Wolfram Mathematica

Adam Dreaver

12:00 AM

Google I/O 2009 - The Myth of the Genius Programmer @ youtube.com/watch?v=0SARbwvhupQ

Oleksandr R.

@acl reminds me of the wide belief in the 1890s that physics was a done deal

acl

@OleksandrR. I am not sure it was as wide as it's made to be nowadays

@AdamDreaver programming is different though...

there's a difference (for the purposes of this article) between someone exceptionally better than their colleagues, and someone who opens up new areas. also programming is made up as we go along, physics (say) isn't.

rm -rf

@acl I think the more important question is: How do I get a 1 page blog post published in Nature?

(this also illustrates what science/academica has turned into...)

acl

@rm-rf invent integration?

(OK not nature, but...)

rm -rf

Ah, that would certainly be one way...

acl

12:07 AM

@rm-rf I have a reasonable first-hand knowledge of how academia works :)

or :(, depends!

rm -rf

D=;

Oleksandr R.

@acl looked at with a modern perspective, it's obvious that a lot was still to be discovered, but I suppose without being there at the time it's hard to understand what typical physicists thought at the time. Certainly things like the Fraunhofer lines had been observed already but quantum mechanics was viewed then, as far as I can tell, much like string theory is today.

acl

@OleksandrR. Well that is wrong on several levels, but it's late and maybe not the best time to do this :)

Szabolcs

Why is Predcitions` in the context path? Is there anyone here still on 9.0.0? Can you test is please?

rm -rf

@acl There is never a "not the best time":

What do you want me to do? LEAVE? Then they'll keep being wrong!

Oleksandr R.

12:12 AM

@acl up to you :P

Szabolcs

1. Make sure the predictive interface is enabled.
2. Restart the kernel.
3. Evaluate 1+1, make the prediction bar come up, ask if the result is a prime using the prediction bar
4. Check is Predictions` is in the $ContextPath

Can someone please test this for me with 9.0.0 (not 9.0.1)?

I'm getting symbol conflicts from that context. It doesn't look like it should be in the context path.

OK night all

Night @acl

Night

Night!

12:16 AM

@Szabolcs Honestly, they could have at least put the symbols that don't start with caps in a private context

Szabolcs

@Rojo I suspect that it's a bug, and that 9.0.0 didn't have this problem. But I don't have 9.0.0 any more

Wait, actually I might have it on a different computer

Rojo

Haha undo-for-mathematica.alaifari.com

rm -rf

@Szabolcs No, not in the context path

Szabolcs

@rm-rf That's on 9.0.0, right? Thanks!

rm -rf

Yes

{"DocumentationSearch`", "ResourceLocator`", "Predictions`",} <-- new contexts in the path in 9.0.1

Rojo

12:20 AM

@rm-rf at least the other 2 have their symbols in caps

Szabolcs

then just $ContextPath = DeleteCases[$ContextPath, "DocumentationSearch`"|"ResourceLocator`"|"Predictions`"]

Actually these only appear when I use the feature, e.g. search the docs, use the prediction bar

it's some DeclarePackage type auto-loading

@rm-rf The other two come up in v9.0.0 too if you search the docs

rm -rf

Ah, ok. I tried it in a fresh 9.0.0 session

Oleksandr R.

@Rojo hahaha... really? They must feel very strongly about it to have made an entire site, complete with forum

Szabolcs

At least there's a place to direct the rants to

I am going to report this rogue context to support. It looks like a bug, doesn't it?

Oleksandr R.

Actually, she has an extremely slick website in general for a PhD student (or for anyone, really)

Szabolcs

12:29 AM

Maybe converting the answer to a comment would have been better. A comment is harmless.

rm -rf

@OleksandrR. derp... she removed attribution to the designer :P

@Szabolcs Nope. Not the place for petitions and signature hunting... Besides, we both know it's pointless.

Oleksandr R.

@rm-rf well, okay, it's a Wordpress site... but most people don't have even that... I know I don't

Szabolcs

I want one now! A pretty one :)

rm -rf

@OleksandrR. She has stuff to post up there... I spent a while learning Haskell, learnt to use Hakyll, designed my own 100% custom site, but... I didn't have anything to put up! :(

the site is still on my hard drive, but now completely broken because hakyll 4 changed completely from hakyll 3.5 and I need to delve deeper into monads to figure things out (aka NOT happening)

@Szabolcs twitter.github.com/bootstrap

Oleksandr R.

