matplotlib-users Mailing List for matplotlib

On 2007年4月27日 23:22:04 +0100
 darkside <in....@gm...> wrote:
> 2007年4月27日, John Hunter <jd...@gm...>:
>>
>> On 4/27/07, darkside <in....@gm...> wrote:
>> > hi everyone,
>> > I'm trying to solve a lineal differential equation 
>>system, and I,m
>> proving
>> > with the mlab.rk4 function.
>> > The problem I've found is that if the solution if a 
>>complex number, I
>> can't
>> > use this functin, because it doesn't accept complex 
>>number, and I can
>> only
>> > get the real case.
>> >
>> > Have anyone treat with this problem?
>> > What do you use to solve differential equation 
>>systems?
>>
>>
>> rk4 was something I wrote long ago to have a simple ODE 
>>integrator in
>> case scipy wasn't installed on my system. You should be 
>>using the
>> scipy.integrate tools
> 
> 
> Thank you very much!!!
> 
> I used scipy.integrate tools, splitting complex and real 
>part of the
> equations, and it worked so good!!
> 
> I tried to use: In addition I have filed a ticket for a 
>complex ODE solver.
> http://projects.scipy.org/scipy/scipy/ticket/334
> 
> But I haven't been able to compile it. It returns a lot 
>of errors when I
> tried. I suposse that it's because of the fortran 
>compiler, but I don't
> know.
Please remove the lines in zvdemo beginning with line 121.
Rename zvdemo
mv zvdemo zvdemo.f
1. g77 -c zvdemo.f zvode.f
2. g77 -o test zvdemo.o zvode.o
3. ./test > test.out
HTH
 Nils
 

Ryan May wrote:
> Hi,
> 
> I'm looking into using matplotlib as the backend for doing visualization 
> for my dissertation work. This would replace some existing Qt/OpenGL 
> code I've written (but isn't too extensible) for 2D plots of weather 
> radar data (think pcolor style plots). I've been using matplotlib for 
> awhile for other projects, and love it, but it has the drawback that 
> it's interactive graphics are not all that quick (especially pcolor). 
> (Note: I know imshow is faster, but my data are on a polar grid.) I've 
> been spoiled by the speed of OpenGL graphics and the ease of interactive 
> data analysis, so I just can't go back. Has anyone tried making an 
> OpenGL backend? If not, can anyone think of any reason why it couldn't 
> be done, maybe using wx's GLCanvas?
> 
> Any comments, or starting points on where I could see how to go about 
> this would be _greatly_ appreciated.
I expect pcolor will be slow regardless of the backend. Have you tried 
pcolormesh? It is a much faster pcolor-workalike, but unfortunately it 
has a major bug. If I remember correctly, alpha doesn't work right and 
resizing a window doesn't work right if some data are masked; things 
that should get erased, don't. I tried to track it down once but got 
lost in Agg internals. Anyway, try it and see if it is adequate for 
what you need to do now.
Eric