Wolfram Mathematica

halirutan

12:00 AM

Has someone every wondered what exactly is set if I do something like f'[x]=Cos[x]?

yoda

@PFonseca no problem :) At least, I know spending all that on mma.se won't be a waste, given the way it's shaping up =)

Mike Honeychurch

presumably the people best placed to make an Mma addition to prettify.js are Wolfram since they already have the list rules used for syntax styling??

acl

@halirutan yep, why don't you post the question? :)

yoda

@MikeHoneychurch are those rules the same as what you'd have in JS, which is most likely going to be a big list of regexes? or can they be easily transferred?

@halirutan I was going to say that it should be in some *values of Derivative, but it isn't

acl

@yoda it's SubValues[Derivative]

halirutan

12:12 AM

@acl, I want this site to reach public beta. Maybe this was a dump question and I got a warning while creating a post, that nice questions are crucial to reach the next stage..

yoda

yup, it's there in SubValues

acl

(there go my upvotes)

@halirutan I think it's a good question

yoda

@halirutan it certainly will reach public beta

acl

what warning?

halirutan

did you created a new question recently?

acl

12:13 AM

yes

yoda

They've canned the sex proposal already – I warned them in my looong post on a51 that dumping mma.se in favour of sex.se was a bad idea :)

acl

well, mma vs sex--easy

halirutan

acl

@hali it's not something easily found out (indeed, SubValues is undocumented)

Mike Honeychurch

@yoda no idea but Wolfram would still be best placed to answer that, i.e. the first place to start.

yoda

12:16 AM

@MikeHoneychurch true. I should ping Arnoud or Brett when they pop in.

acl

@halirutan well, @yoda didn't know it. sounds like a good enough question to me

oh wait, it is a special case of what's discussed here

22

A: What is the distinction between DownValues, UpValues, SubValues, and OwnValues?

In Mathematica, all functions are really just patterns, and there are different kinds of those. Let's start with OwnValues, which is the pattern type of a variable as you know it from other programming languages. The symbol having the OwnValue has, as the name suggests, intrinsic, "own", value. ...

halirutan

About the pretty printer: Had someone already a look at the google-prettyprint

yoda

@acl I didn't know about SubValues until 2 days ago from David's answer. I figured it out after I realized that derivative curries its variables

acl

damn

yoda

is "currying" the right term for it?

halirutan

12:17 AM

maybe a skip this question

I looked up Down, Up and Own values of all symbols and didn't find it.

Never heard about SubValues

until now

@yoda, I thought "currying" is when you get a new function by applying only some of its parameters..

acl

also, even if you know about SubValues, it may take some thought before realizing what is going on in this case.

halirutan

So in the case of Mma it's maybe not exactly currying.

yoda

@halirutan ah right. So probably a "composite" function or something then.

halirutan

I mean what's anyway the difference between f[x][y] and f[x,y]?

yoda

@acl Exactly. I first thought of D[f[x], x] and looked at values for D. Then thought of Derivative, but only after I saw the full form did I realize that it was a different form, and needed to look in SubValues

halirutan

12:25 AM

cool, we really can do curring whith this:

add[x_][y_] := x + y;
add /@ {1, 2, 3, 4}
Through[%[2]]

yoda

12

Q: Currying with Mathematica

One may implement a limited form of Currying in Mathematica, using this construct: f[a_][b_][c_] := (a^2 + b^2)/c^2 Allowing one to do, for example: f[4][3] /@ Range@5 {25, 25/4, 25/9, 25/16, 1} There is a problem: Attributes only apply to the first (set of) argument(s). Consider: Cl...

halirutan

@yoda, hmm I see. As always someone else found it first

Szabolcs

@yoda What are moderator tools good for? I don't see any difference (except the site is actively bugging me to review edits)

halirutan

It would anyway not be as nice as in Haskell

Szabolcs

@yoda I thought last time you mentioned it would give access to some statistics with pretty graphs

yoda

12:29 AM

@Szabolcs that's the moderator as in diamond tools. These are the equivalent of the 10k tools on SO

Szabolcs

I only have 9.5k on SO

so it's completely new now

exploring

yoda

On a site like mma.se, the 10k tools are probably useless. However, there's a lot of info on SO because of the sheer volume

J. M.

@yoda "Not yet that useful", you mean. :)

yoda

It isn't of much use for us on SO anyway, but in general, it lets you see questions with close votes, delete votes, suggested edit queue, moderator flag queue, etc

