 2 and 3 has different print commands, decides it's better to make their own print library instead and uses that in its examples

its hello world example

#-----------------------------------------------------------------------
# stdio.py
#-----------------------------------------------------------------------

import sys

def writeln(x=''):
    """
    Write x and an end-of-line mark to standard output.
    """
    if sys.hexversion < 0x03000000:
        x = unicode(x)
        x = x.encode('utf-8')
    else:
        x = str(x)
    sys.stdout.write(x)
    sys.stdout.write('\n')
    sys.stdout.flush()

#-----------------------------------------------------------------------
# helloworld.py
#-----------------------------------------------------------------------

import stdio

# Write 'Hello, World' to standard output.
stdio.writeln('Hello, World')

#-----------------------------------------------------------------------

# python helloworld.py
# Hello, World

[bright haven]

UGH

[low condor]

Introduction to Programming in Python: An Interdisciplinary Approach

I need to leave an Amazon critical comment at some point, ward people off this

but yeah, throughout the entire book you see stdio.write('blah blah') instead of print('blah blah')

[mossy stream]

I think my rotation maths are alllllll wrong

[bright haven]

not only is it weird to invent a new function for printing, but the implementation of the function is weird and too long.

[wooden jewel]

from __future__ import print_function anyone?

or just teach py3 lmao

[steel bear]

what on earth is this lol

[low condor]

The book was released in 2015

[steel bear]

I guess its aimed for C/C++ programmers

[wooden jewel]

jesus

[low condor]

it should have been written in Py3

[low condor]

The title is Introduction to Programming in Python: An Interdisciplinary Approach

[steel bear]

yeah agree

[steel bear]

hmm, weird lol

[wooden jewel]

I learned Python in... 2012-ish? and back then it was "yeah just use py3, but there is an older version around here are some things to watch out for"

And I'm pretty sure the website I used was also around pretty long before that

(copyright 2010, so yeah)

(https://cscircles.cemc.uwaterloo.ca/ for those wondering)

[low condor]

https://introcs.cs.princeton.edu/python/home/ <- the website for the above horrid code

[bright haven]

from __future__ import print_function appeared in 2008

[humble copper]

heresy of the highest regard

[low condor]

I still rank the goto library as the highest heresy, but this is up there for sure

[mossy stream]

ohmyfuckinggod

hmm okay, part 2 shouldn't be too hard

I feel bad for copy pasting the code every time

[low condor]

Sometimes I read from input twice cos I can't be bothered to refactor ¯_(ツ)_/¯

[mossy stream]

I literally do cp day_19/part_{1,2}.py

2.76s, not too bad

alright, day 18 and 19 done

the aoc is a full day job, I swear

let's do 20 and 21 tomorrow

hopefully they aren't too bad

I have a certain Sally to murder

[sonic silo]

it wasnt so much a keyword

rather

it was a global object thing

that had lots of overloads for the sake of syntax sugar

p funny nonetheless

[low condor]

So, this wasn't the one on pypi

This was another where the syntax was goto whatever

No parens

Can't find it though...

[sonic silo]

the one i was talkin about is in the stdlib

iirc

[peak dock]

the what

i am quite sure theres never been a goto in the stdlib

[low condor]

Similar to this but not exactly this
https://pypi.org/project/goto-label/

[sonic silo]

bytecode rewrite?

where do functions store their bytecode?

whats its interface within python itself?

[mossy stream]

well, she's dead

that was much easier than I thought

[glad tartan]

!e def func():
import dis
dis.dis(func.code)

func()

[marsh current]

@glad tartan :warning: Your eval job has completed with return code 0.

[No output]

[low condor]

this was the library I think: https://github.com/toanh/csinsc . The goto implementation was taken from an April's Fool joke and used in a library for students to use in their homework

[glad tartan]

!e def func():
import dis
dis.dis(func.code)

func()

[marsh current]

@glad tartan :white_check_mark: Your eval job has completed with return code 0.

