Wolfram Mathematica - 2014-07-03

« first day (897 days earlier) ← previous day next day → last day (3885 days later) »

halirutan

00:12

@Szabolcs Spare your efforts. It is indeed the highlighting pass itself which slows things down. It is the setting and storing of text-attributes, which is done so many times in a large file with many, many symbols that need to be colorized.

Gabriel

@acl have you tried the ipython notebook? I have been using it a lot lately ... feels like the notebook with better underlying technology ... but missing a lot of features! Still as a mathematica user that dabbles in python (Pandas is just to amazing ... makes me sad about Dataset every time I use it) I found it much more welcoming than a full featured IDE (unless I needed all the power of wing).

I wish the mathematica frontend had the vim "modes" of ipython and the super duper sweet mathematical display (I also really like the whole write in markdown and then it displays, the programmer in me I guess).

1 hour later…

Nasser

01:16

I just noticed something strange answering a question. I think it is numerical issue. or may be I am missing something. Why does `x = -15*0.1 + 6*0.1; x==0.9` return False, while

  -15*0.1 + 6*0.1==0.9

return True?

I think there was a post on these issue, but could not find it now.

Nasser

01:27

Sorry, forget the above. I had typo. It should be -0.9. All ok now. they both give same result. I need new glasses.

halirutan

01:39

@Szabolcs Thanks again. The problem seems solved now. Read here.

bobthechemist

02:30

I'm sure you've all been waiting: Check out the latest installment of controlling an LCD screen with Wolfram/RPi, where I visualize differential equation solutions in 80x16 pixel low def!

Rojo

02:46

@Nasser Well, it was a numerical issue :P

belisarius

@Rojo The only numerical issue that matters right now is ARG-BEL 2-0

Nasser

@Rojo yes it was. In the comment above, I was just typing 0.9 in one place and -0.9 in the other. I need to go buy new glasses tomorrow when the stores open.

Rojo

@belisarius Pero Argelia ya perdió :P

Nasser

Are they still playing this game? most boring game ever invented this soccer thing.

:)

A bunch of big boys chasing a ball around for 2 hrs

belisarius

@Nasser Are you from Cameroon? :D

Rojo

02:49

Haha

brama

03:06

@MichaelE2 Yes. I figured it after reading your suggested link. Thanks to you.

Nasser

setting Opacity for 3D graphics objects slows down the animation so much, using it is useless. Is there something else to use to make transparent cylinder say, if needed, but not using Opacity?

may be I should ask this on main board.

Rojo

@Nasser If it is really 100% transparent, perhaps the animation would be smoother if you omit the cyllinder

Nasser

@Rojo no, I set a slider to allow me to adjust opacity. When it is 1, the animation is very fast, when it is less than 1, it slows down allot. I know why, but need a solution without this slowing down. Makes using it useless for Manipulate.

I do not think Manipulate is good for heavy duty graphics 3D simulation really, performance wise, but it allows me to quickly draw things, that is why I use it.

I think WRI should work to optimize this part. I do not know if they use openGL or windows direct3D, but it is slow.

Rojo

@Nasser Perhaps adding ContinuousAction->Dynamic[opacity==1] to the other controls so that when the user changes the opacity, it uses False?

Nasser

@Rojo, sorry, I am missing what you say. I allready changed the opacity to say 0.5, then the simulation is now running full throttle on its own, but it slows down. I am not moving any controls any more. It is one 3D object has opacity setting to less than one, the whole rendering slows down.

Rojo

03:20

@Nasser I doubt you can magically make Opacity work faster, but who knows. I was only suggesting something to make it usable, so that it doesn't try to render it again every differential change you do to your sliders, but only once you choose a new value

Nasser

@Rojo you can see yourself here, 12000.org/my_notes/mma_demos/gyroscope_2/index.htm if you change the opacity, it slows down. keep it at 1 it goes fast again

Rojo

Let's see

Uh, I need the plugin. Time to install it

Nasser

@Rojo You can download the CDF

here is the link, page above it: 12000.org/my_notes/mma_demos/index.htm no need to install plugin

I assume you have Mathematica :)

@Rojo the problem is not while the slider is moving. It is after the slider has already been set. halirutan explained to me why it slows down. due to the many extra layers it has to process when opacity is not 1.

