@Szabolcs yep we knew many folks would appreciate it - glad u happy ;)

For some fun :-) github.com/jloughry/BANCStar

@Szabolcs wikipedia article

You guys in Germany never sleep! :)

@Szabolcs "In the 5.1c release the only legal characters are the numerals 0–9, the comma, the minus sign and the carriage return. " that's crazy

comments are strictly prohibited!

@Szabolcs @halirutan is probably working, I'm just cooling off after a day of banging my head against mathematica

@Szabolcs pfft, only real programmers use BANCStar, and real programmers don't write comments

but seriously can you imagine waking up every day and going to work to work with a series of numbers and commas that's meant to be a program?

> Everything in the system is global

> New projects always started off with the programmer searching for a handful of working storage numbers that could be "borrowed" long enough to complete the calculation, then restored to their original values before the rightful owner noticed that they were gone.

It tells something about how the decision to buy and use this system was made ... but they didn't mind paying 7 people to do the development

> Labels within the code are absolute, so any time a new page was added or removed, all downstream GOTOs must be found and re-targeted.

It's like writing the input form of a CompiledFunction directly

@Szabolcs it's much worse actually

Guys how to I get all chess symbols?

[BlackKing]

like that

in a nice simple way ;-)

@Szabolcs at least there you're not limited in the number of variables...

@VitaliyKaurov Us Fireox to copy and paste from the table here: en.wikipedia.org/wiki/Chess_symbols_in_Unicode

Firefox let's you Command/Control select one column only

@Szabolcs CharacterRange[FromCharacterCode[9812], FromCharacterCode[9823]]

Or funnier ...

?? [Black*]

\ missing

@VitaliyKaurov I never knew that was possible

yes, the ??\[Black*]

Got a Raspberry Pi with the infrared-sensitive camera recently, trying to find interesting projects for it.

Here's a section of a whole-night time lapse of our terrace:

dropbox.com/s/a34u58qdcmllel9/trimmed-nightlapse.mov

@Szabolcs Yep, I'm working. Have a colleague who is in Brazil right now. We're chatting about project stuff.

Now I know what's eating the bird seed at night

@Szabolcs ROFL - is that racoon really your neighboor?

Yes, it broke my wife's bird feeder

@Szabolcs Sounds like you have too much time :-)

@Szabolcs this would be a cool GIF

@Szabolcs with a fun WL applications

@halirutan I would guess he was asleep at the time :)

Yes, I was asleep!

BTW this is the upper floor, not the ground floor.

@Szabolcs @acl It was more the trying to find interesting projects part :-)

I want to use Mathematica to detect when there's something interesting in the picture and take a video. That includes trying to detect whether a bird is not just a boring sparrow.

@Szabolcs - post on Community if u come up with anything - this could be so much fun. Then I totally get racoon - who won't get Pissed if treated like a bird ;-)

OK, you're right, actually there's plenty I should be doing, and that includes non-work projects too. MATLink is really due an update.

@Szabolcs btw, containers.Map() <=> Associations[]! Should make it even more useful.

yes, you mentioned that, and it sounds good

We need to look into how feasible it is. It is likely to require some undocumented MATLAB for an efficient implementation/

MATLAB is increasingly relying on classes and object-oriented stuff which is much more difficult to handle using MATLink.

@rm-rf I just discovered that MATLAB has an RPi toolbox ... but it seems to be Windows-only :-(

I no longer use MATLAB seriously... not even on a semi-regular basis. I'm trying to shift more towards python.

@rm-rf how come?

@acl It started out as a slow and natural shift into mma... ended up working almost entirely in mma for 2 yrs, but then realized that by using only mma, you're drawing a box around yourself (even more so than MATLAB) — you face hurdles in collaborating, getting hired (except for those rare jobs that either don't require a $language or use mma), etc. It's still a great prototyping tool, so I'll continue using it for quick stuff, small projects, etc.

@rm-rf And a language like Java or C was no option beside Mma? I mean, Py is often not an endsolution too.

@halirutan Oh, they certainly are options. I just happened to go with python since I already have some experience with it.

@rm-rf It certainly is a nice language.

I lately had to realize (when some guys asked for my help) that it is too slow and too memory hungry for volume processing.

The "all fluff and no substance" nonsense marketing from WRI is not helping either... It's becoming harder to sell it to people, because you talk to them about your real world experience developing large packages, about how it is great at prototyping, etc., but Google for examples and you get lots of fluff... the latest WRI blog post being a good example.

"Ok, it's cool that you can do $FancyStuff by calling one function, but is it fast?" No. "Is it stable?" No. "Is it easy to learn?" No.

@rm-rf Then I looked how all the image-processing libraries are doing it and realized that they just call C code from Py.

