←2010-02-17 2010-02-18 2010-02-19→ ↑2010 ↑all
00:00:33 <oklopol> idempotency is a special case of absorption
00:00:54 <cpressey> "So this can be used to make algrebraic structures which include absorption like idempotent semirings (which are also called dioids), but rings are not one of those structures." <-- I only mean that rings do not include the axiom a + a = a that is found in idempotent semirings
00:01:02 <augur> heyo
00:01:03 -!- Gracenotes has joined.
00:01:30 <cpressey> If R is a ring, it's not necessarily true that a + a = a for all a in R.
00:01:39 <oklopol> yeah it can't be true
00:01:50 <oklopol> because + forms an abelian group
00:01:51 <cpressey> Which is what I thought that statement said.
00:02:05 <cpressey> Rings are not one of those structures. That has a + a = a for all a.
00:02:07 <oklopol> i thought you meant absorption in multiplication
00:02:29 <cpressey> Idempotent semirings don't have that.
00:02:47 <oklopol> "In algebraic terms, the first program absorbs the second during sequential composition." <<< sequential composition = multiplication
00:03:09 <oklopol> so you are saying rings can't have ab = a
00:03:21 <cpressey> I may be misusing the term "absorb" there.
00:03:35 <oklopol> did you mean a + b = a?
00:03:50 <oklopol> then that makes no sense in the context, because sequential composition is not addition
00:03:54 <cpressey> I mean, in the case that the program b doesn't halt, a * b = b. Not in the general case. If both a and b halt, a * b = c.
00:04:14 <oklopol> okay, so you were talking about multiplication
00:04:26 <oklopol> and you can have ab = a in a ring.
00:04:40 <oklopol> so what you're saying on the page does not make sense.
00:04:48 <cpressey> OK, so I'll probably reword it.
00:05:47 <cpressey> Can you have ab = a in an infinite ring?
00:06:36 <oklopol> take an idempotent matrix
00:07:44 <oklopol> also it should be simple to construct an example with a!=b, for instance with matrices
00:08:00 <oklopol> i assume you know matrices form a ring, if you don't, they do
00:09:25 <oklopol> heyo augur
00:09:27 <cpressey> OK. Well, that's actually very encouraging for the possibility that programs could form a ring with * being sequential composition. You'd just have to have a * b = b for all b that never halt :)
00:09:34 <augur> oklopol hey
00:09:49 <augur> wanna hear something funky cool about english grammar that noone has a great explanation for?
00:10:20 <oklopol> cpressey: there's one slight problem tho, the addition must form an abelian group, how were you planning to invert parallelization ;)
00:10:32 <oklopol> parallel is rather naturally abelian, i guess
00:10:38 <oklopol> augur: yes!
00:10:51 <augur> consider these two sentences
00:10:53 <oklopol> i'll try to explain it without a second of thought
00:10:55 <oklopol> or not
00:11:16 <augur> "Which article did mary file ___?"
00:11:17 <augur> and
00:11:26 <augur> "Which article did Mary read the book before filing ___?"
00:11:42 <augur> where the ___ denotes a gap/missing word associated with "which article"
00:11:48 <augur> the first is good, the second is horrible
00:12:04 <cpressey> augur: Can you give an example for ____ in the first case?
00:12:15 <augur> cpressey: what do you mean?
00:12:45 <cpressey> augur: Can you provide a word associated with "which article" which you can include in your first sentence in place of the ____ ?
00:12:46 <augur> the ___ just denotes the space that "article" would normally be found it were this NOT a question
00:12:47 <augur> e.g.
00:13:03 <augur> "Mary read the book before filing the article"
00:13:55 <augur> in the statement, you find "the article" after "filing", whereas in the sentence you dont. so i'm just putting a ___ to denote where the article phrase would be in a declarative sentence
00:14:32 <augur> or really, you can see it as (necessarily) empty position that you understand to be associated with "which article"
00:15:39 <cpressey> Well, in your first sentence you have the word "file" and in your second sentence you have the word "filing". If your first sentence was "Which article did Mary filing ___?" it would be equally horrible. Moreso, actually. So I'm not sure how it's even a fair comparison.
00:15:45 <oklopol> the second is "Before filing which article did Mary read the book" with which article in the beginning?
00:16:11 <augur> cpressey: thats irrelevant to the point
00:16:14 <oklopol> maybe i'll read what's been said
00:16:24 <augur> because the non-question sentence "Mary read the book before filing the article" is fine
00:16:31 <cpressey> Completely lost today, I guess I am :)
00:16:33 <augur> its not about the verb, its about the gap
00:16:49 <cpressey> augur: See, I never bought that question-formation works like that.
00:16:59 <augur> whether you buy HOW it works is irrelevant
00:17:08 <augur> characterize it in a dependency grammar sense for all i care
00:17:33 <augur> the point is that theres this dependency between "which article" and "filing" in the second sentence which is BAD
00:17:43 <augur> but between "which article" and "file" in the first sentence which is FINE
00:17:46 <oklopol> augur: the sentence looked horrible to me at first, but now i'm kinda getting used to it :P
00:17:58 <augur> oklopol: thats ok, thats irrelevant to the point too ;)
00:18:00 <augur> so!
00:18:03 <augur> moving on to the interesting fact
00:18:11 <augur> lets mash these two sentences together
00:18:19 <augur> "which article did Mary file ___ before reading ___?"
00:18:22 <augur> completely fine.
00:18:44 <MissPiggy> what did mary file!
00:18:51 <oklopol> yeah
00:18:53 <augur> the gap in the "before VERBing ___" clause is suddenly completely fine
00:19:00 -!- bsmntbombdood_ has joined.
00:19:26 -!- bsmntbombdood has quit (Ping timeout: 252 seconds).
00:19:33 <oklopol> interesting
00:19:35 <cpressey> I don't believe there was ever a gap there. But I'm sure you'll say that's irrelevant :)
00:19:36 <augur> somehow the adverbial clause gap becomes _acceptable_ when theres also a main clause gap
00:19:47 <augur> cpressey: are you fucking listening?
00:19:54 -!- jcp has quit (Remote host closed the connection).
00:19:56 <augur> the term gap is a purely descriptive tool
00:20:10 <cpressey> Did I say it wasn't?
00:20:12 <augur> whether there was a gap there is irrelevant. describe it, theorize about it, HOWEVER you want
00:20:17 <oklopol> cpressey: you did understand the gaps aren't variables, they just describe where the article would be if it wasn't a question?
00:20:25 <oklopol> i mean you can't actually put anything there
00:20:47 <oklopol> and by the article i mean "the article"
00:20:48 <augur> oklopol: in chomskyan linguistics, the gaps actually _are_ variables, sort of. ;)
00:21:10 <oklopol> well you know what i meant :)
00:21:33 <augur> i mean, if you did this with CCG there's no such thing as a gap; it all just builds up by function application
00:22:08 <augur> the point is not how you characterize it, the term gap is a purely descriptive phenomena. normally "file" and "read" have direct objects after them, but in questions they dont, so you can describe that as a "gap".
00:22:37 <augur> if you want to take that idea seriously as a part of your theoretical apparatus, fine, if you don't, fine, but the descriptive term is just a descriptive term, used to name a phenomena
00:23:14 <augur> call it "an atypical location of Direct Object" or a "question position of Direct Object"
00:23:15 <augur> i dont care.
00:23:51 <augur> now that i've given cpressey the smackdown he so rightly deserved
00:23:57 <oklopol> i was just checking that he had an epiphany after asking "cpressey: augur: Can you provide a word associated ..."
00:24:04 <augur> oklopol: noone reeeaaally knows how to characterize those sentences
00:24:49 <oklopol> cool.
00:25:05 -!- Gracenotes has quit (Ping timeout: 240 seconds).
00:25:08 <oklopol> of course, i'd appreciate this more if i knew what things do have a good characterization :P
00:25:18 -!- FireFly has quit (Quit: Leaving).
00:25:45 <oklopol> "i'd be more interested in this linguistics stuff if i knew anything about linguistics"
00:25:57 <augur> :p
00:26:24 <lament> i'd be interested in this interesting stuff if i knew anything about interest
00:26:59 <augur> well, we have lots of great ways of describing all sorts of good sentences and bad sentences and why they're good vs. bad
00:27:00 -!- Gracenotes has joined.
00:27:01 <cpressey> Should I try to explain that I don't think the concept "where the article would be if it wasn't a question" doesn't really make sense when you
00:27:07 <cpressey> 're analyzing questions?
00:27:10 <oklopol> sure
00:27:18 <augur> cpressey: no, because it doesnt matter
00:27:21 <oklopol> although you might want to /ignore augur
00:27:22 <oklopol> ;)
00:27:39 <augur> my use of that description was purely to make it easier to understand the issue at hand
00:28:18 -!- BeholdMyGlory has quit (Remote host closed the connection).
00:28:32 <cpressey> Right, and no alternative explanation would help shed any light on it.
