The Nineteenth Byte - 2021-08-24 (page 1 of 5)

« first day (3860 days earlier) ← previous day next day → last day (987 days later) »

12:01 AM

0

Q: "-rot" transform

Background -rot transform (read as "minus-rot transform") is a sequence transformation I just invented. This transform is done by viewing the sequence as a stack in Forth or Factor (first term on the top) and repeatedly applying -rot +. -rot is a command that moves the top of the stack to two pla...

code-golf sequence array-manipulation integer

Dude coinheringaahing

@NewPosts @Bubbler How short is your crazy APL/J/Jelly solution?

Just so I have a point of reference to get double :P

@Dudecoinheringaahing My first working solution was 36 bytes in J

Dude coinheringaahing

Ok, so there's not some crazy array-lang trick involved? :P

Not an arraylang trick per se, though I used a J-specific builtin

I think Jelly has the same thing

Can we take the input in reverse? — Dude coinheringaahing 9 mins ago

12:14 AM

@Dudecoinheringaahing No. — Bubbler 1 min ago

ah well. "yes" would have shaved a byte off my answer lol

Dude coinheringaahing

@Bubbler Your challenges have a tendency to make me say "blegh" about my Jelly answers to them :P

:P

Dude coinheringaahing

I'm sure there's a more elegant way, and I was so happy about the ḣ2S;ɗḢo, before it got ruined :/

btw, the J builtin I abused is this

12:17 AM

@Dudecoinheringaahing Imagine using letters

Dude coinheringaahing

@Bubbler œ? in Jelly, I believe

Where should I post if I find a minor typo in a SO blog post?

Dude coinheringaahing

Ping Catija if it's a minor one (here or the Tavern should be fine), open a bug report on MSE if its more than a couple of mistyped words

there's an interesting pattern there but I can't see any nice way of getting it with some sort of scan

[a, b + c]
[a, b + c, a + d]
[a, b + c, a + d, b + c + e]
[a, b + c, a + d, b + c + e, a + d + f]
[a, b + c, a + d, b + c + e, a + d + f, b + c + e + g]
[a, b + c, a + d, b + c + e, a + d + f, b + c + e + g, a + d + f + h]

Gotta visit the tavern I guess

Dude coinheringaahing

12:27 AM

Tavern on the Meta

MY GOD, IT'S FULL OF STARS AND BACKLOG! General friendly chit-...

does jelly have 'uninterleave'?

vyxal does, but there's already a very short vyxal solution

Dude coinheringaahing

Œœ

@rak1507 but you can't really compare it to Jelly, because it uses the fact that it's already operating on a stack

yes

tio.run/##y0rNyan8///opKOT//@PNtQx0jHWMdExjQUA sorry, should this not do something?

Dude coinheringaahing

Use arguments not input

12:32 AM

oh, it should be an arg

yes

out of practice

@Dudecoinheringaahing is using a for-loop len(input) - 1 times any shorter somehow?

Dude coinheringaahing

Not really

Jelly's always-on implicit print hurts it here

shortest way to do [1, 2, 3, 4, 5] -> [1, 3, 4, 5]?

searched for 'pop' (didn't find anything)

Dude coinheringaahing

Jelly, 4 bytes Ḣ;ḊƊ

really

Dude coinheringaahing

12:36 AM

The Ɗ is to keep it as a single link, isn't always required

ok, whatever

Dude coinheringaahing

@rak1507 How would Vyxal do it?

no idea, I don't know vyxal

Dude coinheringaahing

Don't you? Huh, I've lost track of who does and doesn't :P

You're probably looking for "swap the first two" instead?

12:37 AM

yeah that'd be better

thanks :P

no idea how to do that either, @Dudecoinheringaahing ?

@rak1507 is that "remove second item"?

indeed it is

Dude coinheringaahing

@rak1507 ḣ2Uo, assuming no elements are 0

can that be assumed? @Bubbler

Nope

12:39 AM

darn

Dude coinheringaahing

ḣ2Uḷ" if there can be zeros

how on earth does Ḣ;Ḋ work btw

Dude coinheringaahing

@Bubbler Ugh

oh, modifies the list

@Dudecoinheringaahing :ẏ1oİ

duplicate, range(0, len(top)), remove 1, vectorised index

actually, ḣḢJ works

12:43 AM

tio.run/##y0rNyan8///hjsVGoQ93bFc6tPXopKOTD7cc3aPvdmgrXPj////…

some potential maybe?

