« first day (1960 days earlier)      last day (2880 days later) » 

1:00 PM
@wizzwizz4 This isn't the best part. 2 years ago, the same guy said, that windows 8 never dies with a blue screen
 
@Bálint O.O It blue-screens when it's loading!
Well, the screen turns blue, at any rate...
 
Hai @flawr
 
@Fatalize Would be interesting if it's much greater than linear, but boy the variance of this plot is all over the place
(I wonder what Peter would think of the challenge)
 
@Sp3000 yeah variance is pretty wild
 
Who here remembers Windows 98?
 
1:07 PM
I would like to have feedback from one of the math experts on the "difficulty" of that challenge
 
HI @BALINT
 
otherwise as you said it comes down to quickly finding primes
 
I did try random 100-digits shortly after the 50 test, and didn't have much luck with that
So that might be a good sign
Not sure if there's any useful tricks you can apply though - at that digit size, sieving can only do so much, and I can't think of anything else
 
@Fatalize un challenge génial! :)
 
@LeakyNun Remains to be proven
@Sp3000 i can't think of anything either, but I posted this because I'm sure some crazy people here will find some heuristics :p
 
1:13 PM
Hi @MartinEnder
 
@Fatalize How is there any heuristics for prime generation?
 
@LeakyNun Well primes are never even for one thing...
In that challenge the prime must have even length
 
@Fatalize You're applying the first step of the sieve
even length has no use
 
@Fatalize Maybe :P
 
@LeakyNun ? What? that already cuts out all primes of odd length
I don't know if there are any clever heuristic based on the few rules
 
1:16 PM
Even length probably doesn't help much tbh :P it's probably easier to go from the small primes to build the big primes than the other way around
 
@Fatalize I don't think I understand what you are saying. How is there a heuristic based on even length?
 
that's why I'm waiting to get feedback from a math expert
 
Something I find interesting about the challenge is that the rules imply that both A,B need to be different
So stealing the top Mersenne primes won't help :P
 
I could also ask on Math.SE to see if people can come up with properties, but then that would simplify the challenge
 
so A, B, A*10^n+B, B*10^n+A are primes...
primes.utm.edu/nthprime/algorithm.php (how to find the n-th prime quickly for p < 100,000,000,000,000,000)
probably isn't useful, as it involves stored data
 
1:20 PM
You don't really need an nth prime algorithm per se, since you've got a pretty wide range to work with
 
@Sp3000 sure
>> Is 239812851035039 prime?
<< Proven composite (fails sprp test base 2)
so it know it isn't prime without factoring it!!
 
Primality testing is really quick yeah, hence my concern for brute force earlier
 
> "Write n-1 = (2^s)d where d is odd and s is non-negative: n is a strong probable-prime base a (an a-SPRP) if either a^d = 1 (mod n) or (a^d)^(2^r) = -1 (mod n) for some non-negative r less than s. "
so here s=1 and d=119906425517519
 
(and probabilistic is even faster)
 
how does it compute 2^119906425517519 mod 239812851035039? Oh, I have a mod-power algorithm in O(log(n)) time...
so that explains it...
 
1:25 PM
:)
 
turns out J only have the 100,000,000th prime available
the problem is, @Fatalize , how do you check the answer?
 
https://xkcd.com/1156/ Not the time, the conversation was actually on-topic.
 
@LeakyNun what do you mean?
 
@Fatalize Let's say, I give you 1770025444645505806015260734501017063700887248716173438373146687507577667272673‌​822081302616001655077
 
Checking's simple right? Just test primality
 
1:29 PM
I just test the number, the two subparts, and the oppositve concatenation
that's easy
 
I give you 4911896143611633780682402453849989800775189204252065726408220729093666935169295‌​333354972115958561596541882926741138174142872698595829944178957090296779606642580‌​976458630868510407440881709556568052691
Test if this is prime.
 
^ how is that a problem? You have to check that it's prime in the challenge anyway
 
I don't think primality testing takes as long as you think it does (or at least probabilistic, with reasonably good likelihood)
 
Then check it.
I claim that it is a prime.
 
