The Nineteenth Byte

"I'm trying a new strategy where I mess up on the logic for an hour then realize I made a typo on line 4"

2

4:58 AM

@RedwolfPrograms me basically. part 2 yesterday, small bug cost me 10 mins

@Lyxal nice haskell infinite quine

1

note to self: how to read

i literally dont read the prompt

.5

.25

gl everyonbe

4:59 AM

i have my chalk ready

im having great lag right now

same what the hell

i can't load p2

okay GG i'm literally unable to view the problem anymore

rank 36 part 1

Same

I got 37 on part 1

17

unless everyone's also getting issues i probably lose p2

looks like CRT to me, maybe?

diophantine?

im stuck too

5:06 AM

chinese remainder

idk tho

Only 1 person has done part 2 so far

Now 3 people

mmm

definitely a bit slow to brute force lmao

"surely the actual earliest timestamp will be larger than 100000000000000"

crt sounds like it may work

brute force for each number

okay yeah it's crt

5:15 AM

I don't know crt :(

same here

i didnt either

but i google

also not reall crt it just 1 by 1 brute force

Mine is 100% larger than 100000177492586

how have 299 people already done this

im just dumb thats how

idk whats wrong wiyh my thing

5:26 AM

rank 361

pain

if only i actually thought about it for longer instead of switching between my three ideas and messing them all up lol

Wait, am I wrong in assuming a number n's n mod a and n mod b don't change when you add a number made by the product of the factors of n?

that's sort of what i've been trying to do

but that's probably still way too slow

Mine finished super quick but was wrong

rip

About a tenth of the minimum it said

5:28 AM

lol

okay so how do i use this "crt" (Chinese remainder theorem?)

No idea

If someone could explain that'd be really helpful

wikipedia is not being helpful on how to actually use these bézout coefficients

ok so

i lied it isnt really crt

tldr do it for 1 thing

now your step is bus # 1

after next? step is bus # 1 * bus # 2

so much spoilers tho so

bye

okay so it's what me and redwolf are trying lmao

5:31 AM

im angery btw my thing is erroring for my thing

My theory is right but my implementation is wrong. Go figure. :p

dumb error

rank 556

I was able to understand how to do it by reading Wikipedia

okay so if i've understood correctly: find the earliest timestamp for bus 1, find the earliest timestamp for bus 2 starting at the last stamp, then same for the third etc?

or have i missed the mark completely

yeah it's... not actually chinese remainder theorem really

@Lyxal well basically you have a bunch of congruence relations

for example with the sample input, you need to find x such that

x is 0 mod 7
x is 1 mod 13
x is 4 mod 59
...

if you have a value that's 0 mod 7 and 1 mod 13, adding multiples of 7*13 will not change the result

5:33 AM

wdym by "x is y mod z"? you mean x = y mod z?

yeah

wow i choked hard part 2

oh well

so for the first x, it is a multiple of 7

wow smh hyper

so you start at 0, which meets the first requirement, then you keep adding 7 until you meet the second requirement (the number is 14), then keep adding 91 until you meet the third requirement, then add 5369, etc

5:34 AM

spoilers

oh

okay

hyper what rank did u get btw

sorry it's well past R100 otherwise i wouldn't say anything

probs 10th tbh

welp

17/361...

5:35 AM

at least u didnt choke as bad as me...