@rak1507 yor Nyan...

don't encourage the cat people.

tio.run/##ATUAyv9qZWxsef//4bijMlXhuLciCsOHxZLFk8OExbwvRsOH////…

14 bytes

Dude coinheringaahing

13 bytes

btw, please can I post this as my own answer with credit? it's sufficiently different imo

Dude coinheringaahing

Yeah, of course :P

12:44 AM

thanks

ah, Z rather than z/

I don't think it's correct

ḣ2Uḷ" annoys me

@Bubbler oh no

@Bubbler you're right

Dude coinheringaahing

Doesn't work

oh right, just have to replace the last C with an H

tio.run/##ASgA1/9qZWxsef//4bijMlXhuLciCsOHxZLFk8OEWkbhuIr///…

should work

Dude coinheringaahing

12:47 AM

looks like it

@rak1507 *replaces with D instead*

ḣ2Uḷ"µŒœÄZFḊ

@exedraj yeah, forgot I wanted tail not head

actually ends up shorter because you don't have to do the rotation twice

Looks good

Dude coinheringaahing

Feels like there's some way of shortening ḣ2Uḷ"µ

probably

I'll get writing the explanation

Dude coinheringaahing

12:50 AM

11 bytes

ooh nice

ok golf langs can be quite cool, even if you have the idea it can be hard and interesting golfing it

Dude coinheringaahing

:P

anything unclear in the explanation?

Dude coinheringaahing

05AB1E should win this, but it's being held back by design issues ಠ_ಠ

@rak1507 I'd typically do both heads first, then the ṭ below in the explanation, but that's definitely a point of preference :P

rip

ok, that makes sense

done

1:04 AM

@Dudecoinheringaahing how many bytes do you have in 05AB1E?

In general I think I write 1 then 2 when I have a 2-1 combo (or just combine both in one line)

CMC: Give your cat control of your keyboard.

@emanresuA can't.

catn't

I don't have a cat so that may prove slightly difficult

Dude coinheringaahing

Ideal 05AB1E would be splat; swap; wrap; uninterleave; cumsums; interleave; remove_head, where each of those is 1 byte (for a total of 7), but .¥ (cumsum) doesn't vectorise, and, because the stack is the wrong way around, it needs `rsr) instead of `s) ಠ_ಠ

1:05 AM

@Dudecoinheringaahing I have an 8 byter in 05AB1E

@emanresuA wstgyk;io

Dude coinheringaahing

Plus, .¥ prepends 0 when it does cumsum ಠ_ಠ

It was my imaginary cat in my head

Dude coinheringaahing

@exedraj Using your Vyxal method?

correct

1:06 AM

HaLpI 'M TrAped inSiDe ThE BoDy Of A CaT ! /??

@Dudecoinheringaahing if length of stack was 1 byte, I could get a 7 lol

Dude coinheringaahing

Yeah, I was expecting that. What annoys me is that 05AB1E should have the perfect builtins and stack model to make this a walk in the park, but it just misses the mark

Why did adnan make length of stack .g?

yea, so close but so far

Dude coinheringaahing

@exedraj Because length is g

1:08 AM

@exedraj not as if getting the length of something is an extremely common operation or anything, nahhhh

Well, length of stack is rarely needed for a high-level language :P

@rak1507 but why does it have to be a digraph

I was being sarcastic

yeah I could tell

Dude coinheringaahing

That's one thing I like about Jelly vs stack languages is that you never have to balance builtins that operate on either the stack or the top element

1:09 AM

@exedraj oh, well, yeah I agree it shouldn't be a digraph that's dumb

11 in jelly is pretty decent really, would be nice if swap first two elements was a single byte, not sure how often it comes up but at least a digraph maybe

Dude coinheringaahing

The thing is that g is well-defined for integers (cast to string, length), so it can't do what O does (sum of ToS if list, else sum of stack)

inb4 stax wins with source compression somehow

lol

Dude coinheringaahing

@rak1507 Head rot (rotate the first n elements one step to the left) would be great tbh

yeah

would save on the separator as well right

Dude coinheringaahing

1:11 AM

So that would be œṙ2, and you could get rid of the separator

9 bytes in Jelly for a stack-centered challenge is pretty good :P

and if it was a single byte it could be 8

Next up: deque-centered challenge :P

Dude coinheringaahing

@rak1507 Yeah, if Œs (or whatever) was special cased as œṙ2 it'd match Vyxal, which has the right stack model and the builtin as a builtin

yeah

> the builtin as a builtin

that lowkey confused me at first

Dude coinheringaahing

1:15 AM

Every 60 seconds, a minute passes in Africa :P

> the builtin as ~~a builtin~~ an element

ftfy

@emanresuA ===============]

(No joke, that was actually what my cat's paw just pressed)

:p

Dude coinheringaahing

@DLosc Don't believe you, I reckon you're the cat :P

@Dudecoinheringaahing [shrug] Who knows? Maybe you're the cat. Maybe we're all the cat...