@JM good point =)

but I still stick by it. I think we have a very good core group of users who generally agree on most things and things get closed/deleted very quickly by the community if it's inappropriate for us, and a moderator's job is going to be pretty easy

Szabolcs

When we go to public beta, and the privilege thresholds increase, we'll lose the privileges we got during private beta, right?

yoda

12:32 AM

yup

roughly double your private beta thresholds for the public beta ones

I kinda think it would be nice to stay in private beta for half a week more so that we can get some more work done on the tags, etc when we still have the rep privileges

but probably everyone else is anxious for public beta

we have 3 days more, so lots can be done in that time

Szabolcs

I'll send an annoucement to MathGroup when we go public

J. M.

In math.SE, we set up a "jury room" chatroom. A "judge" points out a question they've voted for closing, three "jurors" cast their additional votes, and the "executioner" casts the closing vote.

Szabolcs

The reason I didn't do it before was that I was afraid the moderator would block it saying he doesn't want a discussion about the new site on MathGroup.

But now I can tell people to just post question on meta.mma.se instead of replying to MathGroup

so he'll let the message through

halirutan

@Szabolcs Did this ever happen? I think they are cool about such stuff.

acl

@halirutan so long as you don't say "matlab"

halirutan

12:36 AM

@acl Why? I never lost a fight against Matlab ;-)

yoda

@Szabolcs why meta?

Szabolcs

0

A: Can we have a central list for all the tagging decisions we make?

My reasoning was this: Sometimes I have a suggestion or idea about tagging / synonyms. I don't want to implement it before talking to others, so I ask a couple of people in chat. But after that I'm still feeling insecure about putting the idea in practice. What if someone has serious objectio...

acl

@halirutan I meant the mathgroup moderator is cool so long as you don't mention matlab. if you do, it becomes "another system" in the message...

yoda

@Szabolcs did you see my earlier suggestion of using a chat bookmarklet and linking to that in meta?

Szabolcs

@yoda If I post an announcement about the new site on MathGroup, people will inevitably have questions. They will also ask about the relationship of the new site with MathGroup (is it going to replace it? is it going to hurt MathGroup by drawing away users?). The moderator doesn't want such a discussion, so I think it's essential that I can point people to a place where they can ask questions about the new site. I think meta.mma.se is the only place suitable for this.

@yoda I didn't, I couldn't follow closely what's happening today

yoda

12:39 AM

@Szabolcs Ah, yes. certainly.

4 hours ago, by yoda

Tags: graphics or graphics3d?

2 hours ago, 2 hours 8 minutes total – 6 messages, 4 users, 0 stars

Bookmarked 19 secs ago by yoda

halirutan

@acl Really? Hmm, ok.

yoda

read down a few messages from there

I just created one more

that will be a good way to document small chat discussions.. ok, the second one was too short (3 messages), but you get the idea

Szabolcs

@yoda While he didn't block any posts just mentioning the A51 proposal briefly, I heard he did block posts the only topic of which was the proposal. Before we launched, there was no place I could point people to ask questions at, so the only thing I did was I put the proposal URL in my signature. But now I can safely send an annoucement.

yoda

That way you can point the user to the exact discussion and they needn't weed through the transcript

acl

@halirutan well, I can imagine what the motivation is, but it still results in silly-looking questions

yoda

12:43 AM

@Szabolcs Does he work for WRI?

Szabolcs

no, he does not

he used to sell a Mma package, MathTensor, I think

otherwise I know nothing about him

he reviews posts every day

it's really a lot of work

I know he is editing and cleaning them up when necessary

I suppose he must put the system on autopilot occasionally. Everyone must have a holiday sometimes.

For example, I used to put a [note: also sent to MathGroup] at the beginning of my replies. I did this because I always CCd the reply to the OP (so they didn't need to wait until it passes moderation), and newcomers were often confused and thought I only mailed the reply to them, but not the group. He always edited these lines out, so I know he spent time cleaning up the posts.

halirutan

@Szabolcs It's a horrible job and he does it since what, 20 years..

Szabolcs

12:57 AM

@yoda added a link to the graphs vs graphcs3d bookmark from the meta post

I like this answer :-) by MrW

we have both assignment (x4) and variable-definitions (x2). Do we need both?

Verbeia

@yoda where did you see the sex.SE proposal was canned? (NB Poker is if anything doing worse)

Szabolcs

Ah? It didn't even make it out of private beta?

J. M.

1:16 AM

They lost some steam, it looks.

Verbeia

@Szabolcs I would think that assignment is the act of assigning, variable-definitions is finding out later on what that assigment did

J. M.

1:30 AM

Completely random: is there a tutorial anywhere on how to make your own color gradients?

