The Nineteenth Byte

Seggan

00:11

why on earth did OEIS decide a random combination of pi and e is so important

It's the sum of all the combinations of addition, multiplication, power and square root product, super obvious and natural to include

Sandbox Posts

0

A: Sandbox for Proposed Challenges

Longest alternating subsequence self-scoring inspired by Longest Increasing Substring Task Given a list of non-negative integers, output the length of its longest alternating subsequence. An alternating sequence is a sequence of numbers which alternates between strictly increasing and strictly de...

Jo King

00:59

CMQ: Does the phrase "N digits of pi" count the leading 3? e.g. is the first 10 digits of pi 3.141592653 or 3.1415926535?

OldSandboxPosts

Sandbox posts last active a week ago: Quarter Precision Floating Point Numbers code-golf, Generate a different sudoku

Radvylf Programs

01:20

@JoKing That's a good question...if you google "100 digits of pi" the first result seems to not include the 3 but I can't be bothered to check them all

I'd say the 3 should count, though

Jo King

→ 1 message yeeted to The Nineteenth Bakery

att

01:43

personally I just never use that wording if it actually matters which

caird coinheringaahing

@JoKing Neither, 100 digits of pi should return a list of 100 randomly chosen integers from 0 to 9

emanresu A

CMC: Post any 100 consecutive digits of pi. You must show that those digits appear in pi. No builtins

(actually I might make that a challenge on main, I call dibs)

thejonymyster

post?

@emanresuA lol omw to find 100 consecutive 0s

emanresu A

glhf

caird coinheringaahing

@emanresuA print('1'*100). NB, assumes pi is normal

emanresu A

01:48

NO.

thejonymyster

whats nb lol

your code has no gender <3_<3

caird coinheringaahing

Note

thejonymyster

wheres the b come from :thinkingemojidiscord:

caird coinheringaahing

It's latin, Nota bene

@emanresuA Prove otherwise, I dare you :P

emanresu A

No assuming pi is normal. And good luck proving it

lyxal

01:50

@cairdcoinheringaahing huh, I always thought it meant "note being"

caird coinheringaahing

@emanresuA Current evidence suggests it is, that's good enough, right? :P

emanresu A

:/

caird coinheringaahing

@emanresuA I would not be surprised if someone immediately trivialises this by finding some super easy to generate pattern

OEIS, wikipedia

emanresu A

Yeah but gl finding 100 9s in a row

caird coinheringaahing

Husk has a builtin for "the digits of pi", it's "trivial" for that to find it :P

Radvylf Programs

01:57

@emanresuA How about "you must provide the index within pi that those digits appear at"

emanresu A

Yeah was gonna add that

sandboxing

caird coinheringaahing

@emanresuA Potentially: given n, output a sequence of n digits. The sequence must appear in pi, and you must provide the starting index for the first 10 possible inputs

thejonymyster

that misses the point

caird coinheringaahing

The problem is that that may just result in people using pi rather than anything else

thejonymyster

taking input forces just grabbing n digits of pi

emanresu A

02:00

0

A: Sandbox for Proposed Challenges

Print 100 digits of π Your challenge is to print any 100 consecutive digits of π. You must give the index at which that subsequence appears, and you may not assume π is normal. You may not use built-ins that calculate π or its digits. The 3 is not included. For example, you could print any of the...

caird coinheringaahing

@emanresuA What's the winning criteria?

emanresu A

code golf

caird coinheringaahing

I dislike code golf in this situation

thejonymyster

the entire point of the question is that its like a weird kc challenge

if you have something else in mind, you should go for it, because its definitely a very different challenge

caird coinheringaahing

The challenge is to find a sequence that is super easy to compress, CG feels like it hands the competition to golfing languages more so than normal

thejonymyster

02:02

ah

but if you dont have it be code golf, what point would there be in compressing it? /gen

caird coinheringaahing

Like, if I find a super easy to compress section, and post it in Python, someone can just copy that into Jelly with no extra effort and be 3x shorter. Even in regular code golf, you have to show effort in using the golfing language, rather than just "It's Python, but shorter cause Jelly"

thejonymyster

languages compete with themselves silly

lyxal

But the point is that once someone finds an optimal section, every other answer is just going to port it

thejonymyster

hm

lyxal

Which takes away the ability to come up with multiple approaches independently

thejonymyster

02:05

depends on how much the algo costs, right?

