00:10:05 -!- Jafet has quit (Quit: Leaving.).
00:14:53 -!- pikhq_ has joined.
00:15:06 -!- pikhq has quit (Ping timeout: 252 seconds).
00:53:21 -!- Frooxius has quit (Read error: Connection reset by peer).
00:53:36 -!- Frooxius has joined.
00:55:02 -!- Jafet has joined.
00:55:07 -!- Frooxius has quit (Read error: Connection reset by peer).
00:55:21 -!- Frooxius has joined.
01:04:12 <Sgeo> Why am I interested in Erlang now?
01:04:54 <kallisti> though now I'm taking a sudden interest in perl 6.
01:30:35 -!- myndzi has joined.
01:35:11 -!- sebbu2 has changed nick to sebbu.
01:44:48 -!- yorick has quit (Ping timeout: 252 seconds).
01:58:27 -!- yorick has joined.
02:14:15 -!- zzo38 has joined.
02:24:24 <zzo38> I changed my CatArrow class slightly. Now its methods are arr1, arr2, (*+*), (&|&), and fanout. The minimum definition is (*+*) and fanout.
02:24:25 <lambdabot> zzo38: You have 3 new messages. '/msg lambdabot @messages' to read them.
02:24:31 <lambdabot> oerjan said 4d 20h 2m 8s ago: Sounds like you could generalize the monoid for your pure BF implementation, e.g. make a paintfuck implementation with the (Sum Int, Sum Int) monoid :P
02:24:31 <lambdabot> oerjan said 2d 17h 30m 37s ago: There is reallyunsafeptrequality#, but as the name says that's really unsafe
02:24:31 <lambdabot> oerjan said 2d 17h 30m 8s ago: *reallyUnsafePtrEquality#
02:33:33 <Sgeo> Presumably, I wouldn't just write a BF interpreter in Erlang, I'd write some sort of BF server? :D
02:42:39 <tswett> http://beta.yle.fi/pressanvaalit_2012/ explain please thank you.
02:47:52 -!- Klisz has joined.
02:50:47 * Sgeo ponders what a BF server would look like
02:58:18 <Sgeo> Esolang server
02:59:44 <zzo38> The only one I know of that uses a server is Network Headache, and the server is no longer online
03:06:24 <Sgeo> It's been implemented before
03:06:31 <Sgeo> I guess no reason not to implement it again
03:06:51 <kallisti> ...I think we know what needs to be done.
03:08:32 <Sgeo> I still don't thoroughly understand the OTP
03:11:33 <zzo38> I made up a CatArrow for Lens: instance CatArrow Lens (,) where { Lens x *+* Lens y = Lens $ \(a, b) -> store (\(a', b') -> (peek a' $ x a, peek b' $ y b)) (pos $ x a, pos $ y b); fanout = Lens $ \x -> store snd (x, x); };
03:12:10 <zzo38> But it is not Arrow because there is no functor from (->) to Lens
03:12:47 <zzo38> Do you think this instance is correct?
03:32:30 * Sgeo wonders what Erlnomic would be like
03:32:45 <Sgeo> Hardest thing I can think of is making it safely hold onto untrusted code
03:42:07 -!- cswords has joined.
04:06:21 -!- augur has joined.
04:11:52 -!- augur has quit (Remote host closed the connection).
04:16:54 -!- augur has joined.
04:17:37 -!- augur has quit (Remote host closed the connection).
04:18:26 -!- augur has joined.
04:50:13 -!- NihilistDandy has joined.
06:37:44 -!- Sgeo has quit (Read error: Connection reset by peer).
06:39:27 -!- Sgeo has joined.
06:42:24 -!- pikhq has joined.
06:45:04 -!- pikhq_ has quit (Ping timeout: 245 seconds).
07:01:30 -!- Klisz has quit (Quit: SLEEP, GLORIOUS SLEEP).
08:04:32 -!- NihilistDandy has quit (Quit: Textual IRC Client: http://www.textualapp.com/).
08:32:31 -!- Frooxius_ has joined.
08:32:31 -!- Frooxius has quit (Read error: Connection reset by peer).
08:32:34 -!- Frooxius_ has changed nick to Frooxius.
08:56:45 -!- cheater has quit (Ping timeout: 255 seconds).
09:13:48 -!- Jafet has quit (Quit: Leaving.).
09:30:17 -!- Jafet has joined.
09:42:54 -!- cheater has joined.
09:44:32 -!- zzo38 has quit (Remote host closed the connection).
10:13:04 -!- zzo38 has joined.
10:13:34 <zzo38> How can I make robots.txt tell to disallow everything except wget, in which case it is delayed half a second per request, unless the user overrides that option on the wget command-line?
10:19:56 <zzo38> Actually I think I figure out
10:44:57 -!- pikhq_ has joined.
10:45:04 -!- pikhq has quit (Ping timeout: 245 seconds).
10:45:44 <zzo38> wget does not understand Crawl-delay but that is OK; it is in there in case a later version does support it
10:53:15 <Jafet> So anyone using some other crawler will have to tell it... to impersonate wget.
10:53:46 <shachaf> Jafet: Well, or just to ignore robots.txt
10:55:11 <zzo38> What others are there anyways, that are used manually similar to wget?
11:14:55 -!- oerjan has joined.
11:46:53 -!- Phantom_Hoover has joined.
12:16:04 -!- zzo38 has quit (Remote host closed the connection).
12:33:08 <oerjan> `run echo "A lens is a monoidal natural transformation between higher-order coalgebra functors" >wisdom/lens
12:33:30 <HackEgo> A lens is a monoidal natural transformation between higher-order coalgebra functors
12:38:26 * Sgeo doesn't know enough to know whether that makes sense or is gibberish
12:39:02 <oerjan> i _think_ it's actually true, although it's half a quote from haskell weekly news so probably said in jest
12:39:28 <oerjan> the dual of an algebra, of course!
12:40:12 <oerjan> i don't know precisely what they are, but i think comonads correspond to coalgebras in the way monads correspond to algebras, somehow
12:40:29 <kallisti> what is the category of an algebra?
12:41:04 <oerjan> algebraic structures and algebra homomorphisms is one, at least
12:41:07 <kallisti> because I thought comonad had to do with "dual category" not "dual monad" so it would make sense that coalgebra had a similar relation.