I used to use this in old versions, but I now want to be able to make my own ColorDataFunction[].

halirutan

@JM where exactly do you need guidance? Creating a gradient or packing it inside a function like ColorData?

J. M.

Second one. How does one make a ColorDataFunction[] of your own?

yoda

@Verbeia stackmobile.com/meta.sexuality.stackexchange.com/questions/view/…

also see the second comment under the question \o/

J. M.

This is interesting... stackmobile can peer within private beta sites? :)

yoda

@JM yup, the API works for private beta. That's how the mod list script worked too

4

Q: Custom ColorFunction/ColorData in ArrayPlot (and similar functions)

This is related to Simon's question on changing default ColorData in Mathematica. While the solutions all addressed the issue of changing ColorData in line plots, I didn't quite find the discussion helpful in changing the ColorFunction/ColorData in ContourPlot/ArrayPlot/Plot3D, etc. TLDR: I...

mathematica

halirutan

1:38 AM

@JM You have seen for instance ColorData["Rainbow"] // InputForm?

Create your own gradient with Blend and thats it

J. M.

@halirutan ColorDataFunction["Rainbow", "Gradients", {0, 1}, Blend["Rainbow", #1] & ] I see, so the Blend[] portion's a puzzle...

halirutan

Or use an entirely different function for the gradient

J. M.

@yoda Yep, I think Sjoerd's solution is what I want. Thanks!

(Actually, I think I'll look into converting MATLAB's colormaps now...)

yoda

That's what I do. I just use the interactive colormap editor to create one I like, resample to a finer mesh and export

Of course, you could do that from within mma, but #lazy

J. M.

1:55 AM

(should make for interesting artwork)

I can't believe it's only been four days... all the activity felt like a week's worth of questions, answers, and rep on other beta sites...

2

rcollyer

2:49 AM

@Verbeia might I suggest posting the list of tagging decisions in one of the answers, instead of the question itself, and put some of the additional reasoning into the question? That way, the info can be made community wiki, immediately, and the question can highlight our reasoning for the question, format of the answer, etc.

rcollyer

3:00 AM

@All, as a note, once we enter public beta the reputations required for various things will increase. In particular, Close goes from 1 to 500, Edit will go from 500 to 1000, Approve tag wiki edits goes from 750 to 1500, Moderator tool access goes from 1000 to 2000, and create tag synonyms stays at 1250.

Current stats according to Area51: 24.2 questions per day, 96% answered, 49 users over 200 rep, averaging 2.3 answers per question, and 397 visits per day.

As said elsewhere, the Area51 stats aren't used much in the determination of how successful we are, but they are nice to see anyway. We have three users with over 1000 rep, so we won't have trouble filling out the upper ranks needed to administer the site. And, it has only been 4 days.

Tim Stone

5:17 AM

@yoda In regard to the syntax highlighting, is there any logical way to distinguish the x in your example from the prob without knowing what's a function, and what's not?

yoda

@TimStone You mean the x inside f in the first example vs the prob in the second?

Tim Stone

Specifically the x and y in x Times @@ y

yoda

ah, that'll be trickier. That x and y are variables and correspond to the x_ and y_ in the LHS

Tim Stone

Ah ha, hmm..

Brett Champion

A lot of the syntax information, at least as far as argument structure for functions go, is in SystemFiles/Kernel/TextResources/English/FunctionInformation.m.

yoda

5:29 AM

@BrettChampion Tim here is helping us with custom syntax highlighting for our site. Mike mentioned that WRI might have a handy list of rules for the highlighting since they had to implement it. Could you enquire if something like that is available and can be released? I don't know if it'll be in a format that can be easily ported to javascript, but it will certainly help Tim

well, what do you know. you're a precog

Brett Champion

I saw Mike's comment skimming the transcript.

I wonder how much difference there is in syntax highlighting vs. complete parsing of the language?

Tim Stone

...is what I have so far, trying to get rid of the invalid blue highlighting. :/

yoda

@TimStone that's looks great!

