Aw, I just rewrote some of my compiled code with Do instead of For but it didn't speed up :(

@s0rce I'm not totally sure your code is correct

compiledUnion = Compile[{{v, _Real, 2}},
   Block[{
     dim1 = Length[v], dim2 = Length@First[v],
     sorted = Sort[v], output = Internal`Bag@Most[{0.}],
     seen, current
     },
    Internal`StuffBag[output, seen = First[sorted], 1];
    Do[
     If[(current = Compile`GetElement[sorted, i]) != seen,
      Internal`StuffBag[output, seen = current, 2]
      ],
     {i, 2, dim1}
     ];
    Partition[Internal`BagPart[output, All], dim2]
    ],
   RuntimeOptions -> {"Speed", "CompareWithTolerance" -> True},

There's mine. It's a bit slower than yours.

@s0rce in your 2nd test code there is a typo: in Pick[test, deleteduplicatesC[test], 1] duplicates should have capital D

@s0rce sorry! Your code is correct, although mine is faster (I forgot to test yours on my computer, so I was comparing incomparable timings... ugh)

@OleksandrR. 5% faster :)

@s0rce yes, there's hardly anything in it. :)

@s0rce shorter but no faster:

compiledUnion = Compile[{{v, _Real, 2}},
   Block[{
     sorted = Sort[v], output,
     seen, current
     },
    output = Internal`Bag[seen = First[sorted], 1];
    Do[
     If[i != seen, Internal`StuffBag[output, seen = i, 1]],
     {i, sorted}
     ];
    Partition[Internal`BagPart[output, All], Length[seen]]
    ],
   RuntimeOptions -> {"Speed", "CompareWithTolerance" -> True},
   CompilationTarget -> "C"
   ];

you should post it

OK

I didnt know a Do loop could work like that

@s0rce All the iteration functions take a list as an iterator argument (or whatever it's called)

oh neat, is that in the documentation

@s0rce Yeah

With[{l = Range@5},
     Table[Print@i, {i, l}];
     Do[Print@i, {i, l}];
]

Do[Print@i, {i, Range[5]}] :)

I just did that

:)

@s0rce the Documentation folder is 2.3GB on my machine

there's probably the solution to the current eurozone crisis in there somewhere :)

and yet still nothing about StuffBag

@s0rce the Bag stuff is undocumented. there's some information here

yah I know, I just joking, thanks for the link

dog walking time

@OleksandrR. If you have a second I was playing with my function to limit to one memory access per iteration but it's not working in the compiled version and I cant figure out why:

@s0rce I think you have fallen victim to common subexpression elimination here. This is the main sort of optimization that the compiler performs and it is very enthusiastic about it! By the way I tried a variety of schemes and I could not improve over the performance of the version I posted. Your masking method is as fast (with an efficient implementation) but apparently not faster, which is a bit surprising to me.

aah

thanks!

do you work for Wolfram?

@s0rce no

I've a quick question which I feel doesn't merit a full question on mathematica.se - how would I make this code as short as possible? The intention is to replace o with e and e with o. My current code is Replace[#, {o -> e, e -> o}] &@. This seems ugly and difficult to understand.

I was thinking of an implementation using XOR, but I'm not sure how.

I apologize this intrusion. Would you kindly answer this question mathematica.stackexchange.com/questions/22106/…

hmm apparently this is where "the cool mathematica users hang"... that sounds fatal

shouldn't it be hang out? or am I misunderstanding some us idiom?

or does droidstack cut off part of it?

@VincentTjeng The problem is, that difficult to understand often goes along with as short as possible. Do you want it easier to understand, you write it a bit more verbose.

Why don't you use the infix operator for ReplaceAll? Thats probably easier to understand:

expr = x^3 + y + x*y^2
expr /. {x :> y, y :> x}

@acl I think that, indeed, your android is cutting off part of the display.

More and more, though, this is becoming the place where the needy users hang out to ask questions about their questions.

@Averroès Please do not do this again.

@rm-rf is there a problem with that? Apart from the redundant chat message I don't see an issue

I think it's an issue.

@MarkMcClure okay. What is the problem with it in your opinion?

One particular user took a lot of acl's time discussing questions in chat recently. Once they have you in chat, it's hard to walk away.