1:20 AM

Lynn Lynn'd it again

ok, how does that work

Dude coinheringaahing

œ? is "Permutation at index x of the items in y (where y defines the sort order of those items)"

wtf

1st permutation is the list unchanged. 0th permutation is prev of it, which is entire list reversed. -1th is prev of that, which is entire list reversed, then last two elems swapped

Dude coinheringaahing

U then reverses that

1:22 AM

Yes

wow, that is insanely high iq

Jelly sus

Dude coinheringaahing

@Bubbler (Was in the middle of typing out what -œ? meant before you ninja'd me :P)

I did think of using œ? but didn't think of generating the reverse permutation

Dude coinheringaahing

Given that Jonathan Allan added œ? and it's 3 variations, yes, very high iq :P

œ¿, œ?, Œ¿ and Œ? :P

My headcanon is that he chose ? to indicate it's confusingness :P

1:24 AM

:P

But yeah, using permutation index is so powerful at times

someone smarter than me feel free to edit in a proper explanation rather than handwaving :P

Dude coinheringaahing

I'd suggest changing -œ? to be "-1th ordered permutation, i.e. second-from-last. This will always be the array reversed with the last 2 elements swapped"

Or you can quote my message verbatim :P

if it helps humanize the events, I came here to say that I thought of U2œ?U first and didn't just look at it and think “oh duh, -œ?U”

yeah yeah whatever 'I am definitely not an all knowing being' mhm sure

Dude coinheringaahing

1:30 AM

That still involves going "How can I swap two elements? Oh, I know! œ?!" :P

@rak1507 oh, the key word is, "This will always be the array reversed with the last 2 elements swapped"

(and then U un-reverses it again)

I'm totally gonna start telling people U stands for un-reverse

lol

Dude coinheringaahing

It's "upend" IIRC :P

@Lynn blame caird for that :P

Dude coinheringaahing

> This will always be the array reversed with the last 2 elements swapped"

1:36 AM

what

I copy pasted it

how did that happen

Dude coinheringaahing

Magic ¯\_(ツ)_/¯

You may have copied it in the 10 seconds before I added in the "reversed" :P

lol, yep

I wanna create this binary matrix and then matrix-multiply it with the input:

[1 0 0 0 0 0 0 0] [a]
[0 1 1 0 0 0 0 0] [b]
[1 0 0 1 0 0 0 0] [c]
[0 1 1 0 1 0 0 0] · [d]
[1 0 0 1 0 1 0 0] [e]
[0 1 1 0 1 0 1 0] [f]
[1 0 0 1 0 1 0 1] [g]

well, that didn't go well, but it's still readable

@Lynn I never expected out-of-range index to work there...

I feel like I recognise that matrix

1:44 AM

The matrix looks like a lower triangular checkerboard matrix with 2nd column repeated

what is that from

oh, right, that solves the problem

-rot transform, obvs

it looks suspiciously similar to something else I was doing a while ago

you could start from [1, 0] and then iterate C,1 (in Jelly terms) to generate the rows

Dude coinheringaahing

@Lynn This is going to be like your Padovan answer, isn't it :P

1:45 AM

[1,0] [0,1,1] [1,0,0,1] [0,1,1,0,1] etc

if you manage to outgolf me I will be extremely ~~annoyed~~ impressed

then dot-product each of those with the input? I hope dot-product has the right behavior when one list is short, fingers crossed

Dude coinheringaahing

@Lynn You're already at 7 bytes at that point tho :/

@Lynn There are 2 dot products, that have different behaviours

ḋ pads with 0, æ. pads with 1

Maybe start with 0,1 and iterate with (Z with some dot)C and do U at the end?

ḋ should work at the very end

probably the worst thing about this approach is that you need to specify the iteration count

Dude coinheringaahing

1:50 AM

@Bubbler That's probably be shorter, as there isn't a digraph for 1,0 (superscript 5, D sometimes works, depending on the chain)

DLosc's answer is making me think this is somewhat Fibonacci-esque, and that ¡ has some crazy short dyadic link

An abomination trying to use base

2:07 AM

14 though, pretty short still

Actually 10

wow nice

lynn is about to golf that down probably :P

Prefixes is OP

so glad my first lesson tomorrow is at 1:45 not 8:30 so I can stay up golfing :P

0

A: "Hello, World!"