00:28:49 <augur> it was the first thing i tried.
00:29:13 <augur> whether or not an alternative explanation would help is also irrelevant, because you're bitching about the first thing i tried
00:29:29 <augur> which obviously WORKED because you know what the point of it was
00:29:43 <cpressey> You tried to analyze "file before reading" as an atomic verb?
00:29:48 <augur> that you dont like describe it in that fashion is IRRELEVANT to the communicative intent of that phrasing.
00:29:53 <augur> what
00:30:23 <cpressey> So that wasn't the first thing you tried?
00:30:36 <augur> explain what you're saying because now _you've_ lost _me_
00:30:53 <cpressey> Shame, I gotta go home.
00:30:58 <cpressey> See y'all later.
00:31:00 -!- cpressey has left (?).
00:31:11 <oklopol> if you think of it as an atomic verb, then it seemingly makes sense, say you can't have anything but "which article did mary kill" is ok, "which article did mary file before reading" is ok, the example with the book has "two verbs"
00:31:18 <oklopol> *say you can't have anything but one verb
00:31:38 <augur> its obviously not a single whole unanalyzed verb
00:31:38 <oklopol> see ya
00:32:10 <augur> "Which article did Mary file before throwing into the garbage out back?"
00:32:26 <oklopol> :P
00:32:38 <augur> "file-before-throwing-into-the-garbage-out-back" is now one giant verb!
00:34:38 <oklopol> yeah well i was thinking file needs an object, and because it doesn't get one, you treat it as a lambda which you'll apply "article later", and the same happens with read, and "file before reading" has two lambdas around before, so you make a new verb that forks its argument to both file and read
00:34:49 <oklopol> not sure that makes much sense, that's how i just feel like i'm reading it
00:35:02 <oklopol> *just how it
00:35:03 <oklopol> *
00:35:05 <oklopol> aiuoerh
00:35:07 <oklopol> *i
00:35:32 <augur> right, it does need a object
00:35:53 <augur> but you dont need a gap in the second part, right
00:36:16 <augur> "which article did mary file after reading this book?"
00:36:26 <augur> and it cant simply be a fork to both lambdas
00:37:16 <augur> i mean if that WERE your explanation for the goodness of the two-gap sentence, then you have to explain why there are two classes of one-gap sentences
00:37:21 <augur> DO-of-main-verb gaps
00:37:22 <augur> which are fine
00:37:33 <augur> and DO-of-adjunct-clause-verb gaps which are bad
00:37:49 <augur> gtg, ill be back in a bit
00:38:15 <oklopol> well, the rule could be that the first verb needs to be the only one that needs the article; makes sense because you only need to remember what's being "whiched" until some other ...subject appears
00:39:12 <oklopol> so file before reading is okay because you'll make those into one verb, file before reading the book is okay, because when you reach the book, you'll "forget about the article", but your original example has the book, but you still need to remember the article
00:39:48 <oklopol> i mean that's just what i'd guess, you can probably easily find a counterexample, and i'm not even sure what it is i'm saying, i don't really have a good model of language in my head.
00:42:51 -!- augur has quit (Ping timeout: 272 seconds).
00:58:00 -!- Gracenotes has quit (Ping timeout: 240 seconds).
01:05:47 -!- augur has joined.
01:05:58 <augur> ok hello
01:06:04 <augur> sorry what was your rule, oklopol?
01:08:14 <MissPiggy> augur,
01:08:14 <MissPiggy> 00:42 -!- augur [~augur@129-2-175-79.wireless.umd.edu] has quit [Ping timeout: 272 seconds]
01:09:56 <oklopol> augur: that you only remember what you're "whiching" until there's a first subject, so you just build up a verb that will be applied to "the article" later, until some other subject appears, then you just forget about the article.
01:10:58 <oklopol> basically you just have "which X does Mary V ..." where V is some verb applied to X, and ... can be any additional info
01:11:05 <oklopol> (like before reading the book)
01:14:30 <oklopol> but V can also be a fork. i have no idea what it would be in general.
01:14:41 <oklopol> but some sorta lambda
01:48:17 <augur> oklopol: what do you mean a first subject??
01:48:32 <augur> the word being gapped here, which is linked to the WH phrase, isnt a subject, its an object
01:48:41 <augur> so please explain in more detail
02:00:56 <oklopol> sorry, i mean "object other than the one being whiched", not "subject"
02:00:57 <oklopol> but
02:02:18 <oklopol> really what i mean is you can just have one verb after the which, without an object. "which X does Mary V ..." is the only possible structure.
02:03:09 <oklopol> so always "which object does subject verb", but you can then continue the sentence anyway you like, you just can't add stuff in between
02:03:31 <oklopol> would you consider "the article mary read the book before filing" okay?
02:03:46 <oklopol> was just wondering if the "which" is necessary
02:04:27 <augur> no thats horrible too.
02:04:35 <augur> its pretty much ANY gapping of that sort
02:04:52 <oklopol> yeah
02:04:56 <augur> using a WH phrase was just a convenient way to demonstrate it.
02:05:53 <augur> see the problem is that the GOOD version has TWO gaps
02:06:15 <oklopol> "the article mary has always wanted to read" and stuff work, but "the article mary has a cousin who read" and friends don't work, i think it's the object that does it
02:06:26 <oklopol> augur: yeah and i explained it with a fork
02:06:30 <augur> so it seems like a strange thing to say that the rule is that once you see a direct object you discard the stored phrase
02:06:54 <augur> secondly, that depends on a specifically linear model, but your fork doesnt, so you have to make them work together
02:06:59 <oklopol> i don't think it's strange. the verb can just consist of multiple verbs
02:07:06 <augur> and thirdly, its not just english that allows it, right
02:07:17 <augur> other languages have parasitic gaps too, but in the OPPOSITE ORDER
02:07:25 <augur> so in japanese, for instance, i believe the gap structure is like this:
02:07:52 <augur> [___ V before] S ___ V
02:08:02 <oklopol> "augur: secondly, that depends on a specifically linear model, but your fork doesnt, so you have to make them work together" <<< i don't understand this
02:08:24 <augur> the way you explained why the bad version is bad depends on the linear order of the words
02:08:36 <augur> it doesnt have anything to do with the structure of the sentence
02:08:52 <augur> and crucially, gapping seems to be structurally, not linearly, defined
02:09:13 <oklopol> i see. can you give me an example where my idea doesn't work though?
02:09:27 <oklopol> i mean that would be much more enlightening
02:11:15 <augur> for english, no, sorry. but im not ENTIRELY clear on your description of the rule, to begin with
02:11:27 <augur> i mean, ok, i can probably give you an example without a parasitic gap
02:11:37 <augur> e.g. a sentence with a gap that linearly crosses another direct object
02:11:44 <augur> and those are trivially constructable
02:13:51 <augur> would that help you understand what i mean?
02:14:00 <oklopol> maybe
02:14:12 <oklopol> i think it would be a counterexample to my rule
02:14:54 <augur> well you could always say that your rule only applies to PGs ;P
02:15:04 <augur> so for example
02:15:07 <oklopol> because in my rule, which i do not claim i understand that clearly myself, you can't have a gap crossing any other object, the gap always comes right away.
02:15:36 <augur> who did [the man that john saw] speak to ___?
02:15:44 <oklopol> hmm
02:15:53 <augur> or well
02:16:01 <augur> that doesnt cross an in-place object
02:16:01 <augur> so
02:16:09 <augur> who did [the man that saw john] speak to ___?
02:16:16 <augur> where "john" in this one is the object
02:16:27 <lament> what are you talking about language?
02:16:28 <augur> as far as anyone can tell, these constructions have nothing to do with linear order
02:16:40 <augur> infact, no language seems to care about linear order, as far as anyone knows
02:16:42 <augur> lament: yes
02:19:10 <oklopol> augur: oh but i'd just apply my rule recursively, "the man that john saw" is okay, gap before direct object, then "who did X speak to" is okay, gap before direct object.
02:20:05 <augur> ok but how do you characterize the domain of applicability then
02:20:26 <augur> and then how do you get the fact that in other languages the order doesnt matter to PGs
02:21:38 <oklopol> can you define PG = parasitic gap, is it just this general concept of gap?
02:21:59 <augur> a parasitic gap is the second gap in "which article did mary file ___ before reading ___"
02:22:11 <augur> where if it were the only gap, it would be bad, but because of the presence of the first gap, it becomes good
02:22:46 <oklopol> i can't explain other languages, because i don't know them
02:22:51 <oklopol> oh wait i know finnish
02:22:53 <oklopol> let's see
02:23:01 <augur> you dont need to know another language that does this :P
02:23:07 <augur> you just have to realize that its not linear order
02:24:04 <augur> and its not about objects, either, you can cross subjects too
02:24:29 <oklopol> "augur: ok but how do you characterize the domain of applicability then" <<< always applies
02:24:48 <augur> no no i mean what is the domain that you recursively apply this to
02:25:05 <augur> what is the object that is recursively analyzed
02:25:16 <oklopol> well i dunno... i just think "the man that saw john" is a normal subject
02:25:16 <augur> its not simply all substrings of the sentence, obviously
02:25:24 <augur> ofcourse its a normal subject
02:26:39 <oklopol> so what we have is an instance of the pattern "who did X speak to", which is okay, and X is okay as well; i guess i'd say if there's any parsing that makes it work, it works.