I personally found it particularly annoying because this users topics seemed quite interesting, so I felt compelled to stay away for fear of getting sucked in majorly.

Discussing a few points of clarification is one thing but it seems to me that the whole point of the main sit is to discuss the questions.

Okay, that's fair. However I still think it's better for users to ask in chat than to post questions that we have to concern ourselves with closing and deleting because they aren't well-formulated or relate to a particular user's misconception. Do you agree with that or would you rather chat remain entirely not for asking questions?

@OleksandrR. No, no problem with that if it's someone who regularly hangs out in chat to socialize/discuss other stuff, help out others or just is a part of the general nonsense that goes on, but if it's a new user who posts it within 15-20 mins of asking the question on main, then yeah... I think there's a problem. If we keep letting that happen, it'll turn into a support site with these constant badgerings.

Holy cow, @rm-rf is past 40K.

@rm-rf okay, fair enough. Maybe I'm just more comfortable telling people to sod off if I don't want to help them than others are. :)

@MarkMcClure in a few days... I'm at 38.6k now. The chat total includes my tally from Apple.se :)

Uhmm anyway, I think there is the potential for abuse in chat because of the "being there" aspect.

Oh, got ya.

2 or 3 more days, I guess.

Oh yeah, I'm up to nearly 17 here, I see.

At any rate, that's just my two cents. Generally, I'm not one to quibble about low quality posts but they seem to be growing in number to me.

@OleksandrR. Also, what @Mark said. Some people are more polite here and feel bad, but I've seen nice chatrooms go bad because of people repeatedly asking for help in chat a minute after posting in main... pretty much every Stack Overflow chatroom is like this

@OleksandrR. Oh, I agree with that... I'm not saying questions should never be linked to or asked or that soft questions shouldn't be asked here at all. Just trying to avoid new/low rep users from making a habit of linking to their questions regularly (I'm ok with it if it's a regular user here or on main)

@rm-rf yes. I agree that if this becomes a regular occurrence it could get annoying. Especially since we already have a question feed in the chat so these messages are completely redundant (but maybe new users don't know this)

Just a user pinging random people (who have never seen the question/posted an answer/commented on the post) soon after posting their question

I personally don't mind it if people do this to me (as long as the question relates somehow to what I have discussed in the past), but I can understand that others may feel differently.

I've had Mathematica material on my personal webpage for probably 20 years. As a result I occasionally get emails asking for assistance. (It used to be a lot worse.) If I had no particularly connection, my response has generally been to refer to mathgroup or, now, stackexchange. I think this is what folks here should do - just say, ask on main.

@OleksandrR. This happens all the time in larger sites and since we're getting there/will get there eventually, it's better to form a culture of not allowing it (or politely telling them to not repeat it) from new users.

(I want to stress that I don't mind any of this in the least bit from chat/main regulars)

@MarkMcClure that would be annoying, certainly. I suppose it's not as bad for me since I don't have a website and I don't give contact details.

@OleksandrR. I'm talking before there was even a stackexchange site to provide contact details. People just found me.

Main regulars tend to be people who have a clue about how to ask a question with the level of detail appropriate to the context and audience, so I'd agree with @rm-rf that's not an issue.

@rm-rf okay. I haven't been in the other chatrooms so I trust your judgment on this. One query, though: won't these requests come mainly from first-time users who would not know that it was discouraged for them to do it? If so, telling them not to do it again may not be that effective, because there will always be more needy users waiting to post their messages.

@OleksandrR. Well, yes... we can't solve the problem entirely (the 20 rep barrier at least stops the spankin' new users). My hope was that if people agree with it, hopefully it'll be someone else telling them off (and perhaps someone who didn't know will find out about it and be the next person to tell a newbie off...). They're informal social rules anyway, not handed down from SE mecca or anything...

@rm-rf make all chat regulars room owners so that they can kick needy users without warning. ;)

I'm also against codifying it into a meta post because the minute you have strict laws for an informal place where people come to relax and hang out (or hang, if you're @acl), it becomes ripe for abuse (e.g. flagging for grudges because something you said violates the rules in its strictest interpretation). I'm not making these up... math has them and while some say it has been effective in shutting certain users up, it has also been misused a lot.