36/556 :^)

okay i completely screwed up my approach for p2

and my p1 was pretty slow today

i didnt know how to do part 2 until i did it

also like 3 bugs in p2 as is usual for me

wait that was easy as hell

i'm just embarrassed now

698 lol

I don't know why I'm wrong

How do you deal with ones where the offset is larger than the number being multiplied?

i think i may have overkilled it

5:37 AM

your residue checking might be backwards like mine was

anyone else use modinv

no

I'm doing current_number % bus_number == bus_offset % bus_number, is that correct?

bro i used js bro

@RedwolfPrograms that's backwards

5:39 AM

What's backwards?

you're finding a timestamp where they departed that long ago with that

@RedwolfPrograms almost. im dumb so i forgot only python (?) keeps positive mod?

tldr

holy shit i'm stupid

bus_offset is wrong

lel hyper

you want to add the index and check for divisibility

5:40 AM

oh god i choked on this prob SO hard

i could've gotten it in like sub 10m probably

unless you can check for congruence with negative numbers lmao

by literally adding two more characters

i hate myself

@HyperNeutrino saaaaaame

i just spent too long between approaches

i couldn't decide how to try

two more wtf

I don't get what you're saying, @UnrelatedString

5:41 AM

funny thing

i got the reduce down waaaay early

just had no idea what to put in it

@RedwolfPrograms add the offset to the id and check the result against 0

i was checking s % m == r without noting that r could be >= m.........

Id?

the bus id

s = current, m = bus id, r = offset

5:42 AM

@HyperNeutrino How'd you solve it?

Same issue I have

when i said add to the id i meant add to the timestamp and not the id

@HyperNeutrino same

i got that

@RedwolfPrograms the same brute-force method but using modinv to calculate the required additions because i thought my original method is too slow, but it's just cuz i set a while loop with an impossible fail cond

spent like 10 mins console logging

5:44 AM

@UnrelatedString So (current_number + bus_offset) % bus_number == 0?

lol

@RedwolfPrograms yes

oh yeah my answer was negative what it should've been so i just subtracted it from the product of all of the IDs lol

@HyperNeutrino where do i get the 91 from?

7 * 13

from example

@UnrelatedString Not working still :/

5:46 AM

send your entire sol?

Good luck reading it, but here

who else has finished btw

how do i interpret this

*run

@RedwolfPrograms try using LCM instead of just a product perhaps? shouldn't help if the result is still too small

I just verified it and I think it's right, why isn't AoC taking it?

5:48 AM

not needed

@RedwolfPrograms what's ur input

check the examples, see if it's too large or too small

& output

Wait, I bet it's the LCM thing

shouldn't be

Can I see one of y'alls as an example?

I think I made a typo somewhere

5:49 AM

no

unless you mean input

otherwise init wn wp to 0n 1n and wrap wp *= BigInteger(n[0])

aren't they all prime

Rahul Verma

I'm using a while, and I guess it's wrong. I'm checking (req - idx) % val == 0

ohhhhh

oh yeah its probably a javascript integer precision thing lmao

5:50 AM

oh lmao

well

well gg

js has 52 bits precision

@RahulVerma should be a plus and not a minus

so like 17 digits

works fine till 1 quadrillion

5:51 AM

Nope, not int precision

huh that should be enough then

weird

Rahul Verma

@UnrelatedString hm, actually I've written +, that's typo. But, it's not finishing yet

so yeah maybe your input's indices aren't prime

my solns are in my hackmd as usual

no cheating tho

5:51 AM

@RahulVerma huh

hmmmm

my solutions are on github to be dug up at leisure

indices don't matter the bus numbers do. but afaict they're all prime for convenience

if product is odd, the chance they're all prime is veryhigh

@HyperNeutrino is there a way to speed up the algorithm described by your approach?

i've also taken the incredibly lazy approach to boilerplate of just making my file for the next day as a copy of my current file before i start writing the solution itself

5:52 AM

@Lyxal you can use modular inverse to shortcut it

I miscounted the freaking offsets

yeah the numbers and not the indices

basically you start at A and add B until it's congruent to C mod D

i am getting those mixed up so much holy shit

@RedwolfPrograms i feel like maybe wn + n[1] might work idk tho

5:53 AM

so you can just take (C-A), multiply by the modular inverse of B mod D, and that's the number of times you need to add B

wait my rank actually improved today?