Brett Champion

@Tim You mean the i for the Table iterator?

yoda

@BrettChampion I think he means the blue in x Times @@ y

Tim Stone

5:34 AM

Indeed. As well as the fs, which are inconsistent with the screenshots provided.

yoda

@TimStone no, that's actually fine.

if you replace f in the first line with F, does it get highlighted as blue or black? If black, then it's perfect

Brett Champion

I need to go in a minute, but I'd be happy with simply coloring symbols differently based on whether it's a built-in vs. something else.

yoda

I kept it black in the first example, as the blue for functions starting with small was part of the "additional" stuff... but blue is correct

Brett Champion

(For which capitalization would be a cheap proxy.)

yoda

Brett Champion

5:38 AM

I'll also note that not even Workbench manages to match the highlighting used in the frontend.

yoda

That's how it looks when input. So what you have is correct. I simplified it for the "basic requirements" because I had no idea how hard it might be to implement

Tim Stone

Well, at the moment it's simply looking for lowercase tokens that don't appear to be part of something else, so it's actually entirely coincidental.

Brett Champion

The color of f will change once the function definition has been interpreted by the system via evaluation; I don't think that makes sense in this case.

I'm off for the night. See you later!

yoda

good night

Tim Stone

5:59 AM

Hmm, I don't think that prettify will allow for the variables to be properly highlighted in the f[x_, y__] := x Times @@ y case.

halirutan

@yoda Yes, good night too. It's here 7.00 in the morning and I need definitely some sleep. Check stackoverflow.com/a/8959323/1078614

yoda

6:10 AM

@halirutan that's pretty neat! Looks like you've used a list of system functions to highlight them in bold, which is great1

but awesome nevertheless :)

halirutan

I thought maybe it would be too slow, but it runs fast

You didn't specified in your post what color you wanted for the defined system functions ;-)

specify

Simon

@halirutan That does look great. It would probably work perfectly for most code that will be posted on the site.

(Comments inside comments are rare)

halirutan

@Simon I have to think about wheter it's possible with this simple lexer to catch such a case

yoda

I know, I didn't specify because it'll be a lot faster to check whether it begins with a capital or lwoer case, than for checking against a big list

Tim Stone

Depends on how large the list is, I suppose.

I think that the SQL keywords list is fairly lengthy, without any significant issue.

yoda

6:14 AM

@TimStone ifile.it/vcb2sna/FunctionInformation.m

halirutan

@TimStone yes it is and because I have seen the other languages using their long keyword list in prettify.js, I decidet to make it right for Mathematica

yoda

The functions there should pretty much cover the built in functions

Tim Stone

Ah ha

Simon

In the FunctionInformation.m file, the functions are arranged by context. So, for example the first string in the first list is "System`" then all the strings in the rest of that list are the functions in that context. Although, I guess for simple highlighting purposes, it is not that important.

Also, the strings at the deepest depth are the Options. For example:
{"AccountingForm", {_, _., OptionsPattern[]}, {"DigitBlock", ...
AccountingForm is the function in the System context. The next term specifies what arguments it can take. The final list is the available options. But once again, for highlighting purposes, it doesn't matter. :)

yoda

Exactly... that's now getting into territory that only the mma parser is aware of :)

Simon

6:29 AM

Not only the Mma parser. There's a couple of others...

mockmma.sourceforge.net

mathics.org

And a couple more that I can't remember the names of and don't seem to be in my bookmarks...

code.google.com/p/symja

yoda

I remember Mathics. I used wxmaxima for a day before I gave up and bought mathematica

Simon

Also: omath.org/w/index.php?title=Main_Page#Mathematica_parser_bugs

I've played with Maxima (even answered some questions on Stackoverflow) but I could never get (or be bothered getting) wxmaxima to work on my system

Anyway... got to run. 恭禧發財!

ni hao

wo hen hao

uh oh.

6:36 AM

Bye!

yoda

bye :)

David Zaslavsky

@halirutan I've done some work with syntax highlighting in other systems and from what I've seen, it seems pretty normal to use a list of keywords, even if that list is very long (PHP comes to mind).

yoda

He did use keywords. I was the one who thought it might be too slow

we can have beautiful code highlighting now :)

David Zaslavsky

Yeah, I was just indicating that I was thinking along the same lines.

yoda

well, beautiful is subjective... I've just not bothered to change the defaults =)

I guess most people are that way

Szabolcs

8:35 AM

@DavidZaslavsky @halirutan Once I made a better Mathematica syntax highlighter for Vim. I remember at that time I concluded it's not really worth bothering with lists for Mathematica. BTW the original highlighter for Vim highlighted everything before a "[". But if it runs fast, why not?

@halirutan How can we test this code? :-)

@Simon The author of omath is here on Mathematica.SE

@Simon 你会说中文吗

@JM walkingrandomly.com/?p=24

Tim Stone

9:23 AM

Bwuahah.

J. M.

Lookin' good!

