@RolfMertig I'm at work and can't check until tonight Australian time, but it sounds like there is something going wrong with ligatures and font encodings. Do you have the same problem with fi?

Typing mma in a notebook automatically enters Mathematica. Where is this set up and how can I control it? I thought it was the InputAutoReplacements Front End option, but that's empty for both the notebook and globally.

@RolfMertig I wasn't unhappy with the Mac versions as I was with the Linux one ... Always had more trouble with the Linux front end than the others though.

One problem with the Mac front end is that it's still 32-bit, so it tends to crash when asked to do too much, such as render too many too large (many primitives) graphics at the same time.

@Verbeia Yes, also crashes. I just sent a bug report to support, quoting the more informative ExportString[Style[" fl", FontFamily -> "Gill Sans"], "PDF"]; as quick crash.

@Szabolcs I used the Linux FrontEnd a lot, 10 years ago or so. But the Windows version is just more stable these days. And MacOSX just has too many problems. Starting from Java support up to these font-ligature-PDF problems. I only use MacOSX when I have to. I don't like it.

@Szabolcs ExportString[Style[" fi", FontFamily -> "Gill Sans"], "PDF"]; is hardly asking too much ...

Hey guys! Should this be moved to Cross-Validated as suggested in the comments? mathematica.stackexchange.com/questions/96232/…

@Sosi If you think you can get a better answer there, go ahead and do it. You can either flag it so a moderator can migrate, or you can delete and re-post.

I'm not sure. My initial thought about posting it here was that this could be some method-related issue

Hi! Does anyone know how to compute periodic orbits with MMA?

@RolfMertig I couldn't reproduce on Mavericks - maybe it's El Capitan specific? Let me know if I can help further.

@RolfMertig That also crashes for me, on Yosemite, MMA 10.2

Interesting that 10.3 will have Tracy-Widom. How long before they actually implement the Painlevé transcendents?

...and are they going to replace Print[] with Echo[] now?

@J.M.isback. I'm guessing they are sacrificing more consistency and conciseness of design to try and maintain backward compatibility within their own code base. I actually support making breaking changes in the design if I can update at my own leisure and when I do update my code it will improve it.

I wonder what the percentage of recent blog posts featuring word clouds is.

@MichaelHale I find them somewhat annoying, but maybe that's just because I think they're being overused.

@J.M.isback. Print is a nightmare.

The majority of the time it is used, it's used by people who don't understand the difference between printing and returning a value

If I developed a nanny mode for beginners, it wouldn't allow print to be used until they understood side effects

@Szabolcs Check here: /Applications/Mathematica.app/Contents/SystemFiles/FrontEnd/StyleSheets/Stylesh‌​eetFormatting.nb

Huh - JM gets back, and the first I hear about it is in a comment on math.SE :P

@halirutan You seem to be the expert around IntellIJ. Is there a way to get it to function more like Mathematica's notebooks? It seems like I have to reopen the project in order to have separate windows similar to Mathematica's Notebooks. (Side question how do you get Ctrl+w to work inside of lisp, it seems to parse the language correctly but supporting selecting with Ctrl+w)

Echo looks like a debugging feature.

@Searke Do you know when 10.3 is going to arrive? Days or weeks?

So in Latin, "hi" means "THESE are loyal to the king"? :-) reference.wolfram.com/language/ref/WordTranslation.html

@Searke Can you explain the reasoning behind EchoFunction? I can't immediately see why it's useful (though I'm sure it is).

Pretty certain this is a duplicate of something, but which one? mathematica.stackexchange.com/questions/96321/…

@Szabolcs Bummer. Google Translate just stops at "these".

@blochwave I know what you mean but I cannot find it either.

@Searke Who does? :) An interesting feature for WordCloud would be: be able to access the mapping between the coordinates in the mask and the coordinates in the final image. Also: use any shape, not just words, without having to encode it in a string. Or even better: access tot he underlying algorithm for arranging words, together with access to the algorithms for the various packing layouts for graphs (related).

This is made with IGraph/M:

It's the High Energy Physics Collaborations network available as ExampleData, Louvain community detection, DrL large graph layout algorithm.