12:41:18 <fizzie> Sgeo: I don't think curl(1) reads robots.txt at all.
12:46:14 * Sgeo is going to write one final piece of Erlang tomorrow than is swearing it off
12:46:35 <Sgeo> Apparently, the hot swapping stuff is very tricky to get right, and if it's avoidable, it's best avoided.
12:47:31 <fizzie> On the other hand, why would you even be doing Erlang if you're not going to be all fancy and hot-swappy?
12:48:01 <Phantom_Hoover> <kallisti> because I thought comonad had to do with "dual category" not "dual monad" so it would make sense that coalgebra had a similar relation.
12:48:14 <Phantom_Hoover> A comonad is a monad in the dual category, i.e. a dual monad.
12:48:42 <Sgeo> fizzie, because I thought it would make hot-swapping easy
12:49:14 <fizzie> A duel monad is deadly.
12:49:15 <kallisti> Sgeo: yes that's the conclusion I reached about the hot swapping
12:50:09 <kallisti> other reasons to use Erlang: it handles distributed programming almost automatically for you
12:50:15 <Phantom_Hoover> kallisti, you know, you're not fooling anyone when you say 'yes' after something's explained to you.
12:50:28 <kallisti> Phantom_Hoover: correct. there is nothing to be fooled about.
12:51:21 * kallisti will say "yes" in a way that implies something different to Phantom_Hoover next time.
12:53:23 <kallisti> Phantom_Hoover: I was not confused about the meaning of dual monad, I was more confused about if coalgebra meant "dual algebra" or "the algebra of a dual something-else." does that make things clearer?
12:53:38 <kallisti> I wasn't sure what co- meant in this context.
12:53:54 <Phantom_Hoover> It makes some things clearer, although they were already quite clear.
12:54:32 <oerjan> co- means you turn some arrows around. exactly _which_ arrows, is anyone's guess.
13:00:07 <Phantom_Hoover> I used to be able to work out categoric duals. But then I took an arrow to the knee.
13:00:47 <kallisti> ...........................................................................................................................................................................................................................................................................................................................................................................................................................................
13:02:06 <shachaf> I wasn't sure what - meant in this ntext.
13:02:51 <oerjan> shachaf: what is this about haskell types being useless and stupid?
13:03:51 <oerjan> "shachaf: Haskell's type system is the perfect mix of useless and stupid."
13:04:11 <monqy> i agree with kallisti
13:04:20 <monqy> ........................................................................................................................................................................................................................................................................................................................................................................................................................................... inde
13:04:53 <shachaf> That was @remembered out of context.
13:05:01 <shachaf> 19:38:03 <Philippa> tigger: for what it's worth, it's a more than fair question to have :-) Just the sort of thing that needs an even smarter type system than Haskell's, or a less safe one
13:05:47 <shachaf> 19:25:58 <dmwit> tigger: (\[a, b, c, d, e] -> construct a b c d e) <$> replicateA 5 parseByte
13:06:04 <shachaf> People were talking about how using variable-length lists like that was unsafe, and the constraint should be verified in the type system.
13:06:10 <kallisti> Phantom_Hoover: something passive aggressively directed at both you and Phantom_Hoover simultaneously while also commending monqy for his like-minded attitude.
13:07:12 <oerjan> well i have to ban you now, otherwise elliott might think i'm no longer power tripping and come back.
13:07:42 <kallisti> it's a wonderful circle of passive aggression. so good.
13:08:01 <oerjan> an inward spiral, more like.
13:08:14 <shachaf> elliott thinks you're power-tripping?
13:08:32 <kallisti> because he got +q'd for spamming the channel with pages of the word "really"
13:08:37 <oerjan> shachaf: well he said that while rageparting, anyway
13:08:48 <kallisti> yes he was likely just angry about it.
13:09:51 <oerjan> shachaf: clearly what they need is the unsafeCoerce in Agda mentioned in another hwn quote
13:10:20 <kallisti> fizzie: I've lately found myself attempting to apply Bayesian inference to any problem requiring any remote degree of pattern recognition
13:10:54 <kallisti> what other techniques are there that are similar?
13:11:17 <shachaf> oerjan: I was quoted in HWN? :-(
13:11:41 <oerjan> how otherwise would i find out about quotes
13:12:12 <shachaf> I like how http://www.haskell.org/haskellwiki/HWN is horribly out of date.
13:12:23 -!- Phantom_Hoover has quit (Remote host closed the connection).
13:12:42 <shachaf> Someone add http://contemplatecode.blogspot.com/ , please.
13:13:26 -!- Phantom_Hoover has joined.
13:13:32 <oerjan> either they just did, or you missed it
13:13:38 <kallisti> hmmm not familiar with ntemplatecodes and their duals.
13:15:01 <oerjan> it's possible some of the other links are broken, of course
13:15:18 -!- derdon has joined.
13:16:01 <lambdabot> shachaf says: Sufficiently advanced misunderstanding is indistinguishable from trolling.
13:16:53 <kallisti> I think a distinction needs to be made between good trolling and bad trolling.
13:17:01 <kallisti> I think some trolling is quite obvious.
13:17:38 <kallisti> there's generally any combination of (in)effective and (non-)obvious
13:18:18 <lambdabot> oerjan says: i only do impractical things
13:18:22 <lambdabot> oerjan says: i only do impractical things
13:18:24 <lambdabot> oerjan says: i only do impractical things
13:18:24 <lambdabot> oerjan says: i only do impractical things
13:18:28 <fizzie> kallisti: Thomas Bayes is your Lord thy God, and you shalt not have any other ones. Why would you want to do anything that's *not right*? (There are quite many ways to actually *do* Bayesian inference for all the interesting problems where you can't ever solve anything, though.) ((I guess you might count the MDL principle a sensible-ish "non-Bayesian" tool.))
13:18:28 <shachaf> I guess that's all you do.