@halirutan Yes, almost all the libraries just call C for stuff that needs to be memory efficient.

@rm-rf Especially, the scaling to large projects is non-trivial.

@rm-rf Do you use a notebook with Python? (IPython notebook?)

@Szabolcs maybe you need to make one like that: youtube.com/watch?v=UjYLRLwphcs&feature=kp

@OleksandrR. youtube.com/watch?v=DXqbevF6dWk

@Sektor a bit risky to pick it up? But not much else you can do I suppose...

@OleksandrR. It probably is a risky move, but a nice throw nonetheless .. :D

@OleksandrR. Have you participated in the Wolfram summer school btw ?

@Sektor no, I haven't. Have you? I take it you mean the NKS summer school?

or "Wolfram Science" as he is now calling it...

@OleksandrR. Nopey, and yes, NKS. I am trying to collect info about it

@Szabolcs Yeah, I try to use it like mma... IPython notebook + PyCharm

@Sektor Talk to @VitaliyKaurov about it :)

@rm-rf That was my next move :D

@VitaliyKaurov Vitaliy, are you around ?

@Sektor yeah, @rm-rf beat me to it, but Vitaliy attended in 2010. The list of attendees for each year is on the website but I don't really recognise any of the names in recent years.

@rm-rf I tried it for half a minute and couldn't even delete a cell

@OleksandrR. Oh, I have already consulted the list of attendees :D

@OleksandrR. But thanks for the suggestions :)

@Rojo btw, did you get the plugin working?

Our XKCD answer used in a public talk at the University of Adelaide :)

@rm-rf Haven't retried yet. I've been busy, and there are many things to set up in my transition to Linux. Every day I win a few fights, but there are many

Today e.g I had to fight my display drivers

@rm-rf I'm a fan of the QCD visualizations from the University of Adelaide. physics.adelaide.edu.au/theory/staff/leinweber/features/…

Hi

I need your support if possible.

Could you let me know if Mathematica (newer versions) is able to correctly compute this one? Sum[(-1)^(n + 1) Cos[3^n x]^3/3^n, {n, 1, Infinity}]

@Chris'ssis What is the correct result ?

@Sektor $$\frac{3}{4}\cos(x)-\cos^3(x)$$

@Chris'ssis Nopey, -((3 Cos[x])/4)

@Sektor Wait, there is a sign problem $$\cos^3(x)-\frac{3}{4}\cos(x)$$

@Se

@Chris'ssis I am waiting ;)

@Sektor This is the correct answer.

Anyone know how to catch the symbol that is producing error $RecursionLimit::reclim: Recursion depth of 1024 exceeded? The code is quite long and I can't take it apart now

@Chris'ssis Nopey, adding assumptions also does not help

@Sektor What do you mean by "nopey"? Don't you agree with my answer?

@Chris'ssis I don't get the results you say I should get

@Chris'ssis And frankly I am not in the mood for pen'n'paper calculations right now

@Sektor I consider myself a mathematician ... It's simply $$\cos^3(x)-\frac{3}{4}\cos(x)$$

@Chris'ssis Okay, I am not saying it is right or wrong. In any case, adding proof to your question would make a valid point that there's something wrong going on :)

@Sektor use the identity $$3\cos(x)+\cos(3x)= 4\cos^3(x)$$

$$-\frac14\left[3\sum_{k=1}^\infty(-3)^{-k}\cos(3^kx)-3\sum_{k=1}^\infty(-3)^{-k‌​-1}\cos(3^{k+1}x)\right]=\cos^3(x)-\frac{3}{4}\cos(x)$$

@Chris'ssis As I said right now I am not using/writing/solving anything new, because I am finishing a project :)

@Sektor Good luck with your project then! :-)

@Chris'ssis Thanks :) !

@rm-rf Not easy to learn? Why? I think, the idea of users learning by examples is great. I always found “man pages” approach awful compared to this. It's like developers now feel obliged to write something sciency-looking instead of something immediately useful…

@rm-rf …I'd like to start with Haskell and, while having this mathematica.stackexchange.com/questions/18548/… set up, the lack of example-driven documentation feels like an obstacle. Also, programing environments suggest to “create a project” when you first run them; notebook interface is much frendlier. Mma was very easy to learn for me, and it's after several years of no programing at all.

@Kuba Naive way would probably be to switch off the message half way through. Then you know if the message comes from the first half or the latter half. Then subdivide that half and so on. You should be able to narrow it down with just a few subdivisions, I think?

@Akater Because the some of the most basic principles in Mathematica are hard to grasp and learnt very late.

@halirutan not to mention issues that you "have to get used to" :P

@halirutan Well, maybe I got lucky: we had a one-semester course with it (studying bifurcations, IIRC), and professor favored the idiomatic approach to language (so, there were no For's, etc.).