Szabolcs

9:37 AM

@Tim Looking great :-)

It'd be nice to have the quotes (") string-coloured as well :)

Tim Stone

Thanks, although there seems to be some trouble in IE...I had to do a few unspeakable things for the function variable highlighting.

Ah, that's actually even easier, I had intentionally not highlighted them per the examples.

Szabolcs

@Tim, Mathematica highlights them all in grey, including quotes. I think @yoda highlighted them in magenta by hand, maybe that's why the quotes are black in his example. (But of course we don't necessarily want the same highlighting that Mma uses by default, we want a useful highlighting, magenta is probably better :)

Tim Stone

Ah, makes sense. I've gone ahead and fixed that now.

Szabolcs

Oh, he's actually right, quotes are not highlighted in Mma, but it's so difficult to tell with this silly grey default!

This is how it highlights it in reality by default. Very little contrast there. I think we should have more contrast as in yoda's example.

Tim Stone

Oh wow, the difference is really hard to notice there.

Szabolcs

9:44 AM

Indeed, that's why I increased the magnification

Another thing, I'm not sure it's clear: ## or ##2 should be highlighted the same as # and #2

Tim Stone

Yep

Szabolcs

10:04 AM

@yoda I am wondering if there is any value in highlighting every word that precedes a [ or a @ (i.e. functions when they are being called). Boldface would be nice for this. Vim used to do this for Mathematica, and I liked it, and found it useful. (The other two function call patterns, i.e. x // f and x ~f~ y are probably too difficult to get right. Especially the latter as it could look like x ~ f ~ y ~ g ~ z, and that's why infix is difficult to read even for people)

What do you think @Tim about how easy or difficult this is?

I made a quick (manual) example of what it might look like:

I realized that array indexing (array[[2]]) may cause trouble. f[x] is a function call while a[[x]] is indexing.

Or even better:

Again, the issue is the difference between a[[i]] and f[x]. It would require bracket matching to get this totally right (though bold-highlighting even [[ ]] can be a compromise)

These are just ideas @Tim, I do not know what is easy to do and what is not.

If bracket-matching is easy (maybe built into prettify), so we can distinguish between a[x, b[ ... ]] and a[[ ... ]], then this would be extremely useful. These double brackets always cause readability problems. But I don't expect it to be easy.

I meant to bolden the @ too there, I missed that ...

Szabolcs

10:25 AM

Alright, here is my proposal:

halirutan

Ok, I have to go now, but we may talk later.

Szabolcs

We could approximately (and I believe, usefully) detect function calls using the following rules:

* If a symbol name (upper or lowercase) precedes `[` or `@` (extra: also `@@`, `@@@` or `/@`) it should be bold

* The corresponding [ and ] brackets should also be bold. But double brackets should not be boldened to differentiate array indexing. This is difficult because the code might look like this: a[b, c[ ... ]] which needs boldening or like this: a[[...]] which doesn't need boldening. It's about matching up a starting [[ with the ending one.

halirutan

@Szabolcs, here is what I did: I downloaded the google-prettify project, copied one of their language-extension-files and wrote down the Mma rules

Szabolcs

* As an extra touch, any symbol name following // should be bold as well (if it is feasible)

@halirutan is it as simple as including prettify in a page that has some Mma-code?

halirutan

everyone can test this by using my js-scripts and a simple testpage

yes,

at least here I just opened my stuff locally in the webbrowser

Up to know I didn't change anything in the prettify-lexer itself. I just wrote the pattern-matching stuff

Szabolcs

10:33 AM

@TimStone Could you take a look at what I wrote above, and give some opinion on whether this is easy to do, not easy but doable, or too difficult to be worth bothering with?

halirutan

Did we agree now to use and adapt the prettify-google-stuff? Did Tim use something else?

Szabolcs

@halirutan SE uses prettify as far as I know. Thank you for doing this (and working on it all night :-)

halirutan

Have to go now. I check in later. @Szabolcs, can you post your complete proposal anywhere and maybe check, what you would like to change in the example I gave?

Szabolcs

@halirutan I have to go for a while now, I'll try out your when I come back! I'm curious how well keyword-based highlighting works.

halirutan

ok, see you then.

Szabolcs

10:38 AM

@halirutan do you think it's visually nicer to bolden the brackets too, to match them up with the heads? Maybe not since you don't bolden all heads

see you

BTW I like it much better when the quotes are also string-highlighted (like in @halirutan's solution), even though Mma doesn't do this by default.

@Tim @halirutan another thing we forgot to mention that strings can contain escaped quotes, like "f(\"asd\")". This is pretty common.