@Szabolcs Do I know when 10.3 will arrive? Yes and No.

There's like a system... for announcing releases. It exists for a reason.

It's very very frustrating when someone says "Oh yeah... it's like next week!"

and then someone in QA finds an issue that needs addressed or something else happens

so ... no one is really supposed to say anything

Got it.

So still under testing, can potentially get delayed.

@Szabolcs well it's always under testing. And then you have decide if some of the minor issues are or aren't worth delaying the release and see what the effects of some attempt to fix them looks like ...

@Szabolcs Echo is for debugging. But more importantly, I think it needs to be for new people as well. Too many new people don't understand Print

And Echo fixes what they don't understand about Print

@Searke Does Subgraph preserve edge weights in 10.3? (If you're allowed to say :)

The only thing I did with Graphs recently was look at the issue you brought up on community.wolfram.com

@Searke Normal@WeightedAdjacencyMatrix@Subgraph[Graph[{1 <-> 2, 2 <-> 3}, EdgeWeight -> N@{5, 7}], {1, 2}]

No. I can't talk about the statuses of bugs

I'm sure other people at the company will

OK.

I used to do tech suport a while ago. Talking to people about what bugs were and weren't fixed was often ... just prone to arguing

My knowledge of graph theory in Mathematica is kinda weak

OK, I was just going to ask a question, but then you probably don't know the answer ...

I understand that weighted and directed graphs have the most problems. esp where some algorithms aren't implemented?

I need to extract the edge list as a packed index array as fast as possible. There are methods here: mathematica.stackexchange.com/a/95751/12

The second one with SparseArrays is very fast, but it's not trivially generalizeable for multigraphs.

I was just about to use that method and implement it in C with LibraryLink. Of course that's a lot of work, so if there's anything better in pure Mathematica, I'd take that instead.

But I don't think there is.

Huh. I guess I should really go through and learn the graph theory stuff for real sometime. I'm surprised that Subgraph would have this problem.

@Searke The graph functions have a few really frustrating (and to an outsider, trivial looking ...) problems. But now that I've gotten pretty familiar with igraph, I can confidently say that the breadth of Mma's graph functionality is quite amazing.

igraph is among the best and biggest open source packages for graph/network analysis, but Mathematica has a lot more functionality than igraph.

It was mostly new in version 8, right? I haven't played around with it much since except for .... my more creative unit tests.

Of course there's still a lot in igraph that's not in Mathematica (and vice versa), so an interface between the two is very useful for me.

I remember being very frustrated at the incompatibilities with Combinatorica and how hard and confusing everything was.

I spent a fair amount of time showing people who to convert to/and/from Combinatorica graph expressions

Since version 10, there's really very little in Combinatorica that's not built-in.

Oh also. I was looking back yesterday and I saw there was a conversation about contexts and the lack of desire to use them.

@Searke The GraphUtilities package has a function to convert between the two kinds of graphs. But there's something frustrating about it: the conversion function auto-loads Combinatorica. I like to load Combinatorica as Block[{$ContextPath}, Needs["Combinatorica`"]] to avoid shadowing system symbols. If ToCombinatoricaGraph didn't add Combinatorica to the $ContextPath automatically, that would be quite useful.

Let me once again reiterate that I too wish there were more of a "core language" context

Isn't it calling the Graph... constructor... is constructor the right word?

Isn't it calling the Graph constructor in the Combinatorica context?

I guess it makes sense.

Do you have access to all the support requests? I asked about this in CASE:118856.

No.

Why is it taking a while?

You know you don't ask particularly easy questions.

@Searke No, I mean I described in more detail how the GraphUtilities package could be changed to avoid adding Combinatorica to the context path every time ToCombinatoricaGraph is used. But it's not a big deal. Actually I don't use Combinatorica any more.

@RolfMertig There is already a post about this problem and it was confirmed as a bug.

It was probably filed as a suggestion

Yes, they did file it, I got a response. This was more than two years ago.

oh I see.

It's not a big issue, I hardly load Combinatorica any more. Never used it much because I didn't have the book to go with it.