@OleksandrR. Yes, actually I've been meaning to do that for a while... You still can't kick people out, but you can move messages to a different room. So the standard room owners' way to handle annoying users/abusive messages (if one doesn't want to unnecessarily raise to flag wars) is to have a second room that is "gallery" (i.e., need explicit access to talk) and migrate messages from here to there, essentially having this clean (or silencing them, whichever way you look at it)

Strangely enough I never got any requests for help on my MathGroup posting address--just 419 scams and offers to buy fake viagra. Although, good luck to anyone emailing me there anyway as I don't check it for months at a time so they will probably end up being ignored by default

Also, unless absolutely necessary or something needs to be handled immediately and none of mma mods are around (in chat or main) would I recommend flagging a chat message... the reason being that chat flags are broadcast to every 10k user on chat (i.e., 10k across the network) and any site's mod can see or handle that flag. Usually it results in a flood of random users/other mods, which can be a little annoying if you're used to thinking of this as "our space"

@rm-rf I think these rules seem reasonable and far more permissive than the unwritten IRC rules. On the other hand I don't really see the need to enumerate them like this since most reasonable people will make an effort not to be a nuisance to others, and unreasonable people... well, they can't be reasoned with anyway. So I agree that doing so is more likely to encourage people to be overly pedantic.

@OleksandrR. Ok, so everyone in the frequently in room list is a room owner... like I said, you can't kick, but you can move to a different locked down room (I think you can even if it's frozen)

@rm-rf don't you have to own the other room too? I can't get to the trash room for instance

Yes, just create one and add the others... The trash room is private and mod only that I use mainly to move sensitive info that is posted here (and sometimes, other random stuff)

you can also name it Trash, won't be a problem :)

@rm-rf mods have access without being owners? Or not?

@OleksandrR. yes, they have access to it without being explicit owners

You'll find the link under the room tab (below the list of gravatars)

btw, there's also this room if you want to post random stuff and try migrating to familiarize yourself with it:

chat.stackexchange.com/rooms/8075/trash

I don't think anyone will have to use this tool all that much — I've not moved any annoying messages in all of the approx 100k messages posted so far; just a handful of sensitive ones. However, we're also growing fast, so better be armed.

@rm-rf I can't make a private room so this one will have to be for non-sensitive (senseless?) things only

Yeah, just ping one of us and we'll do it

I wouldn't be too agressive with the definition of "senseless" though ;)

→ 1 message moved to Trash

Okay, I moved one as a test. Seems straightforward enough.

I am trying to print dodecahedrons around a space but the above.

Table[Translate[PolyhedronData["Dodecahedron"], RandomReal[30, {3}]], {2}]

@hhh It would be more appropriate to ask that question on the main site, where there are better tools to format code, include images, typeset the math, etc. Plus, that's rather the point of the main site! :)

@hhh I can't make sense of what you are asking. Why is your image skewed like that? Maybe you meant something like Graphics3D@Table[Translate[PolyhedronData["Dodecahedron", "Faces"], RandomReal[10, {3}]], {10}]?

@OleksandrR. Moved to: mathematica.stackexchange.com/questions/22136/…

@OleksandrR. that's not really the issue. Personally, I have no problem handholding the particular person @MarkMcClure mentioned; but if people start using this place as a real-time personal problem-solving space then that's not OK.

It's enough to have people behave like entitled brats on the main site.

@acl okay. Where would you draw the line? Is it okay to ask something in here that isn't related to a posted question? Does it matter who's asking?

@OleksandrR. I don't know where to draw the line. But I know the extreme limits.

@OleksandrR. I'd like to say it doesn't matter who is asking, but I at least make a conscious effort to judge everything without taking into account who is saying/doing it. Of course, not always successfully (and not always appropriate).

@acl I was asking mainly with respect to the conversation between R.M. and me on Saturday about passing filtered options between functions. He had mentioned options processing so I asked him if he knew of a good approach for that and we discussed some approaches. It didn't relate to a posted question but our discussion helped me to solve my problem. Acceptable or not?