Rojo

@Nasser I am downloading. But even if it takes time after set, and that is not ideal, why does that make it unusable?

If you want opacity, pay the price. Otherwise, don't choose it. But it is always responsive, right?

Nasser

@Rojo it slows down the simulation alot. It makes it suddenly slow down. it does not feel right, when it was fast, and suddenly it slows it. This is supposed to reflect physical simulation.

It is always responsive, but changes speed. And the change of speed is significant.

Rojo

03:26

@Nasser Ah, got it. I missed the fact that it was an animation

Sorry

Nasser

@Rojo, yes. It is supposed to show real actual physics of what happens. If it slows down suddenly, it makes the animation not to appear real.

Rojo

@Nasser How are you doing the animation?

Nasser

@Rojo Manipulate?

Solve the Euler 3D equations, using fixed delT, plot, solve again, etc....

It runs for ever

I have a patent on this method :) I call it the tick method.

Rojo

@Nasser I see. So you use a fixed delT and it runs as fast as the processor

Nasser

@Rojo exactly ! full throttle

I control speed by having slider to change delT. This gives the appearance of speed change. But it is always running full CPU speed

1 hour later…

belisarius

04:56

user image

Rojo

05:07

Totally unrelated to anything. This guy has very nice videos. youtube.com/channel/UCKY00CSQo1MoC27bdGd-w_g

belisarius

@Rojo ha! unrelated to what, exactly? That's a pure 42!

brama

@belisarius I found the reason for the error in chat.stackexchange.com/transcript/message/16387831#16387831. It is that the "inlining of compiled, listable functions is not possible" as mentioned in chat.stackexchange.com/transcript/message/16349840#16349840. I was using a listable function whose input is a scalar and using last/@kr gives me a vector

Rojo

@belisarius Haha, true

belisarius

@brama Good!

@Rojo The sound is annoying :(

Rojo

@belisarius the voice that autoreads? You get used to it :P. Perhaps english captions help?

It's an alien speaking English, be understanding

belisarius

05:20

@Rojo I don't understand. All that work making the pictures/animations and creating the script ... and then using that stupid autoreader

Rojo

@belisarius Agree

belisarius

@Rojo Anyway, it seems the first episode has the worst synthetic reader. The rest seem much better. Thanks for the link!

Rojo

@belisarius Enjoy :)

rm -rf

06:12

@Gabriel I like these about ipython notebook as well... To be fair, the mma notebook came first and was indeed revolutionary at the time (and inspired ipython nb). Given that they've build up a large user base over 25 years, one can't expect them to change the fundamentals of their product interface, no matter how popular some of the new age features might be.

Unless of course, if it involves breaking your users' code by introducing useless and impotent Quantity entities everywhere and "injecting computation" in every crevice... that's A-ok!

Rojo

@rm-rf Did you ever tried to implement anything similar to vim modes, to navigate the notebook?

rm -rf

@Rojo No. I seriously considered it at one point, but I didn't think there'd be much use for it (and it is not trivial). People have plenty of problems using it as is and learning vim-like usage involves an additional overhead as well.

belisarius

@rm-rf Hi toad! Long time no see you.

rm -rf

@belisarius Sorry, been very busy lately :)

and hi!

belisarius

@rm-rf Same here. Damn. Work should be against religion.

Kuba

06:31

Hello

Gabriel

@rm-rf totally I am just happy someone finally is copying it! I was slow to get notebooks, but I hate going without them now. Still the mathematica implementation is starting to feel a bit dated, I hope they and pep it up a bit without pissing everyone off

1 hour later…

Kuba

07:59

How to make Mouseover working for whole area, even for the part that is under second Disk[]

Graphics[{
  Dynamic[{If[CurrentValue["MouseOver"], LightBlue, Blue],
    Disk[{1, 1}]}],
  Red, Disk[]
  }]

@Pickett any ideas? :)

You can add Deploy :P but this is a bug and I would prefer a stable solution :)

@Pickett I'm counting on you, where are you? :p @acl maybe?