@rm-rf interesting. I didn't know such a thing existed. I'm completely out of touch with web technologies to be honest (I find them boring)

rm -rf

12:35 AM

It actually is pretty nice and you can set it up pretty fast too

@OleksandrR. Hakyll isn't really a web technology... it's simply a static site generator written in haskell. You still have to do the CSS, JS part, but for the HTML, you create templates and use hakyll to insert stuff in the templates and build cross-dependencies

So you have a 1 time effort and then for everything else, you can use markdown

Oleksandr R.

Okay; I wasn't really talking about Hakyll specifically when I said that. I meant that although I'm aware that there are many similar things, I don't know any of them by name :P

rm -rf

Ok... you replied to it explicitly, so I assumed you were :)

Oleksandr R.

I also don't have any desire whatsoever to write cross-browser-compatible CSS and JS...

rm -rf

pfft... I always stick it to the IE man

And no, I don't have such a desire either... "Can't load my page? Too bad, you probably use a sucky browser. I'm already not interested in you seeing my page."

Oleksandr R.

Is there one, in this day and age? As an Opera user, what annoys me though is that people tend to test their site on Firefox and Chrome (both of which use MANY proprietary and nonstandard extensions), say, "eh, looks okay" and consider it done

rm -rf

12:44 AM

There are sites that look very different on chrome and ff too... I've never really cared anyway. Somehow, it feels to me like I've "missed out" on the browser explosion in the last 5 yrs.

With Firefox 1.0, I knew enough about it to write a few extensions and custom color themes for it, but that was all that could be done with it back then. One can do a hell lot more with browsers these days

(ok, if you go back far enough, Netscape was just a cardboard box)

Oleksandr R.

Was it ever much more than that though? That's why I switched to Opera... Netscape 5... ugh

rm -rf

Well, now I can play angry birds in chrome... that counts for something

Oleksandr R.

I suppose I shouldn't complain too much about proprietary extensions, though. I once used a Cray pointer in Fortran.

rm -rf

Not sure what they are, but I'll take that to mean something that would've made Stallman explode with rage

Oleksandr R.

A Cray pointer is pretty much a C pointer, but in Fortran. The main disadvantage is that there's no way to tell when it's null. Anyway for my purpose it was much more convenient than the Fortran 2003 POINTER

gfortran supports Cray pointers, so I guess Stallman isn't that concerned about it

Rojo

12:57 AM

I suddenly felt like answering the FoldWhile question with the worse implementation I can think of. Amazing to what extent I not want to work or sleep

rm -rf

@Rojo I once wrote an answer to a "how do I implement Fold" question and Leonid said that it had hidden recursion with exploding complexity... I didn't attempt this one because I'm sure my brain is wired to come up with something very similar :P

Oleksandr R.

Here's a user who reminded me a bit of Stallman... "it's my computer, damn it, so decompiling closed source programs is completely morally justifiable!" Of course, Stallman says that after having spent 35 years writing his own OS and conducting political advocacy in that direction, whereas this guy seems to be a bit of a script kiddie (what some would uncharitably refer to as a "lamer", I think...)

rm -rf

heh... it's been closed a while. Time to start accumulating the delete votes...

Oleksandr R.

His latest question is on just the same topic but less overt about it

Rojo

@rm-rf Damn you

I was confused

My ugly foldWhile returned If[0 + 1 < 5, 5]

because you made me Remove[Plus]

rm -rf

1:11 AM

HAHAHA!!!

lol

Rojo

@rm-rf Grr. Ok, fixed and my answer is up

@rm-rf I'm trying to scare Leonid as you asked

belisarius

@OleksandrR. He's 19. I only hope life to be kind enough with him so he could get the time to learn.

Oleksandr R.

@belisarius actually I'd thought the exact same thing when I saw his responses to some of my comments

Asking about it is totally excusable and what he does in his own home is none of my business. On the other hand I don't want these questions to alienate anyone

I just hope he doesn't figure out how to decode the Encoded packages... or at least not if he's going to write about it on his blog afterwards

Szabolcs

1:28 AM

@rm-rf That's the problem: you ("one", but I don't want to say "one") either have time to hack together a site and do programming-type work on it, but then don't have content; or you have content, but no time/patience for the programming side. I think I have content now, but I really don't want to hack anything low-level or use anything that may not be around in 3 years

I guess it's wordpress then .. ? Tried it once, didn't love it.

belisarius

@OleksandrR. That's exactly what he has to learn: "Do your business, and enjoy it. Do it with all your might ... but don't expect others to approve it"

Oleksandr R.

