01:55:54 <HackEgo> 1/2:xyzzy//Nothing happens. \ bitcoin//bitcoins are coins that have been drilled through with a bit, and can be strung together in long chains. This practice dates to ancient China, and the Chinese remain experts in bitcoin manufacturing. A chain can support up to 21 million coins before breaking. \ bride theory//Bride theory is a theory involving

01:55:56 <HackEgo> 2/2:a headhunter who dresses in yellow. \ tarafurnar//Tarafurnar is the event after which you'll be seen. \ acab//ACAB is short for "all codecs are broken".

03:48:14 <oerjan> zzo38: thanks, those looked so similar in my browser font that i didn't notice there _was_ a difference

03:51:18 <HackEgo> Your pal Ørjan is oerjan's good twin. He's banned in the IRC RFC for being an invalid character. Sometimes he publishes papers without noticing it.

03:51:51 <HackEgo> * \ ☾_ \ c# \ myname \ orin \ örjan \ ørjan \ pico \ sewerjan \ twint-e \ typoerjan \ אrjan

03:53:47 <HackEgo> Örjan is the diæresed twin. He will punctuate your vöẅëls, and maybe a few other unsuspecting letters.

03:59:47 <HackEgo> 1/2:amiga//Amiga is Spanish for a female friend. \ pirate//Pirates are humourously nautical persons. Their grammar is friendly and plural. \ apt-get//apt-get installs whatever you wanted, plus whatever Mark Shuttleworth wanted. \ zork//Zork is like York, except for the first letter. Uaneb invented it. \ lie bracket//Politicians try to stay within t

03:59:52 <HackEgo> 2/2:he lie bracket: Not so many lies that voters cannot stand it, but not so few that they think you have nothing to give them.

07:22:03 <HackEgo> /home/hackbot/hackbot.hg/multibot_cmds/lib/limits: line 5: exec: wgrep: not found

07:22:23 <HackEgo> wisdom/sbus:SBus is the standard bus in many a Sun SPARC-based system, capable of coping with thirty-two (32) bits in parallel, at rates of around 16.67 to 25 MHz. There is a 96-pin connector, and the cards lay parallel to the motherboard, like toppings on a sandwich. \ wisdom/2600:2600 Hz is a tone made by Captain Crunch's whistle. \ wisdom/john:J

07:22:54 <HackEgo> 1/1:/hackenv/bin/`: eval: line 5: unexpected EOF while looking for matching ``' \ /hackenv/bin/`: eval: line 6: syntax error: unexpected end of file

07:23:20 <HackEgo> wisdom/john:John is not Hziulquoigmnzhah of Cykranosh, the dread spawn of Cxaxukluth, according to http://smbc-comics.com/index.php?id=4072

07:57:11 <wob_jonas> M:tG stuff: http://www.manaleak.com/mtguk/2017/06/unstable-magic-the-gathering-spoilers-news-updates/ spoilers for the upcoming Unstable unset.

08:01:12 <wob_jonas> I expected they'd do more reprints from black-bordered world, but no, only one aside from basic lands.

08:14:03 <shachaf> Do you know this puzzle: I think of a polynomial of any degree with natural coefficients. How many points do you need me to evaluate it at to figure out what it is?

08:19:02 <wob_jonas> you just need to pick your favorite transcendent real number and evaluate it there

08:20:22 <wob_jonas> and in that case, two points for natural coeffs and impossible for integer coeffs I believe

08:32:07 <int-e> I don't know; I suspect that if you input something like sum 1/2^2^k then the answer will include all the information one needs.

08:45:03 <int-e> In fact any computable transcendental number greater than 1 will work. The point is that if you use a number greater than 1, then the integer part of the result gives you an upper bound on the coefficients and the degree, so that when you reach the fractions, only finitely many candidate polynomials remain; you can evaluate all those at your number and eliminate those whose value does not equal...

08:45:09 <int-e> ...the given one. Because the number is transcendental, and disequality of computable reals is semidecidable, only one polynomial will remain in the end.

08:46:05 <shachaf> So is convolution with a kernel, the kind that people talk about in image processing, a special case of the usual convolution?

08:46:37 <int-e> using a "nice" computable real like that sum 1/2^2^k one I suggested should make it much more efficient though :)

08:51:20 <int-e> and the kernel, with the center at (0,0), and perhaps a switch from k(x,y) to k(-x,-y).