02:27:40 <augur> right, but what im asking is how do you know what to look at when you're asking what is ok
02:27:53 <augur> you're segmenting the sentence into two domains to investigate
02:27:58 <augur> who did X speak to
02:27:59 <augur> and
02:28:02 <augur> the man that saw john
02:28:04 <augur> not, say
02:28:07 <augur> who did the man that
02:28:07 <augur> and
02:28:11 <augur> saw john speak to
02:28:33 <oklopol> "if there's any parsing that makes it work, it works.", segment it into any amount of domains, if it works with some segmentation, it works.
02:29:17 <augur> you cant just say if theres ANY segmentation that makes it work
02:29:36 <oklopol> why not, is it obviously false?
02:30:00 <augur> "which file did mary read the book before filing ___"
02:30:04 <augur> segment this into two parts
02:30:18 <augur> which file did Mary VP before filing ___
02:30:26 <augur> and VP = read the book
02:30:32 <augur> thus in VP theres no violation
02:30:37 <oklopol> i mean i don't think this phenomenon happens on the same level as parsing, i think i already know it's "who did [the man that saw john] speak to ___?" when i start thinking about gaps, so the recursion works
02:30:39 <oklopol> hmm
02:30:40 <augur> and in the other one theres no violation
02:31:23 <oklopol> there's no violation in VP, but there is in "which file did Mary VP before filing ___", therefore it doesn't work.
02:31:29 <augur> no there isnt
02:31:36 <oklopol> sure there is, VP has an object
02:31:43 <augur> yes but so does X in yours!
02:31:56 <oklopol> hmm, yes
02:32:02 <oklopol> let me think, maybe you're right
02:33:55 <oklopol> oh but umm, even though X does have an object, the "type of X" is clearly just a subject. the type of "Mary VP before filing" is not a verb, and does not fit the pattern "which file did mary verb before filing"
02:34:07 <augur> yeah but what does that have to do with it
02:34:17 <augur> this is why i said whats the domain of these analyses
02:34:41 <augur> because if i can just choose any object in the syntax to separate along, e.g. the VP, then theres no problem
02:34:55 <oklopol> i'm saying there is a violation in "which file did Mary VP before filing ___", because it's not of the form "which file did mary V ...", which is the only allowed pattern.
02:35:13 <augur> what do you mean the only allowed pattern
02:35:53 <oklopol> that all sentences that start with "which file did mary " must be followed by a verb without an object, forming a lambda that's then applied to "file".
02:36:51 <oklopol> and "Mary VP before filing" isn't a lambda in this case, i don't know why it isn't, my guess was that VP contains an object, i just know verbs are, and certain forks are.
02:37:18 <augur> thats not true tho
02:37:22 <augur> which file did mary ask john about
02:37:46 <augur> which person did mary talk to john about
02:38:03 <augur> who did mary give the pot to
02:38:06 <augur> etc etc
02:38:28 <augur> not questioning the subject does not mean questioning the object
02:38:43 <augur> what building did mary read the book in
02:40:00 <oklopol> the first ones i think roughly follow my idea, the last one doesn't
02:40:49 <oklopol> now it's okay to have another object before the gap because it isn't a verb that's applied to building, there's a preposition
02:42:00 <oklopol> anyway if you feel like i'm on the wrong track, i'll just trust on your intuition even if i don't trust in your counterexamples.
02:42:17 <oklopol> probably i haven't solved the problem if linguists haven't.
02:42:18 <augur> who did mary tell john about
02:42:29 <augur> tell obviously can take a person as a direct object
02:42:31 <oklopol> well again
02:42:32 <augur> and who can be that object
02:42:33 <oklopol> tell takes two args
02:42:44 <oklopol> "so tell john about" is just a one-arg verb
02:42:55 <augur> what
02:42:55 <oklopol> so i still don't think that's a counterexample
02:42:59 <augur> no its not
02:43:05 <augur> it IS a counter example
02:43:12 <augur> your rule does not rule it out
02:43:15 <oklopol> "to tell john about X" is a verb with one argument
02:43:25 <augur> "to tell john about X" is not a verb
02:43:30 <oklopol> oh.
02:43:38 <augur> its a whole fucking phrase :P
02:43:48 <augur> its also an infinitival subjectless sentence
02:44:02 <augur> and again, Japanese.
02:44:15 <augur> or Irish.
02:44:55 <oklopol> i'd just say tell is a verb that takes two right arguments, in "to tell john about" you've just curried the first one
02:45:49 <pikhq> What's this about Japanese?
02:45:55 <oklopol> i mean what the fuck does it even matter if that's not something that usually makes sense, if it makes sense in this case, i have fitted your counterexample in my model.
02:46:21 <oklopol> or more like it already fit in it, i just probably have a weird model of language in my head because i don't know any of your fancy linguistics stuff.
02:46:22 * pikhq logreads
02:47:10 <pikhq> augur: Yeah, that is the gap structure of Japanese.
02:47:56 <augur> oklopol: ok, let me put it to you this way
02:48:18 <oklopol> "who did mary give the pot to" how is this not a direct application of my idea? "give the pot to" is a verb with one argument
02:48:22 <augur> give me a full _algorithm_ that is not ambiguous or vague.
02:48:56 <augur> oklopol: fine, but then the VP in the bad sentence is ALSO a verb with one argument
02:49:15 <augur> also, "give the pot to" is a two argument verb
02:49:20 <augur> the object of giveing-pot-to
02:49:22 <augur> and the subject
02:49:24 <oklopol> oh wait right
02:49:33 <oklopol> i didn't notice, because it doesn't matter
02:49:38 <augur> right
02:49:44 <augur> you just shift the argument number up one
02:49:49 <oklopol> but umm
02:50:14 <oklopol> you are right about the VP thing, my rule is pretty vague
02:50:19 <augur> the point is, the linear ordering doesnt matter
02:50:29 <augur> because other languages with different linear ordering have the SAME constraint
02:51:36 <augur> dont try to solve it oklopol, its not an easy problem to solve :p
02:51:57 <oklopol> okay i can't really explain why "Mary does something before filing ___" is not okay while "mary tells john about ___" is, i can just explain why the latter one clearly is (currying).
02:52:11 <oklopol> yeah true
02:52:25 <augur> well its not just "does something" right
02:53:06 <augur> its "does something" where that phrase, whatever it is, doesnt have a gapped item
02:53:12 <augur> its not even direct object, right
02:53:21 <augur> "which book did mary tell john about ___ before reading ___"
02:53:42 <augur> "which book did mary see ___ burn to pieces after reading ___"
02:53:45 <augur> etc
02:53:55 <augur> tho interestingly, not the other way around
02:54:22 <augur> "which book did mary read ___ before ___ burnt/burning"
02:54:45 <oklopol> wait that's not okay?
02:55:53 <augur> not on the reading where the book just spontaneously burnt to pieces
02:55:54 <oklopol> "augur: its "does something" where that phrase, whatever it is, doesnt have a gapped item" yeah as i said, it's okay if it results in a one-arg verb, otherwise not. gap = one-arg verb
02:56:01 <augur> if it was mary doing the burning then sure its fine
02:56:15 <augur> "which book did mary read ___ before ___ fell/falling shut"
02:56:28 <oklopol> oh right
02:57:03 <augur> im not saying it doesnt have to do with the VP being a one-arg/two-arg thing right
02:57:10 <augur> im saying it doesnt have to do with the LINEAR ORDER
02:58:24 <oklopol> okay maybe i'm not understanding correctly what you mean by linear order, you mean it can't be the case that it actually wants the gap to be just to the right of the "which article" thingie, but that the reason must have to do with something else?
02:59:23 <augur> i mean it has nothing to do with the dependency crossing an object
02:59:46 <augur> lets try to construct a different example
03:00:35 <augur> "mary read the book before filing yesterday the article" is bad where yesterday modifies "read"
03:00:50 <augur> "mary read before filing yesterday the article" is fine where yesterday modifies "read"
03:00:52 -!- coppro has joined.
03:01:10 <augur> so in this case "the article" is to the right of its gap
03:01:28 <augur> so the dependency doesnt cross another direct object
03:01:34 <augur> and yet its still unacceptable
03:02:23 <oklopol> the sentence is not of the form "the X that ...", i don't directly see what these have to do with each other
03:02:44 <augur> neither is the original!