So

GraphUtilities is officially an obsolete package

No developer is doing to touch it unless it does something truly awful

I don't think it was decided whether it was intentional or not for it to load the package

just that the function itself was part of an obsolete package and so the suggestion was moot.

Yes, it makes sense.

ToCombinatoricaGraph though... is really more of a linking function

than a part of the old package itself

I think the honest reason is that interest in Combinatorica has dropped off too much to warrant worrying about it

@Searke In fact Combinatorica was never really very usable without the Combinatorica book, which didn't come with Mathematica. And now it's nearly fully replaced.

I didn't find the documentation very good

I recently tried to use Graph stuff while testing TextStructure and some related functions

I have to say I found it harder to use than I remember

I thought there'd be an easier way to find the leaves of a tree for example

than to Select vertices by their degree.

@Searke For directed trees, I would use Pick[VertexList[g], VertexInDegree[g], 1].

Sorry, didn't read your message.

For undirected, you may need to filter the root vertex, as that can have degree 1 too.

@Searke Can you edit the documentation? If yes, here's a suggestion: add this example in the GraphLayout documentation page:

SetProperty[
 ExampleData[{"NetworkGraph", "AmericanCollegeFootball"}],
 GraphLayout -> {"EdgeLayout" -> "HierarchicalEdgeBundling"}
 ]

I think it looks really cool and should be advertised more.

I've been looking at "EdgeLayout" -> "HierarchicalEdgeBundling" for 2 years, and could never figure out how to use it because I always specified a "VertexLayout" to go with it. If I do that, it just seems to have no effect.

Or maybe it's not well documented on purpose because it's still being developed? The documentation does mention that "EdgeLayout" -> "HierarchicalEdgeBundling" is a possible setting, I just didn't realize that it also implies a vertex layout ...

I doubt it will get in anytime soon, but I forwarded the suggestion to the hopefully the right person.

Thank you!

BTW I (mostly) re-implemented this layout from scratch, just for fun. I will post it on Community in 1-2 weeks (after I release the next IGraph/M version, which takes care of some boring sub-tasks).

Sorry, Pick[VertexList[g], VertexInDegree[g], 1] is wrong. It must be Pick[VertexList[g], VertexOutDegree[g], 0]

@William About your question

I'm not exactly sure, which feature of a notebook you are missing, but a Mathematica notebook and and IDE editor are fundamentally different.

You can of course simply detach single editor windows from the main IDEA and move them around, but is it really that what you seek?

@William About the Lisp editing in IDEA: I'm not aware of any working LISP plugin for IDEA and without it, only basic text-file editing features are working.

Basic matching of paired braces is no indicator that IDEA understands the syntax. It doesn't, sorry. There is an old Scheme plugin (called Schemely), but it says in its description:

> This code is currently unstable and only suitable for enthusiastic alpha/beta testers.

That is probably the closest you will get to LISP editing. Other than that, there is some kind of placeholder for a LISP plugin that needs to be implemented in the Jetbrains repository.

@SquareOne @SquareOne Thanks. I missed that. Need to follow closer, but this gets harder with all the uninteresting questions.

Hi all

How are all the mathematicians doin :D

@Mehrad Well, you'll find a lot more Mathematicaticians here.

@halirutan in mathematica wiki?!! :(

Wiki people are mean talking to electrical engineers though :P

especially the ones with really noisy data looking for some mean methods to smooth them out

@Mehrad No, I just wanted to make sure you know, that this room is about Mathematica and not math.

not gonna lie. I never used Mathematica. but plenty of Octave

However we are keeping an eye open for some other software package for our R&D team

@Mehrad Octave is open-source and free. Those two are big advantages.

And Matlab is proprietary and extremely expensive

We might settle in some middle grounds

@Mehrad The same is true for Mathematica.

It's probably not as expensive as Matlab with all its packages but still.

Not gonna fire away and ask How does it compare to other packages since I don't wanna get a "let me google it for you" answer :D

but that was the first thing I have noticed in the wiki page you directed me to. Proprietary and licensed for commercial use.

@Mehrad I could only help you on the Mathematica side anyway, because I wrote only 2.5 lines of code in Matlab some years ago.