Also, I expect people to prove the validity of their approaches, and not pull them out of thin air
 
I'll check it and tell you how long it took once I finish making some avocado Vegemite toast
Or not
 
wait I think the copy paste failed
I don't have Mathematica but it probably can do it quickly
 
@Fatalize I get it
It doesn't work.
wolfram alpha can't check it
@Fatalize You probably used a different number.
 
you can't input that number it seems
since it's free and online it's probably limited
 
1:35 PM
exactly
checking if a 200-digit number is prime is...
it may be fast to prove that it is composite
there is only probable-prime test, not probable-composite test..
 
julia> @time isprime(911896143611633780682402453849989800775189204252065726408220729093666935169295333354972115958561596541882926741138174142872698595829944178957090296779606642580976458630868510407440881709556568052691)
  0.001248 seconds (1.42 k allocations: 75.504 KB)
false
Still not seeing a problem
 
@Sp3000 You missed the 4 at the beginning
 
julia> @time isprime(4911896143611633780682402453849989800775189204252065726408220729093666935169295333354972115958561596541882926741138174142872698595829944178957090296779606642580976458630868510407440881709556568052691)
  0.024024 seconds (4 allocations: 160 bytes)
false
Oops
 
lol
 
@Sp3000 Try this: 5802166458563979118118402595044024839822613606951693823249368750582247183653682‌​429882273371034225069773999682593823264194067085762451410312598613405099769716012‌​7301547995788468137887651823707102007839
 
1:40 PM
julia> @time isprime(58021664585639791181184025950440248398226136069516938232493687505822471836536824298822733710342250697739996825938232641940670857624514103125986134050997697160127301547995788468137887651823707102007839)
  0.004226 seconds (5 allocations: 2.602 KB)
true
 
lol even faster
 
(hmm I wonder what Julia uses)
 
wait, why did your first call take 1.42 k allocations?
 
@Sp3000 me too
 
but the rest only took 4 or 5
 
1:41 PM
number is longer and it's prime but it checks it faster
 
Dunno shrugs
 
But there is no definite-prime algorithm except trial division...
 
the allocations/memory used is weird
 
@LeakyNun See: AKS
 
Oh, really
@Sp3000 Thanks, I learnt something new today
 
1:42 PM
any prime generation would also work for prime checking
(not recommended though)
 
@Sp3000 what is its runtime complexity?
 
Julia seems to use a probabilistic test for BigInts with a false positive rate of about 0.25^reps
Of AKS? Can't remember, but it's polynomial
The AKS primality test (also known as Agrawal–Kayal–Saxena primality test and cyclotomic AKS test) is a deterministic primality-proving algorithm created and published by Manindra Agrawal, Neeraj Kayal, and Nitin Saxena, computer scientists at the Indian Institute of Technology Kanpur, on August 6, 2002, in a paper titled "PRIMES is in P". The algorithm determines whether a number is prime or composite within polynomial time. The authors received the 2006 Gödel Prize and the 2006 Fulkerson Prize for this work. == Importance == AKS is the first primality-proving algorithm to be simultaneou...
 
> Probabilistic primality test. Returns true if x is prime; and false if x is not prime with high probability.
@Sp3000 Well, trial division is O(sqrt(n)) (if division is O(1))
 
Sorry, polynomial in the number of digits (i.e. log)
 
Õ(log(n)^6) according to wiki
 
1:47 PM
Impossible....
 
I'm uploading a new youtube video soon. finally made an intro and outro for my channel
 
Õ(log(n)^6) = O(log(n)^6 log(n)^k log(n)^6)
where k is fixed
complexity is in the abstract
 
@Fatalize Using that algorithm one could generate big primes?
 
@Bálint the perspective changes TOOO quickly on the left, and then has almost no difference when you slide more to the right
 
2:01 PM
If it is indeed polynomial time, then why is the largest known prime only 22,338,618 decimal digits long?
 
polynomial time is not linear time
 
oh, ok
 
also that big prime is definitely not tested with AKS
 
Should probably also note that Mersenne uses different algorithms
^^ that
 
yeah, they have specialized algorithms for Mersenne primes
 
2:06 PM
A number on the order of 10^22338618 is frikkin' huge anyway.
 