03:03:01 <oklopol> well yeah it was "which"
03:03:02 <augur> "which article did mary read the book before filing" is not of the form "the X that ..." either
03:03:02 <oklopol> same thing
03:03:14 <oklopol> "the article mary read the book before filing"
03:03:14 <augur> and we already established that it has nothing to do with the "which"
03:03:30 <augur> oklopol
03:03:42 <augur> i just showed you that the CONSTRAINT STILL APPLIES
03:03:48 <augur> and yet your description of the required sentence DOESNT
03:04:15 <augur> you have failed to explain the phenomena because you have no explanation for the RIGHTWARD filler-gap badness
03:04:27 <oklopol> my description was only about cases where you have some sort of "the X that ..." or "which X ..."
03:04:34 <augur> yes and thats the problem
03:04:38 <oklopol> i agree
03:04:45 <augur> your description ONLY explains the leftward filler-gap case
03:04:49 <augur> but not the rightward one
03:05:14 <augur> and more generally, its OBVIOUSLY got nothing to do with the direction --- the violations are EXACTLY the same regardless of direction
03:05:24 <oklopol> i don't even think "mary read before filing yesterday the article" is a good sentence.
03:05:34 <augur> well, you might need a heavier NP
03:05:50 <augur> "mary read before filing yesterday every TPS report that had been left on her desk in the last week"
03:06:11 <oklopol> oh right, i see
03:06:15 <oklopol> silly me
03:06:18 <pikhq> Yeah, that parses just fine.
03:06:45 <pikhq> It just seems very unusual with that syntax structure and having only a single noun as the object of "read".
03:06:55 <augur> 5yeah
03:07:14 <pikhq> Make it longer, and it's magically idiomatic.
03:07:23 <pikhq> ... Not the correct word in this context.
03:07:23 <pikhq> :P
03:07:31 <augur> rightward extraposition is almost always done for weightedness of the phrase being moved
03:07:35 <pikhq> Magically normal-seeming.
03:07:37 <pikhq> There we go.
03:09:32 <augur> you also need the right intonation
03:09:35 <augur> thats another thing, right
03:09:46 <augur> sometimes without the correct intonation a sentence will sound horrible
03:09:51 <oklopol> i have to go do some homework, i admit i have no idea how to solve the problem in that sort of sentence, cya.
03:10:01 <pikhq> Oh, certainly.
03:10:08 <augur> oklopol, dont worry, noone else does either :)
03:10:16 <augur> pikhq: presumably its because intonation indicates structure
03:10:33 <augur> so without the right intonation, you're just not conveying the right structure, so it IS bad
03:10:43 <pikhq> augur: Yeah.
03:10:57 -!- Sgeo_ has changed nick to Sgeo.
03:11:04 <oklopol> i usually draw parentheses in the air with my hands with complicated sentences
03:11:05 <augur> thats the problem with language, we cant enforce a structure, we have to use this linear mechanism
03:11:10 <augur> but the question is ALL about structure
03:11:25 <augur> oklopol: sure, but that also sort of doesnt work with the average person
03:11:41 <pikhq> Y'know, it's kinda funny... It seems to me many people have this odd idea that intonation is limited to various Asian languages... (Y'know, since Chinese is tonal and all)
03:11:58 <pikhq> Even though there's probably not a language *without* intonation being significant in some way.
03:12:18 <pikhq> (if there is a language without significant intonation, its name is Lojban.)
03:12:20 <augur> pikhq: no no intonation isnt what people think is limited to many asian languages
03:12:26 <augur> rather, TONE is
03:12:29 <augur> well not just but
03:12:32 <pikhq> augur: Right.
03:12:33 <augur> english doesnt have TONE
03:12:37 <augur> but it has intonation
03:12:41 <augur> the two are not the same
03:12:43 <pikhq> augur: Intonation is merely ignored.
03:12:51 <augur> indeed
03:13:16 <pikhq> augur: Sorry. Minor bit of confusion there.
03:13:20 <pikhq> I really should know better.
03:13:30 <oklopol> it isn't tone that questions have in the end?
03:13:48 <augur> oklopol: well, intonation and tone are closely related
03:14:06 <augur> in that both are primary tone, right
03:14:15 <augur> so questions have high intonation at the end of them, usually
03:14:33 <augur> but tone is a property of words or syllables
03:14:44 <oklopol> true
03:15:08 <oklopol> i mean i don't know what primary tone is, but otherwise
03:15:15 <augur> er
03:15:35 <oklopol> also the tone at the end of questions is mostly ignored as well
03:15:36 <augur> the tone underlying the harmonics
03:17:20 <oklopol> okay sleep and/or homework
03:17:21 <oklopol> ->
03:17:38 <augur> see ya
03:21:10 -!- SimonRC_ has joined.
03:27:20 -!- SimonRC has quit (Ping timeout: 265 seconds).
03:27:54 -!- bsmntbombdood__ has joined.
03:29:59 -!- bsmntbombdood_ has quit (Ping timeout: 252 seconds).
03:38:24 -!- oerjan has joined.
03:39:54 -!- Asztal has quit (Ping timeout: 265 seconds).
03:54:41 -!- yiyus has quit (Ping timeout: 256 seconds).
04:01:56 -!- MizardX has quit (Excess Flood).
04:02:05 -!- pthing has changed nick to Pthing.
04:03:00 -!- oklopol has quit (Ping timeout: 240 seconds).
04:03:32 -!- MizardX has joined.
04:07:10 -!- yiyus has joined.
04:09:00 -!- jcp has joined.
04:11:53 -!- MizardX has quit (Excess Flood).
04:12:16 -!- MizardX has joined.
04:27:29 -!- augur has quit (Ping timeout: 252 seconds).
04:31:32 -!- coppro has quit (Remote host closed the connection).
04:32:20 -!- coppro has joined.
04:39:13 -!- MissPiggy has quit (Quit: Lost terminal).
05:08:13 -!- bsmntbombdood__ has quit (Ping timeout: 272 seconds).
05:16:51 -!- MizardX has quit (Ping timeout: 256 seconds).
05:25:09 -!- bsmntbombdood__ has joined.
05:27:26 -!- augur has joined.
05:40:09 -!- bsmntbombdood__ has changed nick to bsmntbombdood.
05:44:21 -!- coppro has quit (Remote host closed the connection).
05:45:05 -!- coppro has joined.
05:54:40 -!- jcp has quit (Read error: Connection reset by peer).
05:55:52 -!- jcp has joined.
06:08:49 -!- Gracenotes has joined.
06:43:08 -!- oerjan has quit (Quit: Verily So).
06:56:43 -!- jcp has quit (Quit: I will do anything (almost) for a new router.).
07:06:08 -!- deschutron has joined.
07:58:15 -!- oklopol has joined.
07:59:02 -!- addicted has joined.
07:59:59 -!- clog has quit (ended).
08:00:00 -!- clog has joined.
08:13:45 <fizzie> `swedish This is an example sentence right here.
08:13:50 <HackEgo> Thees is un ixemple-a suntunce-a reeght here-a. \ Bork Bork Bork!
08:26:33 -!- MizardX has joined.
08:27:30 -!- Asztal has joined.
08:30:20 -!- MigoMipo has joined.
08:33:21 -!- deschutron has left (?).
09:03:30 -!- oklopol has quit (Remote host closed the connection).
09:03:54 -!- oklopol has joined.
09:07:38 -!- gm|lap has quit (Quit: 2 hour UPS expired. Shutting down laptop.).
09:10:30 -!- addicted has quit (Ping timeout: 265 seconds).
09:11:18 -!- addicted has joined.
09:44:41 -!- kar8nga has joined.
10:19:24 -!- FireFly has joined.
10:21:44 <AnMaster> hm
10:22:26 -!- tombom has joined.
10:22:27 <AnMaster> what about a piet/bf polygot? I do think it would actually be possible, since bf ignores anything but +-<>[],.
10:22:57 <AnMaster> so as long as the image header or such requires that there is unbalanced [] in it or similar it should be possible
10:23:21 <AnMaster> s/such requires/such doesn't require/
10:24:11 <AnMaster> a piet quine also sounds quite interesting
10:26:47 -!- Asztal has quit (Ping timeout: 272 seconds).
10:44:14 -!- oklopol has quit (Read error: Connection reset by peer).
10:44:45 -!- oklopol has joined.
10:54:14 -!- BeholdMyGlory has joined.
10:58:21 <fizzie> AnMaster: If you use the PPM image format, a piet/bf polyglot is boringly trivial: the PPM format is ascii-based, the "header" is just the string "P3", all the data is just decimal digits and newlines ignored by bf, and lines starting with # are PPM comments, so you can stick any brainfuck program there.
10:59:26 <AnMaster> fizzie, hm, what format is usually used for piet?
10:59:31 <AnMaster> png? bmp? ppm?
10:59:38 <fizzie> I don't really know, png would be my guess.