Tarfish, 160 bytes x++++++++++++++++++++++++++++++++:+:+++++++++++:++++++++++++++++++++++++++++:.+++++++++++++++:++++++++++++++:.+++++++:::..+++:.{{{.{.{{.}}.:++++++.:.--------..; Tarfish is an esolang of mine that is a more tarpit-style version of ><>. This answer doesn't really use that much o...

2:15 AM

Oh yes, that answer.

It's... verbose.

you don't say

this is 9: J>2C;¥\Ḋḋ tio.run/##y0rNyan8/9/LzsjZ@tDSmIc7uh7u6P7//…

damn

nice one, wouldn't have thought of using a matrix product

I guess I should post it

Go ahead

2:32 AM

codegolf.stackexchange.com/a/233671/3852 why did I type all that it's 4am

nice answer

ohh J>2 that is clever

Dude coinheringaahing

On mobile, so I can't try it, but could you replace >2 with ’ and C with vectorising logical not?

Or would that mess up the first row?

You mean J’¬;¥\Ḋḋ?

Try it online!

if you read the rows backwards, as binary numbers:
1, 110, 1001, 10110, 101001, 1010110, ...
1, 6, 9, 22, 41, 86, 169, ...

a formula generating those numbers is `floor(pow(2,k)/3) xor 3`

but J‘2*:3^3BUḊḋ is pretty long

@Dudecoinheringaahing I need the actual 1s because they are the right arguments to ; in the scan

@Lynn Which also happens to be the output for the fourth test case.

Dude coinheringaahing

2:45 AM

@Lynn :(

CMC: Create a Jelly program that never halts. You may not use any quicks. Shortest wins

Segfaults due to infinite recursion (e.g. no TCO) count as halting. You should hit the 60s timeout on TIO

i think 9!!! works lol

wait, you can do like ⁹!!

Dude coinheringaahing

That will halt tho

@Dudecoinheringaahing I have a very boring 13 byter

oh wait, i misread it as just that it should hit 60s lol

Dude coinheringaahing

@exedraj Eval as Python?

2:55 AM

;p

Dude coinheringaahing

Fair enough, but I bet there's a pure Jelly way :P

v keeps segfaulting on me

I got it down to 10 bytes with string compression

7 bytes kinda using quicks but not in the main program: “1 1¿”V

lol bruh

“ƬÄ¡ṾEı» for 10

no wait

I forgor the eval as python part

well if you can cheat with quicks like that then just do ”ÇV

2:59 AM

lol

Browncat Programs

3:11 AM

I really don't know why my school has five spanish teachers, a sign language teacher, and no french, german, italian, or japanese

Because Texas?

Sounds pretty reasonable for a U.S. state adjacent to Mexico IMO.

Browncat Programs

Seems like there could be slightly more options though :/

True. Maybe change out one or two of the Spanish teachers for something else.

@BrowncatPrograms budget funding issues?

Browncat Programs

I doubt it

3:27 AM

Lack of specialisation?

Browncat Programs

Oh hey, I'm starting to get email spam from actually good colleges for once

Ooh nice

Top Ten Reasons You Should Attend MIT! Number Seven Will Shock You!

Browncat Programs

3:42 AM

I'm aiming to go to a college somewhere around the top 20, since most of the ones higher than that that I've looked at don't accept many AP tests

1 hour later…

4:45 AM

I've found more suffering for everyone

Even me wtf

not funny didn't laugh

the video I linked is actually cursed

@exedraj yes it is

@exedraj well for me it was pretty funny

@exedraj wtf

@hyper-neutrino yes

4:50 AM

i am going to go listen to the original song to attempt to clean my mind of this ungodly eldritch horror

@hyper-neutrino use soulless 4 as mind bleach instead

oh hyper's back in his λ form

Me: Looks at ngn/k's implementation to learn how reshape works.
Also Me: Looks pretty readable: https://codeberg.org/ngn/k/src/branch/master/h.c

it looks liek an esolang; i cant understand anything except the include header

oof there is like 50 macros defined

This is the way of Arthur Whitney.

CMC: Given a list of 9 numbers, reshape it to 3 by 3 matrix. Don't use reshape builtins

Define builtins

5:00 AM

Any builtins?