@OleksandrR. I would say yes, because it's a discussion. Not you saying "why does nIntegrate[sin(x),{x=5..12}] not work?"

although I wouldn't mind that, once or twice. it's someone insisting ("please do a little more for me") that I'd have a problem with, because I view this as somewhere I can have informal chats about mathematica.

look, I would like this space to be more or less like a cafeteria at a physics dept at a university. people chat about lots of things, but if o-level students made a habit of spending time there and asking about their homework, you'd stop going there to relax.

(well, replace "physics dept at university " by something else, you see my point)

@acl okay, I understand that. :) Let's go with that, then.

oh, am I a room owner now? cool :)

what can I do now that I couldn't before?

aha, I can pin my messages!

what else?

@acl pin messages, move messages, see deleted ones (the moving aspect was the reason for making the regulars room owners)

@OleksandrR. yes just saw that.

I thought we could kick users too but turns out this isn't like IRC

and Wolfram language it is... Sigh

→ 3 messages moved to Trash

ha. thanks @OleksandrR.

@Rojo I'm going to stay optimistic. Perhaps this signifies an implementation of the language outside of Mathematica itself. At the Euro WTC, Brian Beckman of Amazon is talking about term rewriting in JavaScript

@Rojo You don't like Wolfram language? Personally, I was hoping for W.

@MarkMcClure yes, I was sort of trying to make the distinction between "people start using this place..." and "it randomly happened once".

@acl I can see that - thanks!

@MarkMcClure also, I realize that not everybody feels as comfortable as I do in telling people "you're wasting my time, go away" (not that I've ever done that here), so that could be a problem.

Well, I was amazed at your patience and quality of responses.

huhu

@acl That pretty much sums up everything that needs to be said on this :)

@MarkMcClure if you're talking to me, thanks!

@acl Yes.

@OleksandrR. when the "pitfalls for beginners" question was asked, what you just said in a comment to hhh's polyedron question is what first came to mind

Too many people around assuming based on nothing that what they choose to write should work as they wish. But someone convinced me I was being harsh so I let it go.

@OleksandrR. Outside Mathematica maybe, but outside anything that starts with Wolfram _?

@OleksandrR. I don't think you can see deleted messages... (except for the ones that were moved (because they're still undeleted, but in a different room (why is this turning into lisp? (help!))))

@rm-rf they have the history link

@Rojo aha, so we could then have a Wolfram Wolfram compiler

@rm-rf see, you need emacs

@OleksandrR. Interesting... I was not aware of this.

@Rojo What do you mean I can't feed LaTeX equations to Mathematica?

page not found

it's better off that way :)

@Rojo I agree it is harsh, actually, but I lose patience after more than a few silly questions

@rm-rf Haha

its totally stupid there are so many ways to type a formula in LaTeX, Mathematica has absolutly no chance for that

Then stop treating it as a typesetting program... your chances of being disappointed will drop drastically

in fact there is a limited possibilty with inputform

LaTeX ?

No, mma.

@MarkMcClure I don't like Wolfram. But I have no clear arguments in favour or against it.

(ok, I'm biased... I don't like the typesetting features in Mma and do my best to avoid it at all costs)

Keep in mind that Mathematica's typset forms are largely acceptable input as well and the ability to do so guided its development.

In fact, Mr Wolfram is what I like the least of the Mathematica universe. So, seing that his self absorption only grows as he gets old makes me fear a little bit for my dear Wolfram language :P

@Rojo I might just stay clear of this. :)

Was it announced today? I don't see anything on their blog...

@rm-rf He announced at the SXSW talk.

@rm-rf i would never typeset anything

in Mathematica

@MarkMcClure Ah, ok.

@DominicMichaelis Ok, I must've misinterpreted you then :) There were a few posts (a long time ago) where the OP was all like: "LaTeX is so advanced at typesetting... mma sucks, waaah", and my response was "Then don't use mma for typesetting!"

@DominicMichaelis Have a look at the PDF labs on this webpage: facstaff.unca.edu/mcmcclur/class/Fall10IntroToMathematica They were all typed with Mathematica.

Not typeset but typed.