13:18:34 <HackEgo> 7) <oerjan> what, you mean that wasn't your real name? <Warrigal> Gosh, I guess it is. I never realized that. \ 17) <oerjan> ehird has gone insane, clearly. \ 19) <fungot> oerjan: are you a man, if there weren't evil in this kingdom to you! you shall find bekkler! executing program. please let me go... put me out! he's really a tricycle! pass him! \ 22) <oerjan> In an alternate universe, ehird has taste \ 23) IN AN
13:18:42 <HackEgo> 7) <oerjan> what, you mean that wasn't your real name? <Warrigal> Gosh, I guess it is. I never realized that. \ 17) <oerjan> ehird has gone insane, clearly. \ 19) <fungot> oerjan: are you a man, if there weren't evil in this kingdom to you! you shall find bekkler! executing program. please let me go... put me out! he's really a tricycle! pass him! \ 22) <oerjan> In an alternate universe, ehird has taste \ 23) IN AN
13:18:50 <shachaf> Is `quote <username> sequential?
13:19:04 <HackEgo> 617) <shachaf> elliott: GHC bug? Come on, it's the parentheses. <shachaf> The more parentheses you add, the closer it is to LISP, and therefore the more dynamically-typed. \ 663) <shachaf> Real Tar is GNU tar. <shachaf> You just ignore whichever features don't make you feel superior enough. \ 701) <shachaf> VMS Mosaic? <shachaf> I hope that's not Mosaic ported to VMS. <shachaf> Hmm. It's Mosaic ported to VMS.
13:19:08 <oerjan> shachaf: `quote doesn't randomize with an argument.
13:19:11 <HackEgo> 131) <coppro> what's the data of? [...] <Sgeo> Locations in a now deceased game called Mutation <coppro> I have no problems with you being interested in online games <coppro> but the necrophilia is disturbing
13:19:33 <kallisti> fizzie: the different techniques are essentially which models/events you use, I guess?
13:19:43 <oerjan> shachaf: there is `pastequotes when you want all of them
13:20:08 <HackEgo> 347) <oklofok> destroying a local copy of the world is kind of like raping a robochick with a shovel tho
13:20:09 <HackEgo> http://codu.org/projects/hackbot/fshg/index.cgi/raw-file/tip/paste/paste.4501
13:20:15 <HackEgo> http://codu.org/projects/hackbot/fshg/index.cgi/raw-file/tip/paste/paste.10378
13:21:37 <oerjan> well, unless he has fled recently, and that's why we don't see him.
13:22:14 <kallisti> fizzie: but the main idea would appear to be that you iterate conditional probabilities upon new events.
13:22:30 <kallisti> are there more specific techniques with... names... I could learn?
13:27:02 <fizzie> kallisti: Some keywords would be the EM (expectation-maximization) algorithm (except you can never get closed-form solutions for anything ever, except maybe for some Gaussians); the variational Bayesian inference; all the Monte Carlo -style sampling-based approaches (Gibbs sampling, Metropolis-Hastings sampling); and then all the things where you do a Bayesian network graphical model and use one of the graph-based inference thingies (exact, which you prob
13:27:23 <fizzie> ... which you probably can't do to anything interesting; or inexact, like "X belief propagation" for all kinds of X (e.g. loopy)).
13:27:39 <fizzie> That's quite a few names.
13:28:54 * kallisti wonders if the AI course at his university is actually good.
13:28:59 <kallisti> I doubt it. nothing here has been good so far.
13:29:08 <kallisti> data structures wasn't bad though.
13:29:35 <fizzie> I happen to like the name "loopy belief propagation" a lot, even if it's kinda fuzzy thing, and you're always kind-of wondering whether it'll converge or not.
13:29:42 <fizzie> Still, it sounds happy.
13:30:39 <fizzie> (It's just belief propagation except your graph's got some cycles. Belief propagation converges to an exact solution in a tree, or some-such.)
13:30:49 -!- atehwa has quit (Ping timeout: 240 seconds).
13:31:00 <fizzie> (Quite a lot of people are very enthusiastic about graphical models.)
13:33:08 <fizzie> As for AI courses, at least ours isn't really "machine-learningy" at all, but it's still sort-of interesting stuff, maybe. Search things (A* and the like) and logic-programming related topics, and then assorted odd bits.
13:33:19 <fizzie> (Also they're going to scrap it after this year.)
13:34:05 -!- augur has quit (Remote host closed the connection).
13:34:38 <oerjan> the robots told them to scrap it, or else...
13:34:49 <kallisti> I think I'm more interested in machine learning than any of that stuff.
13:35:04 <kallisti> which I've been learning a little about.
13:35:04 -!- sebbu2 has joined.
13:35:05 -!- sebbu2 has quit (Changing host).
13:35:05 -!- sebbu2 has joined.
13:35:08 <fizzie> Very possible. People keep asking me what's replacing it, and I have to be all "well uh..."
13:35:13 <fizzie> (Since nothing really is.)
13:36:31 <kallisti> I'm also using LogicT in my portal chess implementation, but I'm really just using it as a glorified (and more efficient) list monad.
13:37:16 <fizzie> It's just that the professor doing the courses from that... uh, professorship, retired a couple of years ago, and the CSE department (which is so much about engineering it's probably best called CsE or something) doesn't feel like keeping those things running; while our "actual computer *science*, guys" department -- formed by forcibly combining the machine learning lab and the theory lab -- doesn't have an equivalent "introductionary" and wide-scope thin
13:38:19 <kallisti> my school is a small engineering school based off of Georgia Tech. so the entire curriculum is focused on engineering mostly.
13:38:36 -!- sebbu has quit (Ping timeout: 240 seconds).
13:39:47 <kallisti> I've decided to get a dual major in math because I feel my theoretical education is lacking. Also it probably looks good on a resume. :P
13:40:08 <kallisti> HOWEVER. I have to continue with calculus which I have forgotten ridiculous amounts of stuff about...
13:41:28 <oerjan> georgia tech, that's where lipton of godel's lost letter is
13:41:43 <kallisti> so I'll be jumping into Calc 3 after like 2-3 years of forgetting about Calc 2 (not to mention my Calc 2 teacher was absolutely terribly and never got beyond the first two chapters, and was later fired I found out)
13:42:17 <kallisti> so I don't think I ever properly learned all of the integration methods that you're typically supposed to learn.
13:42:44 <kallisti> and I definitely know nothing about convergence/divergence of infinite series.
13:42:52 -!- atehwa has joined.
13:43:31 <kallisti> but, I think I can learn these things pretty fast.
13:44:09 * kallisti is much better at differential equations than integration.