lyxal

Algorithm? It's a KC challenge - it's just printed fixed text

caird coinheringaahing

Exactly, it's a classic enklact challenge

thejonymyster

@lyxal algo to get the things being printed? what did you mean by "optimal section" if not "section that can be algorithmically generated with the least code"

caird coinheringaahing

Someone will find a super easy to compress section, post that, and every other answer will port it cause any other approach will be longer

thejonymyster

easy to compress in what way?

caird coinheringaahing

02:07

Hell, there are current algorithms for generating the digits of pi, you could feasibly run one of those until it finds a section easy to compress

@thejonymyster All the same digit, the digits 0123456789 10 times, anything with a low Kolmogorov complexity

thejonymyster

the highest known repdigits at like, less than 10 digits long

good luck finding those goldilocks sequences in like, this decade

its 100 digits

caird coinheringaahing

It's similar to a challenge being "Output a 100 character excerpt from Shakespeare". Once someone finds something easy/short to output, everyone will copy it

lyxal

^

A better idea for the challenge would be to come up with some sort of compression score and make it a code-challenge

caird coinheringaahing

^

lyxal

E.g (and a very rudimentary eg): how many digits the 100 digit sequence is in base 255

caird coinheringaahing

02:10

@thejonymyster The challenge itself isn't flawed - the scoring criteria is. It lends no effort to doing anything more than waiting for someone else to find a "good" section

You aren't doing anything besides copying someone else's method, because they've shown that basically everything else is terrible compared to what they found

(obviously, not "you, you", more "you" as in the general golfer)

thejonymyster

lol ik hm

would making it dramatically more digits change that :P

caird coinheringaahing

No

If anything, fewer digits would

lyxal

Yeah, because then you could leverage more language-specific constants to obtain the digits of pi

caird coinheringaahing

Plus, e.g. 20 digits is less likely to be "all the same digit", but also more likely to lead to patterns that are easily reachable, and can compress decently well

→ Begone previously posted Sandbox post

Bubbler

03:49

The world record for calculating pi is 10^14 digits

Wheat Wizard

@cairdcoinheringaahing This seems more interesting though.

Bubbler

and somehow there's a Pi API, so you might be able to DoS it to get some interesting sequences :P

oh, there's actually a collection of downloadable files

emanresu A

@cairdcoinheringaahing Different sequences will be more optimal for different langs though. And I think it'll be quite hard to find an "optimal" sequence.

DLosc

04:11

@RadvylfPrograms CMC: Narayana's cows sequence

Pip, 13 bytes, outputs infinitely: W1o::i::y+:Po

Alternate 13 bytes, outputs the nth number: Lao::i::y+:oo

emanresu A

... triple swap???

Vyxal, 7 bytes

generators are nice

DLosc

@emanresuA Indeed! (Well, there are two swaps, but three variables involved.)

- Print o
- y gets y + o
- Swap i and (new value of) y
- Swap o and (new value of) i

emanresu A

hmm

DLosc

which means y gets i, i gets o, and o gets y + o

emanresu A

04:26

My one's just "given [1, 1, 1] (111) as the initial vector, and a triadic lambda that adds its first and third arguments, generate an infinite list

DLosc

That's five builtins in funstack.hs: 1 Plus Const compose iterate

emanresu A

How are the golflangising plans for that going?

Also what's Const? Is it a dyad that returns its left/right arg?

DLosc

@emanresuA Yes, left arg

@emanresuA I'm currently almost done with a rewrite of how functions are implemented (well, done with the function implementation, just porting a few more builtins to the new system). After that, I'm not sure whether to push forward with adding a SBCS mode, adding more builtins first, or rearranging some stuff with the data types.

As it stands, it's very powerful for certain integer/list challenges, but is still lacking a lot of important builtins and has very little in the way of string functions.

emanresu A

Do you even need a sbcs? An encoding method would probably be sufficient

DLosc

@emanresuA Yeah, true. That's kinda what I meant. It's just easier to think of it as being 5 bytes if you can see the code represented as 5 characters.

I also want to make pretty pictures to explain all the modifiers, like BQN has. :)

> Some people consider it reasonable to name combinators after types of birds.... There is something wrong with these people.