08:53:00 <shachaf> In general I expect a convolution to look like this: (f * g)(x) = A_{i,j | b(i,j) = x} { c(f(i), g(j)) }

08:53:58 <wob_jonas> shachaf: there has to be a sum in there (unless you're a mathematical physicsist and using tensor notation)

08:54:33 <int-e> f(x+i) k(i) which is probably how it's implemented, and corresponds to the switch from k(x) to k(-x) I suggested.

08:55:19 <wob_jonas> and you either assume the kernel and image are zero in the rest of the places, or you say the sum only goes when both i and j are in the respective domains

08:55:39 <int-e> shachaf: f is the image; g is the kernel, possibly reflected, and one of the points is that g(j) is zero almost everywhere.

08:56:47 <wob_jonas> int-e: either that or it converges to zero exponentially fast as j goes to infinity and -infinity, and when computing we may approximate the small values with zeros

08:57:04 <shachaf> Concretely convolution in the finite case where b = + usually corresponds the antidiagonals of some matrix, right?

08:57:20 <wob_jonas> people often convolve images with bell curves, but since they don't need infinite precision, it's practically zero far enough

08:59:14 <int-e> wob_jonas: yes, but I believe they tend to do that by iterating rectangular filters like [1/3,1/3,1/3]

08:59:15 <wob_jonas> shachaf: the matrix is the outer product of two vectors, the first vector is column vector of values of f, second vector is row vector of rosw of g

09:01:30 <wob_jonas> and then there are these fast methods for computing such convolutions even if the image and kernel are both big, of which the most famous is called fast fourier transform, but there are other transform methods

09:01:59 <int-e> I'd expect [c, b+2c, a+2b+3c, 2a+3b+4c, ..., 5a+6b+7c, 6a+7c, 7a] in an actual implementation, and possibly truncated

09:14:00 <wob_jonas> depends on which direction you convolve. some call the plus version convolution, some call it correlation and the minus version convolution.

09:14:31 <wob_jonas> just like which direction of fourier transform is forward is also a terminology difference

09:15:06 <wob_jonas> you have to read documentation carefully to find out what conventions it uses for the formulas

09:15:30 <int-e> to actually get a result of length 9, you should pad 'img' with two zeros to the left and to the right.

09:15:37 <shachaf> But I suppose you can extend them both to Z and define them aas 0 everywhere else.

09:18:12 <int-e> subtraction corresponds to r = 0; for (int i = 0; i < ker.length; i++) r += img[x+i]*ker[i]; res[x] = r;

09:19:24 <shachaf> I like convolution in contexts where you don't have inverses so you can't even say f(i)g(x-i)

09:20:43 <wob_jonas> oh, and the whole thing is made cool by the fact that GPS devices do 1d convolution

09:28:40 <wob_jonas> basically GPS satellites can't be each assigned different frequencies like radio stations, so you can't pick their signals out with just simple electronics, their signal is really weak, and they aren't just round-robin broadcasting either, they just all talk at the same time,

09:29:42 <wob_jonas> so to pick out the voice of one satellite from the others and all the other noise that's actually stronger than the signals, GPS clients digitize the signals they receive and convolve it with what the GPS is expected to say to find out when they received that signal, and they do the convolution in microchips digitally

12:40:53 <mroman> that way it's probably possible to have scouting bots scan the whole field into team memory

17:22:44 <int-e> for numbers >3, you can just start from the nearest square... except when that square is 256

17:29:49 <mroman> but this is why I believe genetic evolutionary algo stuff will never find certain solutions.

18:46:45 <int-e> mroman: how about this, iiisisdddddddsddsiiisddddddsdddddddsddsiiiisiiiii produces -2 on a 32 bit deadfish implementation

18:48:34 <int-e> (Unlikely to be optimal, but this involves quite a bit of manual effort so I will stop there. Also, doing that for 64 bits seems to be out of reach for me right now)

18:50:57 <zzo38> Some implementations may use decimal arithmetic, such as the one for Famicom and the one for MIX, and possibly some others too.

19:26:16 <int-e> so much fun... "iiisisdssissdsiississdddsiissiiiisiiiii" produces -2 as well, and is optimal.

19:27:13 <int-e> iissdddsiiiiiisisdsiisisdsiisiiiiii produces 255, and iissdsdssisiiiisiisssdddsddds produces 257, all assuming 32 bit deadfish

19:35:04 <int-e> and, finally, a Hello, world! program (with final newline): iiisdsiiiiiiiiossssiiisisioiiiiiiiooiiioisssiiisddsdddddoddddddddddddosssiiisiisddoddddddddoiiioddddddoddddddddossssiisiisdddodsssiiisio

20:06:42 <int-e> As a sketch, assuming unbounded natural numbers, you can first make the number big (so that the +1 of the 3n+1 becomes negligible for the next phase), then adjust the log_2 to have the right fractional part (the precision required for that is bounded by the target number; if it's n, then you want the fractional part between the fractional parts of log_2(n-1) and log_2(n)), and then divide by 2...