13:45:08 <oerjan> ...you need the latter for the former :P
13:45:48 <oerjan> although probably the exercises for the former will not use the hard integrals
13:46:18 <fizzie> oerjan: Or alternatively they'll be so hard you can't do closed-form solutions.
13:46:48 <kallisti> however a dual major is quite a lot of extra classes. definitely an extra year.
13:46:55 <itidus20> kallisti: maybe you need to re-learn calc 2 right now
13:47:19 <kallisti> hm I might still have my calc book actually.
13:47:30 <kallisti> somewhere. those things are hard to sell.
13:48:56 <itidus20> i was thinking that if you took a unit line, and curved it into a circle with endpoints meeting, and then scaled that circle up until the radius was a unit circle
13:49:09 <itidus20> then pi would be the scaling factor of that circle
13:49:42 <kallisti> that is basically how pi is defined. pi = C/d
13:50:19 <kallisti> a unit circle has a circumfrence of pi.
13:50:58 <itidus20> but it leaves me as puzzled as ever as to why it is irrational :D
13:51:07 <itidus20> but the whole world would collapse if pi was rational
13:51:27 <oerjan> a dual major is just a major in the opposite category, right?
13:52:15 <kallisti> itidus20: pi is irrational because it isn't expressable as a ratio of two integers.
13:52:33 <itidus20> Phantom_Hoover: that was my way of ascertaining that circles can exist
13:52:47 <itidus20> rather than the idea that every circle is a set of straight lines
13:53:08 <kallisti> basically this is saying that both the circumfrence of a circle and its diameter cannot both be integers.
13:53:16 <oerjan> itidus20: the proof that pi is irrational is fairly advanced
13:53:18 * oerjan doesn't actually know it
13:53:32 <oerjan> e is much easier, i understand
13:53:55 <Phantom_Hoover> http://en.wikipedia.org/wiki/Proof_that_22/7_exceeds_%CF%80
13:54:19 * kallisti knows a number of people who were convinced that 22/7 = pi
13:54:20 <fizzie> We had this "Matrix Computations" course, it was the funniest. There was a one lecture's worth of motivation of how Google is just a numerical linear algebra problem (apparently you can formulate PageRank in terms of an internet-sized matrix), real highfalutin stuff; and then the rest was just matrix decompositions and "show this and that for a Hermitian positive semidefinite P" and even few bits of floating-point error estimation of all things.
13:55:57 <fizzie> Even formulate PageRank as an eigenvalue problem of that huge honking matrix, if I recall it right.
13:56:29 <oerjan> we are all in that matrix
13:56:47 <fizzie> Locked in it, one might say.
13:57:18 <itidus20> ... rears it's ugly head again
13:57:48 <oerjan> it never really went away
13:58:41 <itidus20> if you have the circumference of 1 circle
13:58:48 <itidus20> and the circumference of another circle
14:00:03 <itidus20> like you can start with the unit circumference... and then scale the circle up to a circumference of 2
14:00:16 <itidus20> but you never know how far you scaled by :-?
14:00:44 <oerjan> i was with you until that last line
14:01:03 <itidus20> i was not with me even tat far
14:01:21 <itidus20> i know its not possible that i can think of anything new in pi
14:01:37 <fizzie> oerjan: You were hoping for some topology, right? They're always turning things to other things.
14:02:02 <itidus20> but still.. the ratio of the circumferences of 2 circles must be quite fascinating
14:02:18 <oerjan> the topology i know doesn't involve much pi
14:02:19 <fizzie> It's approximately as fascinating as the ratio of the diamaters of the circles.
14:03:07 <fizzie> Yes, well, C1=pi*d1, C2=pi*d2, C1/C2 = (pi*d1)/(pi*d2) = d1/d2, you know.
14:03:27 <oerjan> the ration of the circumferences is the same as the ratio of the diameters. i think archimedes proved that, or something.
14:04:05 <oerjan> thus proving pi existed as something common for all circles
14:04:22 <fizzie> *My* circles have a pi of 4, so there.
14:04:27 <itidus20> so its not a useless notion :D
14:04:36 <itidus20> only an extremely redundant one
14:05:06 <fizzie> (Sadly my circles aren't very round, either.)
14:06:09 <oerjan> we call those pi = 4 circles "squares" around here
14:06:36 <fizzie> I recently read about the ballsiest balls humankind has ever produced.
14:06:52 <fizzie> "The gyroscopes in GP-B are the most perfect spheres ever made by humans. (The experiment actually carries four gyroscopes for redundancy.) These ping pong-sized balls of fused quartz and silicon are 1.5 inches across and never vary from a perfect sphere by more than 40 atomic layers. That means that if these gyroscopes were the size of the Earth, the elevation of the entire surface would vary by no more than 12 feet!"
14:07:17 <fizzie> Sadly I couldn't find a "buy now" link.
14:07:40 <oerjan> and _still_ the experiment essentially failed (not totally, but enough that other experiments had surpassed them in quality before the results were out)
14:07:41 -!- cheater has quit (Ping timeout: 244 seconds).
14:07:50 <fizzie> And I guess it might get dented easily. But still.
14:07:58 <fizzie> Could be; I was just interested in their balls.
14:08:34 <itidus20> 1am and cold.. time for my supreme commander jacket
14:08:44 <oerjan> the problem was something about unforeseen electric or magnetic fields on the balls
14:09:25 <oerjan> they did spin it as a success, though
14:11:21 <fizzie> "Science journals have expressed negative overviews, citing the $760,000,000 cost and that prior experiments had made 130-150 times more accurate measures over six years earlier."
14:11:26 <fizzie> That's quite some success.
14:11:39 <fizzie> But at least they had balls.
14:13:33 -!- kallisti has quit (Ping timeout: 255 seconds).
14:13:41 -!- augur has joined.
14:14:10 <itidus20> so could we make a topology where C/d = 3?
14:16:02 <Phantom_Hoover> I asked oerjan and oklopol and both said they didn't know.
14:16:36 <oerjan> you can sort of think of a hexagonal tiling as having that
14:17:31 <oerjan> but can that be made arbitrarily small...
14:20:13 -!- oerjan has quit (Quit: Later).
14:29:44 <fizzie> How about just starting to speak of circles in spherical geometry instead? If you look at circumference/diameter, you just need a circle of suitable size to get c/d = 3.