I think I messed up my offsets. I messed up the input parsing.

well

i dropped to 34th yesterday cuz i choked rly hard on p1

@RedwolfPrograms maybe. i didn't do that and just did arr.map((element, index)=>element&&[element, index])

Okay it's still wrong

5:57 AM

wn + n[1] also not it?

wait

ignore me

It should be - n[1]

At least for what I'm doing

@HyperNeutrino what's A, B, C and D in this situation?

@Lyxal ~~A = index of line. B = line number. C = current. D = current lcm (product since bus lines are all prime)~~

A is your current working value, B is the product of your previous moduli (the value to add so that it doesn't break your working value), C is the offset of the next bus, D is the next bus index / modulus

yea that

6:02 AM

...you know, I find it funny that with how the global leaderboard works, we have two people on the private leaderboard between two people who are on the global leaderboard, who are not on the global leaderboard themselves

:( im sorry im 80 points short of leaderboard

I validated my answer. it is correct. WHY WON'T AoC take it?

@RedwolfPrograms input pls

and output

wait ascii you're somebody1234 right

so i can validate

yes

6:04 AM

you're beating me locally?

nice :D

probs

along with manish

also

>:( its cuz i started like 7 and 12 mins late on 2 days

oof

smh my head. also on mobiel for one day

6:04 AM

Input: gist.github.com/RedwolfPrograms/…

i didn't do day 1 or 7 on that day

and i started day 2 and 3 like 20 minutes late

@RedwolfPrograms i'm getting something that starts with a 5

Oh, I am...not

same (as unrelated string)

Let me reread the challenge

6:06 AM

i could compare the sizes of our results

@UnrelatedString same

but don't know how to run your code lmao

5xxxxxx5xxxxxx8

Stupid ten minute delay...I have the right answer but need to wait six minutes :(

theres ur answer

6:07 AM

:(

rip

f

honestly it might not cost you even one rank lol

at the pace this is going

Top 150 in part one and probably top 150k in part two :p

never mind it already cost you two ranks

6:08 AM

lel 1720 finished part 2

*10

a hour in. a lot worse than 5k

and another 10

Apparently y'all were right about something and I thought I had it implemented differently so it was reversed and it...wasn't

riiiip

6:10 AM

I had tried it earlier and was like "yeah, they're wrong, it still doesn't work", but my input parsing was still incorrect

the unshift?

what the hell is an unshift by the way

Yeah I don't know how I managed to forget indexes existed

anti pop

wouldn't unshifting just be
shifting in the other direction

6:10 AM

hahah big rip

ah

that is correct

It pushes into the beginning of the array

so... push

er yea

6:11 AM

Push is the end of the array

but to beginning explicitly

its front push my bad. shift is front pop

btw @HyperNeutrino so... it is crt

Only one more minute to wait

good luck

let's hope it's not wrong

I checked it against your result with my input, and it matched. I don't see how it could be wrong :p

Unless you're all conspiring to massively troll me, which probably isn't it

6:13 AM

yeah it is CRT i'm just having some difficulty using my brain cells rn lol

Yay, it worked!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

let's go

Took me over an hour but still better than day 7 :p

:p

Day       Time  Rank  Score       Time  Rank  Score
 13   00:04:25   104      0   01:13:35  1909      0

So close to top 100 in p1

6:14 AM

what was day 7

hey i was close to top 100 too

The bags one

oh that problem was kinda annoying

i can't believe i've still only hit top 100 once this year

especially when i kept typing clear gold as the root instead of shiny gold...

rip

6:15 AM

i think today was the closest i've gotten since i did actually get top 100 on day 2 lol

@HyperNeutrino have i understood things correctly?

recursive invmod monkaw

i didnt do to shabby on day 7 surprisingly

that looks right

@HyperNeutrino i found the function here

6:18 AM

lmao, nice

@HyperNeutrino well it isn't.