@rm-rf i just mean that I can't think of that Mathemtica can interpret all those commands and won't get confused

or any other program

@DominicMichaelis ah, yes true... especially with tricks like \phantom{} and what not

and a \crushed and some empty groups for other atoms and all those fancy stuff

Note that the last one, in particular, has a little typeset math: facstaff.unca.edu/mcmcclur/class/Fall10IntroToMathematica/labs/…

I once squeezed space for a 4 page letter by using the savetrees package (check it out, if you haven't) and compressing all my equations... the publishing department undid all that and made my figure tiny instead :(

(not blaming them... totally expected it. The squeezing was just to get past the electronic submission)

And this guy here has a slew of typeset math: f521ffe9-a-b04e5bb8-s-sites.googlegroups.com/a/unca.edu/…

@rm-rf see, for APS journals, they now just count words!

Hmm @acl how do you feel about pinging? For instance I just saw @-rm -rf make a remark that browser chrome may not properly display the wolfram documentation online because of javascript settings. Now I wouldn't reply in a comment that my javascript settings are to allow javascript, as that would seem to be off topic for that Q&A. Of course -rm -rf is here and well... its not an issue.. but.. Well I guess it just seems a awkward if everybody goes "@somebody here is something totally unrelated"

all the time

@acl Time to pull out a GRE English reference book then ;)

@rm-rf or write in german!

i can write in german :D

@acl DasIstEvenGanzBetterisch!

I feel it is counterintuitive as well as inconvenient that I cannot just ping people from a room that I make myself

@JacobAkkerboom I don't have much problem with anything really, I was just saying that keeping this relaxed would be good.

good :) I agree. Hm I guess also discussions about "bad questions" made me a bit nervous about obliging to "rules" that may not be easy to see

@JacobAkkerboom That is not a problem at all... chat is certainly the place for such stuff and don't be afraid to ping. We were just discussing about random/new users bugging regulars to answer or look at their question. A lot of the discussion was preemptive rather than reactive. :)

@JacobAkkerboom The rules are 1) There are no rules 2) Since you insist on reading past point 1, the rule is "Don't be an ass" :)

Haha (I laugh because that was funny (stop reading my thoughts!))

woohoo can I be room owner too? :O

@rm-rf @JacobAkkerboom yes, eg I meant don't do this (although evidently people will answer your question anyway)

@rm-rf so I was reading about linking to a part of the documentation page.. is there also a way to link to a section in an answer?

You mean, link to a section within an answer on this site?

Yeah

no

besides, you'd need those only for Leonid's answers :)

Ah alright. I was thinking about these "canonical questions" as somebody wanted to take a column from a matrix which I thought must have been asked thousands of times. Then I remembered the meta Q&A about canonical stuff and about referring to parts of a larger answer (or something :P)

not really thousands I guess

i finally find a way to make a dyadic product ( w.w^T )

@DominicMichaelis KroneckerProduct[Subscript[a, #] & /@ Range[3], Subscript[b, #] & /@ Range[3]]

oh thats much neater than my one :D

my problem was that i gave mathematica always wrong dimensinos

dimensions*

Is it possible to launch the kernel on OS X so that I have a terminal where I can read debugging output?

The debugging output is coming from a library so it's not easy to redirect it

@Szabolcs Applications/Mathematica\ 9.app/Contents/MacOS/MathKernel

@Szabolcs Just start it

I mean while using it with the Front End!

Sorry for being unclear :)

Aren't all kernel messages directed to the FE (or you can make them)?

no

@Szabolcs you could start it in console and connect to it by creating a new kernel launch configuration.

What the kernel writes to stdout doesn't go to the FE

Yes, that's the non-automatic ugly way ... do you know if Termianl.app has any command line options so I can get it to launch the kernel indirectly? Let me google for that ..

@Szabolcs What do you mean indirectly?

@rm-rf Just starting M normally and when the kernel is started, it is started in a terminal and then the FE connects to it.

I suppose

@halirutan but I thought Szabolcs didn't want to do that...

@rm-rf He wants to see the terminal

So instead of just starting a kernel in launch mode, he will start a termin and in it a kernel which is then used .

@halirutan Ok, I'm confused... that's what I mentioned above (which should be used with rlwrap for maximum goodness)

but then, I don't do very much of headless kernels/connecting/detaching/snooping/etc, so I could be making no sense...

@rm-rf The difference is: in one case you would have to open a terminal and start the kernel there, in the other case you just type 1+1 in the FE and where it usually just starts a kernel, it starts now a terminal and in it a kernel so that Sz sees all stdout messages.