Lol

lyxal

04:39

too right

couldn't agree more

DLosc

04:56

@emanresuA 6 bytes?

emanresu A

That... works? TIL

DLosc

@emanresuA I was confused about how ⁺ worked, so I thought your original code was starting with 11. I tried substituting 11 and it still worked. So then I took the next logical step to 1. :P

emanresu A

⁺ gets the next char and adds 101 to its codepage index

⁺<nl> is 111

DLosc

@emanresuA @lyxal The tooltip on the Vyxal keyboard doesn't mention anything about adding 101. Might be a good thing to clarify.

lyxal

Just gotta wait for approvals on the PR :p

PyGamer0

08:58

feedback?

mousetail

09:39

Are there 183 vyxal answers on the site? Seems low

emanresu A

There should be about 1.2k

mousetail

That makes more sense

I'll need to check my script again

getting 508 now when I include posts that have vyxal, with a comma at the end

AJFaraday

Could I ask for a quick judgement call?

I've got a code challenge on the go. Starting with this and I've got a few ideas for later iterations... The big question is, is this first run too simple a problem to solve?

https://ajfaraday.github.io/the_fog_game/

mousetail

@AJFaraday what is the goal of the game?

AJFaraday

To explore all the spaces with the lowest cost.

Bubbler

09:52

The costs of E and N seem to be reversed

AJFaraday

The challenge would be writing a client class to play the game.

Good point, Bubbler. I recently switched those round because enhance sensor seems so much more valuable. Seems I made a mistake in the process

mousetail

The game is always the same though right?

There is a single optimal solution, not really a optimal "algorithm"

AJFaraday

So that's my first question. I think the second iteration would be running it through a few different board sizes and shapes. Different starting positions.

So is it worth doing it with a fixed board as a first instance, or is that too simple?

(I'm thinking that would allow some feedback on the API, game dynamics etc)

Adám

58?

mousetail

58 seems to be the best score if you enhance the sensors all the way

AJFaraday

09:58

Hmmm, I got 59

mousetail

59 if you turn right first

AJFaraday

Oh, I see, I went east/west

mousetail

I did the same thing first time lol

AJFaraday

Yeah, it's definitely going to get to a point of reacting to changing circumstances.

Adám

Enhancing sensors is way overpowered.

mousetail

10:00

maybe if the sensors created more a circle rather than a cross it would make the no-sensor strategy more viable, since you don't leave as many 1 cell gaps

AJFaraday

There is that, which is why I knocked the cost up to 10 (tho apparently, not acutally)

Adám

how about that adding a layer on a triangle?

AJFaraday

How do you mean?

mousetail

Or maybe successive sensor costs should get more expensive

5 10 20 etc.

Adám

X
XX
XXc
XX
X

AJFaraday

10:01

Interesting thought. Would certainly make movements more important

Adám

And turning.

three turns would clear the square.

AJFaraday

It could almost be entirely sensing forward in a line...

But you're right that it would make more sense.

I'm a little concerned that it would explore a lot more with each enhancement, 'tho that would go well with the increasing costs.

Really helpful ideas, thanks!

There's still the question of whether or not to do a one-board challenge... I suspect with the sensor tweaks it would still be worth doing

The end-game, you might have guessed, would be to make it a KotH challenge. Replace exploration with claiming land and score based on who has the most land when the challenge ends...

mousetail

10:38

KOTH would be interesting

Very similar to this though codegolf.meta.stackexchange.com/questions/2140/…

DialFrost

Hiya

lyxal

o/

DialFrost

Gah exams tomorrow morning :(

lyxal

You ready for them?

DialFrost

11:06

Uh, not so much :P

Btw @lyxal once you changed pfp can you lend me urs? ;3

lyxal

Okay Dialxrost :p

3

Ginger

G'Yonkus

Parcly Taxel

good evening everpony

DialFrost

@Ginger ?