001 |   2           0 LOAD_CONST               1 (0)
002 |               2 LOAD_CONST               0 (None)
003 |               4 IMPORT_NAME              0 (dis)
004 |               6 STORE_FAST               0 (dis)
005 | 
006 |   3           8 LOAD_FAST                0 (dis)
007 |              10 LOAD_METHOD              0 (dis)
008 |              12 LOAD_GLOBAL              1 (func)
009 |              14 LOAD_ATTR                2 (__code__)
010 |              16 CALL_METHOD              1
011 |              18 POP_TOP
... (truncated - too many lines)

Full output: https://paste.pythondiscord.com/osumutenec.txt?noredirect

[restive imp]

I think I'm starting to lose my mind with this advent calendar 😔

[sonic silo]

ooh

[restive imp]

hours of adjusting my code an figuring stuff out, and in the end it still doesn't work

[sonic silo]

oh god

why does the bytecode have its own methods

[glad tartan]

I have some examples on how to create a function from scratch here https://github.com/P403n1x87/surgery

[sonic silo]

nice

[glad tartan]

In 3.11 every code object has a co_qualname attribute (I contributed this feature :))

[sonic silo]

since builtins like sum are written in C that means they mustnt have bytecode 🤔

ye

oh wait

thats not

ok yeah i was right

[glad tartan]

!e sum.code

[marsh current]

@glad tartan :x: Your eval job has completed with return code 1.

001 | Traceback (most recent call last):
002 |   File "<string>", line 1, in <module>
003 | AttributeError: 'builtin_function_or_method' object has no attribute '__code__'. Did you mean: '__call__'?

[sonic silo]

yep

[glad tartan]

!e import sys; print(sys.version_info)

[marsh current]

@glad tartan :warning: Your eval job has completed with return code 0.

[No output]

[glad tartan]

!e import sys; print(sys.version_info)

[marsh current]

@glad tartan :white_check_mark: Your eval job has completed with return code 0.

sys.version_info(major=3, minor=10, micro=1, releaselevel='final', serial=0)

[glad tartan]

bleeding edge

[sonic silo]

this is like

dam

i should investigate about this

pushes that into the long queue of "things to investigate for later"

[glad tartan]

https://github.com/MatthieuDartiailh/bytecode

[sonic silo]

can i not manipulate it directly?

or is this just some nice API over an already existing but lmited API?

[glad tartan]

code objects are immutable, as well as function objects. This package is a nice way of manipulating bytecode by using abstract instructions where you don't have to resolve jump offsets yourself

[sonic silo]

hm

does that mean the wrappers and decorators return new functions?

which read the original function and make a new one off its bytecode?

[sleek cave]

um

if this is the right answer

okay it's not

[humble copper]

i hate this.

||took 5 hours to do, with each attempt at running the program taking 15-20 minutes ||

[low condor]

oof

[humble copper]

||because i dont want to waste time optimizing||

[low condor]

by not wasting time, you end up wasting time 😛

[bright haven]

why is this marked as a spoiler? 🙂

[humble copper]

i thought i was gonna end up mentioning how i did it, so i spoilered just in case

[low condor]

this is Advent of REDACTED

[humble copper]

should i unspoiler

[ruby epoch]

For day 5, we don’t actually need to draw the board, do we?

[bright haven]

the answers are always short strings that you type into the site

[sleek cave]

BRUH

||30 minutes of debugging only to find i forgot to heapify my priority queue every time i mutate it lmao||

programming in a nutshell: stupid mistakes

[sonic silo]

mood

[sleek cave]

||i made a very shitty a* that probably won't solve part 2 of day 15||

time to optimize

[peak dock]

||you shouldnt need to heapify repeatedly after you've built the heap, are you using heappop/heappush/etc to mutate the heap? if so, those functions maintain the heap invariant so you dont have to heapify again||

[sleek cave]

oh, it's a hint on optimization

sorry, no hints, can't read!!

[robust heart]

There's something I don't understand with day 16

How do you define where the subpackets are

In the second example ie

00111000000000000110111101000101001010010001001000000000
VVVTTTILLLLLLLLLLLLLLLAAAAAAAAAAABBBBBBBBBBBBBBBB

I understood that the next subpackets must have a length of 27

But how do you define where A stops, where B begins, and if there's a C subpacket

[velvet stag]

you always know where a packet stops

from the length

or the 0 marker if it's a literal

[sleek cave]

you always know how many subpackets are there or how long the subpacket data is using the length bits

[velvet stag]

and yeah you don't know in advance

you can't just say the packet is from here to here and read that part, no way around it

you need to start reading blindly

[sleek cave]

||https://docs.python.org/3.5/library/heapq.html#priority-queue-implementation-notes saves my life||

[robust heart]

oh so I have to cut my packet into bits and see which ones fits?

[velvet stag]

uh

i don't know what you mean

[robust heart]

VVVTTTILLLLLLLLLLLLLLLAAAAAAAAAAABBBBBBBBBBBBBBBB I have to try where A extends till it's correct, right?