14:45:19 -!- augur has quit (Remote host closed the connection).
14:52:10 -!- Frooxius has quit (Read error: Connection reset by peer).
14:52:37 -!- Frooxius has joined.
14:56:19 -!- itidus20 has quit (Ping timeout: 245 seconds).
15:00:37 -!- Vorpal has joined.
15:09:21 -!- Phantom_Hoover has quit (Ping timeout: 276 seconds).
16:12:39 -!- derdon_ has joined.
16:15:31 -!- derdon has quit (Ping timeout: 245 seconds).
16:34:49 -!- Phantom__Hoover has joined.
16:43:23 -!- kallisti has joined.
16:43:24 -!- kallisti has quit (Changing host).
16:43:24 -!- kallisti has joined.
16:46:07 <kallisti> I think >>= probably easier to understand in terms of fmap and join
16:47:10 <kallisti> the problem is that newcomers perhaps won't recognize the immediate connection on why >>= appears to be "central" to Haskell's framing of monads.
16:48:18 <kallisti> perhaps one angle of approach would be to explore so-called monadic functions, and how they frequently arise in various, seemingly unrelated problems.
16:49:18 <kallisti> of course m a -> m a is an equally important case, so functors should be discussed in detail as well, probably before monads. This approach isn't really anything new. Most well-written tutorials begin with functors first.
16:51:06 <Vorpal> strange, ubuntu decided to update firefox to the last version on the 10.04 LTS release. (And no I don't have the backport repo enabled)
16:51:09 <kallisti> I think the list monad is actually a good candidate as a starting example. Firstly, because it transitions well from discussing Functors, where the easiest example is lists, because by then you're probably already familiar with map. Secondly, it's not quite as trivial as monads like Maybe, and not as magical as IO. It does something pure under the hood.
17:31:56 <kallisti> so to describe the denotational semantics of Haskell
17:32:22 <kallisti> do you normally describe the semantics of Core or STG, and thus equivalently Haskell
17:32:53 <kallisti> or, would you denote the semantics of Haskell itself.
17:33:18 <kallisti> I think Core would probably be a better candidate than STG
17:34:08 <kallisti> because it's not quite as different from Haskell as STG is.
17:36:38 <kallisti> actually I guess that isn't really necessary
17:37:23 <kallisti> probably more work if you want to be rigorously correct
17:48:34 -!- augur has joined.
17:51:19 -!- Phantom__Hoover has quit (Read error: Connection reset by peer).
17:54:05 <kallisti> ah, I just learned what monotone means. :>
18:00:06 <fizzie> Vorpal: What was it before in 10.04?
18:00:35 -!- Klisz has joined.
18:01:49 <fizzie> Since the post-3 versions aren't really "major versions" in which you could sensibly stay in K.x instead of going to (K+1).y, at least if you expect any security patches.
18:14:56 -!- pikhq has joined.
18:15:04 -!- pikhq_ has quit (Ping timeout: 252 seconds).
18:22:28 <kallisti> http://upload.wikimedia.org/wikibooks/en/math/9/d/7/9d767770a5813fd4292965450c4b3092.png
18:22:36 <kallisti> what does the big arrow thing mean in this?
18:24:29 -!- pikhq has quit (Ping timeout: 240 seconds).
18:24:35 -!- pikhq has joined.
18:25:14 -!- ais523 has joined.
18:25:56 <fizzie> Are you trying to hide you're reading wikibooks/Haskell/Denotational_semantics there?
18:26:25 <fizzie> Guilty pleasures and all that.
18:28:15 -!- ais523 has quit (*.net *.split).
18:28:17 -!- Sgeo has quit (*.net *.split).
18:28:17 -!- cswords has quit (*.net *.split).
18:28:17 -!- Nisstyre has quit (*.net *.split).
18:28:18 -!- lambdabot has quit (*.net *.split).
18:28:18 -!- kmc has quit (*.net *.split).
18:28:18 -!- clog has quit (*.net *.split).
18:28:19 -!- quintopia has quit (*.net *.split).
18:28:20 -!- quintopia has joined.
18:28:21 -!- clog has joined.
18:28:30 -!- cswords has joined.
18:28:40 -!- Sgeo has joined.
18:28:45 -!- kmc has joined.
18:29:25 -!- ais523 has joined.
18:29:51 -!- Nisstyre has joined.
18:31:57 -!- lambdabot has joined.
18:33:31 <Deewiant> kallisti: If g is the Y combinator and f factorial, it's the same as the arrow in Haskell's lambda abstractions like (\x -> x). (Does this arrow have a name?)
18:34:22 <fizzie> Deewiant: U+21A6 RIGHTWARDS ARROW FROM BAR. Alias names: z notation maplet.
18:34:47 <Deewiant> fizzie: I meant the arrow in \x -> x.
18:34:49 -!- _Slereah has quit (Ping timeout: 260 seconds).
18:35:09 <Deewiant> fizzie: (And yes, I know it's HYPHEN-MINUS followed by GREATER-THAN SIGN.)
18:35:22 <fizzie> I had *just* finished writing that.
18:35:32 <kallisti> Deewiant: so it defines a function
18:35:42 <Deewiant> kallisti: That's what it looks like.
18:36:04 <Deewiant> kallisti: So the pic says "f equals g(f) equals \n ->" etc.
18:38:05 <Deewiant> > let g = fix; f fac n = if n == 0 then 1 else n * fac (n-1) in g f 5
18:38:13 <kallisti> yeah it's not quite fix though
18:40:21 -!- pikhq_ has joined.
18:40:29 -!- pikhq has quit (Ping timeout: 244 seconds).
18:41:21 -!- _Slereah has joined.
18:44:09 -!- Zuu has quit (Ping timeout: 252 seconds).
19:31:26 <ais523> I got a reply back from Graue, apparently he doesn't have time to reconfigure MediaWiki any more
19:53:19 -!- Ngevd has joined.
19:54:29 -!- Phantom_Hoover has joined.
19:55:48 -!- Ngevd has quit (Read error: Connection reset by peer).
20:00:21 -!- calamari has joined.
20:00:36 -!- quintopia has quit (Ping timeout: 252 seconds).
20:09:01 -!- quintopia has joined.
20:34:44 -!- Zuu has joined.
20:36:42 -!- Ngevd has joined.