Ginger

in Off-Topic TNB, 3 mins ago, by lyxal

Gaming

DialFrost

11:16

Hehe

@lyxal and of course, it had to be a literature paper

mousetail

Ginger Gaming

DialFrost

11:31

Mouse Gaming :3

mousetail

I'm terrible at gaming though, my reaction speed is measured in minutes

DialFrost

11:42

mine in hours

mousetail

11:56

Yet you reacted within one hour of me

DialFrost

Hehe, well good night guys! Gtg sleep early and study for tmr :3 Bai

Ginger

12:14

@mousetail epic gamer moment

thejonymyster

TIL theres no closed form for number of partitions

idk why but i thought that it would have been easy :P

mousetail

@thejonymyster Yes there is, n!

thejonymyster

typo

i guess its like, i thought itd be just finding all the places to insert a split between N elements, which is doable closed form, but then you have to get rid of all the equivalent ones, which makes sense that it isnt easy

Ginger

question: how many bits of data can fit in 4 bits?

this sounds stupid, but is there a way for me to magically compress more bits into that space?

I doubt it

mousetail

@Ginger No, it's a famous unsolvable problem

If you could do that you could compress data infinitly

thejonymyster

12:28

calling it a "famous unsolvable problem" makes it sound like its not trivial :P

Seggan

@emanresuA Fig, Or10. appears at position 17,387,594,880

Codidact Posts

New Posts - Challenges - Code Golf: 8 coexisting queens

posted on September 29, 2022 by trichoplax‭

This is a fixed output challenge. Output a textual representation of a chessboard hosting 8 queens, none of which are attacking each other. There are 92 ways of arranging them, 12 if rot...

Ginger

wow, an actual codidact post

crazy shid, that

Taylor Alex Raine

12:47

Hey - quick question - are we okay with users adding tags to old questions that invalidate existing answers and then deleting those pre-existing answers

I had two 5 year old answers to this question removed after a user added the ascii-art tag

Parcly Taxel

When is a built-in disallowed for a code golf question?

Only when it explicitly says so?

Ginger

@TaylorAlexRaine I don't think this is illegal but it's certainly a bit of a dick move

the user should've left a comment instead of just flat-out deleting the answers

but I'm not an expert on site policy so don't take that as gospel

Taylor Alex Raine

like, I spent legitimately like 10 hours on these questions so I am kinda upset about it, yeah

Parcly Taxel

Um... anyone?

Taylor Alex Raine

@ParclyTaxel codegolf.meta.stackexchange.com/questions/11441/…

meta post about the topic - it's complicated

thejonymyster

12:52

@TaylorAlexRaine what, was your answer graphical output?

Taylor Alex Raine

no - excel VBA that outputs to a worksheet

implemented all the borders, differnces of the borders with color and all other reqs

Parcly Taxel

For example, codegolf.stackexchange.com/questions/252302/…

thejonymyster

@TaylorAlexRaine i dont see how thatd be invalid then

can you like, flag for undeletion?

Parcly Taxel

PARI/GP's polrootsreal completely whacks the question

Taylor Alex Raine

@thejonymyster - I can, just wanted to know if there was a policy changes or anything about that

Ginger

12:55

I'd advise you to do the flag then

thejonymyster

@ParclyTaxel thats fine, what i usually see is people posting both the builtin along with a non builtin solution, but unless theres a specific reason not to use a builtin theres no like, sitewide ban

Parcly Taxel

hmm...

thejonymyster

@TaylorAlexRaine there isnt a way for a tag to invalidate an answer, not sure why your answer got deleted

ascii art tag doesnt even have any default rules in it, and default tag rules typically reflect expected allowances rather than new restrictions (iirc)

lyxal

thejonymyster

yes, but the question itself already specifies that the output is that text and not a thing which looks like that text

but also according to taylor the text is already output as it should be? unless i misread

lyxal

13:00

that's what the output looks like

thejonymyster

oh yea then that wouldnt be valid regardless of the tag unfortunately

it is nice looking but definitely not the text

Wheat Wizard

You shouldn't loose any rep from the deletion fwiw.

thejonymyster

"Your challenge is to print/output/return this text" is the reason its invalid, not the ascii-art tag. probably since the post got boosted by the tag edit, it got more attention and a mod looked in. the person who added the tag was not jo king

if you end up re-attempting the challenge, id leave the old invalid solution in to show it off too

Wheat Wizard

Or the challenge could have been boosted by something else and both the flagging and the retag happened separately.