Haven't done many bit manipulations

[white elbow]

I don't quite understand day 20

and I have a weird thing that when I comment out the print_image function I made I get a different answer

[pallid quail]

I am on day 4 - (literally been writing pythin for about a month, so its slow going).

I had a quick question. For the win conditions for the bingo boards - right now I am just thinking of writing out a bunch of if statements that check for each index for the various win conditions. Its going to be huge - wondering if there is a better way to approach this.

[white elbow]

you can use for loops

[pallid quail]

But, the loops will still need an if statement for checking each index in a possible "win condition"

[white elbow]

I made 2 loops for horizontal and vertical, and in the loops I said if a == b == c... : return True

or whatever your loop variable is named

[pallid quail]

I guess I could make a dict of the various win conditions, then wrote a loop that checks each board for each possible condition...

[white elbow]

and at the end of the function I have a return False

[low condor]

@pallid quail ||I used numpy to let me check the vertical conditions||

[glad tartan]

they normally create closures

!e import dis
def decorator(f):
def wrapper(*args, **kwargs):
return f(*args, **kwargs)
return wrapper

dis.dis(decorator)

[marsh current]

@glad tartan :white_check_mark: Your eval job has completed with return code 0.

001 |   3           0 LOAD_CLOSURE             0 (f)
002 |               2 BUILD_TUPLE              1
003 |               4 LOAD_CONST               1 (<code object wrapper at 0x7fd59d0fe290, file "<string>", line 3>)
004 |               6 LOAD_CONST               2 ('decorator.<locals>.wrapper')
005 |               8 MAKE_FUNCTION            8 (closure)
006 |              10 STORE_FAST               1 (wrapper)
007 | 
008 |   5          12 LOAD_FAST                1 (wrapper)
009 |              14 RETURN_VALUE
010 | 
011 | Disassembly of <code object wrapper at 0x7fd59d0fe290, file "<string>", line 3>:
... (truncated - too many lines)

Full output: https://paste.pythondiscord.com/evoqozehax.txt?noredirect

[glad tartan]

see the MAKE_FUNCTION call there

[sleek cave]

hint: ||the win condition can be generalized to "5 marked numbers in a line, vertically or horizontally"||
hint 2: ||this can be determined by keeping track of marked numbers' coordinates all your 5 by 5 boards, since 5 in a line vertically means their x's are the same and 5 in a line horizontally means their y's are the same||

[woven sable]

I love this package from how I've used it so far

[violet raft]

for the transparent oragami, is it by chance that the chosen x and y was just perfectly half of the paper?

or could it be like folding at x = 8

when the length of the paper is 10

[woven solar]

it can be

[noble skiff]

someone told me it was always at the halfway point, but I never tried to verify that or took that into account in my coding

[violet raft]

ok, thx

[placid lake]

it’s always at the half point, yeah

[humble copper]

historically, have days 21-25 been easier than 16-20?

or are they the hardest ones

[woven sable]

They have, since people usually start travelling for the holidays iirc

actually, i'm going to test this out into actual statistics!

[sleek cave]

ugh another bruh moment day

my thought process is clouded by a thick substance of confusion

can't even implement ||a*|| 😔

[woven sable]

google sheets doesn't let me label the years but here's all AOC years ever and how many % of pepole have completed themv ersus at the beginning:

[woven sable]

Hmm, from the stats it said that on average from all the years, 13.03% people completed days 16-20 on avg but only 9.18% completed days 21-25 on avg

[humble copper]

that orange line on day 11

[woven sable]

yeah, I wonder what year that was

also fyi @humble copper ,here's a chat of the average amount of people who complete deach day from day 16 through 25 in all years

[pallid quail]

Ooh! I hadn’t thought of treating them as x y coordinates.

I was just dealing with a dictionary for all of my rows and triggering a win if all cells in a row are marked, or when considering all rows of all cells at a specific index position are marked.

Made for some convoluted logic.

[sleek cave]

ah, i see

[low condor]

@sleek cavewhat day are you on?