@aditsu Fixed
 
CMC: bijective base. (a challenge of bijective base exists, but is not completely the same)
that means, instead of using 0 to n-1, we use 1 to n.
For example, we use 123456789X for base 10
100 would be 9X
110 would be XX
 
2:22 PM
gcj round 3 is happening, unfortunately none of us is in it
 
One year, one year... :P
 
I'll do gdcj round 2 tomorrow :)
 
Have fun :)
 
2:38 PM
Julia looks very interesting
 
@aditsu Fine, @Audits.
 
-3
Q: The teacher is always wrong flip his answer

CrazyPythonYour dumb teacher is always wrong, so flip his answer. Change these responses True -> False Yes -> No 0 -> 1 Yeah -> Nope and vice versa. Keeping punctuation and capitalization count. Test Cases: True -> False True. -> False. Yes -> No Yes. -> No. yes -> no yes. -> no. 0 -> 1 1 -> 0 Yeah ->...

 
@Lynn It is.
@Bálint what happened to the OMGWTFBBQKITTENZ!!!!!1!!!!11eleven effect for the hypercube?
 
Fun fact, I wanted to post a challenge like that
 
@MartinEnder Look, what's not clear about the challenge?
 
2:39 PM
@LeakyNun s/X/A/
 
@EᴀsᴛᴇʀʟʏIʀᴋ What do you mean?
 
@Bálint pls bbqkittenz pls
 
@Lynn are you aware of the Julia room?
 
6 hours ago, by aditsu
XYZ sliders: nice, good stuff, W sliders: OMGWTFBBQKITTENZ!!!!!1!!!!11eleven
> OMGWTFBBQKITTENZ!!!!!1!!!!11eleven
pls
 
@EᴀsᴛᴇʀʟʏIʀᴋ I still don't know what you mean, try changing the perspective
 
2:41 PM
A button for hyperdimensional BBQ kittenz to fly across the screen.
 
@MartinEnder Nope!
Where is it?
 
@EᴀsᴛᴇʀʟʏIʀᴋ I didn't change anything about the sliders nor the cube's rendering
 

 Julia

Discussion about the Julia language and golfing in it — julial...
@Bálint pls make teh kittenz tho
 
@EᴀsᴛᴇʀʟʏIʀᴋ Just visit nyan.cat
That's pretty close IMO
 
Why is nobody interested in J?

 J

Discussion about the J language (jsoftware.com) (code.jsoftwar...
 
2:44 PM
@Bálint but but but 4D BBQ KITTENZ
@LeakyNun Because J is too much golf not enough interesting
 
Whoops
@LeakyNun idk
 
@Sp3000 Unbroken. One of the interpreters isn't playing nice...
 
Ah k :)
 
@EᴀsᴛᴇʀʟʏIʀᴋ Draw me one, and I swear I do it
 
2:47 PM
@LeakyNun I found it very hard to understand
 
@aditsu Come, I (we) can teach you :)
 
I'll consider it, don't have time now
 
back
@VTC that video.. ._.
Just got around to watching it.
 
3:08 PM
Video?
 
Him lagging the hell out of his minecraft and posting it to youtube.
 
@Sp3000 The reason I spelled it with two I's is because I'm bad at spelling. ;_;
 
Ah k, no worries. Was just curious
 
Also
> The input will be padded with spaces so that the length of each line is the same.
 
Yeah, I was confused about that because 1) I think the comment that lead to that edit asked whether they can assume input will be padded with spaces (as in a "may" rather than a "will") and 2) The test cases themselves aren't actually padded
 
3:19 PM
Fast prime checker for n < 3,317,044,064,679,887,385,961,981
I'll implement n >= 3,317,044,064,679,887,385,961,981 later
 
35 mins ago, by Sp3000
Ah k :)
 
Ah Miller Rabin, nice :)
 
5 mins ago, by Sp3000
Ah k, no worries. Was just curious
 
Says 3 is composite though
 
10 secs ago, by Sp3000
Ah Miller Rabin, nice :)
 
3:22 PM
@Sp3000 I just noticed a bug
 
Yeah, I'm pretty sure someone could impersonate me in chat if they really tried: Ah k logs
 
hm
martin is the next most
Martin is SP3000 confirmed.
 