thejonymyster

ah yeah that too

mousetail

13:17

Huffman coding for Vyxal: https://gist.github.com/mousetail/001c139871c7e91c8c8181b6339a804a

Preliminary research makes me think there are defiantly gains to be made this way but the sample size of only 505 answers is still way to small.

thejonymyster

that image is hugenormous

very cool though

mousetail

Yea that's the issue with binary trees, the width grows exponentially

lyxal

did you try getting the data from SEDE

mousetail

I used the API

I'll try your query

thejonymyster

how do you use the encoding?

mousetail

13:22

I did not, I used UTF-8 and just filtered out to use only characters that appear in the code page

thejonymyster

no like... how would one use / implement the encoding?

mousetail

So if the first bit is 0, you take the left branch of the tree, if it is right, take the right branch

if you reach a character decode to that character and start from the top

if you reacht the special END key you stop parsing

(this is needed because you often won't end perfectly at a byte boundary)

thejonymyster

ahh gotcha

so theres no "N01" anywhere because that takes you to ⇧

mousetail

Exactly

thejonymyster

ok i understand it now hooray ty

what do the (seemingly) decimal numbers mean

like the (2.1) near ⇧

mousetail

13:27

After the letters? That's the entropy of the letter

it should be as close as possible to the number of bits needed to represent that letter

thejonymyster

but doesnt that always increase in a system

mousetail

basically log2(frequency_of_letter)

@thejonymyster it should always increase top to bottom

thejonymyster

lol

@mousetail wait wouldnt you want it to be inverted somehow?

mousetail

sorry -log2

thejonymyster

til -log

will mathematic notation ever cease to be absurd and self contradictory

mousetail

13:31

It's just a negative sign

in other words log(1/frequency)

thejonymyster

right but. negative sign prefixed on an expression typically represents...

well........... a negation of that value :-)

mousetail

which in this case is exactly what is happening

thejonymyster

so you want a negative number of bits

or wait

*remembering how log works*

mousetail

logs are weired

thinking about entropy: not even once

thejonymyster

and i suppose frequency is between 0 and 1?

mousetail

13:33

yes

thejonymyster

i missed that part somehow, all makes sense now

i was thinking it was like a sin² situation

mousetail

It makes sense right? If there was a character with frequency 1 you would need 0 bits to represent it since it was the only option

thejonymyster

no yea it does now lol, i was thinking of frequency as like. a value that could just keep going up, for some reason

mousetail

Anyway, what should I name my encoding? I'm thinking μxial maybe

thejonymyster

whyxal

mousetail

13:37

Hard requirement is to include a latin or greek letter

thejonymyster

ah damn i was gonna say hyxal

for Huffman vYXAL

@mousetail this looks good but im wondering what the meaning behind it is

mousetail

The mu is often used to represent enthropy and is also a uppside down h for huffman

thejonymyster

is the "ial" a typo

mousetail

I guess

More like never having known how the language is written

lyxal

νγχαι?

thejonymyster

13:40

vyxial....

lyxal

@mousetail vyμ

thejonymyster

nice, +1

lyxal

it has the mu for enthropy (entropy?) and fits with the names of things like vylight (our syntax highlighter) and vyxapedia (our tutorial site)

thejonymyster

now you just gotta be careful so that people dont pronounce it vim

mousetail

How would you even pronounce

vymu

thejonymyster

13:44

doesnt seem that hard

mousetail

depends if you pronounce the y like a vowel or consonant I guess

lyxal

maybe vyι then, pronounced vyota

because the iota is the smallest letter

and the vyι is the smallest (shortest) encoding

thejonymyster

@mousetail i mean.... yea. thats like saying "type" might be hard to pronounce, depending on if you pronounce the y like a vowel or a consonant

@lyxal this is more fun though :P

lyxal

@mousetail what do you plan to do with vyμ (or vyι depending on what name you choose)?

as in, how do you plan to use it/do you intend for it to be used?