@halirutan aaah... that makes sense :)

Here's what I mean:

This doesn't work yet because I'm probably giving the wrong MathLink options

sorry about the type in LinkProtocol, already corrected that

I have a symlink to MathKernel in ~/usr/bin/math and everything after --args is supposed to be passed to the program

Hi @Szabolcs

Hi @OleksandrR.

First start your kernel in the terminal. Then evaluate $ParentLink = LinkCreate["name"], where name is some name for the link

Then start the front end as Mathematica -mathlink -linkmode connect -linkname name

Hmm ... I think the problem is with how I'm trying to pass command line arguments on OS X

I'll put this aside for a while and will just use the kernel from the terminal

@Szabolcs I got it.

Create an applescript

tell application "Terminal"
	activate
	set shell to do script "/Applications/Develop/Mathematica804.app/Contents/MacOS/MathKernel -Linkmode listen -LinkProtocol SharedMemory -LinkName \"blub\" -mathlink" in window 1
end tell

Store it somewhere for instance here in /Users/patrick/Documents/Kernel.scpt

@Szabolcs Then you create a new kernel configuration in Mathematica

Arguments to MLOpen is then

-LinkMode Connect -LinkName "blub"

and Launch Command is

osascript /Users/patrick/Documents/Kernel.scpt

@Szabolcs sorry, it's my mistake. Apparently this just ends up launching the FE with two kernels attached (or 3 in the case of version 9) but no obvious way to access the one you just connected. What you have to do is add a new kernel in the Kernel Configuration Options and put -LinkMode Connect -LinkName NAME in the arguments to MLOpen box. Then you can connect to your running kernel from the front end normally.

@OleksandrR. Thanks! I should have known the answer to this one ...

@halirutan Thanks, I will use this one! I should really learn a but about AppleScript

@Szabolcs Yep, you should. Learn for me too.. I just googled this one. Was my first applescript too ;-)

@Szabolcs applescript is the language I have most trouble writing. it's hard to predict the right form to use. I don't know what whoever designed it was thinking (but it is easy to read)

@acl Have you heard about Inform 7? It is a rather unique programming language designed for writing interactive fiction. This is actually the source code of a game. Actually this system works quite well, at least judging by the number of games written in it, compared to more traditional systems like TADS.

Some said that if perl is a write-only language, Inform 7 must the the prime example of a read-only language ... I guess AppleScript is the same.

@Szabolcs I have, yes, but never tried it. I've only ever played such games, never wrote them.

Good to know I'm not the only one playing them :-)

@Szabolcs unfortunately not any more... and never really played games that much

Actually I haven't played for years. But when I did play, I quite enjoyed them. They're also good for learning English.

@Szabolcs yes: "Sherlock: The Riddle of the Crown Jewels" is how I learned that one "knocks" on a door...

never finished it, because I didn't know enough english and couldn't work out what to do (and was too proud to ask my parents)

@Szabolcs on an unrelated note, I've seen advertisements for everything from washing machines to computers by people leaving recently, but no bikes....

that's why I did not forward you any

If there's someone who wants to sell a bike, I'll definitely take it!

guess you are not interested in a nice, top-loading washing machine? slightly used?

Thanks again for checking

No, I don't think so :-)

@Szabolcs it's not a problem, I've simply subscribed to a list and get everything.

@Szabolcs apparently, the institute's washing machines (in the guesthouses) were recently stolen

think again

@acl O_O you're kidding, right?

@Szabolcs nope

(or were they vandalized? anyway, there was an announcement, and general panic amongst my bachelor colleagues--ie, all of them)

no, stolen I think

anyway

Wow, there must be some pretty strong burglars there. A washing machine is not an easy thing to just carry out the window

@Szabolcs they were in a basement. so, up a stairs, out through the door.

@Szabolcs maybe they were pranksters. some people in the year ahead of me at my school built up the main entrance overnight (a proper wall, with bricks etc, and even painted it) as a prank. maybe this was the same.

I should try to find more details, I was at a conference when it happened

@acl I would have loved to see that :D

OK, I need to go home for dinner

@Szabolcs it was absolutely hilarious

See you later!

see you