oh :(

it outputs 6176089 for the sample

haven't run / debugged it yet

i can take a look

are invmod args in the right order

6:20 AM

yes

yea ignore me

i feel like the bit thats being added dont look right. (jk)

why is it outputting 7 for the input

am i running it wrong

also ew lyxal don't for in len range

if for in array can do the trick

I updated the gist: gist.github.com/Lyxal/7d419c2c5a95f2c26a7a0f2f09ca2337

@ASCII-only as i said yesterday: once an hour passes, I don't care about elegance i only care about getting it working

wait what are these values

6:29 AM

at this point i honesly dont know. what exactly is the offset of the next bus?

the offset is the index, the modulus is the bus ID

okay so i fixed your solution by changing 0 to 1 in your outer for loop

because the invmod shortcut breaks when the existing value is already valid, i suppose

also the resulting value is -1x what the value should be (it's positive, but you can just do -n % m)

idk why or how to fix it smartly, but just do brute force lol

it's O(N max M) times the "smart" modinv sol, which is tiny

methinks i better try this again tomorrow

because right now, i have no clue what tf Im doing

Q: Iterate over the neighborhood of a string

fair enough

lol

Anush

12:09 PM

hello!

New Main Posts

12:20 PM

0

Input A string S of length between 2 and 20. The only letters in the string will be a or b. Output All strings within Levenshtein distance 2 of S. You must output all the strings without duplicates but in any order you like. Example If S = aaa then the output would be (in any order) a aa ab ba aa...

code-golf

2:05 PM

@xash Is it possible to make a ' character in convey? I can't seem to put it in a string for obvious reasons, and doing math with characters seems to convert them permanently to numbers.

2:41 PM

started way too late lol

I started on time, but then, after struggling for half an hour, gave up and went to bed. After waking up, I was able to finish in 5 miniutes

I'm starting now lmao

just finished p1

P2 is much harder

I don't even remember what p1 was about, lol

Which bus to actually ride

It makes much more sense than part 2, which is a random math challenge devoid of any real connection to the story

2:52 PM

agreed

3:08 PM

@pppery hm yeah very large number

My answer was around 210 trillion

holy moly

maybe I should start at a huge number

3:23 PM

@xash Also .s seem to be very broken.

3:39 PM

@pppery ok now I'm starting to understand why you said this

@WheatWizard '…\'…' escapes now correctly. You mean the whole f. thing?

Yeah, sometimes . is applied to the character after it, or above it.

Maybe I just don't understand.

hello everyone

@xash For example this program https://λ.land/convey/run.html#eyJjIjoiPj4+dlxuXiAgdlxuLC4tLFxuWyAwWyIsInYiOjEsImkiOiIifQ==

I took the easy route when implementing it: first all functions are parsed, with . being a 1-1 function. Then, whenever a function has a . in its output, it is applied to it. So {f.} and }.f{ are the same.

3:48 PM

Ok, that is very confusing.

It is kinda helpful for this case: https://λ.land/convey/run.html#eyJjIjoiICAgfVxuey8rP1xuIHYuXlxuID4sXiIsInYiOjEsImkiOiIxIDEzIDYgOCJ9

For it to be so simple, +. would have then be one function taking two tiles, so that ? still sees the left as empty … maybe there is a cleaner solution.

Another downside to its non-intuitiveness is that f and f. always need to have the same number of in- and output ports.

Oh and I think & is still broken

https://λ.land/convey/run.html#eyJjIjoiez5cInYgICAxXG4gIDA9diY+QCRgdlxudjw8PDxeICs/LDxcbj4+Pj52XiBcIjBeXG4gICAgPl4gfSIsInYiOjEsImkiOiIxIDIgMSAyIDEgMiAxIDEgMSAxIDIgMSAyIDEgMiAxIDEgMSAyIDEgMiJ9

If you remove the queue from this it changes the output. It looks like some things are getting lost in the queue.

4:13 PM

@WheatWizard I missed a ! in my fix yesterday. Sorry for that. Works now … uh, better at least.