10:59:42 <AnMaster> right
10:59:53 <AnMaster> fizzie, a piet quine sounds more interesting
11:01:32 <fizzie> Yes. Or maybe even a multistage quine, where you have a .png file that's a Piet program that outputs a .bmp, which is a Piet program that outputs a .tiff, which is a Piet program that outputs the original .png.
11:01:41 <AnMaster> hehe
11:02:03 <fizzie> Sounds nontrivial to write, though.
11:05:00 -!- addicted has quit (Ping timeout: 260 seconds).
12:19:12 -!- FireFly has quit (Quit: Leaving).
13:03:56 -!- kar8nga has quit (Remote host closed the connection).
13:26:49 <oklopol> nono you should make a chain of piet programs p1 ... pn s.t. pi outputs p(i+1), pn outputs p1, and they form an animation of some sort
13:27:00 <oklopol> like a dancing dude
13:27:36 -!- oerjan has joined.
13:32:11 <fizzie> oklopol: I am sure that sooner or later that sort of thing (incl. a Piet browser plugin to run them) will deprecate Flash for doing animated web ads.
13:34:08 <oklopol> probably also interactive stuff
13:50:48 -!- paramananda has joined.
14:00:09 * oerjan vaguely suspects paramananda is in the wrong channel
14:00:32 <oerjan> this is for esoteric programming languages, nothing to do with swamis of any kind :D
14:02:48 <oerjan> i don't think we have yet found a good esotericism channel to point people to
14:04:26 -!- lament has quit (Ping timeout: 252 seconds).
14:04:28 <oerjan> hm actually according to http://irc.netsplit.de/chat/esoteric.php, the #esoteric at DalNet looks promising
14:04:42 <oerjan> *DALnet
14:05:37 <oerjan> well, apart from having just 1 user, that is
14:05:53 -!- lament has joined.
14:06:39 <oerjan> alas, http://irc.netsplit.de/channels/details.php?room=%23esoteric&net=DALnet shows that being the mac
14:06:42 <oerjan> max
14:08:25 <oerjan> the rusnet one looks more promising actually (8 users), but you'd have to know russian
14:10:38 <oerjan> (topic transliterates as "ezoterika, mistika, magiya")
14:30:47 -!- paramananda has left (?).
14:40:46 -!- FireFly has joined.
14:56:54 <Sgeo> Who said anything about swamis?
14:57:37 <oerjan> `google paramananda
14:57:43 <HackEgo> Swami Paramananda (1884-1940) was one of the early Indian teachers who came to the United States to spread the Vedanta philosophy and religion in America. ... \ [13]Biography - [14]Works - [15]Books on and by Swami Paramananda
14:58:11 <Sgeo> Ah
14:58:36 <oerjan> the -ananda is usually a dead giveaway :D
14:58:51 <Sgeo> Well, I have to get ready to go to school
15:08:50 -!- cpressey has joined.
15:13:43 -!- Pthing has quit (Remote host closed the connection).
15:29:33 -!- deschutron has joined.
15:33:33 -!- Suprano has joined.
15:33:57 -!- Suprano has left (?).
15:37:50 <deschutron> hi fungot
15:37:51 <fungot> deschutron: i don't know anyone who speaks latin would like to join in r.a.w...?
15:38:10 <oerjan> reductio ad whateverum
15:39:18 <deschutron> haha
15:39:36 <deschutron> sure, fungot, count me in
15:39:36 <fungot> deschutron: deleted by fnord white user fnord 19:56, 20 october 2006 ( utc
15:40:02 <deschutron> fungot: i was deleted?!
15:40:03 <fungot> deschutron: ' ' '
15:40:21 <deschutron> fungot: fungot
15:40:22 <fungot> deschutron: operation blue star: this was the first black man to receive a kbe. but, unfortunately the error function. but is the constitution the enabling act, or
15:40:57 <oerjan> ^style
15:40:58 <fungot> Available: agora alice c64 ct darwin discworld europarl ff7 fisher ic irc jargon lovecraft nethack pa speeches ss wp* youtube
15:41:52 <deschutron> so fungot can be set to a style, can it?
15:42:17 <deschutron> ^style lovecraft
15:42:18 <fungot> Selected style: lovecraft (H. P. Lovecraft's writings)
15:42:36 <deschutron> fungot: hey there, how's it hangin'?
15:42:58 <deschutron> fungot: fungot
15:43:10 <deschutron> ^style speeches
15:43:11 <fungot> Selected style: speeches (misc. speeches from Project Gutenberg)
15:43:39 <deschutron> fungot: what is the only thing to fear?
15:53:14 <oerjan> fungot: stop ignoring deschutron please
15:53:16 <fungot> oerjan: gentlemen, flushed with hope, burning with animosity, they agree in their votes, so to speak, or for iphicrates in the field, and in the course of athenian education, the administration of justice.
15:53:52 <oerjan> (it has a five response consecutive limit per person)
15:54:02 <deschutron> i see
15:54:38 <deschutron> fungot: what is the only thing to fear?
15:54:40 <fungot> deschutron: the great french soldier, de fnord fnord fnord atque fnord fnord, saw the royal family: so as on one hand, and caused the seal of that great people were peculiar to themselves, the interest became still more absorbing. the fnord understanding, the rudest hand, is not a good hat, and asked me who had been talking prose all their lives without knowing anything of the sort since he built the tower of babel. the foreign
15:55:31 <oerjan> fnords to live by
15:55:34 <fungot> tr~lx@115-166-55-84.ip.adam.com.au: what an ignorance of human nature, that whoever takes a view of the court of louis. he was in truth a monster in jurisprudence. it signifies nothing whether a court for this purpose too the druids performed all their ceremonies in the bosom of france itself; we have been speaking had been members of large communities, and of which he had knocked down osborne and bidden defiance to
15:55:34 <deschutron> lol "the great french soldier, de fnord fnord fnord atque fnord fnord"
15:55:54 <oerjan> what the heck?
15:56:51 <deschutron> he must have noticed me quoting him
15:57:06 <deschutron> adam.com.au is my isp
15:57:52 <oerjan> well there's a bug in that...
15:58:02 <oerjan> also, i thought this channel censored colors
15:58:31 -!- scarf has joined.
15:59:31 <oerjan> ^bf +[.+]
15:59:32 <fungot> <CTCP>.. !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~ ...
16:00:11 <scarf> heh, my client interpreted that as a CTCP
16:00:16 <oerjan> mine too
16:00:16 <scarf> with colour codes in!
16:00:40 <oerjan> oh i got no color (other than inverse video). hm...
16:00:43 <oerjan> ^bf ++[.+]
16:00:44 <fungot> .. !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~ ...
16:00:51 <deschutron> whoa
16:01:32 <oerjan> in fact that was what i wanted to test. why did the one above (just befor scarf joined) have colors?
16:01:40 <deschutron> i got no colour on that
16:02:06 <deschutron> i got no colour on the one before scarf joined
16:02:06 <oerjan> huh, i did. blue on black.
16:02:35 <oerjan> maybe that's just bold + inverted in my terminal
16:02:37 <deschutron> i'm using pidgin
16:02:38 <oerjan> or something
16:02:40 <oerjan> irssi
16:04:10 <oerjan> oh well, it was a strange bug - fungot giving a malformed response to deschutron without deschutron saying anything (afaict)
16:04:12 <fungot> oerjan: speusippus. to sup with fnord. i have that inward and dignified consciousness of my own meditation. with eyes dimmed with fond and melancholy tears, she leaned over the fnord, if more and better food is any standard of melioration. they work more, it was the duty of such a compliment; and therefore, in a general way to know why he was in power, and that
16:04:28 <oerjan> fnords to live by
16:04:47 -!- MissPiggy has joined.
16:05:19 <oerjan> perhaps it was that the long response it was already giving got broken up into two lines somehow...
16:05:27 <deschutron> "the great french soldier, de fnord fnord fnord atque fnord fnord"
16:05:55 <oerjan> deschutron: i saw that after fungot's weird response, last time
16:05:56 <fungot> oerjan: the objectors, education among the people for that object. they wish us to do is a mysterious and searching question which those may answer who know what it is.
16:06:34 <deschutron> oh
16:06:50 <deschutron> i sent it before i received his weird response
16:09:04 -!- kar8nga has joined.
16:10:04 <deschutron> by the way oerjan, i like your language CHIQRSX9+
16:10:11 <scarf> I like that last fungot quote, btw
16:10:12 <fungot> scarf: nothing is more natural than that the parliament has possessed great power in the democracy under napoleon. napoleon might probably have lacked the information that certain monsters of virtue ever had existence. but it is not in the power of the keys of the fortress on his knees is daily studied, no sectarian ill-will nor narrow human dogma is permitted to do it.
16:10:40 <oerjan> thanks
16:11:41 <deschutron> i came across it recently.
16:12:11 <oerjan> oh, food ->
16:13:04 <deschutron> food?
16:14:21 <fizzie> oerjan: Yes, there's a sort of a "known bug" in that sometimes fungot gives partially corrupted spurious replies. It's proven to be a bit of a heisenbug.