@Pickett neat :)

@Akater I even can give you an example from some minutes ago where a friend asked me how he can make a call like

Map[PDF[NormalDistribution[3, 2], #] &, {1, 2, 3, 4}]

@Kuba That never was a problem in my case. :-) I could never get used to what I see in other languages (except for Lisp but I only read about it, not really use it — due to white-on-black console environments, again). So I suspect if you don't have common programing habits it's never a problem.

faster, because he suspected that the PDF is recalculated in each iteration.

@Akater Although the solution is simple and knowing the evaluation process, Attributes, etc, it is perfectly clear why this happens, I guess many users struggles with something like this.

@Akater I only know basics of C++ so I'm not biased but the documentation is a mess. Standard, but more sophisticated contructions are not explained and the details matter.

I had no idea that function can do that and With will fail. Maybe I'm just not bright enough :P

Well, it is documented but is hidden as tiny font comment somewhere in docs

@Kuba Yes, and exactly those little problems will leave even experienced users with no clue. Scoping is anyway an interesting topic in mathematica.

@Kuba I was looking at your example in the comment. Did you realize this:

@halirutan yes, it fails :P

@halirutan but I'm not sure why

:15922357

? :)

Row[{ArrayPlot[tt, Background -> Blue],
  ArrayPlot[tt, ColorFunction -> (GrayLevel @@ Reverse[#] &),
   Background -> Blue]}]

Hehe..

@Kuba Additionally, setting a ColorFunction explicitly makes that the raster data is compressed.

Look at the pure expressions of the above example.

@halirutan good point.

@halirutan

Row[{ArrayPlot[tt, Background -> Blue],
  ArrayPlot[tt, ColorFunction -> (GrayLevel @@ (1 - #) &),
   Background -> Blue]}]

quick fix, not sure if stable :P

@Kuba And do I have to understand this? :-)

I mean, why is it that inconsistent?

I mean the doc clearly states that

> An array of pairs of numbers is treated as a gray level with opacity:

@halirutan moreover try it with mixed scales: tt = Table[{.1 i, j}, {i, 0, 10}, {j, 0, 10}]; :P

@halirutan well, I think it's the legacy of different people developing different stages of mma without clear strategy :)

Last time when I was here we talked about our community. I just want to say I find it fascinating that the author of a decent-looking package manager is a law student mathematica.stackexchange.com/questions/49111/… (that's what his profile says). :-)

@Kuba the problem is, that not GrayLevel is used directly. It's as in your example inverted, meaning 0 is white.

@halirutan @Kuba Got a minute ?

@Sektor Yes, but I'm going to leave soon. What;s the problem? (hope it's not tough one)

:)

@Kuba I have a set of data (7x601) and am trying to plot a 3D histogram of every sub-set

The problem is when I pass the data to Histogram3D it actually tells me that its not a valid data set :D

@Sektor Dim is 7x601x1 or x3?

7x601

I was originally trying to create a waterfall out of the 7 different histograms

@Sektor I'm not sure why Histogram3D what would x - y refer to?

@Kuba Yes, I just realised it doesn't make sense :D :D :D

@Sektor :)

Thanks :D :)

@Sektor No problem :P

Nevertheless suggestions how to create waterfall in 3d out of the 2d histograms ?

something like this out

@Sektor Like this:

i = 0;
SmoothHistogram[RandomReal[NormalDistribution[1, 3], {5, 500}]] /.
 Graphics[content_, ___] :>
  Graphics3D[
   content /. Line[pts_] :> (i++; Line[Append[#, i] & /@ pts]),
   BoxRatios -> 1]

@halirutan Exactly ! :D I have done this with the lines, but histograms .. :D

@Kuba Yes, thank you :)

@Sektor make up your mind :P

@Kuba I will try both of them :D

@Sektor @halirutan @Akater See you later guys :)

@Kuba Bye.

@Kuba Later :) and thanks again :)

@halirutan Is it possible to use MathGL3D with the current versions of Mathematica ? I tried to plot some rather large data sets and the front end just chokes...

@Sektor No. The transition to the new lighting model introduced in V6 or 7 was not finished.

@halirutan So, in order to produce huge plots I need to upgrade my cpu/mem and can not exploit my gpus ?

@Sektor If the FE and the display of the Graphics is the problem, you could try to export it to something which can be used in other programs... MeshLab, Blender, etc..

I guess it is only the display which slows everything down and not the underlying data.

@halirutan Yup, the display

@Sektor Is it a possibility when you tweak your graphic with low settings and when you have set everything up, you render it with quality?