1 hour later…

belisarius

09:12

Graphics[{Dynamic[{If[CurrentValue["MouseOver"], LightBlue, Blue],
Disk[{1, 1}]}]}, Epilog -> {Red, Disk[]},
PlotRange -> {{-2, 2}, {-2, 2}}]

Kuba

@belisarius nice, I was not expecting this

belisarius

@Kuba Me neither :)

Kuba

@belisarius what about

Graphics[{Dynamic[{If[CurrentValue["MouseOver"], LightBlue, Blue],
    Disk[{1, 1}]}],
  Dynamic[{If[CurrentValue["MouseOver"], LightRed, Red], Disk[]}]}]

belisarius

@Kuba But that one does a different thing. What do you want now?

Kuba

@belisarius the same, the common area should trigger both of them

belisarius

09:23

@Kuba Oi! I don't know

is that possible?

Kuba

@belisarius you can use that bug e.g. :p

Deploy@Graphics[{Dynamic[{If[CurrentValue["MouseOver"], LightBlue,
      Blue], Disk[{1, 1}]}],
   Dynamic[{If[CurrentValue["MouseOver"], LightRed, Red], Disk[]}]}]

but I don't want it :)

belisarius

@Kuba I remember you've posted something bout it

Kuba

@belisarius yes, then it was about how to prevent it.

belisarius

:)

Kuba

I want this behaviour but not as a side-effect, because I might want "normal" for other objects in graphics

belisarius

09:28

@Kuba Perhaps with nested Eventhandlers,

and PassEventsDown

not really sure

Nah, event "mouse over" isn't there

Kuba

@belisarius you can mimic it with:

Graphics[{
  DynamicModule[{over = False},
   EventHandler[
    Dynamic@{If[over, LightBlue, Blue],
      Disk[{1, 1}]}, {"MouseEntered" :> (over = ! over),
     "MouseExited" :> (over = ! over)}, PassEventsDown -> True]],

  DynamicModule[{over = False},
   EventHandler[
    Dynamic@{If[over, LightRed, Red],
      Disk[]}, {"MouseEntered" :> (over = ! over),
     "MouseExited" :> (over = ! over)}, PassEventsDown -> True]]
  }]

@belisarius but the problem is that those EHandlers are not nested but "next to" each other so passing event does not work

belisarius

@Kuba I see. I hate Dynamic

Kuba

@belisarius Dynamic is great, GUIs in MMA suckssss

Nasser

@belisarius would you have an idea why WRI made Dynamics instead of normal callbacks for GUI as is done in all other UI frameworks?

@Kuba I only use Dynamic for GUI. Why else would one need it?

belisarius

@Nasser I remember I've read something Leonid has posted about that. Not awake enough now to remember it

Kuba

09:40

@Nasser I meant GUI interactive controls

Nasser

Ok, thanks. I always wondered who came up with Dynamics. It is unique to Mathematica really.

I wonder if Dynamic is Wolfram's own idea? or someone else at WRI thought of it.

1 hour later…

Nasser

10:45

I love simulation of gyros :)

user image

acl

11:30

@Gabriel hmm as a matter of fact, no.

m_goldberg

11:46

@Nasser. The name I associate with user interface stuff at WRI is Lou D'Andria.

m_goldberg

12:26

@Nasser. I got to thinking about what you said about callbacks. It seems to me that going the normal callback way would not fit well with functional programming. Callbacks do their work through side effects. Functional programming wants to eliminate side effects, making for a paradigm clash. Of course side effects must happen for a GUI to work. Dynamic is WRI's way of implementing side effects while maintaining a functional looking syntax.