Simon

11:09 AM

@Szabolcs 我不会说中文

And thanks for the tip about who was working on omath - I don't remember noticing that before.

Verbeia

@yoda that's interesting about sex.SE - and very good news that the SE powers-that-be have seen what has happened with mathematica.SE

Szabolcs

12:40 PM

@MrWizard

Mr.Wizard

1:23 PM

@Szabolcs glad it didn't get stuck there :-)

Re: syntax and function calls, don't forget ~infix~ ;^)

Also, what's with the Cantonese? (if that's what it is)

Szabolcs

2:11 PM

It's Chinese new year today

Michael Pilat

2:22 PM

Hi all! Congrats on getting the new SE site started!

Szabolcs

Hi @MichaelPilat

@MichaelPilat Why didn't you commit to the site?

rcollyer

@Szabolcs He did, eventually.

halirutan

I'm back. @Szabolcs, escaped quotes in strings should be no problem since this is very common in other languages too. About the brackets: I matched and highlighted all braces the same way. The only thing which is bold are the defined keywords like functions and symbols. Btw, the highlighting colors and faces are just an example. I tried to follow Yodas suggestions. I personally don't like pink strings and keywords are usually blue.

The blue keywords are crap in the case of Mathematica. For me it would always feel like this is an undefined symbol.

But all these settings are just done in the css-file for the highlighter and can be changed easily.

Szabolcs

3:12 PM

@rcollyer Actually I was just wondering if he needed an email invitation

@halirutan Ignore all I said, it's just little details. If it can be implemented, a truly useful feature would be differentiating between single and double brackets (x[[i]])). This needs bracket matching for cases like a[b, c[...]], but if nested comments work, this should be already implemented in prettify, I suppose ...

Brett Champion

@Szabolcs @halirutan My preference is to not to use bold for System` symbols. It feels really jarring to me, in a way that (well-chosen) colors don't.

Unfortunately, I need to go now..

Szabolcs

@Brett Perhaps you're right. The look of bold fonts is very system-dependent anyway, so I can imagine it won't look good on every computer

@Brett that was just an idea anyway. What I'd really like to have is highlighting to differentiate between [...] and [[...]]!

I saw someone once use a string argument to an interpolating function to get some data about it

I can't find it in the docs anywhere now

Am I mistaken? Does anyone know how to use this?

Interpolation[{1, 3, 4, 3, 1}]["Domain"] <-- this works but where is it documented?

halirutan

3:45 PM

The problem is, that for the double brackets you need the surrounding context which you don't have in a lexer

Szabolcs

@halirutan Why do we need the context? Isn't it enough to match up brackets? I.e. match up each [ with ] and each [[ with ]]?

halirutan

How do you want to differentiate between f[g[1,2]] and f[[1,2]]

Szabolcs

what I just said: match brackets

J. M.

@Szabolcs I wouldn't call it "documentation", but: check the package DifferentialEquations`InterpolatingFunctionAnatomy`

halirutan

then you need recursive regexpr

Szabolcs

3:47 PM

I've seen this implemented on SO in Mathematica. I don't know how efficient it can be

halirutan

comments are done like (* eat everything until *)

Szabolcs

@halirutan yes, that's true, it has to be something recursive. That's why I said I don't know if it's possible to do this in prettify. I don't know how it works. But if nested comments can be handled, this can be handled as well surely. Though it may be too slow to be practical.

@halirutan so prettify can't handle nested comments?

halirutan

look at:

Szabolcs

I'll be back in 5 min. I understand it needs to be recursive and I know how to do it generally (the algorithm). But I don't know how prettify works and if it's feasible to do it with prettify's framework.

if you can only cutomize it with regexps, then perhaps not

yoda

@Szabolcs ooh good catch. I turned off my default stylesheet to use mma's default, but I forgot that I had globally changed the color of strings =) my bad :P... but I guess I changed it for a reason – grey and greyer is not easy on the eyes

halirutan

3:50 PM

code.google.com/p/google-code-prettify/source/browse/trunk/src/…

yoda

@Szabolcs Remember Mr.W's question on that... not an easy problem, I think.

halirutan

this is the extension for tex in the google-prettify

yoda

13

Q: Matching brackets in a string

What is the most efficient or elegant method for matching brackets in a string such as: "f @ g[h[[i[[j[2], k[[1, m[[1, n[2]]]]]]]]]] // z" for the purpose of identifying and replacing [[ Part ]] brackets with the single character forms? I want to get: With everything else intact, such as t...

parsing string mathematica

I don't know how easy that would be to convert to regex

@Szabolcs Do you really want gray and grayer for comments and strings?