@belisarius that's a good way to put it

belisarius

:)

@OleksandrR. I had 19 once upon a time too

Oleksandr R.

Me too, and not so long ago (or so it seems)

Murta

Someone know if there is some option for tab indentation in Mathematica, so I could select some text, press tab, and have it indented instead of deleted?! Some shortcut for that?

Oleksandr R.

1:41 AM

@belisarius: smbc-comics.com/comics/20130203.gif

belisarius

@OleksandrR. Good one!

rm -rf

@Murta The only known shortcut is to use the Workbench =)

Oleksandr R.

I linked that partly in retaliation for R.M's xkcd meme. From now on whenever someone stars an xkcd, I'm going to link some SMBCs in the hope someone will star that

rm -rf

Alternately, you can be room owner and pin anything you want :D

Oleksandr R.

@rm-rf no thanks, I don't feel like taking any actual responsibility here :P

Plus, it's no fun if you star your own messages. Like laughing at your own jokes

rm -rf

1:46 AM

Lol... all you get is your name in italics and free supply of starbucks

belisarius

@OleksandrR. So are there un-actual responsibilities here?

rm -rf

@OleksandrR. No, only pin... at which point it becomes obvious who pinned what. (pinning and quickly unpinning is a backdoor to self-starring)

Rojo

@OleksandrR. Why wouldn't you laugh at your own jokes?

Oleksandr R.

@belisarius I'll gladly take on the non-actual responsibility to show up here whenever I damn well like and say anything or nothing that may come into my mind. In fact I think I already did

belisarius

@OleksandrR. 42?

Oleksandr R.

1:49 AM

Nope. In addition to xkcd I also have a vendetta against HHGTG.

Szabolcs

I feel this should be undeleted: mathematica.stackexchange.com/a/19133/12

Polygon should be mentioned in an answer to that question. It's simpler than FilledCurve.

FilledCurve must be mentioned too, but doesn't make the polygon answer obsolete.

@rm-rf Maybe you could be so kind to ping @m_goldberg here in chat and point him to my comment above? I can't ping him because I'm not a mod and he hasn't been here recently.

belisarius

@Szabolcs You should make a super-pinging palette

Szabolcs

:P

Rojo

@rm-rf used your trick in the second version

Murta

2:10 AM

@rm-rf I really don't like Workbench!.. :(

Rojo

Bye all

Rolf Mertig

2:32 AM

@Murta Why not? Use it as a plugin to Eclipse. It just works. Needs some time to get used to, but there you have multiple undo ...

belisarius

@OleksandrR. What is wrong with xkcd?

Rolf Mertig

Plus you can put your code easily into code versioning, and then onto [googlecode] (code.google.com/p/packageinstaller/source/browse/…) or even your own repository (e.g. RhodeCode).

m_goldberg

5:11 AM

I got a ping about an answer I deleted. I'm not sure how to respond. I deleted the answer because it duplicated the one of the methods given by Vitaliy Kaurov in his answer. I only noticed the duplication after I posted. I think it should be deleted even if it has an upvote. Don't think we need duplicate answers.

@Szabolcs I added a message about this. Sorry I didn't make it an answer to this one. I never used chat before and don't really understand how it works.

rm -rf

@m_goldberg Ah, I see the method in Vitaliy's now... feel free to delete it again and it won't be undeleted

m_goldberg

@rm-rf I wil.

Szabolcs

5:28 AM

@rm-rf I cannot.

rm -rf

@Szabolcs ok

Ajasja

7:35 AM

@Murta I'd give WB another try. I fought it for a long time, but now I've learned to love it:)

P. Fonseca

9:07 AM

Hi all

I'm looking for examples of uncertainty analysis by fuzzy alpha-cuts, done on Mathematica. I have no problems doing it, but I'm looking for tricks, code structure good practices...

Anyone?

(and I have no money for the existing package sold by Wolfram... although I don’t know its price…)

halirutan

3:29 PM

Oh my goodness.. The C-Compiler does not work anymore in 9.0.1.
@Szabolcs Do you have the same behavior?