[sleek cave]

day 15. i'm really struggling today but i think it's just me not having enough energy today

||my a star algo is so badly implemented||

but nonetheless, after literally spending 20 minutes on how to expand the input by 5 in both dimensions (you can thus get an idea of how confused my brain is today, to be fumbling on such a simple part of the problem), i'm running the solution

i expect the solution to be calculated in about an hour

so it's just patience from here lmao

[low condor]

hey, as long as you get the solution, who cares, right? 🙂

just as long as it completes in an hour and not 500 years, that's totally fine 😛

[sleek cave]

lol, thanks a lot, i totally agree with you
i need to get rid of this unrealistic expectation where i think i can pass the puzzles efficiently each time, or else i might just get more and more irritated while solving it like today lmao

idk why, sometimes i see a puzzle and i go "hey, i can definitely solve that" then that mindset starts yeeting me off into the grand canyon of confusion and despair (exaggerated, but it's kinda like this)

[pliant comet]

i think i spent longer than this lol 😔

[low condor]

I don't think it's necessarily unrealistic, more that if you don't get a perfect solution, don't let that be the enemy of good

[sleek cave]

yeah, but like today, i keep on getting mad at myself because i just can't implement a*

that's what i mean by unrealistic - i'm eventually going to not figure out the answer at one of the puzzles (because eventually there'll be one that exceeds my capabilities), but i'm still irritated when that happens

[low condor]

that's fair. Well, if you get the solution, you can always come back to it afterwards

also, if a puzzle does defeat you, and it happens to the best of us, the best thing to do is accept it and see how it should be done

that way you learn from it

I had to do that with 2019 day 18 this year, no matter what I tried I just couldn't do it

[violet raft]

i don't understand what counts as a dot for day 13

are these like dots that are being folded onto another dot?

or a # thats being folded onto a dot, so it's visible on the back side of the paper

or is it a dot that's replacing a #

[low condor]

||if a # folds onto a ., that leaves a #||

||if a # folds onto a #, then there's still a # there. And the same with . .||

[violet raft]

ok, so am i counting how many times ||"if a # folds onto a ., that leaves a #"|| this happens for part 1?

[low condor]

||you're counting how many # are left with one fold||

[violet raft]

oh, wow

i got confused lol

i forgot # were "dots"

and thought "." were dots

[low condor]

easy mistake lol

[violet raft]

ty lol

[low condor]

'.' are literally dots 🙂

[sinful pawn]

well... technically... they're "periods"

•, though, is a dot

[low condor]

The official Unicode name is FULL STOP

which, as a British-man, I approve of rather than period 🙂

[sleek cave]

that's it. I'm switching to saying full stop now lmao

\s

[woven solar]

I think an exponential fit would be a better idea

since it's plausible that every task removes x% of the playerbase

[woven sable]

Dumb question, but what's an exponential fit?

[woven solar]

Fitting an exponential curve a*e^(b x) instead of line

it can be done, say, by doing a linear fit on x, log(y) - if you get a,b such that log(y) ~= a+b*x, then y ~= e^a * e^(b*x).

[woven sable]

Ah

[woven solar]

here, you could fix a=0 since day0 is always 100% and only fit b

[sleek cave]

me brain has now recharged after today's defeat on day 15. I'll reattempt that day when I have time. looking forward to day 21!

[hidden copper]

<@&518565788744024082> Good morning! Day 21 is ready to be attempted. View it online now at https://adventofcode.com/2021/day/21. Good luck!

[sinful pawn]

oops

[stable heron]

interesting

but ive just given up

[crystal shard]

Heh

[stable heron]

but hey this isnt that hard...

[crystal shard]

No, seems easy

Lol

[stable heron]

lmao

[humble copper]

this seems easy

[potent pumice]

its been 6 minutes! where are the stars

[crystal shard]

I am doing school work

[potent pumice]

not you specifically, i mean the screenshot you took

[crystal shard]

Oh, people solved it already

only 27

[potent pumice]

now 51

57

68

[woven solar]

wtf do you mean

the lb filled in 20:44

[potent pumice]

yes, i sent those at 00:18?

[humble copper]

interesting part 2

...and i have no idea what im doing

[humble copper]

||i wrote down the different rolls that could happen, im assuming this has something to do with how to solve pt2?||

[pearl moon]

On day 21....will a brute force solution work for part 2? I glance at part 1 and see it's easily brute forced but also sense there should be a closed form solution. I just don't know how to easily work that out

[woven solar]

||no, for part2 it won't||

though ||part2 is very different from part1 for day21||, so it's okay to use a straightforward solution for part1

[onyx plank]

ok so part 2 is getting out of fucking hand lmao

part 2 be like "but what if it were 5D Chess With Multiverse Time Travel?"

[woven solar]

part 3: players can now choose to instead go back in time to an earlier state

[pearl moon]

Ahh ok. I'm just afraid of writing out a brute force solution for part 1 which would be a waste of time...like when if you just write an optimized solution for 1, 2 is cake

[sonic silo]

5d gambling???