20:16:49 <int-e> mroman: the approach is stupid though; for example, 257 is produced by iciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciciccciciciccciciciccciciciciciccccicicciccicicicicicicicicccicciciciciciccccciciciccccicicccicicciccciccccccciciciccicicicc

20:16:54 <HackEgo> [wiki] [[Esolang:Introduce yourself]] https://esolangs.org/w/index.php?diff=52940&oldid=52937 * Mroman2 * (+108)

20:17:05 <HackEgo> [wiki] [[Hardfish]] N https://esolangs.org/w/index.php?oldid=52941 * Mroman2 * (+698) hardfiiiiish

20:19:39 <HackEgo> [wiki] [[Hardfish]] https://esolangs.org/w/index.php?diff=52942&oldid=52941 * Mroman2 * (+36) /* Some Constants */ starting c is a nop obviously.

20:19:55 <rdococ> I'd talk here more often, but I'd run the risk of self-contradicting people who at the same time don't harbor any hate for me at all, and have a grudge against me.

20:20:54 <rdococ> Of course, I wonder how the world would react if I was also an extremely unforgiving, confusing person.

20:22:42 <rdococ> I'm not unforgiving, though, and I don't contradict myself in the span of a few minutes.

20:33:59 -!- MrBismuth has quit (Quit: https://www.youtube.com/watch?v=xIIqYqtR1lY -- Suicide is Painless - Johnny Mandel).

20:36:39 <HackEgo> [wiki] [[Hardfish]] https://esolangs.org/w/index.php?diff=52943&oldid=52942 * Mroman2 * (+129) /* Some Constants */ more constants

20:41:50 <HackEgo> [wiki] [[Hardfish]] https://esolangs.org/w/index.php?diff=52944&oldid=52943 * Mroman2 * (+173) /* Some Constants */ more constants

20:46:52 <HackEgo> [wiki] [[Hardfish]] https://esolangs.org/w/index.php?diff=52945&oldid=52944 * Mroman2 * (+55) /* Some Constants */

20:52:46 <int-e> and a Hello, world!, assuming we add o: iciccicciciccioccicicccccccoicicciccicicciooccccccicccciciccioicciccoccccccccciccoicccccicioiccicicciciccioiciccccccccccicciocicicciccccioccccicccicocciccciccio

21:01:23 <zseri> what's with the line " 27 - cciccciccicicci" in [[Hardfish]], I think "cc" at the begin is a nop?

21:03:46 <int-e> mroman: but the program actually enumerates values, so if two prefixes produce the same value they only count as one

21:05:08 <int-e> zseri: people have the same complaint about DNA, most of it is, as far as we can tell, redundant, though sometimes we find a purpose for parts that were previously thought to be unused

21:06:49 <int-e> mroman: and it turns out that *a lot* of values collapse. Here's a list of the intermediate set sizes: http://sprunge.us/IIjH (there are two sets in each iteration, one for programs ending in 'c' and one for programs ending in 'i')

21:10:39 <HackEgo> [wiki] [[Hardfish]] https://esolangs.org/w/index.php?diff=52946&oldid=52945 * Mroman2 * (-89) /* hardfish */ change a lot to make it more intelesting.

21:11:17 <HackEgo> [wiki] [[Hardfish]] https://esolangs.org/w/index.php?diff=52947&oldid=52946 * Mroman2 * (+5) * typo

21:12:22 <HackEgo> [wiki] [[Hardfish]] https://esolangs.org/w/index.php?diff=52948&oldid=52947 * Mroman2 * (+136) + let's support the o instruction as well.

21:13:01 <int-e> in particular, it raises the question how to write Hello world with as few 'o' instructions as possible :)

21:14:59 <int-e> (it's possible with a single 'o' if one can find a code segment that produces 72 from 0, and can also produce all the other necessary values... and that may be tough to put into practice.

21:28:35 <mroman> but yes... whether it's possible to have less o's than output characters is interesting

21:29:06 <mroman> although I have a hunch that these programs are probably longer than the ones with more o's in them

21:30:35 <Taneb> Am I right in thinking that simply typed lambda calculus + a fixed point combinator is Turing complete?