My system is OSX 10.8.2 and I have both, gcc and Intel compiler installed:

<< CCompilerDriver`
CCompilers[Full]

{{"Name" -> "GCC",
  "Compiler" -> CCompilerDriver`GCCCompiler`GCCCompiler,
  "CompilerInstallation" -> "/usr/bin",
  "CompilerName" -> Automatic}, {"Name" -> "Intel Compiler",
  "Compiler" -> CCompilerDriver`IntelCompiler`IntelCompiler,
  "CompilerInstallation" -> None,
  "CompilerName" -> Automatic}, {"Name" -> "Generic C Compiler",
  "Compiler" -> CCompilerDriver`GenericCCompiler`GenericCCompiler,
  "CompilerInstallation" -> None, "CompilerName" -> Automatic}}

Oleksandr R.

@halirutan I thought the CCompilerDriver` package hasn't changed at all since 8.0.1?

halirutan

@OleksandrR. I have no idea. It worked in all versions including 9.0.0

And currently I have no time to debug. I just have to work :-(

Oleksandr R.

Oh no, pardon me, there have been some changes. Still, 9.0.1 is working fine for me on Windows, both with Visual C++ and MinGW-w64

halirutan

@OleksandrR. I'm really wondering what unit-tests they have at Wolfram Inc. You can never get all cases, but OSX is a pretty solid installation and I haven't done anything nasty.

Oleksandr R.

For one thing the message I made fun of regarding ILP64 has been changed slightly. It's still wrong, though. :)

Oleksandr R.

3:49 PM

@halirutan $SystemID is what?

halirutan

@OleksandrR. "MacOSX-x86-64"

Oleksandr R.

And the SystemFiles/Links/MathLink/DeveloperKit/MacOSX-x86-64/CompilerAdditions directory is present and correct?

halirutan

Oleksandr R.

OK. I am just looking at things that changed between 8.0.4 and 9.0.1 that might have broken things.

halirutan

@OleksandrR. I could compare 9.0.0 and 9.0.1 myself but today I just wanted to work.. let me make a quick diff..

Oleksandr R.

4:03 PM

Yes, probably better to compare 9.0.0 with 9.0.1 but I don't have 9.0.0 installed any more

Szabolcs

@halirutan I don't have that problem here, same system.

halirutan

@Szabolcs Hmm

Szabolcs

To list the installed compilers (not the supported ones), just use CCompilers[], not CCompilers[Full].

Also, I wonder what compiler it really uses internally. OS X ships with both clang and an old gcc

cc calls clang.

@halirutan I simply installed Xcode from the app store and then went to preferences -> downloads to install the command line tools. I have Xcode 4.6 now and cc --version in the terminal gives me Apple LLVM version 4.2 (clang-425.0.24) (based on LLVM 3.2svn).

gcc --version gives i686-apple-darwin11-llvm-gcc-4.2 (GCC) 4.2.1 (Based on Apple Inc. build 5658) (LLVM build 2336.11.00).

halirutan

@OleksandrR. @Szabolcs Aehmm, the problem solve itself.

Szabolcs

What compilers do you have there?

halirutan

4:11 PM

It's working now and I simply restarted Mathematica

Oleksandr R.

@halirutan oh well, all's well that ends well...

Szabolcs

Does anyone else feel that 9.0.1 was a bit rushed? They left in that testing palette and leave things in $ContextPath that don't belong there

But there seem to be lots of fixes too

What I'm hoping for is more frequent bugfix releases ... if 9.0.2 comes in two months then I'm happy

halirutan

@Szabolcs I would say that they just had to keep the speed they were using when throwing out 9.0.0

Szabolcs

I think the 8.0.0 --> 9.0.0 gap was the longest ever between two major releases. They were clearly under pressure.

And I really wanted to see 9, even if it still had problems ...

Oleksandr R.

@Szabolcs only 2 years... 2.0 to 3.0 was 5 years, 3.0 to 4.0 3 years, 4.0 to 5.0 4 years, 5.0 to 6.0 3 years

so if anything it was under-done...

Szabolcs

4:17 PM

but there were 4.1, 4.2, 5.1, 5.2 at that time

Oleksandr R.

Sure but do they count as major releases? I'm not sure.

Szabolcs

They had new features.

wolfram.com/mathematica/quick-revision-history.html

For example it seems GraphPlot came in 5.1 (not sure), that's a major new feature

Oleksandr R.