That would be terrifying if Martin could simultaneously be the top user and the third user.
 
3:34 PM
@LeakyNun It auto-updates as I type and keeps dying every time I try to test a number starting with 1 :/
 
Looking good :)
 
4:02 PM
Back from camping
(yet again)
Lol, haven't used a keyboard (or monitor, for that matter) for 5 days
 
@muddyfish What's the compression algorithm that Pyke uses?
 
@LeakyNun I don't think it has one
 
@quartata right, it uses dictionary..
 
No, I mean I don't think it has any form of compressed strings
 
0
A: The teacher is always wrong, flip his answer

muddyfishPyke, 51 45 bytes .d̾ॗǣ/ற㻯dc2cDMl4+D.F\.+)+T`]1+Dm_+Y@ Try it here! .d̾ॗǣ/ற㻯dc2c - {'yeah': 'nope', 'true': 'false', 'yes': 'no'} DMl4+ - ^ += map(^, str.title()) D.F\.+)+ - ^ += map(^+".") ...

@quartata
 
4:12 PM
Huh
oh hey that's a challenge cinnamon gum can win
Shit nevermind requires case sensitivity
 
@quartata No, it's case sensitive
 
@LeakyNun I mean that it wants true -> false but True -> False. Probably should have explained that better :P
 
@quartata How does Cinnamon Gum work?
 
Is there a word that describes how many things a function returns? Like the opposite of arity?
Or is the concept of a function returning more than one thing not common enough for its own formal word?
 
@LeakyNun Right now the lookup table mode isn't particularly clever
It just maps strings to strings (like a big dictionary)
 
4:18 PM
 
Maybe in the future I could have a partial replacement thing
 
@quartata For example?
 
@LeakyNun true&false;0&1 creates a dictionary {true: false, 0: 1}
Then it just does dict[input()]
Now something you can do is have something like true&True&false which would give {true: false, True: false}
 
@quartata Then why can't it be used in this challenge?
 
But that doesn't help here
@LeakyNun Well of course it can be but I'm saying it wouldn't win
 
4:21 PM
oh, lol
but it would be fun, wouldn't it.
 
I suppose I could see how much it compresses to
 
@quartata Compress?
 
Well yeah that's the point
 
@mınxomaτ cool
 
The first byte is the mode (l in this case)
 
4:22 PM
how does it compress?
 
The rest is a Bubblegum-esque string that gets processed depending on the mode
@LeakyNun It can be deflate lzma or bb96 like Bubblegum but I'm probably going to swap bb96 out for lzstring soon
 
First attempt, highly ungolfed:
True&False;False&True;True.&False.;False.&True.;true&false;false&true;true.&false.;false.&true.;Yes&No;No&Yes;Yes.&No.;No.&Yes.;yes&no;no&yes;yes.&no.;no.&yes.;0&1;1&0;Yeah&Nope;Nope&Yeah;yeah&nope;nope&yeah;Yeah.&Nope.;Nope.&Yeah.;yeah.&nope.;nope.&yeah.
255 bytes
 
Yeah that's literally the only way to do it
 
@Dennis Can you please pull 05AB1E?
 
@quartata I thought you said you can compress it
 
4:24 PM
Right you compress that string
But I'm saying the lookup table itself can't be made any shorter. There's no tricks
 
@quartata You mean compress the lookup table using bb96?
like compress the 255-byte long string
 
Yes
 
@Adnan Pulled.
 
Thanks :)
 
@quartata Well, then what would it become?
lzstring could be useful here, since there's so many duplicates
 
4:25 PM
I don't know I'm finding out
zopfli is very slow so
 
@LeakyNun Pyke uses a dictionary found here (Warning: very big file)
 
@muddyfish how do you look it up?
 
I assume it's by the code point of the character
hence the weird unicode chars
 
Enter words string-seperated. It's indicated by the code-=point
1st-char - no words to look at, rest - index in dictionary as unicode
 
4:29 PM
Well, since it's essentially sorted by frequency, why not use prefix code?
 
prefix-code?
 