4 hours later…

Leonid Shifrin

16:00

@m_goldberg Mathematica's event / UI model implements a version of functional reactive programming, which is a way to avoid the "callback hell" and make event management sit well with functional programming.

bobthechemist

Anyone familiar with the vim-mathematica plugin and able to help me get the autocomplete feature to work?

rm -rf

@bobthechemist What's happening?

brama

Does any one know why elegant and efficient functions such as Clip, ArrayFlatten, and ArrayPad etc. are not compilable?

bobthechemist

@rm-rf Well, I'm trying to get it to work on my RPi (never used it before on any other platform). unzipped to .vim and I do get syntax highlighting but (a) the highlighting doesn't look right and (b) CTRL-X XTRL-O doesn't find autocomplete options with sitting on Dyn or Man (for example).

rm -rf

@bobthechemist Ah, Rojo had a similar problem... it has to do with where the plugin files are installed. Does your directory structure look like .vim/vim-mathematica/stuff?

bobthechemist

16:12

@rm-rf yup, .vim/vim-mathematica-master/stuff and I've also tried .vim/bundle/stuff

rm -rf

@bobthechemist Ok. The first is the incorrect way :) If you're using plain vim w/o a package manager, then the structure should be .vim/stuff... those stuff folders are shared by all plugins

If you're using pathogen, then the second is also incorrect and the structure should be .vim/bundle/vim-mathematica/stuff (assuming the pathogen function is called in .vimrc)

I do see a small hiccup with the ctrl-x + ctrl-o on my end, but let me check a few things first... I might have installed a few other plugins since then that interfere with the shortcuts

bobthechemist

@rm-rf Not using pathogen, so should the plugin be in .vim/vim-mathematica?

or do the three subdirectories (in this case after, ftplugin and syntax) go in .vim?

rm -rf

@bobthechemist Yeah, the subdirectories all go in .vim. Every plugin will have a similar folder structure and they'll all share the same subdirectories

bobthechemist

@rm-rf I still must be missing something .vim now contains the subdirectories after, ftplugin and syntax; the latter two contain mma.vim files and after contains syntax/mma.vim. I've got syntax on and filetype plugin on as the first two lines of my .vimrc yet CTRL-X CTRL-O states that omnifunc is not set

rm -rf

16:29

@bobthechemist Ok, we'll come to omnifunc in a bit, but does the highlighting look good now?

bobthechemist

@rm-rf I'm seeing highlighting, but I don't think it is correct. For example with f[x_]:=Plot[Sin[x y],{x,0, 4 Pi}] the only highlighting I get are brackets and numbers.

rm -rf

@bobthechemist Can you try restarting vim?

Also, what does :scriptnames show? Do you see the files in .vim/ftplugin/mma.vim, etc.?

bobthechemist

@rm-rf restarted - it looks like the plugin is not being loaded. :scriptnames gives me plenty of scripts loaded from /usr/share/vim/vim73 but no reference to ~/.vim.

rm -rf

@bobthechemist hmm... do you have a .vimrc?

bobthechemist

@rm-rf Yup, (and it's referenced in :scriptnames). Contains syntax on, filetype plugin on as the first two lines, then a few sets (number, autoindent, background)

rm -rf

16:41

@bobthechemist Can you also set nocompatible as the first line in your .vimrc and try restarting?

bobthechemist

@rm-rf no change.

rm -rf

@bobthechemist hmm... brb in 20 mins

bobthechemist

@rm-rf no worries, thanks for the suggestions

Öskå

17:09

@rm-rf Are you aware that the ad "There are x unanswered questions blabla" doesn't have any image anymore? :)

bobthechemist

@rm-rf Doing some digging, I note that there's a file /usr/share/vim/vim73/filetype.vim that contains a Matlab *.m entry. I wonder if this is causing part of the confusion.

and :setfiletype mma does the trick for highlighting.

rm -rf

17:38

@bobthechemist Ah, yes :D Forgot to mention that

@Öskå Pushing a fix now. It's because of this

bobthechemist

@rm-rf now all that's left is the autocomplete. Would this be the same problem and vim needs to know I'm not looking at a matlab file?

rm -rf

17:51

@bobthechemist if you've set filetype=mma for the file and your .vimrc has filetype plugin on, then omnicompletion should work

Can you paste the output of :version?

brama

@PatoCriollo I realize you live in Atlanta as I do. Do you work at GaTech?