OK, well I'm not disputing they had new features (in fact I think by that standard 7.0 should have been 6.1 and so on) but wasn't sure what you considered a major release

In any case I agree that 9.0.1 came quickly and it's obvious that they were fixing bugs right up until the release date

Szabolcs

The versioning system is a bit messy

halirutan

@Szabolcs Yes, consistently non-existent, compared to other software products.

Szabolcs

4:27 PM

1.2.x releases should't really introduce new features and especially they shouldn't change the API, but they did. E.g. the Radon function's output format changed sometime between 8.0.0 and 8.0.4

halirutan

@Szabolcs From 9.0.0 -> 9.0.1 a new feature was introduced: Autocompletion ;-)

Szabolcs

Actually this is what happened: dsp.stackexchange.com/questions/374/river-detection-in-text/…

@halirutan Did it actually stop auto-completing some things? E.g. try auto-completing anything from from NumericalMath context

halirutan

@Szabolcs Yes. That worked before. I know it because I had to put the context completion in my camel humps.

But now they store everything in the FE. Maybe it was too much memory.

Szabolcs

@halirutan So previously the FE queried the Kernel for each completion but now it doesn't?

halirutan

@Szabolcs Yes. Thats why I cannot make the camelhumps working. There is no traffic between FE and Kernel I could interfere with.

Szabolcs

4:39 PM

@halirutan How does the FE learn about new symbols? When they're defined?

halirutan

@Szabolcs I assume it parses the packets.

@Szabolcs Oh wait. There is a new Packet:

CallPacket[FrontEnd`UpdateKernelSymbolContexts["Global`", {"Global`", "PacletManager`", "QuantityUnits`", "WebServices`", "System`"}, {{"Global`", {"xxx"}, {}, {"xxx"}, {}}, {"System`", {}, {}, {"$RecursionLimit"}, {"ToExpression"}}}]]

Nevertheless, all the magic happens behind closed doors now.

OK guys, lunch is waiting. See you later.

CBenni

Isnt there any way to calculate $limsup$s with mathematica?

I can do it for constant functions >_<

but there is no LimSup command

rm -rf

4:57 PM

There is a Limit and a Max command...

Szabolcs

5:11 PM

@CBenni Is MaxValue what you need? (check the docs)

CBenni

@Szabolcs no, not really. It can be used to calculate these values, but it does not help me, I want to calculate the $\limsup$s, not the $\sup$s, and Mathematica cannot find the limit for a function like MaxValue[{f[x],x>x0},x]

Szabolcs

6:22 PM

@CBenni It seems Limit can do this sometimes. It give an interval as a result sometimes, e.g. Limit[(1-Exp[-x])Sin[x], x -> Infinity] give Interval[{-1,1}], so the lim sup is +1. Is this what you need?

gives** I don't know why that s always goes missing at the end

Arnoud Buzing

@halirutan Just to confirm: there is no issue here, correct? Or there was, but you only encountered this issue once and went away after a restart of the kernel?

Szabolcs

6:46 PM

Hi @Arnoud! There's a new problem in 9.0.1: the Predictions` context somehow got into the context path. It's causing symbol conflicts with a soon-to-be-released package of a colleague of mine. What we were wondering about is how long before this is expected to be fixed? If the wait is comparable to the one between 9.0.0 and 9.0.1 then it's no big deal. But if it's many months away then he would need to put a workaround in the package.

As I understand some code in the predictive interface can be updated through paclets, but perhaps this doesn't apply to the code that caused the bug?

Arnoud Buzing

@Szabolcs This should be addressed with a paclet update that will be going out today.

2

Szabolcs

@ArnoudBuzing That's great news, thank you!

halirutan

7:53 PM

@ArnoudBuzing We'll see. I didn't do anything unusual.

Currently, Compile to C seems to work properly. When I have the behavior again, I try to find out how to reproduce it.

@ArnoudBuzing But have you seen this magnificent piece?

yesterday, by halirutan

Try to define a function großÄ[x_]:=x and then start to write it and use autocompletion.

Rojo

8:19 PM

@halirutan That's exactly what I was talking about yesterday :P

halirutan

@Rojo Yes, but I hadn't really noticed it. Today, it got clear what you meant.

Arnoud Buzing

8:57 PM

@halirutan That's pretty bizarre (and obviously a bug). Checking into it now.

Szabolcs

CUDAQ[] is crashing for me in 9.0.1. I will now try CUDAResourcesInstall[Update -> True].

halirutan

@Szabolcs I can try this tomorrow.