It was fun days we used to use Matlab for free in the uni not knowing how much would it cost us using it the moment we leave

Somehow I get the feeling the syntax is very similar in all these packages :D

sometimes when I can't find my answer for Octave I dive into Matlab helps

@halirutan and I appreciate the offer :)

@Mehrad I read about Octave some time ago and it seems to be more advanced than all the open-source Mathematica clones.

@halirutan Oh yeah. take top end specialty packages of Matlab out and it's basically the same

I use it in quite a light weight for compiling the data from our experiments and never felt any difference. But it surely doesn't have Simulink

They have also did a good job in the GUI they have introduced in the later versions so it's easier to use than ever

@Mehrad Can you do analytical calculations? Like for instance hmm..

calculate the integral of say cos(phi)^2*phi?

As I said, I am not the best person to ask but I don't see why not

@Mehrad No problem. Should probably just try it.

Doing it atm :D

 d = cos(pi)^2*pi
d =  3.1416

:I

@Mehrad I meant phi not pi :-)

Something like this:

Oops :D

It's quite early over here in Australia. So don't mind my sleepy eyes :D

I've done it in MATLAB once. It doesn't fit into MATLAB's paradigm naturally as it does for Mathematica, for example you had to declare which variables were symbolic because MATLAB doesn't treat variables as symbols by default. But it worked.

But I doubt very much that it is state of the art like Mathematica or Maple; probably it was more like SymPy.

@Pickett well said

1.2.5 Integrating Differential Equations

Octave has built-in functions for solving nonlinear differential equations of the form

dx
-- = f (x, t)
dt

with the initial condition

x(t = t0) = x0

and I guess this is as far as Octave goes :D

But does it solve them symbolically or numerically? I.e. can you input an equation in terms of variables (x,y,z) and get back a mathematical expression?

@Mehrad But this is again numerics. Solving a differential equation with initial conditions is different from deducing the analytic solution.

exactly

@Pickett Do you have any experience with SymPy?

I wanted to highlight solving in the help example but i forgot

I wonder how fast it can be if it is really entirely written in python

@halirutan Not in Python, but it is what Mathics uses. I have not tried Mathics extensively either, and I'm pretty sure it's not that fast or great.

@Pickett I tried Mathics once, but it was in no usable state for serious work.

@halirutan No, but I guess the symbolic part might be the part that is somewhat usable. Supposedly it's just a different (better as it seems) syntax for SymPy.

The server is down I guess.

@Pickett Give me a second I have a link that works I think

I have been trying to contribute to Mathics lately.

@halirutan @Pickett Here is one that works mathics.angusgriffith.com

halirutan I asked the lisp question here

@William that's the same link

@halirutan I used a different link, the link @William gave worked for me.

@Pickett Yes, for me too.

Good, I believe the site is maintained by the old maintainer(although I haven't gotten a straight on github yet)

@halirutan The link is here for the lisp question sorry now I understand.

@William If you don't mind using emacs, then you should use it. It's written in Lisp and should support it very well.

@halirutan emacs probably would work with I think its expand-select function(or something don't remember exactly) just never work intuitively for me. I am on Windows although it appears both Vim and emacs seem to have ports.

I have never enjoyed the edit mode vs regular mode in both vi and emacs. It makes pasting and copying just slightly more annoying.

@William Emacs as well as vim is very different from common editors.

It disappoints that no lisp or any language supports(Haskel might I haven't checked) supports anywhere near as many functions as Mathematica. Mathics is a nice attempt, but it will likely never hit maturity or speeds that I would like. Lisp isn't really a viable option simply because IMO it doesn't support arguments types.

@William Well, maturity is an evil word, because it kind of implicates usable as Mathematica. Now guess how much energy and men-power it will take to create a piece of software that is even nearly equivalent to a program that has been developed since 1990.

Not to mention the very specialized knowledge of the many mathematicians that worked on the kernel. Mission impossible. Even with nicest piece of algebra library under the hood like SymPy.

@halirutan True maturity is evil. Let's just start with a decent Pattern Matcher.

Writing "guaranteed correct" algorithms is, as noted, highly nontrivial.

@William That's quite nontrivial itself.