Szabolcs

@yoda I don't care about the colours at all.

halirutan

As far as I can see this currently, the way you add your language to prettify is to supply a list of regexp which are tried in this order and if something matches, it eats and labels this. To make the matching brackets you have to count (which is equivalent to "be aware of the context") wich is in the linked post done with the n++

yoda

3:59 PM

@Szabolcs Never mind... I just read further up and saw that you didn't like the gray either

Szabolcs

@yoda I just gained the "create synonyms" privilege. So if anything needs to be done, I can do it, but only before we go public and the thresholds increase.

halirutan

I'm open for suggenstions but what you would have to express is match [[expr]] where expr can be everything from the language again.

yoda

@Szabolcs still needs to be voted on by users who have rep in those tags =) You can only suggest now

since tags are fragmented now, and not under a single mma umbrella, it'll be harder to find enough users

Szabolcs

Oh, I see. Anyway, I didn't want to change something personally, just noted that we can do that now as a community. But if several users need the same rep, we need to wait.

yoda

I'd wait... in a few days, temporary moderators will be appointed and then you can do it all in one click

so the community can suggest and you can approve in a click or you can set them yourself... either way

I don't think they need the same rep to vote... but there is some requirement for min rep in the tag... tag synonyms and support for that is pretty flaky and complicated...

halirutan

4:05 PM

Btw, is planned to include something like MathJax to be able to display formulas?

J. M.

@halirutan We already have MathJax. Try it out.

halirutan

Ah, ok. But in Mathematica.SO it's not included, right?

yoda

@halirutan not on Stack Overflow. It will never be implemented on SO

J. M.

emphasis on "never".

halirutan

Do I have to understand this? I mean for things like image-processing it's sometimes crucial to have a formatted formula.

If a site has a math-section it would be the first thing to include.

J. M.

4:08 PM

Well, SO isn't primarily a mathematical site, so...

yoda

@halirutan right, and all those questions form 0.00001% of SO's traffic, and enabling mathjax can only be done for the entire site (from what I hear)

J. M.

Here, we have it. Use LaTeX as seen fit.

yoda

They won't implement it for little to no gain

J. M.

Besides, loading the JS underlying MathJax incurs a slight performance penalty.

yoda

Exactly. that's the primary reason rather than "benefits only a small percentage"

halirutan

4:12 PM

At least we have it here. I tested it and it's working like a charm. I really like it.

Szabolcs

0

Q: Programmatically copy code so that all output is commented out

Sjoerd suggested that when we want to post input lines that alternate with output lines, instead of including the In/Out labels, we could simply comment out the output. Read his suggestion here: How should I include code samples in posts? I have seen several people use this style in the past...

front-end notebook

halirutan

@yoda but isn't this global turn on not the same for prettify? I mean as far as I can see prettify has one, global css style for code-coloring. This would mean we cannot change colors on SE without changing is for every language.

yoda

@halirutan well, this would be everywhere for Mathematica.se... not other SE sites

J. M.

Think of each SE site as its own island...

Szabolcs

MathJax is very slow on my old machine, I can feel how typing into the edit box slows down. I'm not saying I don't want LaTeX, in fact I want it very much.

But it also causes an incovnenience that's quite noticeable here, even with MathML rendering in Firefox

halirutan

4:21 PM

@Szabolcs there is of course overhead. It has to render the formulas. Usually it just displays the tex-code and renders in background. At least it had this behavior when I used it for my own projects

J. M.

@Szabolcs: you might be interested in this.

(meta.math.SE is quite the treasure trove, I'd say.)

halirutan

@JM Does this mean we could (if someone wants) hack the whole google-prettify and not only extend it for Mathematica in the usual way?

Szabolcs

@halirutan I could painfully feel it today when writing a question for scicomp.SE ... even though I only had one tiny formula

but the post was pretty long and as I said, my computer is not too new (Pentium M)

J. M.

@halirutan That's a pretty big "if"...

halirutan

@JM The problem is not the "if", the problem is, that JavaScript is such a nice and beautiful language ;-)

Szabolcs

4:25 PM

Welcome @belisarius

belisarius

@Szabolcs Greetings all!

halirutan

Hi

belisarius

Hi Szabolcs!

@halirutan Hi!