:o

[cosmic herald]

I'm confused with the Day 5 puzzle

[cosmic herald]

In this drawing, it shows the numbers at pos 7, not 8 (like the co-ords say)

Am I missing something?

[woven solar]

8,0 -> 0,8 is a diagonal vent

it's not shown at all, as you ignore these in part1

[cosmic herald]

OH!

I've spent longer than I'd like to admit on this one

Thanks

[mossy stream]

What's the catch for today, hmmm

It seems a bit too easy

[sonic silo]

part 2 is funky

[candid berry]

it's the easiest one in a while, though

[muted aurora]

I'd say yesterday's was easier tbh

[candid berry]

Hm. Well, I found today's easier

[bright spindle]

I had about the same amount of trouble with yesterday's and today's, though today it was mostly because I misunderstood the part 2 description.

Yesterday it took me a while to understand the "trick"

[candid berry]

Took me over an hour to finish both parts of yesterday's, took me 20 minutes to finish both parts of today's

[bright spindle]

I figured out a workable solution to both parts very quickly, but I misunderstood a requirement for part 2 so I got stuck debugging for like an hour before I spotted it

||For some reason I didn't grok that you had to add up three rolls for part 2 as well. But once I got it, it was easy to adapt my solution to it.||

[quick glacier]

In today's part 2, do you still throw the die three times on a turn?

[bright spindle]

||Yes. I misunderstood this and wasted like an hour.||

[quick glacier]

For some reason I didn't read the most recent post on this thread before I asked. Figures.

[bright spindle]

Heh

[quick glacier]

How do I make spoilers, btw?

[bright spindle]

|| like this ||

[bright haven]

or /spoiler the whole message

[quasi fjord]

||hm didnt know of this mehtod, knew of || this || tho||

[bright spindle]

Oh, nice

[proud cosmos]

I wonder why my numbers are so low :(

More xamples pls.

What a mess to debut

Actually just not sure what could be wrong lol

I guess I can go through everything one by one till I see anything wrong, sounds a bit questionable though

Mayr I'll just go do the obvious solution method instead should be a lot easier

[sleek cave]

drop from p1 to p2 today, interesting

if it's exponential growth again I will cry

[proud cosmos]

:)

[sleek cave]

oh god, it is isn't it

part 1 seems too deceptively easy

[bright spindle]

:))))))

[sleek cave]

okay, I need a minor spoiler: is today's part 2 harder because you need an efficient solution or it's something else
because if it's the former I might do day 20 first

[bright spindle]

||Part 1 and part 2 are like, almost nothing alike||

[sleek cave]

so it's not the former, yay :D

[bright spindle]

||I had to write a new solution from scratch||

[sleek cave]

I struggled so hard with day 15 yesterday, couldn't figure out how to implement an efficient algorithm

[bright spindle]

||I mean, it can't be inefficient||

[sleek cave]

so I was hoping today isn't an algorithm puzzle

[sleek cave]

oof

[bright spindle]

||But I don't think it's super hard to figure out either||

[sleek cave]

alright, thanks!

[humble copper]

||recursion|| cant be too bad of an idea for this, right?

as long as ||i remember to multiply for the number of times a specific final roll can occur||

[simple sorrel]

wasup everybody

??

[bright spindle]

||Recursion apparently does work well for this problem. I did an iterative solution, but other people have had success with recursion.||

[simple sorrel]

hello?

[humble copper]

||thats good to hear, i wouldnt expect a recursive solution to be too far of a reach, because the games (i think) cant go on for anywhere close to 1000 total turns||

[woven oriole]

||Cache||

[proud cosmos]

You're amazing my try at that struggled so hard and was wrong by a lot lol

The other way is just so easy

Still haven't figured out what's wrong

[sonic silo]

||p sure the vast majority of people went with a recursive||

[bright spindle]

||Depending on when you recurse, it's more like < 2*21 or < 2*21*3 turns||

[sonic silo]

||at worst it would be actually 7 turns, for 21 unlucky dice rolls||

[bright spindle]

||https://github.com/dementati/aoc2021-rs/blob/main/src/day21/mod.rs it's pretty slow though, 5 seconds for part 2 and this is in release mode Rust||

[proud cosmos]

Hmm

I'll look at it later in curious if it's similar to min, doubt

[bright spindle]

It's a shame that it's slow because I thought it felt neat.