Szabolcs

It is still crashing after CUDAResourcesInstall[Update -> True]. @Arnoud do you know anything about this problem? I have $Version === "9.0 for Mac OS X x86 (64-bit) (January 24, 2013)". Later today I'll try removing the config directory and starting anew, btu not right now.

belisarius

Curious ... 2 consecutive votes on a rather old answer of mine ... What could it be?

Szabolcs

Someone has linked to it

belisarius

9:07 PM

@Szabolcs Nope I got it. @Rojo linked to a question of his that in turn is linked to my answer :)

linking is believing

Szabolcs

Sorry, there's nothing wrong with CUDALink. It was my mistake. Forgot that I disabled the nvidia card ...

belisarius

@Szabolcs In Spanish Envidia means "envy" ... and are "homophones"

Arnoud Buzing

@Szabolcs It wasn't working for me either ... And then I remembered I have a AMD Radeon in my Mac ...

belisarius

@rm-rf I flagged a comment of yours while trying to do ... nobody knows what. Sorry

Arnoud Buzing

@Szabolcs So there is no issue here, correct?

rm -rf

9:23 PM

@belisarius Doesn't seem like you've flagged anything...

~~Or was it in chat?~~ nope, not chat either.

belisarius

@rm-rf I swear. In a deleted answer. Perhaps that is the reason you can't find it

@rm-rf Whatever. I just wanted to say that it was an error

rm -rf

@belisarius I don't see any flags... Maybe comment flags are disabled on deleted posts. In any case, no worries :) That comment was probably useless anyway...

There's no penalty for a flagged comment or a declined flag, so don't worry about it... worst that could happen is the comment gets deleted or the flag gets declined :)

belisarius

@rm-rf Nope, it was on the filling curve question (asking the user to undelete) The comment is useful ... if he can see it :)

rm -rf

@belisarius Oh, that... I undeleted it at first. mgoldberg was here yesterday and he said that Vitaliy had already mentioned that approach in his answer and that MG didn't want to leave a duplicate answer (which I agree with). I told him to delete it again and that we'll leave it alone

belisarius

@rm-rf Ok!

rm -rf

9:29 PM

I do see that there was a flag on the comment, but no sight of the flag itself...

belisarius

@rm-rf A shadow of a shadow

rm -rf

Anyway, I deleted my comment... your flag has been marked valid =) Enjoy!

belisarius

@rm-rf You made my day :)

Ajasja

@Szabolcs How can one get the update? (Or is it automatic)?

Could someone tell me why Clear@b; b = b + 5; Exceeds the recursion limit?

rm -rf

@Ajasja What was the answer you expected?

Ajasja

9:38 PM

I can understand why Clear@b; b = b + 5 exceeds the recursion limit

Set returns b+5 which again evaluates b etc...

@rm-rf But I guess I expected the ; to suppress the first b+5

rm -rf

The presence or absence of ; shouldn't change the definition in this case...

Whatever happens in b = b + 5 happens here too... just that the output is suppressed

ssch

still a bit odd that during evaluation of the right hand side it already knows that b = b + 5

Ajasja

@rm-rf Yes, but actually probably the first part is OK. Clear@b; b = b + 5 ; ?b returns b = b +5

So the problem actually comes from Set returning b + 5

@ssch I don't think it does. But after Set returns b + 5 then the definition takes effect. So My purely theoretical question is: Is it possible to convince Set not to return b + 5. But probably there is := for that purpose :=)

ssch

yea, you are right

Ajasja

Oops, got to go -- wife calling. Bye!

Arnoud Buzing

9:59 PM

@halirutan Confirmed with the developer, he is looking into the issue now.

halirutan

@ArnoudBuzing Thanks. That saves me from trying to reproduce the behaviour.

Rojo

10:36 PM

@belisarius, hey. What did I link to what?

ssch

10:52 PM

Is there some function similar to Tuples that only returns the sets that are unique up to sorting? Like Union[Sort/@Tuples[ ... ]] (halfway between Subsets and Tuples)

Rojo

@ssch Not sure I understand

Can you complete the ...?

ssch

@Rojo Union[Sort /@ Tuples[{1, 2, 3}, 2]] gives {{1, 1}, {1, 2}, {1, 3}, {2, 2}, {2, 3}, {3, 3}}

instead of Tuples[{1,2,3},2] that gives {{1, 1}, {1, 2}, {1, 3}, {2, 1}, {2, 2}, {2, 3}, {3, 1}, {3, 2}, {3, 3}}

Rojo

@ssch Gotcha

ssch

I'm just goofing around with project euler and hoping to save myself some time :)

Szabolcs

@Ajasja I don't know. I believe it is automatic.