20:42:31 -!- oerjan has joined.
20:54:37 -!- Taneb has joined.
20:57:28 -!- Ngevd has quit (Ping timeout: 248 seconds).
21:06:48 -!- azaq23 has joined.
21:13:30 -!- calamari has quit (Quit: Leaving).
21:14:38 -!- oerjan has set topic: Now open for Vampires | Take a trip to see our Castles | Get your esoblood here! | Claims elliott shortage hurts productivity "Absurd, bleh" | Glorious optators "worried, bleh" about sudden population decline | Based on the power of immortality! | Spotted marsh elliott spotted in other marsh | http://codu.org/logs/_esoteric/.
21:16:58 -!- TeruFSX_ has quit (Ping timeout: 248 seconds).
21:26:32 -!- augur has quit (Remote host closed the connection).
21:26:47 <pikhq_> It's kinda weird how slow conversation is in here sans elliott.
21:31:52 <ais523> btw, I think Backslash Calculus's BF-like representation is hilarious
21:31:59 <ais523> and secretly hope it's designed to troll Phantom_Hoover
21:32:28 <Phantom_Hoover> <pikhq_> It's kinda weird how slow conversation is in here sans elliott.
21:33:23 -!- Taneb has quit (Read error: Connection reset by peer).
21:33:26 -!- _Slereah has quit (Ping timeout: 252 seconds).
21:37:33 -!- _Slereah has joined.
21:44:57 -!- pikhq has joined.
21:45:02 -!- pikhq_ has quit (Ping timeout: 252 seconds).
21:45:19 * oerjan hazards a guess on 14/1 or 15/!
21:47:17 * oerjan swats oklopol -----###
21:47:41 <oklopol> why would you guess something about a date and not tell me what it is
21:48:05 <oerjan> 22:32 Phantom_Hoover> <pikhq_> It's kinda weird how slow conversation is in here sans elliott.
21:48:08 <oerjan> 22:32 Phantom_Hoover> Check the tunes logs sizes for 14/1 and 15/1.
21:48:12 <oerjan> 22:44 oklopol> when did elliott huff?
21:48:38 <oerjan> 22:44 * oerjan hazards a guess on 14/1 or 15/!
21:49:06 <oklopol> btw you accidentally pasted some of my and Phantom_Hoover's lines too
21:49:16 <oklopol> but i guess you noticed that too
21:49:23 -!- ChanServ has set channel mode: +o oerjan.
21:49:25 <oklopol> anyway nm, i'll keep thinking about your guess
21:49:49 -!- oerjan has kicked oklopol Trying to get this dense object out of the channel before we get swallowed by it.
21:49:59 -!- oklopol has joined.
21:49:59 -!- oerjan has set channel mode: -o oerjan.
21:50:33 <oklopol> "<HackEgo> 69) <oklopol> Warrigal: what do you mean by 21?"
21:50:56 <ais523> hmm, I don't really mind the channel being more relaxed
21:51:12 <HackEgo> 69) <oklopol> Warrigal: what do you mean by 21?
21:51:21 <ais523> it's strange without elliott, but I respect his decision to leave, and the channel's still useful for its intended purpose and at least one of its secondary purposes
21:51:36 <oklopol> yeah elliott was a bit too serious about esolangs really, it became all about the mone
21:52:12 <oklopol> well without elliott, who's going to keep my quote rate up?
21:52:48 <oerjan> oklopol: i think if he were here, then that one above would get killed in one of his quote purges
21:53:32 -!- quintopia has quit (Ping timeout: 255 seconds).
21:53:41 <oklopol> yeeeeah elliott would totally remove one of my quotes
21:55:54 <HackEgo> 76) <Warrigal> What do you call the husband of my first cousin once removed? <apollo> Warrigal: "Hey, Sexy."
21:56:06 <HackEgo> 173) <pikhq> Vorpal: YOU ARE AMERICAN
21:56:06 <HackEgo> 688) <Phantom_Hoover> I'd insult you behind your back, but I don't care which side of your back I insult you on.
21:56:55 <HackEgo> 74) <Warrigal> Invalid! Kill! Kill! <Aftran> I get that feeling too.
21:56:55 <HackEgo> 116) <DoctorDog> I am an inherently pornographic being.
21:57:14 -!- quintopia has joined.
21:57:21 <ais523> bleh, those are pretty much all bad, apart from 688
21:57:22 <oklopol> "<tswett> A dynamical system is minimal if every point's orbit is dense? That sounds... like a strange thing to say." yeah, when would this ever be worth mentioning
21:57:25 <ais523> opinions on which I should delete?
21:57:39 <HackEgo> 617) <shachaf> elliott: GHC bug? Come on, it's the parentheses. <shachaf> The more parentheses you add, the closer it is to LISP, and therefore the more dynamically-typed.
21:57:52 <HackEgo> 673) <Phantom_Hoover> You know what annoys me about Deep Space 9. <Phantom_Hoover> It wasn't in deep space. <Phantom_Hoover> It was orbiting Bajor.
21:57:58 <oerjan> i didn't think they were that bad...
21:58:39 <oklopol> oh oerjan mentions it in his thing?
21:58:50 <HackEgo> *poof* <Warrigal> What do you call the husband of my first cousin once removed? <apollo> Warrigal: "Hey, Sexy."
21:58:55 <oklopol> wow i wonder how similar our maths are/were
21:58:56 <ais523> FIVE QUOTES ENTER, FOUR QUOTES LEAVE
21:59:32 <oklopol> oerjan: did you ever do stuff with the cantor-bendixon rank? i solved an open problem about 2D SFTs and that thing today
21:59:44 <oklopol> (and another problem that was actually hard)
21:59:54 <oerjan> cannot say i recall ever hearing the term
22:00:28 <oklopol> do you know what the derivative of a topological space is? in any case it's just that you remove the isolated points
22:00:41 <oklopol> do you know what a 2d sft is?
22:01:02 <oerjan> i'd guess sft is "shift of finite type"
22:01:04 -!- augur has joined.
22:01:15 <oklopol> that's always used in symbolic dynamics as i know it
22:01:19 <oerjan> which i've heard, although not done work on
22:01:35 <oklopol> well that and sofic shifts is usually what i do
22:01:39 <oerjan> since iirc they're like the opposite of minimal
22:02:00 <oklopol> but usually you at least work with transitive ones
22:02:25 <oklopol> which is a bit like minimality...