[humble copper]

||my recursive solution is really slow, which doesnt exactly make sense||

[proud cosmos]

It should be pretty fast

[bright spindle]

||Does it work?||

||I mean, does it finish, even if it's slow?||

[humble copper]

||appears to be stuck on the first step||

[bright spindle]

Oh, ok

[humble copper]

||its going, but really slowly||

||to be fair, that could be because i have probably the worst cpu known to man||

[glad tartan]

does this spawn a new universe at every die throw?

[muted aurora]

Hmm that's fishy, mines like 100ms in Python

[bright spindle]

Yeah

[bright spindle]

Yeah, maybe it's the recursion strategy or maybe there's something about the iteration model.

[sonic silo]

p sure that

[glad tartan]

I'd try spawning just the ||7 possible branches and use their weight to cut down on the possible different universes||

[sonic silo]

||hes just not doing memoization||

which is why its taking so long

[glad tartan]

mine is like that too

[bright spindle]

||I am doing memoization, that's what the cache is for||

[muted aurora]

Mine's ||just recursive|| too, and my C solution is also fairly similar at ||~5ms||

[sonic silo]

||my bad||

[muted aurora]

||using spoilers is fun||

[sonic silo]

||maybe||

[bright spindle]

||Not sure what that means. What 7 possible branches? What weight?||

[humble copper]

||7 possible final throws, 3-9||

[bright spindle]

||Otherwise it'd never finish||

[muted aurora]

I'll ping you in spoilers

[glad tartan]

||e.g. there are 7 universes where the sum of 3 die throws is 6 etc...||

[sonic silo]

have you tried doing profiling?

[bright spindle]

I don't really have time for that, I was already late to work when I finished this solution

Maybe I'll polish it later, but my parents are coming to stay tomorrow and I've got shit to do before then

AoC is very fun, but it's kind of naturally at the worst possible time in terms of having free time

[somber yacht]

why everyone talking in spoilers?

[bright spindle]

If you celebrate Christmas, that is.

[bright spindle]

We were discussing today's problem, and we don't want to spoil it for those who haven't completed it yet.

[somber yacht]

Oh ok lol

isn't this for non spoiler things tho?

[bright spindle]

It's not a big deal to talk about spoilers here if you spoilertag them, but if you want to discuss full solutions it's just more convenient to to do it in #advent-of-code-spoilers-archive

[muted aurora]

It's very easy on the spoilers channel to have stuff spoiled for you even if you just want a different completely unrelated question answered, so I use spoilers on here to answer specific questions people have usually

[bright spindle]

Yeah, that's true.

[hollow wharf]

waaat, my day 20 input starts with a true

which mean my result is infinity since any pixel outside that gets enhanced will flip to lit?

wtf

[humble copper]

||perhaps that may be why the number of times you have to enhance is even? 👀||

[hollow wharf]

the prompt literally doesn't say anything on how to solve this

oh, hmm

but like

whyyy

[sleek shuttle]

hi

what's advent of code i've never done it before

[rapid wolf]

you get a puzzle every day of december up to the 25th

[woven oriole]

||yes because at the end the infinite region flips back to dark pixels||

[humble copper]

currently fighting for its life right now because of today's part 2

hovering somewhere around 7000-7300 mb

[rapid wolf]

I do wonder if that will finish

[humble copper]

it didnt

[hollow wharf]

the fuck is that prompt

time to do maths ig

ooooorrrr

hmm

I am finding interesting results

I hardly see how such big numbers can be achived

[somber zodiac]

O_O

[frigid mural]

idk if this is still relevant, but for day 20 ||the input also ends with what you would call a False, meaning that the infinite area of unlit->all lit on enhance 1, then goes back to all unlit on enhance 2||

[fossil cipher]

random thought of the day... have bugs been found in people's inputs in AoC? I imagine every generated input is run through a solution to check it works... but nothing is perfect.

I can imagine it's extremely frustrating if your result is correct, but expected answer is wrong 😄

[rapid wolf]

I assume there are enough testers that someone would spot the error

[sleek cave]

oh god major drop from part 1 to part 2

[bright haven]

occasionally something goes wrong: see the message just above the 2020 leaderboard: https://adventofcode.com/2020/leaderboard

[bright haven]

and the 2018 leaderboard: https://adventofcode.com/2018/leaderboard

[sonic silo]

nah

wait

yeah

year 2020 day 23 did have some bugs iirc

[hollow wharf]

Yeah I thought about that after but I can't really be bothered :D

[frigid mural]

F

my solution uses a pretty quick fix for it, idk whether itll be hard to implement for urs

||just have a default value for unseen coords and have it alternate between days||

[hollow wharf]

Yeah that's what I did

[crisp coyote]

I didn't realise that special case but somehow my solution worked anyway lol

[hollow wharf]

Does your input not start with a #?