Is there a compilable way to rewrite this without creating a copytensor? p1 = RandomReal[10, {5, 4}]; P1=ReplacePart[#, abcd, 1] & /@ p1

bobthechemist

VIM - Vi IMproved 7.3 (2010 Aug 15, compiled Feb 20 2013 06:48:10)
Included patches: 1-547
Modified by [email protected]
Compiled by [email protected]
Huge version without GUI. Features included (+) or not (-):
+arabic +autocmd -balloon_eval -browse ++builtin_terms +byte_offset +cindent
-clientserver -clipboard +cmdline_compl +cmdline_hist +cmdline_info +comments
+conceal +cryptv +cscope +cursorbind +cursorshape +dialog_con +diff +digraphs
-dnd -ebcdic +emacs_tags +eval +ex_extra +extra_search +farsi +file_in_path

part 2
user vimrc file: "$HOME/.vimrc"
user exrc file: "$HOME/.exrc"
fall-back for $VIM: "/usr/share/vim"
Compilation: gcc -c -I. -Iproto -DHAVE_CONFIG_H -g -O2 -fstack-protector --p
aram=ssp-buffer-size=4 -Wformat -Werror=format-security -U_FORTIFY_SOURCE -D_FOR
TIFY_SOURCE=1
Linking: gcc -Wl,-z,relro -Wl,--as-needed -o vim -lm -ltinfo -lnsl -lse
linux -lacl -lattr -lgpm

rm -rf

@bobthechemist Hmm... it's a new enough version so it should work. Ok, let's try this manually. Can you enter the following:

bobthechemist

@rm-rf hold on, might have figured it out

rm -rf

set omnifunc=syntaxcomplete#Complete

and then press Ctrl-X Ctrl-O in insert mode?

bobthechemist

17:58

@rm-rf Yeah, that's working.

I think I need filetype plugin on in my .vimrc; and it looks like that needs to be before I ever set(force) the filetype to mma.

rm -rf

@bobthechemist Yes. I thought you said above that you had it in your .vimrc...

In any case, adding that to your .vimrc and setting ft=mma should do it.

bobthechemist

@rm-rf it was, but then I started "debugging" ;-)

rm -rf

vim remembers the ft for a file once you set it. I think there's even a way to get it to recognize all .m files as mma instead of matlab (if you never use matlab), but that might be overkill

bobthechemist

I'm ok with turning it on when needed. Thanks for the help.

rm -rf

No problem :)

acl

18:46

@rm-rf or anybody else, any experience on what I should use for parallel python?

trivially parallelizable, like ParallelTable

do I go for multiprocessing?

Szabolcs

19:04

@acl I don't know and I have no experience with this, but I am curious now. IPython is supposed to have some similar functionality: ipython.org/ipython-doc/dev/parallel/parallel_intro.html

PatoCriollo

@brama no @Brama, I work across the street from you, the bubbly building in North Avenue.

acl

@Szabolcs OK, I need to read it to work out if I can easily use that on our cluster

Szabolcs

@acl I'll send you a notebook I just made ...

@acl Do you use IPython notebooks?

@rm-rf Do you know how to render MarkDown cells in IPython notebooks, so I see the rendered HTML, not the MarkDown source?

rm -rf

19:23

@Szabolcs Yes, just change the cell type to markdown and then type in markdown

I have to reload the notebook sometimes for it to pick up the typesetting changes, but that's about it. Reloading the page doesn't alter the kernel state

Szabolcs

@rm-rf But after typing in MarkDown and moving on to a new cell, I still see MarkDown code, not rendred HTML

@rm-rf OK, got it

acl

@Szabolcs I've never used them no

Maybe I should try although at the moment I'm very happy with writing stuff in an IDE/editor

Rojo

No v10?

Ok

Gabriel

19:38

@Szabolcs is this a notebook showing parallel stuff? Would love to see it as well if so. Really digging ipython/python at the moment. Just wrote some cython simulation stuff that has speed up something I have been doing in mathematica by 100 times! (mostly because I can call low level random number distributions from GSL that get C compiled which I couldn't do in mma). Really in love ... at least until v10 comes out ;)

Szabolcs

@acl If you have ipython installed, just do ipython notebook to launch the notebook interface. Then you can open the file I sent.