Jelly: s3, split into chunks of 3

(Jelly does not have reshape in the APL sense)

@PyGamer0 Vyxal: 3ẇ

(we don't have reshape in the APL sense either)

@exedraj split into 3 chunks?

chunks of 3

> `s` Split into chunks
`#` Reshape (APL/k)

^ which is more useful?

5:05 AM

@PyGamer0 but if you want something more fun: 0 3"3 6"6 9"Wƛ?ni

@PyGamer0 both

@exedraj what does that do?

@exedraj nice

@PyGamer0 input[n] for n in [[0, 3], [3, 6], [6, 9]]

because indexing a list with a list treats the index as a python slice

look in the flags :P

yes I saw that

very cool

@exedraj i updated the flags

5:11 AM

very good

5:32 AM

CQ: CLike Structures in esolangs?

@PyGamer0 ⎕⊂⍨9⍴3↑1 in APL then?

When a language properly supports declaring and using C-like structures, it is no longer an esolang

If you remove the "declaring" part, it just becomes a tuple

> it is no longer an esolang

???

which can be simulated relatively easily in many esolangs

5:58 AM

how do i implement (APL's) reshape in python?

@PyGamer0 k6 has that as (0N;n)# or (n;0N)#, in k9 it's n^

6:09 AM

@PyGamer0 First adjust the length of the list of elements, then split into chunks from the last axis to the first

6:22 AM

Try it online!

@Bubbler ^ gtp3 with my modifications

doesnt work like ⍴

also horribly breaks on higher axis

wdym breaks on higher axes it doesn't even work on your sample case

your list comp is definitely wrong; your loop should either have a step argument or you should be multiplying i by the interval size

> your list comp is definitely wrong; your loop should either have a step argument or you should be multiplying i by the interval size

not mine

that is generated by gtp3

in The Edge of Propinquity, 12 mins ago, by exedraj

def reshape(shape, l):
    if len(shape) == 1:
        return l
    else:
        return [reshape(shape[1:], l[i]) for i in range(shape[0])]

in The Edge of Propinquity, 12 mins ago, by exedraj

apparently

well that's impressive for something made up by a computer then

it's still totally wrong but like. it's sensible enough that it didn't look immediately blatantly dumb

It's the codex model

Trained on open source

Same stuff that powers copilot

6:30 AM

aight i think this works? idk tho

Yeah, that's pretty impressive for machine generated code. It got the basic approach (almost) correct after all

@hyper-neutrino yes it does thx

lol, nice

to be clear there's almost certainly a much more intelligent way to do this

than to turn the list into a deque and just keep rotating it

but i am too lazy and it is 2:40 AM

its fine for now

6:47 AM

@hyper-neutrino Sounds like itertools.cycle

1 hour later…

7:59 AM

how do i make reduce_first(

as a complement to this

@PyGamer0 Your reduce is actually reduce_first I think

@Bubbler oh yeah

so how do i do reduce?

8:17 AM

@PyGamer0 Recurse until you've got a list of numbers, then reduce on it

@PyGamer0 Implementing APL?

@Adám no

actually i might do that in the future

but not now

@Bubbler is this it?

now i need vectorise1( vectorise2( vectorise3(

i think i will copy that from vyxal or something

@exedraj why is there a custom vy_zip(?

where is jelly's generic vectorise function?

8:40 AM

@PyGamer0 because normal zip doesn't pad shorter iterables with 0

@PyGamer0 Looks good

Golfing = 'economizing characters

@exedraj you can use itertools.zip_longest

Try it online!

The wikia has someone called 'redwolf'. Who's that, never heard of them ;p

user image

one 3 too many

LEEET

8:57 AM

dyadic vectorisation done

@emanresuA 1+3+3+3+7 = 17 which is prime

True

@emanresuA and oeis

@PyGamer0 and monadic also done

9:13 AM

Good job

ty

Fun Fact: I made all the functions in tio.run first before copy pasting them.

That's actually quite a good idea :p

2 hours later…

11:32 AM

Better than bogosort - gobosort!

Which sadly hasn't been created yet

Dude coinheringaahing

11:43 AM

@Bubbler It has mold, which is similar :P

> Jelly
> mold

@Dudecoinheringaahing is that deliberate? :Þ

No, Jelly's moldy.

« first day (3860 days earlier) ← previous day next day → last day (987 days later) »

Transcript for

Aug '2124

The Nineteenth Byte

The Nineteenth Byte: General discussion for codegolf.stackexc...

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

site design / logo © 2024 Stack Exchange Inc; legal

mobile