[sleek cave]

same here

okay why is every puzzle's part 2 just severe optimization now, leave my naive inefficient algo alone 😔

d20p2 requires optimization once again

time to do math

[frigid mural]

o_O

[crisp coyote]

it does

[pliant comet]

||if you solve it using a 2d array instead of a set of points, you don't need a special case for that (when i tried using a set to solve it, i used a defaultdict and defaulting to 0 caused problems because of the # at the beginning)||

[hollow wharf]

Damn

[rapid wolf]

I was trying to come up with a clever way to do it with sets

[frigid mural]

feels like a spoiler tag here?

[pliant comet]

oop sorry

[clear nimbus]

for you what was the hardest day until today in 2021 AOC?

[sleek cave]

sigh saw the spoiler

[sleek cave]

19, i didn't even look at it

[peak dock]

19 imo

[fossil cipher]

Does it? Mines a mess and isn’t optimised in any way on purpose lol

[minor cave]

According to the stats, def day 19

[peak dock]

Iron man dies

[clear nimbus]

do you think this year has been more difficult that the others?

[crisp coyote]

ah yeah I used a ||2d numpy array||

[low condor]

probably a touch harder than last year, but nothing like 2019 and before

[fossil cipher]

This has been my best year. But I think the hardest so far. I’ve only done so well due to motivation from you lot 😄

[hollow wharf]

||I totally used a 2D array and totally got problems||

[sinful pawn]

this channel has slowly evolved into spoiler heaven

I am so tempted to click them all

[clear nimbus]

i'm about 1 week behind because of school, can anyone explain if in day 15 you can walk backwards or only upwards and to the left?

[peak dock]

Can walk in all 4 directions

[crisp coyote]

I've done 20 but am still stuck on 18 for some reason. I've written 150 lines of unit tests and still can't work out what's wrong 🥲

[peak dock]

Rip

[clear nimbus]

oh... because I basically used the same algorithm as I did in project euler level 67... well it doesn't work ...
https://projecteuler.net/problem=67

[hollow wharf]

Have you checked that your algorithm finding the numbers to increase when exploding is correct?

[fossil cipher]

18 and 19 are dead to me.

[hollow wharf]

That was the problem for us

[frigid mural]

exploding is more important than splitting, thats one a lot of people tripped over, it says in the problem that if there can be an explode you do that first

[sonic silo]

thats cool

i wanna eventually do project euler tasks

they nice

[clear nimbus]

yeah, but unfortunately it worked with all the inputs including sample, except for the last one

because it just not correct since you can move in more than 2 directions

[crisp coyote]

I'm pretty sure i've got my traversing stuff right, although there might be a subtle bug somewhere... These are my tests ||https://paste.pythondiscord.com/oraloqesin.py|| and literally only the last one is failing, which is a full addition I copied from the problem and has like 20 reduction steps so will be pain to debug

[clear nimbus]

there could potentially be a case where going aroung and passing through like 3 "ones" is better than through 1 "nine"

[humble copper]

i think ill just go sleep now

[muted aurora]

till next year's AOC you should just implement your own full stdlib 😅

[sonic silo]

hm

i end up writting everything from sctach anyways

[crisp coyote]

yeah I think I got that right. Might be something to do with the order i'm traversing over the tree though, idk... time to add more tests lol

[humble copper]

yep im gonna copy a ton of functions i did this year and put them in a custom library for next year

[sonic silo]

i think the only thing i would want is a hashable vector

[muted aurora]

why scratch? Use a real language

[sonic silo]

2d to 4d vector

[sonic silo]

lol

but still

just 2d, 3d and 4d vectors

hashable vectors that is

[muted aurora]

Join my on the dark side in C

[sonic silo]

some op overloads and thats it

[muted aurora]

I mean I'd just overload a tuple and redefine the +- etc operations

[sonic silo]

🤔

right, you can just inherit a tuple and redefine ops

[muted aurora]

ye, inherit is the word

[sonic silo]

but still

thats still making vectors lmao

[muted aurora]

yeah yeah

would be super convenient I agree

[rapid wolf]

julia I have found treats vectors in sets properly

[sonic silo]

thats it tho, i dont even use my parsing functions

[muted aurora]

i wrote a helper to parse a grid for me

[sonic silo]

yeah, same

[muted aurora]

got to use it a surprising number of times so far

[sonic silo]

i DID use intmatrix which is just a parser for converting a 2d grid of single digit numbers into a numpy array

and i do have ONE single snippet

made exactly for writting a variable called "data" that's assigned to an empty multiline string

data = """\
| <- cursor, copy data here
"""

this way i can just.. well. paste any samples from the exercise directly

[sleek cave]

ggs!