16:14:23 <fungot> fizzie: revolution, the french monarchy so low in the scale upon which they will intrigue to obtain, or of negligence scarcely less culpable. mr mitford has nevertheless told without any qualification, and at once the grand distinction to be asked to come amongst you. this is the subject of representation; much on the art of cementing the ties of party had superseded those of country, have been more pleasing. i shall introduce
16:17:08 <oerjan> deschutron: i'd been forgetting to eat
16:17:42 <oerjan> fizzie: i was wondering if it was related to having very long replies that get split somehow?
16:18:02 <deschutron> right.
16:18:21 <deschutron> i know the feeling
16:18:44 <fizzie> oerjan: It is possible that the reply-generation overflows something, yes. I don't remember if I have a hard limit for the number of tokens there, it has a stopping probability that grows as the number of already generated tokens increases though.
16:19:37 <oerjan> well does the formula for the stopping probability ever reach 1?
16:20:01 <oerjan> otherwise it's at least possible to grow without limit
16:21:00 <fizzie> I'm not sure, it's not completely trivial to read from the code. And I think it still will only stop if the model permits that, so if there's a nonzero-probability "loop" (context-wise, I mean) of nodes in the n-gram graph that do not have the "can stop" flag, it's possible to get stuck.
16:21:23 <fizzie> Might also be that there's a sensible limit of tokens, but the token → text conversion overwrites something.
16:21:24 <fizzie> ^style
16:21:25 <fungot> Available: agora alice c64 ct darwin discworld europarl ff7 fisher ic irc jargon lovecraft nethack pa speeches* ss wp youtube
16:22:19 <fizzie> I have a feeling the speechification might have a larger average token length that something like IRC. But that's just a guess.
16:23:06 <oerjan> sesquipedalian loquaciousness, you mean
16:23:14 <fizzie> Yes, that.
16:31:01 -!- deschutron has left (?).
16:42:48 <cpressey> oklopol: I reworded http://catseye.tc/cpressey/louie.html#Potro -- hopefully it's clearer now.
16:48:15 -!- KingOfKarlsruhe has joined.
16:49:02 <oklopol> tbh i'm still not completely happy with it :P
16:49:21 <oklopol> just thinking about "Unfortunately, we would need ..."
16:49:23 <oklopol> i mean
16:49:30 <oklopol> couldn't you just append them, and it halts it if halts
16:49:56 <oklopol> you don't actually have to know, because you can't invert programs anyway
16:49:59 <cpressey> Maybe.
16:50:18 <cpressey> True, I think I am still thinking along those lines.
16:51:04 <cpressey> Although, I suspect it's something to do with distributivity. I'll try to work it out.
17:01:27 <MissPiggy> try using it in a sentence
17:04:20 <cpressey> Well, I'm kind of working from an insight I had while working out Cabra. But I'm starting to think whatever that insight was, it was depending on + meaning parallel execution.
17:06:55 <cpressey> Say program a always halts but program b never halts.
17:07:24 <cpressey> In a ring, (a + b) * c = (a * c) + (b * c)
17:07:37 <cpressey> But b never halts, so b * c = b
17:07:45 <cpressey> But that still seems fine.
17:07:59 <cpressey> Just a little weird I guess. (a + b) * c = (a * c) + b
17:09:04 -!- oklogon has joined.
17:11:00 -!- oklopol has quit (Ping timeout: 264 seconds).
17:12:19 <cpressey> Hm. Damn, this might actually work!
17:14:38 <cpressey> No, not sure about that now. Say a never halts, b never halts, c always halts. Then: a * (b + c) = (a * b) + (a * c) = a + a. But: (a + b) * c = (a * c) + (b * c) = a + b.
17:15:04 <cpressey> No, wait. Maybe that's OK.
17:15:24 <cpressey> Nothing says a * (b + c) = (a + b) * c :)
17:17:07 <cpressey> OK, so. I admit, it looks like * in a ring could be pretty well suited to being sequential execution.
17:19:03 <cpressey> I would guess the bigger problems, now, are with making + commutative and invertible and non-absorptive.
17:21:08 <cpressey> Having a + b = b + a where a always halts and b never halts pretty much drives the semantics of + towards something parallel.
17:23:33 <cpressey> No, wait again.
17:23:37 <cpressey> Damn, need more coffee.
17:26:20 -!- sebbu2 has joined.
17:28:57 -!- scarf has quit (Remote host closed the connection).
17:29:33 -!- sebbu has quit (Ping timeout: 256 seconds).
17:33:42 <cpressey> Is it OK for every element of a ring to be its own additive inverse? a + a = e for all a?
17:33:53 -!- coppro has quit (Ping timeout: 240 seconds).
17:34:26 <oerjan> yes, those are rings of characteristic 2 iirc
17:34:40 <cpressey> Nice. Then I might have an idea.
17:36:04 <oerjan> (if you have a 1, and 1+1 = 0, then it follows automatically for everything else)
17:37:32 <oklogon> you and your "iirc" :P
17:38:43 <oerjan> http://en.wikipedia.org/wiki/Characteristic_(algebra)
17:39:50 <oerjan> oklogon: well i did bump into it the other day during one of our discussions
17:40:18 <oerjan> i'm not sure if i knew the term applied to non-fields before
17:41:31 <oklogon> hmm well true i've never heard it with anything but fields
17:42:57 <oerjan> it would seem to include at least boolean rings among other things i already knew about
17:52:21 -!- kar8nga has quit (Remote host closed the connection).
17:55:24 <cpressey> Hm, problem: additive inverses don't seem to play well with absorption. Say a never halts. Then a * (b + c) = a. Then a = (a * b) + (a * c) = a + a. So a = e.
17:55:52 <cpressey> Although, that does suggest the possibility of just making e = bottom :)
17:57:18 -!- coppro has joined.
17:59:27 <cpressey> *That* derivation is why it's so nice to have an algebra that gives you a + a = a.
18:03:18 -!- KingOfKarlsruhe has quit (Remote host closed the connection).
18:05:03 <oerjan> i suppose doing something in parallel with something that never halts is pretty equivalent to doing it alone, intuitively, if you go by something like what results you eventually get...
18:08:21 <Wareya> multithread with one process checking the other to see if it ever loops
18:09:28 <oerjan> cpressey: also, e is intuitively additive 0, while bottom is intuitively multiplicative 0
18:10:08 <cpressey> oerjan: Yeah. I banged my head repeatedly against this particular wall while designing Cabra.
18:16:20 -!- augur has quit (Ping timeout: 252 seconds).
18:21:05 -!- coppro has quit (Quit: zoo time).
18:34:16 <cpressey> Hate to keep asking these things, but is it OK for every element of a ring to be idempotent wrt addition? a + a = a for all a?
18:34:43 <oklogon> nope
18:35:00 <oklogon> a + a = a <==> a = 0
18:35:08 <cpressey> Ah, because then what is a' in a + a' = 0. Right.
18:35:10 <oklogon> by adding the additive inverse
18:35:25 <oklogon> (to get from left to right)
18:35:43 -!- jcp has joined.
18:35:49 <cpressey> Well, that
18:35:56 <cpressey> That's why it's hard to get a ring.
18:36:58 <cpressey> Actually - forget "all a". Say there exist idempotent b and c, where b + b = b and c + c = c. Can b =/= c?
18:37:12 <cpressey> I don't think so.
18:37:55 <cpressey> a + a = a <==> a = 0, so b = c.
18:38:13 <oklogon> yes, b = c = 0
18:38:56 <cpressey> So there can only ever be one element where a + a = a. But if there is any "absorbative" element x where forall y, xy = x, then x + x = x.
18:39:15 <cpressey> So there can only be one "absorbative" element.
18:39:33 <cpressey> (There is probably a better word for "absorbative")
18:39:58 <MissPiggy> projections ?
18:41:24 <cpressey> Maybe projection... not sure.
18:41:51 -!- Asztal has joined.
18:42:16 <cpressey> Anyway, that means there can only be one program that never halts. Or, all programs that never halt are considered equivalent.
18:42:17 <oklogon> i don't see why "x where forall y, xy = x" wouldn't be possible,
18:42:48 <oerjan> cpressey: "zero"
18:43:08 <oklogon> oh i completely ignored the x + x = x
18:45:22 <cpressey> oerjan: Yes.
18:45:25 <oerjan> cpressey: also, if two programs x and y never halt, then x = xy = y
18:46:01 <oerjan> or wait
18:46:07 <oklogon> is there a name for an algebraic system with an infinity?
18:46:46 <cpressey> But for all programs x, there needs to exist a unique x' such that x + x' = 0. If bottom is zero, then x' is not unique.
18:46:52 <oerjan> oklogon: i've got this on my watchlist: http://en.wikipedia.org/wiki/Wheel_theory
18:46:58 <oklogon> non-halting could be infinity
18:48:02 <cpressey> oklogon: If addition is commutative, you'd need negative infinity too.
