The Nineteenth Byte

1:30 AM

Oops, I did it again.

I played with computer science, got lost in the game, and next thing you know it, I've created another esolang. esolangs.org/wiki/Stisp

5

This language is a "Lisp implementation helper." The idea is that it's easy to implement Stisp, and it's easy to implement Lisp in Stisp.

thejonymyster

you have my full attention

Tanner Swett

Here's an implementation of a very tiny subset of Lisp in Stisp:

(CONS xh xt) -> eval[] = xh -> eval_cons[xt];
(QUOTE x) -> eval[] = x;

xh -> eval_cons[xt] = xh -> eval[] -> eval_cons_2[xt];
h -> eval_cons_2[xt] = xt -> eval[] -> reverse_cons[h];

t -> reverse_cons[h] = (h . t);

See esolangs.org/wiki/Stisp for the full spec.

thejonymyster

one day i will understand this SDFG orz

Tanner Swett

The basic idea is a little bit like Haskell. Lines 4 and 5 of that are basically the same as

eval_cons xt xh = eval_cons_2 xt (eval xh)
eval_cons_2 xt h = reverse_cons h (eval xt)

However, the form of an expression is tightly restricted. Basically, function application is denoted using this arrow-and-brackets notation, and it's illegal for function applications to be nested.

The purpose of all the restrictions is to make Stisp easy to implement without any recursion.

thejonymyster

2:21 AM

LDQ: if a user tries to feed multiple args to a unary operator, what options do i have (besides erroring and informing the user of their mistake)

Tanner Swett

The other obvious option I can think of is to just use the first one.

thejonymyster

and just act like the other args dont exist?

oh i did think of something though

Bubbler

Mapping over the args is also an option

thejonymyster

oh yeah

that seems reasonable too

the idea i was about to say didnt actually make much sense lol

Sandbox Posts

3:04 AM

0

A: Sandbox for Proposed Challenges

An algorithm to find even sublime numbers code-golf math number-theory A perfect number is a positive integer whose sum of divisors (except itself) is equal to itself. E.g. 6 (1 + 2 + 3 = 6) and 28 (1 + 2 + 4 + 7 + 14 = 28) are perfect. A sublime number (OEIS A081357) is a positive integer whose ...

DialFrost

3:42 AM

Hi guys

You can't award yourself your own bounty right on your own question?

Bubbler

3:52 AM

I'm pretty sure "You can't award yourself your own bounty" on any question

forest

You can't award a bounty to yourself, period.

DialFrost

4:10 AM

Oh, but is it different on SO?

Cuz I just saw someone award himself/herself 300 rep

Bubbler

Link?

DialFrost

3

A: React native typescript number 0 read as NaN

As it turns out, this is some kind of issue with VSCode. The NaN output only appears when using the VSCode Debug Console or the active VSCode debugger (hovering over the variable). The logcat output is correct. Furthermore, console.log(typeof variable) shows the variable type to be number (which ...

His rep graph shows +300 too

Bubbler

> This answer has been awarded bounties worth 300 reputation by Jared Smith

jack != Jared Smith

Someone else can award a bounty on your answer to your question

DialFrost

Oh whoops

my bad

I messed up the names

Sorry!

pxeger

6:00 AM

> ◉ Yearling × 2

I did not realise you could earn it more than once

I guess it makes sense though

null

8:18 AM

Hi there

null

8:32 AM

How to golf the program n/(;{.1=88<\4/zip{&}*!&"YNeos">2%n}%?

Because I saw @lyxal , @lyxal

Number Basher

9:06 AM

worst code I ever wrote without intending it to be bad:

from functools import reduce

def radix_sort(array):
    number_of_digits=len(str(max(array)))
    for digit in range(number_of_digits):
        bucket = [[] for _ in range(10)]
        for i in array: bucket[(i // (10 ** digit))%10].append(i)
        array=reduce(lambda x,y:x+y,bucket)
    return array

print('\n'.join([f'{i//10} {i%10}' for i in radix_sort([int(input().replace(' ','')) for i in range(int(input()))])]))

(the last line especially, )))])]))

Adám

@NumberBasher Wouldn't it be nice if all functions could use Python 2's print syntax?

from functools import reduce

def radix_sort array:
    number_of_digits=len str max array
    for digit in range number_of_digits:
        bucket = [[] for _ in range 10]
        for i in array: bucket[i // 10 ** digit % 10].append i
        array=reduce(lambda x,y:x+y,bucket)
    return array

print '\n'.join [f'{i//10} {i%10}' for i in radix_sort[int input().replace(' ','') for i in range int input()]]

mousetail

9:23 AM

@NumberBasher [print(f'{i//10} {i%10}') for i in radix_sort([int(input().replace(' ','')) for i in range(int(input())]

null

9:38 AM

Again: how to golf the program n/(;{.1=88<\4/zip{&}*!&"YNeos">2%n}%? (GolfScript)

Anyone? :(

Adám

I fear GolfScript isn't as popular as it once was.