[humble copper]

next year im not making it this difficult for me

[violet raft]

could someone give me a hint on day 14 part 2

[sleek cave]

hint on general approach: ||try using recursion, working your way backwards until the problem is reduced to a bunch of base cases. at least this is how i solved it||
hint 2 (more specific, might not want to look at it now): ||the base cases said above are searching polymers and polymer substrings in the original template||

[violet raft]

ty, i'll keep these hints in mind!

[sleek cave]

you're very welcome!

[frigid zephyr]

I just finished day 18 (snailfish) feeling like a ||DFS|| god, and now I see day 19... what on earth

[rapid wolf]

day 19 is hard

[muted aurora]

oh boy I'm sorry, yeah 18 and 19 were tough

19 was something else

[rapid wolf]

it gets easier after that

[muted aurora]

20 and 21 have been pretty jokes tbh, just some ||annoying tricks to look out for||

[rapid wolf]

yeah, I think my favourite were the snailfish thus far

a tough problem with multiple approaches, each with its own pros and cons

but none of them were dead ends

[frigid zephyr]

snailfish was quite fun tbh

[muted aurora]

yeah i did it both the popular ones, one was hell

[low condor]

I found d18 easier than p2 today, but yeah, d19 hardest for sure

[frigid zephyr]

I might leave d19 and come back to it later

[muted aurora]

I think it was implementation vs problem solving difficulty, day 18 I knew what to do, was just annoying to implement

today I spent a lot of time for p2 just thinking about approaches

[frigid zephyr]

What were the common approaches to d18 people used? I used a bunch of ||DFSs (some recursive some not)|| and got it to work

[low condor]

probably best to discuss that on the spoilers channel 🙂

[muted aurora]

The two most common were, generally speaking ||using a recursive structure such as a Tree, or just parsing it as a flat list/tokens||

[frigid zephyr]

oops, I forgot where I was

[low condor]

^ i did the second one of that

[robust heart]

I'll never get this done by the time Christmas strikes

But at least i'm halfway done

Way better than my previous attempts

[humble copper]

take your time

if you don't feel like completing all of them in a short time then you dont have to

[hollow wharf]

so I heard we caught up to the AoC

that's cool

[robust heart]

I was too cocky when I assumed I could end them in a day. Too cocky.

[hollow wharf]

all of them? in one day?

[humble copper]

im trying to do them all within 24 hours after they release
its really hard but its almost over

just a few days to go

(and on top of that, without using any imports at all, even standard library imports. i've come to the realization that maybe i hate myself)

[woven sable]

this ^

[low condor]

if you aren't using standard library then you definitely hate yourself 🙂

Python is meant to come with "batteries included" but you've removed the batteries 😛

[rapid wolf]

do everything using just lists and ints 🙂

[humble copper]

days 18 and 19 were fine honestly

[woven sable]

I found Day 20 and 21 quite hard tbh

[humble copper]

16 was torture

[rapid wolf]

16 was pretty straightforward for me, but I also have quite a bit of experience writing parsers

[low condor]

I liked 16 a lot

I tend to enjoy parser-style questions

[rapid wolf]

same

[humble copper]

i was half-joking during the first days, but after 21 days of it i've realized that for me it's a great way to understand how the language works, and how to properly understand problems

[low condor]

the ones I tend to struggle with are the ||part 2 here's something with huge numbers || type questions

[low condor]

oh yeah - joking aside, you'll definitely learn a lot from it

[minor pike]

work has been busy so I'm very behind

still stuck on day 3 part 2 ;(

part 1 came easy, idk why part 2 is so hard for me

[rapid wolf]

yeah, I like the this is a pretty hard to implement thing kind of problems, like snailfish, more than the ||optimize thing|| like CO2 scrubbers

[low condor]

whenever you get a chance, do keep at it!

[rapid wolf]

still need figure out a recursion scheme for snailfish.

[low condor]

a recursion scheme? Like, a way to represent it?

[humble copper]

this isnt really easy to define but i really like the problems where i can make... unusual... solutions

[low condor]

I like those questions but purely from an observation perspective