@Sektor For instance this

@halirutan Haven't tried that

    gr = ParametricPlot3D[ {Sin[u] Sin[v] + 0.05 Cos[20 v],
   Cos[u] Sin[v] + 0.05 Cos[20 u],
   Cos[v]}, {u, -\[Pi], \[Pi]}, {v, -\[Pi], \[Pi]}, MaxRecursion -> 4,
   PlotStyle -> None, Axes -> None, Mesh -> True, MaxRecursion -> 7,
  PlotPoints -> 50]

is really fast once it is calculated. When you insert the PlotStyle {Orange, Specularity[White, 10]}

you see how performance drops.

yup

well the only thing i had one was a colorfuncton

will check and see if it is a bottleneck

@Sektor Yep, good luck.

Thanks for the suggestions :)

Does anyone here have practical experience with processing large datasets with Mathematica?

"Large" means "too large to fit in memory all at once".

Here's a toy problem, to illustrate the type of problem I might want to solve: there's a very long table of numbers in plain text format with two columns. Find all rows where the first column stops increasing (i.e. if it goes like 1 2 3 2 then find the third row). These mark "section boundaries". In each section, find the first occurrence of a zero in the second column (if it exsits).

@Szabolcs I've used OpenRead and Read successfully in situations like that. You have to switch to more procedural programming style, but it was a pretty smooth user experience I thought.

I mean, that's the most straightforward way. But is there anything better? Is there a better way to store data, for example? Does it make sense to put it in a database (I have no experience with this)

If you are going to be accessing it a lot and not always the whole thing then it definitely makes sense to put it in a database. I'd only use OpenRead/Read for single-pass analysis. It will definitely be easier to jump to the different sections with an appropriate database schema than just a single linear file.

What shall we do with this? It's not a question and it doesn't fit on the site in this form, but I don't want to discourage the OP.

@Öskå This looks weird to me because the stuff inside the outer Dynamic creates another expression with Dynamic.

@Öskå You get the error for the same reason as here: Graphics[Locator@Dynamic[{0, 0}]]

When Dynamic is used inside a Locator, it will automatically assign to the expression inside the Dynamic as the Locator is moved.

@rm-rf, can you just run commands in pycharm or it's like workbench to MMA?

@Rojo it has a command line built in

@Szabolcs I believe that all the stylesheet related posts should be posted here. It goes for the question that you linked, and for this one.

@Szabolcs Let's see

@Öskå That's what I suggested in the comment.

@Szabolcs Ok, a terminal where you can run python, right?

@Szabolcs Ooh sorry I didn't click the link the OP provided

@Rojo Yes, but I didn't really use it so I don't know how it works ...

@Szabolcs And thanks for Locator explanation. The problem is that by removing the outer Dynamic the locator can be moved, but the Arrow doesn't follow.. :)

@Öskå The main problem is what Sz said, that you are effectively writing Locator@Dynamic[{0,0}].

Dynamic in a locator needs to wrap a variable where the values are stored

mhm.., I guess I will just go with Manipulate :)

@Öskå Locator@Dynamic[{0,0}] is not okay because it wants to assign to {0,0}. DynamicModule[{pos = {0,0}}, Graphics[{Locator@Dynamic[pos]}]] is fine.

@Szabolcs Takk :)

@Öskå ikke noe problem

Huh. This chat had messages like "6 seconds later" between comments for some reason, but those disappeared on reload.

Can someone outline a good approach to group datasets, based on their plots having 'plots' of the same shape ? Extreme examples would be datasets that plot like: horizontal line, a diagonal line (uptrend), diagonal (downtrend) and a square wave.

I've been working on the SAX approach but am stuck after standardizing and grouping datavalues and turning them into Strings. SAX then clusters them by string dis/similarity. The only built-in Mathematica String dis/similarity I can find is based on substitutions. I think the SAX approach requires that the difference between ABC and FBC isnt MMA's "1" (1 substitution) but "4" (letters different). Is there an alternate suggestion?

@PlaysDice There are multiple string similarity functions: EditDistance, DamerauLevenshteinDistance, HammingDistance, SmithWatermanSimilarity, NeedlemanWunschSimilarity. None of them give 4 for "ABC" and "FBC" though.

@halirutan anything better than just PDF[NormalDistribution[],{1,2,3,4}]?

@SjoerdC.deVries The actual use-case was a bit more complex.

In this case I simply suggested to define pdf[x_] = PDF[NormalDistribution[3, 2], x];. The difference of := and = is IMO not clear for many everyday-users. You just learn := is for functions and = for variables and you can live with this for a long time.

It's a bit like in this question where a rather simple problem turns out to require injecting code which definitely looks awkward for most users.

@PlaysDice @MichaelHale Can you upvote the question here so that the user can join chat with >20 rep? (@Rojo @kirma)

Sweet. My first phone vote.

@MichaelHale Hehe, thanks!