Try on CR ;-)

null

:(

GolfScript is the only golfing language I can write program in

Adám

You find yourself in the same situation as many an expert COBOL programmer.

8

null

:(

emanresu A

10:03 AM

@Adám ~~ruby~~

PyGamer0

11:16 AM

@Adám print isn't a function in 2.x though

Adám

Right.

mousetail

@PyGamer0 I think he want to replace every function with a keyword

PyGamer0

@Adám looks nice tbh

null

11:48 AM

Python 2.8!

Unrelated String

wait does it

null

Oh nope

Unrelated String

peps.python.org/pep-0404

yeah no

emanresu A

I love that it's #404

null

PEP Not Found

Let's prove it wrong should we

mousetail

11:52 AM

Release your own langauge and name it python 2.8

null

Correct.

Now for some non-golf questions

You are give two set of point. Select two points, one from each set, so that their distance is maximized.

Lowest complexity wins.

mousetail

What type of distance?

null

given*

@mousetail Euclidean

points*

mousetail

Hmmm I'm not sure if a more efficient solution exists

null

I know if the "one from each set" restriction is removed

It is UVa 1453

But I'm

> not sure if a more efficient solution exists

in this case

Wheat Wizard

12:04 PM

What is a point here?

How many dimensions?

null

Two

mousetail

UVa 1453 seems to be different

null

@mousetail Well that's the same thing but with some extra useless fluff

Actually the idea here came from there

mousetail

I imagine you may be able to build a Btree in O(n) time then somehow eliminate some pairs that way but I don't see a way that is guaranteed to reduce the number of potential pairs to something like n log(n). Imagine something like this has to be the way though

null

Hmm

I'd imagine that you get the convex hull for both sets and then perform some magic to get the answer

But IDK

mousetail

12:19 PM

en.wikipedia.org/wiki/Strassen%27s_algorithm still blows my mind

I'd have sworn it impossible

m90

12:37 PM

@null Indeed; simultaneously rotate two parallel lines around the two sets: i.stack.imgur.com/bRMjr.png

null

Ohhhhhhhhhhhhhhhhh

I get it

pxeger

12:59 PM

> It is an ex-release

there a lot of quality Monty Python jokes in that PEP

Number Basher

2:19 PM

@mousetail has an extra line

DLosc

3:22 PM

@DLosc Silly me, BQN totally has flatten: it's just monadic ∾, no need to fold. :P 18 bytes.

thejonymyster

3:35 PM

if i pick a random number from 0 to 1, and then whatever i get out of that, use that as the probability of heads for a coinflip, is it still a 50/50 chance?

intuitively i want to say yes, but i just want to make sure im not missing something lol

att

yes

integral from 0 to 1 of p dp / 1 = 0.5

thejonymyster

oh interesting

that makes sense

right i guess im just picking a random point in a square, and seeing which side of the diagonal im on

neat

att

yep

that's assuming your random number is uniform, ofc

thejonymyster

right yeah all that yadda yadda :P

att

just to be sure ;)

thejonymyster

3:44 PM

"Yes, my infinite monkeys are all uniformed. Generating random numbers is serious monkey business."

Bubbler

4:00 PM

Feedback?

Steffan

lgtm

DLosc

4:27 PM

Every time I see lgtm I think it's lmgtfy, which makes my brain try to read it as "let's google that ... m???"

PyGamer0

lol

Adám

5:11 PM

@DLosc same here.

thejonymyster

only recently has it settled as "looks good to me" for me finally

Sandbox Posts

0

A: Sandbox for Proposed Challenges

Find "Millionaire" score code-golf string I was looking through the uploads of GAMES Magazine to archive.org, and in the first publication I found a challenge called "Millionaire: a word-and-number prize competition". In this game, each word is assigned a score in the following way: Turn each le...

thejonymyster

6:09 PM

LDQ: previously i asked about users attempting to pass multiple args into a unary operator; now i ask this: What can I do if the user tries to give a binary operator more than two operators? I know if it's commutative, just reducing by the arg is reasonable (like how addition can be a binary operator but also add(1,2,3) makes perfect sense)

in particular, i am thinking about range, but in general i find the question interesting and wonder if it can be generalized :P

cause for range i can imaging range(1,5,3) being like [1,2,3,4,5,4,3] but that doesnt seem generalizable

DLosc

6:32 PM

@thejonymyster Interesting question. I'd say reducing is reasonable if the operator is associative, even if it's not commutative: for example, concatenation.