18:48:56 <oklogon> yes, the part where i ask for the name of such a structure was my main point
18:49:42 <cpressey> oklogon: Sorry, thinking of it in the context of my current train of thought. I don't know if such things have a specific name.
18:50:33 <cpressey> Kleene star is a kind of infinity, maybe.
18:54:09 <oklogon> oerjan: i'll have to check that out in-depth, would the extension of reals be the extended reals?
18:54:33 <oklogon> wait, kleene star is a value?
18:54:58 <cpressey> No. I should say, it's a kind of infinity-generating-operator-thing.
18:55:10 <oklogon> :)
18:56:14 -!- augur has joined.
18:56:55 <cpressey> Say 0 is bottom. Then a + 0 = a (where a =/= 0) means something like "run a in parallel with loops-forever, take a's result when done." a * 0 = 0 means something like "run loops-forever after a". Both are good.
18:57:00 <cpressey> But then, for a there is some a' where a + a' = 0. But what could you run in parallel with a that would equate to loops-forever? Maybe something that interferes with a (and only a) in such a way that causes a to loop forever.
18:57:00 <oerjan> oklogon: i'm not quite sure what it is, although there is at least both distinct 1/0 and 0/0 iirc
18:57:59 <oklogon> what's a watchlist
18:58:27 <AnMaster> oklogon, on irc?
18:58:35 <oklogon> on wp
18:58:38 * AnMaster has not read context
18:58:54 <oerjan> cpressey: you could think of 0 more as "never gives a result", in which case looping forever may not be mandatory
18:58:58 <AnMaster> oklogon, oh that, a way to "subscribe" to edits to certain pages
18:58:59 <cpressey> I believe it makes Wikipedia message you somehow whenever the page changes.
18:59:21 <AnMaster> (that description is basically a simplification, but you get the idea)
18:59:36 <oerjan> i think that may be an option, although i just visit my watchlist page...
18:59:38 <cpressey> oerjan: OK, then a' is the "result killer" of a (and only a)
18:59:47 <AnMaster> cpressey, yes it will also email you (there is a setting iirc for that)
18:59:59 <AnMaster> (grammar fail)
19:00:26 <AnMaster> oklogon, hm have you used your current nick before?
19:00:30 <AnMaster> I don't remember seeing it
19:01:04 <oklogon> it's new
19:01:06 -!- oklogon has changed nick to oklopol.
19:01:09 <oerjan> cpressey: what you might want then, is for results to form a group
19:01:10 <AnMaster> oklopol, it sounds like the name of a noble gas
19:01:21 <oerjan> and a' gives the inverse result of a
19:01:25 <AnMaster> I guess I'm thinking of argon and such
19:02:20 <cpressey> oerjan: yes, that sounds like a good way to proceed.
19:03:09 <cpressey> Hm. Maybe food will help me think. bbl.
19:03:20 <oerjan> clearly an oklogon is a polygon with strange non-euclidean angles
19:03:24 <fizzie> "oklogon" must be some sort of an irr{egular,esponsive,ational,everent} shape, by way of analogy from polygon.
19:03:32 <fizzie> Gaah, oerjan strikes faster again.
19:04:14 <fizzie> And that was supposed to be irresponsible, not some strange "irresponsive". A compound fracture, I mean, failure.
19:06:11 <oerjan> a fractal compound
19:06:32 <lament> farctal
19:11:28 <cpressey> oerjan: Problem: We say b + 0 = b means "when b finishes, take its result". What if we have b + x, and b finishes before x. Do we take b's result or do we wait for x? If x is 0, we can't wait for it. But x might not be 0, so we have to wait for it.
19:11:43 <cpressey> I'm just reliving Cabra :)
19:11:50 <cpressey> Away for lunch for real this time.
19:12:21 <oerjan> the result only becomes known asymptotically
19:13:18 <oerjan> i suppose this could be a problem if it never stabilizes
19:28:46 -!- kar8nga has joined.
19:40:57 <cpressey> Or another way to put it would be, you need an oracle. if you have that, or you restrict the set to programs that always halt, I don't think there's a huge problem.
19:41:24 <oerjan> mhm
19:42:22 <cpressey> Say 0 maps all tapes to blank tapes, and programs refuse to run when the tape is blank. Obviously the initial tape can't always be blank, but whatever. 1 is the identity function on tapes.
19:42:59 <cpressey> Then, a + b -> run both and add their tapes. a * b -> run a then b.
19:43:13 <oklopol> err yes, obviously the -gon is from polygon
19:43:19 <oklopol> you were both slow
19:43:35 <oklopol> i was also oktagon at some point
19:43:40 * oerjan quickly swats oklopol -----###
19:43:42 <oklopol> possibly not here tho
19:44:56 <oerjan> cpressey: that won't distribute (a + b) * c properly unless c is linear
19:45:04 <cpressey> oerjan: Re wheels: it reminded me of an algebra I considered once, over the reals - {0}. Division, multiplication, addition are defined everywhere, but subtraction is not.
19:45:30 <cpressey> oerjan: Hm, I think you're right.
19:46:17 <oerjan> addition? how does that work?
19:46:37 <oerjan> or do you mean positive reals
19:47:23 <cpressey> Sorry, yeah. Addition of numbers with the same sign. (If subtraction is not defined, then neither is addition of different-parity numbers.)
19:47:44 <oklopol> addition of the numbers with the same sign? so it's a partial algebra
19:48:23 <oklopol> usually in an algebraic structure every function should be defined for all inputs
19:48:27 <cpressey> s/parity/sign/. don't know what i'm saying :)
19:48:38 <oklopol> yes that was clear
19:48:47 <cpressey> Well, I was inspired by division by zero being undefined in the reals.
19:48:54 <oklopol> (parity and sign are the same group)
19:49:55 <cpressey> |a| + |b| and -|a| + -|b| are defined, and you could consider them two new operations.
19:49:56 <oklopol> well, Q\0 and R\0 are both abelian groups w.r.t. multiplication, it's just adding the addition doesn't make, afaict, make it any known structure
19:50:26 <oklopol> but then you already have subtraction because -|a| + -|b| can be negative
19:50:29 <oklopol> unless it's partial
19:50:38 <oklopol> (some functions defined only for some inputs)
19:51:14 <oklopol> well anyway i need to go to sleep
19:52:38 <cpressey> Well, |a| + |b| and -|a| + -|b| are defined everywhere, but |a| + -|b| and -|a| + |b| are not defined everywhere. Maybe the terms "addition" and "subtraction" are misleading in this context.
19:53:06 <cpressey> 'Night oklopol.
19:54:54 <cpressey> oerjan: From the Cabra doc: "I will go so far as to conjecture that, in fact, any semantics for parallel composition a+b (in an otherwise Cabra-like language) that combines results from both a and b will not be right-distributive."
19:55:41 <cpressey> That might be hard to get around, even for only the programs that always halt.
19:56:10 <oerjan> hm right the problem is how to run c on the results
19:56:21 <oerjan> if it can do anything with them
19:59:04 <oerjan> the * operation needs to distribute the multiple/negated results from the left part to different invocations of the right part
20:17:45 -!- oerjan has quit (Quit: Good night).
20:46:27 <AnMaster> <fizzie> "oklogon" must be some sort of an irr{egular,esponsive,ational,everent} shape, by way of analogy from polygon. <-- possibly discworld related?
20:49:34 <AnMaster> cpressey, couldn't you do addition with different sign as long as you don't have the same value on both sides?
20:50:17 <AnMaster> unless I completely misunderstood you
20:50:42 <AnMaster> after all 1 + (-42) = -41, but 1 + (-1) = 0, so the former case should work but the latter not?
20:51:12 -!- gm|lap has joined.
21:00:07 -!- jcp has quit (Quit: I will do anything (almost) for a new router.).
21:06:39 <cpressey> AnMaster: Yeah. It is defined, just not defined everywhere. Specifically a - a isn't defined. A bit like a / 0.
21:07:41 <cpressey> Well, I updated the Potro bit of my LoUIE page again, FWIW with the algebra fans asleep. :)
21:08:01 <AnMaster> cpressey, indeed
21:10:00 <cpressey> I thought it was a nice "dual" of a field in some sense.
21:13:14 <cpressey> It might be a partial algebra, but if that's true, fields are partial algebras too.
21:56:48 -!- Gracenotes has quit (Quit: Leaving).
22:27:24 -!- kar8nga has quit (Remote host closed the connection).
22:32:48 <oklopol> oerjan: i solved it.