@Gabriel Can you email me? You can keep the email body empty if you like

bobthechemist

@Rojo Word from my source is mid-July

Gabriel

@Szabolcs done. Thanks so much

Szabolcs

@Gabriel BTW you know a lot more about scientific computing with python than me so it might be better for you to look at ipython.org/ipython-doc/dev/parallel/parallel_intro.html than to look at the 1 page notebook put together by a novice based on that

Gabriel

@Szabolcs yeah checking that out as well. Been so long since I really used python I feel like a beginner anyway ... still find myself using fun[] instead of fun() all to often ;)

@Szabolcs @acl by the way you can start the kernels from the web interface in ipython by going to the "clusters" link in the start menu thingy (where your list of notebooks is)

Nasser

19:53

debugging Mathematica GUI code is so hard. So, run something and get some error on the GUI saying say "coordinate blalab should triple number blala" with no line number, no clue where it is coming from in the code. I spend so much time looking at the code trying to guess where it is from. Things like this sometimes makes me want to go back to Matlab, which its easy to use debugger.

In Matlab, I would set break point on error, and run the code and it will stop right there at the error. Never took me more than 3 minutes in Matlab to find any problem. In M, I spend 20 times the time just on debugging.

If Matlab GUI was not so terrible, I would switch right now.

Szabolcs

@Gabriel Thanks, that's good to know!

Nasser

if M would at least show the piece of code that generated the error, that would help.

Szabolcs

I'm worried about getting stuck in Mathematica (even though it will very likely remain my preferred tool). So I really need to learn some other tools well.

brama

@PatoCriollo cool....not too far...May be we can meet sometime.

Nasser

@Szabolcs what is your work area?

I am student in mechanical engineeing now, everyone uses Matlab at school.

in engineering departments at least, even in math depts. All teachers know only Matlab.

Job security for them :)

acl

20:37

@Gabriel good to know thanks

@Szabolcs ok that looks neat. I wonder how easy it is to drive "engines" on another machine from my laptop

I have always had trouble doing that with mma

Szabolcs

20:52

@Nasser I'm a physics postdoc, doing complex systems stuff (= not really physics, but same methods as in physics). At the moment, network/graph related stuff.

Szabolcs

21:03

@acl It is definitely possible, there's a little bit about it in the intro tutorial. If you figure out how to do it, please let me know too! (Also your findings about performance.)

Gabriel

@acl agreed. I have seen some amazing videos of people doing this, but would like to know how it goes for a regular user!

@Szabolcs the danger of being stuck in Mathematica is something I have been thinking about lately as well. Definately my favorite environment at the moment -- but I worry in a couple of years it becomes a cloud service like the adobe sweet does ... unlikely, but all the cloud stuff got me thinking I need to have a backup plan

acl

21:22

@Szabolcs OK, I'll try to get to it this weekend

Gabriel

21:42

@Nasser totally. The line debugger in matlab is awesome. I miss having a proper debugger in mathematica all the time.

acl

@Gabriel there is the built-in frontend debugger, as well as this

Gabriel

@acl I will have to try that package, the built in debugger I find very brittle hard to use. I loved how in matlab you just clicked on the line etc ... something that makes less sense in a notebook environment I guess, but really nice from a users perspective. I use workbench when I really need to debug something for this reason

acl

@Gabriel Yes, I also prefer the workbench for that. But overall debugging mma code like this isn't really so easy

Gabriel

@acl yeah. The dark side of more functional styles ... makes debugging a bit more involved

acl

@Gabriel "a bit" indeed :)

Nasser

22:29

@acl I do not think one can debug Manipulate in workbench. Workbench for debugging non-GUI/front end code. At least never heard anyone do that. Since I work with Manipulate, can't use workbench.

acl

@Nasser right, I don't know if you can use it for manipulate. maybe not

Nasser

23:25

What is "WPC" I see it mentioned on community in some posts. is it Wolfram Platform Cloud?

or "Wolfram Performance Cloud" ? too many new names lately, it is getting confusing.

Nasser

23:39

found it. it is "Wolfram Programming Cloud". writing WPC is easier than having to type W-Cloud. I learned a new 3 letter word today.

« first day (897 days earlier) ← previous day next day → last day (3885 days later) »

Transcript for

Jul '143

Wolfram Mathematica

Welcome! This is the main Mathematica chat room for mathematic...

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

site design / logo © 2025 Stack Exchange Inc; legal

mobile