So I guess the top three options in my mind are 1) ignore the extra arguments, 2) reduce (which wouldn't work for your three-point-range example, unless you back it up with some type shenanigans), or 3) special-case what you want each operator to do.

"Ignore the extra arguments" is what funstack.hs currently does, but I'd like it to do something more useful if there's a good option.

Radvylf Programs

@thejonymyster Associative makes more sense than commutative

A commutative non-associative operator doesn't really make much sense to reduce by, whereas a noncommutative associative one is still useful

(E.g., reducing by concat is very common and useful)

Average would be an example of a binary commutative non-associative operator that doesn't make sense to reduce by

Since averaging one by one will bias heavily toward later items

pxeger

6:50 PM

@thejonymyster I think you'll have to special-case them, or at least, categorise operators into groups of behaviours (e.g. one category for operators which reduce, etc.)

Radvylf Programs

7:05 PM

Yeah, I agree with ^

DLosc

"And you know why four plus minus one plus ten is fourteen minus one? Because addition is commutative, right"

Radvylf Programs

@thejonymyster Do you have a link to previous discussion of this? I was considering supporting variadicity in an upcoming golfing language, in a similar way to this

@DLosc I propose "Let's Google This Motherf***er"

DLosc

17 hours ago, by thejonymyster

LDQ: if a user tries to feed multiple args to a unary operator, what options do i have (besides erroring and informing the user of their mistake)

Radvylf Programs

Ah, thanks

DLosc

@RadvylfPrograms :/

Radvylf Programs

7:12 PM

@thejonymyster Applying a reasonable variadic operation before applying the unary one

E.g., for negation, product could make sense

Same with square root (that would make geometric mean easy)

Then again in a golfing language this would rarely add golfiness, and in a practical language it would be really confusing

DLosc

For unary, it makes a fair amount of sense to wrap everything in a list and map the operation. But that doesn't generalize well to higher arities.

thejonymyster

@DLosc right, thats what prompted the question lol

also thanks all

DLosc

7:33 PM

I suppose here's one possible generalization (don't know how useful it would be): given an operator of arity a and n arguments, with n > a, apply the operator to all length-a slices of the argument list and return a list of the results. E.g. Plus(4, 5, 6, 7) -> [Plus(4, 5), Plus(5, 6), Plus(6, 7)]

Or if T is some arity-3 operator, T(4, 5, 6, 7) -> [T(4, 5, 6), T(5, 6, 7)]

thejonymyster

why just slices? why not all combinations :P

DLosc

Sure, that's another possibility.

thejonymyster

7:51 PM

actually yeah i guess that could work, i was like half joking lol

Seggan

8:09 PM

@thejonymyster how does the parser even parse that?

Radvylf Programs

Presumably some sort of spread/splat operator

DLosc

I can't speak for thejonymyster, but in the language I'm working on, the program is a function with a well-defined arity, so I have to figure out what happens when you give it too many or too few arguments.

DLosc

9:07 PM

Does anyone know if there's a way in BQN to check whether an identifier is undefined?

Steffan

CMC: Given n, output the length of smashing the FizzBuzz sequence from 1 to n together. So 5 => 11 because smashing becomes 12Fizz4Buzz which is 11 characters long. Standard sequence rules apply, so you can output the infinite sequence or whatever.

No OEIS.

Seggan

@Steffan i still havent found a good fizzbuzz in fig :(

thejonymyster

@Steffan i think this has a closed form solution

DLosc

The thing that makes it tricky is that the numbers get longer

emanresu A

@Steffan CMC: This but handling up to 1 quadrillion

thejonymyster

9:21 PM

i feel like the fizzbuzz part doesnt actually add that much to this :P

but i digress lol

Seggan

@thejonymyster i concur

50 rep to who can provide it

DLosc

@Steffan BQN, 32 bytes: +´·{×◶⟨≠•Fmt𝕩,⊢⟩+´4×0=3‿5|𝕩}¨1+↕

Feels pretty clunky

Seggan

yay kotlin native worked on the first try

in my universe, nothing ever works on the first try

Radvylf Programs

9:36 PM

@Steffan I'm like 100% sure this is a dupe

emanresu A

There's thet bytewise fizzbuzz

Steffan

@RadvylfPrograms There's one with the inverse of that.

Radvylf Programs

Ah

Steffan

23

Q: How did I end up with this FizzBuzz?