22:03:14 <oklopol> in any case there was this open problem whether there's a countable 2D SFT such that for some ordinal n, the nth derivative is empty, but it's not empty for any finite ordinal
22:03:44 <oerjan> yes, i recall certain transitive systems other than minimal ones can be given bratteli-vershik diagrams
22:04:21 <oklopol> we also constructed a 1D subshift with rank omega^omega (rank is when the fixed point of derivation is reached)
22:04:48 <oklopol> (the cantor-bendixon rank that is. sometimes it also means when the thing becomes empty but that's just stupid.)
22:05:05 <oklopol> i still don't remember what a bratteli-vershik diagram is
22:05:41 <oklopol> have you done subshift stuff?
22:05:57 <oerjan> what was a subshift again
22:06:01 <oklopol> we've talked about this but i understood like 0
22:06:09 <oklopol> closed shift-invariant set
22:06:26 <oklopol> with the product topology on S^\Z
22:06:31 <oerjan> the toeplitz systems count as subshifts, then
22:06:52 <oklopol> like, every cell has a period, so to speak?
22:07:17 <oklopol> that's what toeplitz sequences are at least
22:07:33 <oerjan> yes, they're the subshifts generated by toeplitz sequences
22:07:34 <oklopol> i mean the set of them isn't
22:08:13 <oklopol> i've only encountered them because some dude ran CA on them with the besicovitch topology and proved some trivial stuff
22:09:49 <oklopol> i mean the space of toeplitz sequences (probably he too used its closure, not that it makes the slightest difference in this case probably)
22:10:04 <oklopol> toeplitz sequences are i've seen quite a few times
22:12:34 <oerjan> iirc, if you identify conjugate systems, subshifts correspond to expansive ones.
22:13:24 <oerjan> minimal dynamic system on a cantor set...
22:13:25 <oklopol> compact metric space plus homeomorphism?
22:14:34 <oerjan> where a cantor set is anything homeomorphic to the usual one
22:14:49 <oklopol> but so your dynamics is given by a homeomorphism?
22:15:19 <oerjan> basically, yes. although i also looked a little bit on using two.
22:15:22 <oklopol> otherwise expansive doesn't really mean anything but just checking
22:15:44 <oerjan> (not in relation to the expansive stuff, though)
22:16:19 <oerjan> we needed to put expansive as an explicit added requirement to our condition on the diagrams to get toeplitz systems
22:17:02 <oklopol> what do you call the branch of math that you did?
22:17:09 <oerjan> there's a picture in the paper of the diagram for a system which satisfies the rest but not expansivity
22:17:25 <oklopol> really this sounds exactly like what we're doing except you're perhaps doing it oh so differently.
22:18:28 -!- augur has quit (Remote host closed the connection).
22:18:34 <oerjan> minimal cantor dynamical systems, with slight generalization to all sides to see if something interesting cropped up?
22:18:58 <oklopol> but so i suppose your dynamics aren't in general cellular automata (that is, not shift commuting if you choose S^Z as the cantor set)?
22:19:02 <oerjan> also a lot about orbit equivalence and k-theory, since that's the fundamental theorem of my advisor
22:20:03 <oerjan> our dynamics _are_ the shifts
22:20:04 <oklopol> i suppose i mentioned there's a paper on k-theory on my desk since it's been there forever and i don't get it
22:20:53 <oklopol> so 2D subshifts, exactly what i worked on today
22:21:24 <oklopol> of course, these were not minimal systems.
22:21:45 <oerjan> more seriously, the dynamics are shifts _if_ we are looking at sequences, which we weren't in general
22:22:09 <oklopol> since minimal systems are transitive and transitive systems have rank 0
22:22:21 <oklopol> cantor-bendixon rank that is
22:22:34 <oerjan> but they're more analogous to shifts than to CA dynamics
22:23:11 <oklopol> if a natural dynamics is to appear
22:24:15 <oerjan> um we were looking at _all_ minimal systems by a single homeomorphism on cantor sets as our base case. no requirement that they were sequences
22:25:37 <oerjan> the bratteli-vershik diagrams are essentially a different way of naturally describing such dynamics, which don't need to be shifts
22:26:08 <oerjan> but will be conjugate to shifts iff the system is expansive
22:26:16 <oklopol> what i don't get is, if you take a homeomorphism with say S^\N from the cantor set, don't you get a homeomorphism on S^\N and some minimal subsystems of S^\N
22:26:26 <oklopol> what do you mean they don't have to be sequences?
22:27:05 <oklopol> maybe i don't know what you mean by a cantor set, if you cannot just simply move to S^\N directly.
22:28:14 <oklopol> i mean how do you represent your points even? and do you use cylinders?
22:28:32 <oklopol> i suppose i need to read your thesis, it seems that there's some fundamental difference in how you do things.
22:28:37 <oerjan> i mean they don't have to be subshifts.
22:29:41 <oerjan> in a sense they are sequences, since you can take the sequence of edges in the diagram corresponding to each point.
22:29:46 <oklopol> so you just have a self-homeomorphism on S^\Z or S^\N or w/e. i mean not that i have any idea what those actually look like ofc.
22:30:12 <oerjan> (an infinite path in an infinite graph)
22:30:44 <oklopol> wait, what are sequences in a sense? sorry for being dense.
22:31:07 <oerjan> the points in the systems described by BV diagrams
22:32:13 <oerjan> although, hm, there need be no common bound on the number of edges at all levels, so you don't necessarily get it into a finite alphabet
22:32:47 <oerjan> in fact i think we proved at one point that if there _is_ a common point, the system must be uniquely ergodic
22:33:43 <oklopol> so there was this rather old open problem whether a CA on S^\Z can be uniquely ergodic without all points going to zero after a finite amount of steps
22:34:03 <oklopol> my colleague mister x proved that in his master's thesis
22:34:27 <oklopol> so we've played with those too
22:35:00 <oklopol> unique ergods that is. (if a system is uniquely ergodic, it has unique ergods, as well all know)
22:37:18 <oklopol> for an obvious definition of ergod
22:38:00 <oklopol> not that being ergodic would then make sense for a measure :D
22:38:22 <oklopol> okay maybe i won't use this terminology ever agian
22:38:33 <oklopol> sooooooo about these BV diagrams
22:38:46 <oklopol> can you be all formal and shit
22:38:51 <oklopol> and tell me all about them
22:41:06 <oklopol> since if there's a unique invariant measure then that has to be ergodic and vice versa because all measures are integrals over ergodic ones IN SOME SENSE I THINK
22:41:36 <oklopol> where vice versa means if there's a unique (invariant) ergodic measure then it's also the only measure.