22:32:56 <oklopol> ->
22:34:46 <AnMaster> cpressey, for R_+ \ {0} shouldn't addition, multiplication and division all be completely defined? Not substraction for a - b where b >= a though
22:35:12 <AnMaster> oklopol, oerjan wasn't connected when you said that
22:35:14 <cpressey> AnMaster: Well, as oklopol pointed out, x + y isn't defined when x = -1 * y
22:35:32 <MissPiggy> AnMaster, hey I think you are right
22:35:38 * AnMaster considers that
22:35:45 <AnMaster> cpressey, wut? I must be too tired
22:36:00 <MissPiggy> x = -1 * y can't happen in R_+ \ {0} (aka R_+)
22:36:04 <cpressey> AnMaster: Addition is subtraction when you have opposite sign values :)
22:36:10 <AnMaster> MissPiggy, well yeah
22:36:14 <cpressey> Oh sorry
22:36:21 <cpressey> Didn't see the + in R_+
22:36:24 <MissPiggy> R_+ meaning all positive reals (although some people use positive to include zero, when it's convenient..)
22:36:27 <AnMaster> cpressey, right
22:36:41 <AnMaster> MissPiggy, well I also added "\ {0}" after that
22:36:47 <AnMaster> which should clearly exclude zero
22:37:08 <cpressey> AnMaster: Yes, your statement about R_+\{0} is true.
22:37:08 <MissPiggy> :(
22:37:17 <MissPiggy> why are you saying that I can tell you added \ {0} by reading
22:37:27 <AnMaster> MissPiggy, what?
22:37:39 <AnMaster> I can't even parse the grammar of that, nor the meaning
22:38:10 <AnMaster> maybe some comma is missing?
22:39:20 <AnMaster> cpressey, would it be possible to make some set where addition, multiplication, division are all defined for all possible combinations of values?
22:39:50 <AnMaster> apart from the empty set that is ;P
22:40:52 <AnMaster> alternative: define what division by zero means
22:41:13 <AnMaster> cpressey, I suggest we "leave it to the reader/ask the user" ;P
22:42:42 <cpressey> AnMaster: Isn't R+_\{0} such a set?
22:43:19 <AnMaster> cpressey, as I said, substraction is not
22:43:29 <AnMaster> oh
22:43:31 <cpressey> Oh, did you mean to include subtraction?
22:43:32 <AnMaster> I forgot that from the list
22:43:33 <AnMaster> -_-
22:43:34 <cpressey> Ah.
22:43:35 <AnMaster> cpressey, yes
22:44:38 <AnMaster> it seems to me that subtraction and division are mutually exclusive when it comes to "defined for all values"
22:44:51 <cpressey> I agree.
22:44:53 <AnMaster> I can't prove this though
22:45:03 <AnMaster> cpressey, well the exception would be for the empty set of course
22:45:24 <AnMaster> but I don't think that even makes sense
22:45:49 <cpressey> Every operation on elements of the empty set is well-defined! :)
22:46:10 <AnMaster> cpressey, you could sometimes define division by zero as the limit when going towards zero I guess
22:46:24 <AnMaster> doesn't work for all possible functions though
22:46:34 -!- MigoMipo has quit (Remote host closed the connection).
22:46:34 <AnMaster> (such as constant functions)
22:47:08 <cpressey> Yeah, I'm not so happy with that.
22:47:11 <AnMaster> or diverging (sp?) functions
22:48:51 -!- Sgeo has quit (Quit: Leaving).
22:49:08 <AnMaster> cpressey, other things not to be happy with: that lim_{x→0}(sin(1/x)) is undefined.
22:49:13 -!- Azstal has joined.
22:49:25 <AnMaster> Same for lim_{x→0}((sin(1/x))') of course
22:50:43 -!- Asztal has quit (Ping timeout: 265 seconds).
22:52:03 <AnMaster> cpressey, why not add a NaN? Like we have i
22:52:12 <AnMaster> as in, make NaN a real number
22:52:16 <AnMaster> XD
22:52:28 <cpressey> NaN + NaN = NaN
22:52:39 <AnMaster> cpressey, same goes for aleph_0 though
22:52:40 <cpressey> Therefore NaN = 0
22:52:48 <AnMaster> cpressey, err no
22:52:56 <cpressey> (in context of ring theory stuff I was talking about earlier only :)
22:53:07 <AnMaster> by that logic aleph_0 + aleph_0 = aleph_0 gives aleph_0 = 0
22:53:14 <AnMaster> which is ridiculous
22:54:52 <cpressey> Earlier we were talking about how in a ring, if a + a = a, a must be 0.
22:56:32 <cpressey> Which might explain why aleph_0 doesn't often appear in the ring of reals.
22:57:01 <AnMaster> cpressey, true
22:57:23 <AnMaster> cpressey, neither does i. Since it by definition isn't a real
22:57:32 <AnMaster> cpressey, but then the complex numbers are not a ring?
22:57:44 <AnMaster> oh wait
22:57:46 <AnMaster> it may be
22:57:56 <cpressey> We could define 1/0 as 1/0, irreducible. Basically, work on pairs of integers.
22:58:07 <cpressey> Probably falls apart somewhere.
22:58:07 <AnMaster> true
22:58:32 <cpressey> I don't do complex numbers. :)
22:58:47 <AnMaster> why not?
22:58:55 <AnMaster> cpressey, you should do quaternions even
22:59:01 <cpressey> I dunno, they just never held any interest for me
22:59:42 <AnMaster> cpressey, what about calculations with alternating current? They are rather useful there
23:00:11 <AnMaster> or maybe you don't do that kind of stuff
23:02:45 <cpressey> Heh
23:03:13 <cpressey> Yeah, complex numbers might come in handy for computational induction.
23:03:22 <MissPiggy> hmm?
23:03:35 <MissPiggy> ahh
23:03:44 <MissPiggy> induction like electromagnetic currents
23:04:19 <cpressey> "Execution of instructions in one program induces execution of instructions in another, nearby program." Yes.
23:04:53 <cpressey> Which is an extremely silly idea. But fun.
23:06:18 <AnMaster> <cpressey> "Execution of instructions in one program induces execution of instructions in another, nearby program." Yes. <-- I first thought "what are you messing around with by doing induction over an uncountable set"
23:06:35 <cpressey> Ha! Ugh.
23:07:03 <AnMaster> hm is that possible btw?
23:07:09 <AnMaster> I can't see how it would be
23:07:22 <cpressey> Well, I've heard of "transfinite induction" but I don't know if it's the same thing.
23:07:30 <AnMaster> I see
23:07:33 * cpressey defers to the real mathematicians present
23:07:49 <AnMaster> cpressey, oerjan is *not* present
23:08:31 <AnMaster> <cpressey> "Execution of instructions in one program induces execution of instructions in another, nearby program." Yes. <-- you should make an esolang based on this
23:08:31 <cpressey> Well then I defer to wikipedia.
23:08:32 <cpressey> http://en.wikipedia.org/wiki/Transfinite_induction
23:08:38 <AnMaster> like, alternating ips
23:08:40 <AnMaster> or something
23:09:13 <cpressey> I'd like to make an esolang of it, but I have too many other ideas, and no good idea where to start.
23:10:15 <AnMaster> cpressey, so throw all the ideas into one esolang
23:10:45 <cpressey> I tried that. It's not pretty.
23:10:53 <cpressey> Also, it never gets finished.
23:16:12 -!- tombom has quit (Quit: Leaving).
23:21:58 <cpressey> from * import *
23:23:42 -!- BeholdMyGlory has quit (Remote host closed the connection).
23:23:58 <AnMaster> cpressey, does that work in python?
23:24:14 <AnMaster> it would be rather interesting if it did
23:27:57 <fizzie> Alas, you can't have wildcards in the module-where-to-import-from.
23:28:07 -!- FireFly has quit (Quit: Leaving).
23:30:39 <cpressey> No, but I was (briefly) considering supporting it in my own language...
23:31:56 -!- coppro has joined.
23:37:03 <AnMaster> cpressey, I can think of reasons not to
23:38:53 <cpressey> Yeah.
23:40:28 <fizzie> But, the awesomeness.
23:40:33 <fizzie> "Just import everything."
23:43:07 <fizzie> What about #include <*> as a GCC extension? Read all headers in the system include directories.
23:44:06 <coppro> :P
23:44:13 <coppro> the go slow button
23:44:50 <Gregor> fizzie: YES
23:44:54 <Gregor> #include "**" too
23:45:00 <Gregor> Include every header file in .
23:45:01 <Gregor> Recursively
23:47:24 <fizzie> fis@eris:~$ ls /usr/include/*.h | sed -e 's%/usr/include/%#include <%' -e 's/$/>/' > test.c
23:47:24 <fizzie> fis@eris:~$ gcc -c -o test.o test.c 2>&1 | grep 'error:' | wc -l
23:47:24 <fizzie> 1144
23:49:26 -!- coppro has quit (Quit: I am leaving. You are about to explode.).
23:50:59 -!- Asztal has joined.
23:53:02 -!- Azstal has quit (Ping timeout: 256 seconds).
23:58:35 <cpressey> Blahaha.
←2010-02-17 2010-02-18 2010-02-19→ ↑2010 ↑all