Szabolcs

We already have 127 questions in ~4 days

500

@Sjoerd, That would be great !

belisarius

4:28 PM

@Szabolcs Perhaps they gave us just 7 bits

Szabolcs

:D

Sjoerd C. de Vries

@500 It will be short, as my wife and kids will be there as well and we'll be in NY for only 3 days. My wife is in psychophysics too, so perhaps we could do a small lab visit (not sure whether she'd like to do work related stuff during vacation, so perhaps not)

belisarius

@SjoerdCdeVries Hi!

Tim Stone

@Szabolcs Hmm, that might be tricky. But, let me see.

yoda

@JM someday, folks should be introduced to the meta.math vs jeff stories... can easily lose a day in there :)

Szabolcs

4:39 PM

@Tim I have no idea what the possibilities are. Take it as a nice to have. Just to differentiate between [ ] and [[ ]]. I know it needs some recursive approach ... If it's difficult or slow then maybe it's not worth it

J. M.

@halirutan That, too.

Sjoerd C. de Vries

@belisarius Hi belisarius

Looks like you're less often here

Busy?

belisarius

4:53 PM

@SjoerdCdeVries Yeah. An insane amount of work piling on my desk.

Szabolcs

@Sjoerd he was not able to commit (he already had 3 commitments), I just invited him now

yoda

I think several of us sent invites... the man is a true SOist at heart.

belisarius

@Szabolcs True, but I am almost away of SO too.

@yoda I received only one from @Szabolcs.

yoda

Really? I sent one and rcollyer mentioned that he sent one too

hmm.. something must've gone wrong

anyway, what matters is that you have an invite now

Szabolcs

different emails?

yoda

4:56 PM

ah.

belisarius

@yoda I can believe that! Probably some procedural/ software error

Szabolcs

1 hour ago, by Szabolcs

Interpolation[{1, 3, 4, 3, 1}]["Domain"] <-- this works but where is it documented?

I'd rather not make it a main question, but I'd really like to know where this comes from (maybe deprecated functionality from an old version?) and what other special arguments it can take

J. M.

@Szabolcs The Interpolating Function Anatomy package is more or less a front end for this functionality. It's where I saw the only mention of these strings.

I pinged you earlier to look at the package. You seem to have missed it.

Szabolcs

Ah

Thanks!

J. M.

(It's not unlike the time the GraphPlot` package was in fact just a front end for hidden functionality in the kernel.)

Szabolcs

5:06 PM

ah, really?

I don't dig into packages much

maybe i should

Tim Stone

@Szabolcs Yeah, I think it's likely not practical to handle arbitrary bracket nesting, which would be required to differentiate between those two cases, in all cases.

J. M.

@Szabolcs It's a pretty good source of programming examples, I'd say. :)

Szabolcs

5:23 PM

@JM actively editing tag wikis?

@JM can you approve mine too?

J. M.

@Szabolcs Yes.

@Szabolcs Done.

I say we take advantage of the fact that we still have the bright orange alerts available to us for the rest of this site's private beta.

Szabolcs

Agreed

yoda

aah.. need to make it to 1k before private beta ends. Then it slips back to 2k. I'm sure the second stretch will be longer than the first

Szabolcs

8:35 PM

0

Q: Can we have MathJax disabled for question previews, and active only for full question pages?

It seems that the questions page (or the front page) does not render code spans in question previews. This means that if a dollar sign appears in a code span (quite common in Mathematica), MathJax will try to render it as math: MathJax is very valuable, and I think it is absolutely necessary ...

bug

Szabolcs

9:34 PM

Is anyone familiar with Sobol sequences?

From the docs: "The first six methods are uniform generators. "Niederreiter" and "Sobol" generate Niederreiter and Sobol sequences. These sequences are nonuniform and have underlying structure which is sometimes useful in numerical methods. For instance, these sequences typically provide faster convergence in multidimensional Monte Carlo integration. "

tutorial/RandomNumberGeneration#57850632

Low-discrepancy sequence

In mathematics, a low-discrepancy sequence is a sequence with the property that for all values of N, its subsequence x1, ..., xN has a low discrepancy. Roughly speaking, the discrepancy of a sequence is low if the number of points in the sequence falling into an arbitrary set B is close to proportional to the measure of B, as would happen on average (but not for particular samples) in the case of a uniform distribution. Specific definitions of discrepancy differ regarding the choice of B (hyperspheres, hypercubes, etc.) and how the discrepancy for every B is computed (usually normalized)...

David Zaslavsky

10:05 PM

@Szabolcs A little, I've just been reading about them in Numerical Recipes...

Computational Science would be a great place to ask questions about such things, if they're not Mma-related

Tags: graphics or graphics3d?

Transcript for