Rojo

11:01 PM

@ssch What problem? :P

I just got home and am having some icecream

I konw I did exactly that more than once

but I don't exactly remember how

belisarius

you linked, you're liable

ssch

@Rojo 254 I thought I'd only check {d1,d2,d3,..,dk} and then I know for each of Permutations[ {d1,d2,d3,..,dk} ]

(by {d1,..,dk} I mean the number n = FromDigits[{d1,..,dk}])

but I'll just change my approach and iterate n in such a way that it's always in increasing order

belisarius

@ssch Link to the problem?

ssch

projecteuler.net/problem=254

belisarius

tnx

@ssch five chained functions is too much

Xerxes

11:10 PM

Here's a fast way of doing ordered tuples: orderedTuples[list_,n_]:=First/@Nest[Join@@(ReplaceList[#,{{L___},{A___,b_Integer,C___‌}}:>{{L,b},{b,C}}]&/@#)&,{{{},list}},n]

err, assuming list is a sorted list of integers

ssch

that it is :)

doesn't seem to give right result though, perhaps I copied it wrong, orderedTuples[{1, 2, 3}, 2] == {{3,3}}

Xerxes

hmm, it worked for me; maybe I copied it wrong?

Rojo

@belisarius Come ooon

What link

belisarius

@Rojo Only if you kiss me

Rojo

@belisarius You should kiss the toad, maybe it turns into a princess

Xerxes

11:15 PM

ugh, there's some extra spaces after C___

belisarius

@Rojo or a spouse ...

Xerxes

whitespace!!! (shakes fist)

belisarius

@Rojo Your second comment on whuber's question

Rojo

@belisarius Ah, your graph based half solution

belisarius

@Rojo nope, thet is the curiosity

the one upvoted twice today was linked by that one

Rojo

11:19 PM

@belisarius Quite indirect

Oh, nice, I'm 4 upvotes away from 20k

belisarius

@Rojo yup funny

Rojo

I'll let it happen by itself, won't answer anything until then

I'll watch it happen, slow, while having icecream

belisarius

@Rojo I'll start cross linking all my answers

Rojo

@belisarius Haha, leonid does that

@ssch There are plenty of ways to get those sets, starting from what you suggested. The question is, which is fastest-less memory hungry

rm -rf

@belisarius No, there is a direct link to that under DL's answer...

ssch

11:29 PM

@Rojo Yes now I just do it stupidly with d={0,0,...,1}; from end find first non-9, increment, let all elments from there to the end be what it incremented to

Xerxes

@ssch ok, this should work on anything:

orderedTuples[list_List,n_Integer/;n>0]:=First/@Nest[Join@@(ReplaceList[#,{{L___}‌,{A___,b:Blank[Head[list[[1]]]],C___}}:>{{L,b},{b,C}}]&/@#)&,{{{},Sort[list]}},n]

Rojo

@Xerxes That would be faster than, say

ssch

@Xerxes Now it gives me only {}

Rojo

orderedTuples[list_] :=
 list[[#]] & /@
  Flatten[Table[{j, i}, {j, Length@list}, {i, j, Length@list}], 1]

or orderedTuples[list_] := list[[Flatten@Table[{j, i}, {j, Length@list}, {i, j, Length@list}]]] ~Partition~ 2

Ah, but that's for 2, and you wanted n

Xerxes

orderedTuples[list_List,n_Integer/;n>0]:=First/@Nest[Join@@(ReplaceList[#,{{L___}‌,{A___,b:Blank[Head[list[[1]]]],C___}}:>{{L,b},{b,C}}]&/@#)&,{{{},Sort[list]}},n]

I dunno, is there some way of marking up code here? it keeps getting garbled

Rojo

11:37 PM

@Xerxes Try backticks

but I'm not sure

Xerxes

@Rojo: no, you're right; that's faster

orderedTuples[list_List,n_Integer/;n>0]:=Module[{sym=Array[Unique[]&,n]},list[[‌#]]&/@Flatten[Table[sym,##]&@@(Append[Reverse[#],Length[list]]&/@Partition[Prepen‌d[sym,1],2,1]),n-1]]

Transcript for