22:41:44 <oerjan> oh right, ergodic ones are extreme points in the convex set or something
22:43:17 <oklopol> i don't really recall the details since we mostly use this stuff as a tool in very simple ways.
22:44:00 <oklopol> like herp derp by birkhoffs ergodic theorem thingie let's take a point where things occur with nice densities.
22:44:21 <oklopol> although sometimes slightly less simple but rarely
22:49:03 * oerjan is wondering if he's starting to get the flu
22:49:33 <oerjan> it's apparently all the rage this week or so
22:50:13 <oklopol> http://www.mscand.dk/article.php?id=153 the fuck are minimal interval exchange transformations?
22:50:48 <oerjan> interval exchange transformation that happen to give minimal systems, i think
22:51:07 <oklopol> "Furthermore, these systems are orbit equivalent to Sturmian systems." you have worked with sturmian words?????
22:51:23 <oklopol> YOU HAVE DONE EVERYTHING THAT'S COOL EVER
22:52:47 <oklopol> "infinitesimal subgroup" :D
22:53:10 <oerjan> hm was that connected to the k-theory stuff...
22:53:18 <oklopol> please stop having been so awesome
22:53:44 <oklopol> i don't know i'm just skimming for cool terms.
22:54:10 <oerjan> hm it was the group elements that integrated to 0 wrt all invariant measures, wasn't it
22:54:23 <oklopol> dimension group :DDDD i played with those like 3 weeks ago
22:54:52 <oerjan> that was the word. they are what gives the k-theory.
22:55:26 <oklopol> a dynamical system has only one dimension group afaik
22:55:38 <oerjan> yes, the K_0 group corresponding to Cantor minimal systems is a dimension group
22:55:57 <oklopol> they have the same dimension group?
22:56:09 <oerjan> and divided out by the infinitesimals, it characterizes orbit equivalence
22:57:07 <oerjan> (dividing out by the infinitesimals ~ identifying any two group elements that have the same integral wrt all invariant measures)
22:57:11 <oklopol> when you say group elements that integ... blah blah
22:57:32 <oklopol> do you mean we have the dimension group as like fiber bundle thingies
22:58:13 <oklopol> i mean we usually define it with matrices for SFTs but for more general subshifts you can also talk about rays and beams and shit.
22:58:20 <oerjan> well iirc you can simply do C(X,Z), the continuous functions from the space to the integers, and divide those out by the functions that integrate to 0
22:59:00 <oerjan> although if you want the full dimension group, you divide out by something less, generated by shifting functions by the homeomorphism
22:59:44 <oklopol> well that continuous functions thing kind of makes sense as a more abstract version of the rays and beams and shit thing.
23:00:19 <oerjan> erm i'm sure there is always a connection to things but fiber bundles was never my thing.
23:00:36 <oklopol> i promised myself i'd actually learn some math this year, but every day i just start working on a new problem and lose interest in anything else.
23:01:19 <oerjan> i'm sure i've read about fiber bundles a number of times, but they somehow don't stick.
23:01:26 <oklopol> i don't mean what fibers mean in algebraic topology, and i don't mean that weird construction that may have a similar nam. only the terms ray and beam are used.
23:02:21 <oklopol> because they look like bundles of fibers. in my head.
23:03:49 -!- azaq23 has quit (Quit: Leaving.).
23:05:40 <oerjan> also my advisor and colleagues has since i left proved that these groups characterize orbit equivalence even when there is more than one (commuting) homeomorphism.
23:06:55 <oklopol> can you define orbit equivalence?
23:07:28 <oerjan> sure, it's a homeomorphism from the space of one system to the space of another that maps orbits onto orbits
23:07:55 <oerjan> but with no requirement that each orbit is mapped in sequence
23:08:11 <oklopol> yeah so conjugacy but lazier
23:09:24 <oklopol> because the def i guessed was that they are mapped in sequence. and then i was like wait this is conjugacy.
23:09:31 <oerjan> this automatically transfers the invariant measures between the systems
23:09:47 <oerjan> well, assuming countable orbits
23:10:19 <oklopol> i'm not sure what you mean by that
23:10:56 <oklopol> like, not having a homeomorphism whose Z-action gives the dynamics but having just R act on the space, or something like that?
23:11:05 <oerjan> or rather, a countable group acting (e.g. Z for a single homeomorphism)
23:12:41 <oklopol> and of course when i said "not Z-action but R-action", i meant the opposite, as you just said you want countable orbits.
23:12:41 <oerjan> it follows easily from this that the C(X,Z)/infinitesimals groups have to be isomorphic for an orbit equivalence to exist. the reverse however is very deep BV diagram magic.
23:13:55 <oklopol> is all this in your thesis?
23:17:21 <oklopol> WHY IS THERE SO MUCH COOL STUFF IN THE WORLD AND SO LITTLE SMART IN MY BRIAN
23:17:30 -!- Ngevd has joined.
23:18:01 <ion> I’m sorry to hear about your brian.
23:22:41 <oerjan> actually a considerable part of this BV and dimension group stuff is in my master's thesis, which unlike my PhD thesis i've got easily available and just converted to pdf: http://oerjan.nvg.org/main.pdf
23:23:09 -!- Ngevd has quit (Quit: Goodbye).
23:23:16 <oerjan> (don't worry about the norwegian on the first page :P)
23:25:05 <oerjan> my advisor told me he keeps my master's around as a basic reference
23:26:36 <oerjan> come to think of it, i may have a certain weakness for theorems with lists of equivalences :P
23:26:42 -!- Ngevd has joined.
23:26:56 -!- Ngevd has left.
23:27:52 <oklopol> this looks nice, i'll see if i understand any of it tomorrow
23:42:53 -!- derdon_ has quit (Remote host closed the connection).
23:46:49 -!- ais523 has quit (Remote host closed the connection).