FizzBuzz is so simple, bet you can do it backwards. In this challenge, you will be given the length of the FizzBuzz string and must give the positive integer that produced that string. Description To break this down, a FizzBuzz string for n is generated by the following algorithm. Start with ...

code-golf

I've got 13 bytes in Vyxal

And should be 12 without a bug

~~seems like half my Vyxal solutions can be shorter without a bug~~

actually, got 12 bytes

or 11 with flag

thejonymyster

@Seggan welcome to the universe where everything works on the first try

Seggan

9:54 PM

@thejonymyster thanks. now how did i get here?

thejonymyster

@Seggan integration isnt closed form is it? cause if not i need to figure out how to integrate floor(log(x+1))+1 :P

@Seggan idk im from the universe where no one can think of a funny response

anyway congrats :-)

Zion mycelia adamancy

@thejonymyster No, integration is generally not included in "closed form"

thejonymyster

thats what i thought u_u

Zion mycelia adamancy

Also, integrating floor(f(x)) is usually pretty easy, as it's just a series of rectangles

thejonymyster

ah yeah

Seggan

10:16 PM

@thejonymyster no

thejonymyster

@Zionmyceliaadamancy easy sure, but not necessarily closed form is it?

Zion mycelia adamancy

Depends on the bounds of the integral

thejonymyster

damn

im doing fizzbuzzlength(x)=int(from 1 to x){floor(log_10(t+1))dt}+f(x) where f(x) is some other junk

not looking good for me :P

Zion mycelia adamancy

10:34 PM

Let \$x = \lfloor \log_{10}(n)\rfloor\$. The total length of concatenated integers up to \$n\$ is \$n - 10^{x} + 1 + \sum^x_{i=1} 9i\times10^{i-1}\$

thejonymyster

i feel like sum also isnt closed form

Zion mycelia adamancy

It's more closed form than integration, but you can probably manipulate that sum into a closed form solution

@Zionmyceliaadamancy Sorry, that should be \$n - 10^{x} + 1 + \sum^x_{i=1} 9\times10^{i-1}\$

thejonymyster

is chatjax busted or is something else going on? it looks all \$-ey

Zion mycelia adamancy

Works for me

Seggan

is floor even considered closed form?

thejonymyster

10:39 PM

floor is close enough lol

nah idk

floor is closed form if mod is closed form

Zion mycelia adamancy

@Zionmyceliaadamancy Ignore me, it was right the first time

Seggan

@thejonymyster hmm same with me

ChatJax++ error: ReferenceError: GM_getValue is not defined

Bubbler

final feedback?

Zion mycelia adamancy

Just wanna test this: \$\mathbb R\$, \$\mathbb 3\$ and \$3\$

Aww, doesn't affect 3 :/

thejonymyster

im not getting any errors i might just not have it installed right somehow lol

Seggan

10:47 PM

i just see a bunch of \$

thejonymyster

is it chrome only

^^

Zion mycelia adamancy

Interestingly, for sufficiently large \$n\$, \$n - 10^x + 1 + \sum^x_{i=1} 9i \times 10^{i-1}\$ is actually an upper bound, not a lower bound :P

Bubbler

CMC: Golf the constant 6086555670238378989670371734243169622657830773351885970528324860512791691264

= 2^126(2^3-1)(2^5-1)(2^7-1)(2^19-1)(2^31-1)(2^61-1)

Zion mycelia adamancy

Jelly, 14 bytes

Seggan

11:03 PM

@Bubbler is there any significance to that

Bubbler

It is the only known sublime number other than 12

Zion mycelia adamancy

Sublime number

In number theory, a sublime number is a positive integer which has a perfect number of positive factors (including itself), and whose positive factors add up to another perfect number.The number 12, for example, is a sublime number. It has a perfect number of positive factors (6): 1, 2, 3, 4, 6, and 12, and the sum of these is again a perfect number: 1 + 2 + 3 + 4 + 6 + 12 = 28. There are only two known sublime numbers: 12 and (2126)(261 − 1)(231 − 1)(219 − 1)(27 − 1)(25 − 1)(23 − 1) (sequence A081357 in the OEIS). The second of these has 76 decimal digits: 6,086,555,670,238,378,989,670,371...

Seggan

is there any reward to finding the next one :P

Zion mycelia adamancy

> there can be no more even sublime numbers unless there are other (presently unknown) Mersenne prime exponents that are themselves Mersenne primes.

Bubbler

You may earn some reputation (not SE rep) if you prove or disprove the existence of odd sublime numbers :P

For evens, yeah it is very constrained by the known list of Mersenne primes

My upcoming challenge has a bit more info on that

emanresu A

11:10 PM

@Zionmyceliaadamancy How did OEIS let that in?

Seggan