Seggan

ooh huffman coding?

still dont get how it works

wouldve loved to make a language using it

lyxal

13:58

basically, it uses binary trees to compress text based on the most frequent characters of an alphabet

and each letter is represented by a unique and unambiguously identifiable string of bits

Seggan

yeah i know that but im trying to figure out how it works, not what it is

i think im getting it using the diagram

thejonymyster

@mousetail @seggan this helped me a lot

Seggan

oooooooooooooooooooooooooooooooooh

now i get it

so 0111111 is ⇧o?

lyxal

hey wait a minute, isn't it supposed to be that more frequently used characters are at the bottom of the tree?

thejonymyster

@Seggan looks like it to me

@lyxal no

lyxal

14:03

sorry, I meant the least frequently

thejonymyster

ok, then yes :P

lyxal

then what the hell is ⇧ doing at the top

our data shows ; is the most frequent character

⇧ is beaten by at least 20 other characters

thejonymyster

48 mins ago, by mousetail

Huffman coding for Vyxal: https://gist.github.com/mousetail/001c139871c7e91c8c8181b6339a804a

Preliminary research makes me think there are defiantly gains to be made this way but the sample size of only 505 answers is still way to small.

lyxal

no but even from 505 answers, ⇧ still shouldn't be at the top

something like , or $ or ; or * or ƛ should be at the top

hey wait @thejonymyster I meant to ask, how's the hurricanes treating you?

mathcat

Something's definitely wrong here

An 8 by 8 "QR code" with 0-error correction and zero redundant bits can store 2**64 bits, right?

lyxal

14:18

in theory, I guess so

2 bits for every cell in the 8*8 grid

wouldn't it be 2*64 then?

mathcat

oh oops

anyway

nobody saw my message

nothing happened, it was just the wind

lyxal

sorry, did you say something?

I felt like you did, but I don't see any messages

mathcat

ikr, weird

lyxal

anyhow

oh hey they added follower counts for organisations on github

cool

ah only organisations you own/are a member of

still, cool

thejonymyster

@lyxal not so bad- being hurled through the skies by gale force winds isnt so bad when youve got a gaming chair

(didnt hit where im at lol but ty)

Seggan

14:25

@thejonymyster this is what ive always been looking for

Fig 2, here I come!

lyxal

We here at Vyxal do not have mannequins

Seggan

whaaat is that

lyxal

idk but it's funny

some sort of github thing

Parcly Taxel

14:47

codegolf.stackexchange.com/a/252339/110698 finally...

I had to make a non-builtin answer

PyGamer0

6 hours ago, by PyGamer0

feedback?

New Posts

15:03

1

Q: Count Futoshiki row solutions

Futoshiki is a logic puzzle where an \$n×n\$ Latin square using numbers from \$1\$ to \$n\$ must be completed based on given numbers and inequalities between adjacent cells. In a solved Futoshiki puzzle any row (or column) forms a linear extension of the poset induced by that row's inequalities....

mousetail

15:15

@lyxal Prove I'm better at making languages than lxyal despite literally ripping his language of with very minor modifications

Seggan

ive already managed to do that with myxal :P

Ginger

there are a lot of people in this room right now

3 rows on my screen

Seggan

2 on mine

mathcat

7 on mine lol

Ginger

!mobile user!

either that or you're using LG's rollable TV in the mode where only the top sticks out AND you've rotated the display 90°

mathcat

15:27

nah I'm using a laptop

oh wait rows not columns

Ginger

with a 480p wide screen??? :p

mousetail

15:57

The vyxal clone will not have a single huffman tree btw but one for each character, since the distribution can change depening on what characters appeared earlier in the string

Though it will be hard to find enough data so there will be some guestimating involved

overfitting protection

Seggan

16:14

thats cool

but how will you determine how to encode the first char?

@mousetail also how did you create this?

mousetail

@Seggan There is a separate table for the first character

@Seggan I scanned all vyxal answers on here, computed the frequency of every character, then repeatedly combined the two trees with the lowest frequency until only 1 tree remained

Seggan

i mean the graphic

mousetail

Using graphviz dot

it's installed by default on many linux distrubtions, you encode your graph in some text like a --> b and it turns it into a svg

Seggan

noice

im thinking of building mine mostly manually

mousetail

You could probably do a lot better manually

The corpus simply isn't big enough for a good data

mathcat

16:53

CMQ: If you could store a program in a QR code, and the program will directly be executed on the scanner's device, what would you program?

mousetail

Install a backdoor on their systems to install a bigger program on a later date

DLosc

@mathcat I mean, there's a pretty obvious answer. :P

3

mathcat

okay true

Transcript for