Huffman encoding
 
I'm just storing it as a json file github.com/muddyfish/PYKE/blob/master/dictionary.json
(I don't know much about compression)
 
@muddyfish the dictionary isn't the problem. I'm referring to the way you access the dictionary
 
@LeakyNun 119 bytes :/
 
4:34 PM
@quartata post here?
@muddyfish Right now, the first 128 elements or so take 1 byte to index, and the next XXX elements take 2 bytes, etc
just make that more extreme
 
ok
I'll look at it now
 
@LeakyNun The permalink is too long
 
I don't know if it would be better
the extreme is that the first element takes 1 bit, the second element takes 2 bits, ...
so you'd have to balance
@quartata 119 bytes, just post the string here
or the hexdump
 
Looks like lzstring compresses it down to 67 bytes
soo much better
Definitely need to get on that
 
@quartata said it.
 
4:36 PM
0000000: 6c1d 4b27 9603 0114 ba0d 9237 a3f1 2bc7  l.K'.......7..+.
0000010: 6caa 4bef bde7 f6f9 60e8 f46e cf25 fea6  l.K.....`..n.%..
0000020: 87fb 5241 3890 8148 4085 9882 7a14 6295  ..RA8..H@...z.b.
0000030: 7d10 0e92 1209 a810 5350 e3e5 1ddd 59dd  }.......SP....Y.
0000040: 19a5 ec58 96e5 e980 fa56 7f3a eb74 4629  ...X.....V.:.tF)
0000050: 3b96 6579 3aa0 1ab4 6ad1 d473 baa9 e765  ;.ey:...j..s...e
0000060: 2903 6cf5 7576 7266 806d 76cc 900a 2249  ).l.uvrf.mv..."I
0000070: b6cc 980a 22c9 0f                        ...."..
 
@quartata what compression algorithms are available atm?
 
deflate lzma and bb96. I tried them all already
 
@quartata You should probably include more compression algorithms. I can help you if you want to
9 hours ago, by Leaky Nun
include this one, for example
each compression algorithm is optimized for some condition
 
I can't really just add more compression algorithms.
 
Why not?
 
4:41 PM
The reason why Bubblegum can have deflate lzma and bb96 is that the latter two will throw errors when they encounter the others' format
 
@quartata easy, use the first nibble to judge the compression algorithm
 
No, I'm already having to use the first byte to tell what mode
 
how many modes are there?
 
8 currently
But I'll be adding way more (I should have 256 by the time I'm done, some slight variations of other modes)
 
the first byte have 256 variations
just add more modes there?
 
4:43 PM
I want to visualize a graphical algorithm, but I don't know wich one.
 
one mode for each compression algorithm
 
I considered that but eh
 
@Bálint Cantor dust
@quartata What are the current modes?
 
I think deflate lzma and lzstring will cover most cases, really.
@LeakyNun Lookup l Sprintf f Generate g Format and generate h Expand p Char lookup e and regex substitute s
 
@quartata ...and what is regex substitute?
 
4:45 PM
re.sub
 
no, I mean
can't you use it for this challenge?
 
Possibly, yeah? I don't even know if it works lmao. Recent addition
I'll try that I suppose
 
How does e work?
 
Instead of looking up the whole string it looks up each character
So for test it would output dict["t"] and dict["e"] and ...
 
I see
 
4:57 PM
Actually, after some further investigation I think I might be able to add lzstring without removing bb96
 
Chat poll: What's your favourite quote?
 
11 secs ago, by Bálint
Chat poll: What's your favourite quote?
 
12 secs ago, by quartata
11 secs ago, by Bálint
Chat poll: What's your favourite quote?
 
@Bálint -._(._.)_.-
 
@CᴏɴᴏʀO'Bʀɪᴇɴ @quartata ಠ್ಗಠ
 
4:58 PM
1 min ago, by Cᴏɴᴏʀ O'Bʀɪᴇɴ
12 secs ago, by quartata
11 secs ago, by Bálint
Chat poll: What's your favourite quote?
 
Hmm nope nevermind
 
@Bálint this one: '
 
Seriously
 

« first day (1960 days earlier)      last day (2880 days later) »