matplotlib-users Mailing List for matplotlib

On 2009-05-23 17:17, Eric Carlson wrote:
> I should read entire posts before sending people down the wrong
> pathways. I just happened to be working on a Python equivalent to MATLAB
> "triplot" stuff when I read your subject line and made the wrong
> assumptions. That program does just plot the edges, as you noted.
>
> I have attached a python program, much of which is a translation of a
> program I found at M*B central, contributed from the outside. Given a
> triangulation, it allows you to interpolate on a regular rectangular
> grid (dx=constant, dy=another constant). In your case, it should allow
> you to use your original triangulation, and should avoid the convex hull
> artifacts of your original griddata plot. I do not know if this new
> program will give you a figure that will look as good as your latest
> based on John's suggestion or not.

delaunay has a linear interpolator implemented in C++ that could be used for 
this purpose, too. The natural neighbor interpolator is only for Delaunay 
triangulations, but the linear interpolator should be usable for general 
triangulations.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
  that is made terrible by our own mad attempt to interpret it as though it had
  an underlying truth."
   -- Umberto Eco

I should read entire posts before sending people down the wrong 
pathways. I just happened to be working on a Python equivalent to MATLAB 
"triplot" stuff when I read your subject line and made the wrong 
assumptions. That program does just plot the edges, as you noted.

I have attached a python program, much of which is a translation of a 
program I found at M*B central, contributed from the outside. Given a 
triangulation, it allows you to interpolate on a regular rectangular 
grid (dx=constant, dy=another constant). In your case, it should allow 
you to use your original triangulation, and should avoid the convex hull 
artifacts of your original griddata plot. I do not know if this new 
program will give you a figure that will look as good as your latest 
based on John's suggestion or not.

Just a few warnings:
1. I use an environment that (like matlab but very unpythonic) imports 
everything, so the attached may need to import one or two more routines

2. I have only tested with generated data, using triangulations created 
with delaunay, but it worked okay as long as my random points did not 
lead to bad triangles.

3. The routines use a lot of loops, so they are relatively slow - should 
be an ideal program for Cython though.

3. method='linear' uses linear interp,and is very stable. It should not 
be influenced in any way by convexity.

   method='quadratic' uses a quadratic method, which does well except in 
the presence of very thin triangles. I suspect that derivative 
estimation at the vertices has problems on the edges and may very well 
be influenced by the convexity.

linear on a fine grid will probably look pretty good for you.

4. to get something as nice as the mayavi sample, you will probably need 
a fine rectangular mesh (maybe 200x200).

I tested the tinterp program in octave for a number of problems and 
found that the quadratic option gave nice results when I did not have 
functional discontinuities, and when I had good triangulations. I have 
been meaning to translate this to Python, so thanks for the motivation.

At the very least, this program may help you appreciate John's 
suggestions more.

Cheers,
Eric




Ondrej Certik wrote:
>> Ok, I made a progress, it seems it's working. Script and picture
> 
> Forgot to attach the script.
> 
> Ondrej
> 
> 
> ------------------------------------------------------------------------
> 
> ------------------------------------------------------------------------------
> Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT
> is a gathering of tech-side developers & brand creativity professionals. Meet
> the minds behind Google Creative Lab, Visual Complexity, Processing, & 
> iPhoneDevCamp asthey present alongside digital heavyweights like Barbarian
> Group, R/GA, & Big Spaceship. http://www.creativitycat.com 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Hi all,

I have developed a little GTK program which uses Matplotlib to render 
plots. Now it is done, I would like to distribute for both Linux and 
Windows platforms but, to make things short, I have some troubles to get 
matplotlib work with GTK backend on Windows.
Basically, here is what I get when I try to use GTK backend with matplotlib:

 >>> import matplotlib
 >>> matplotlib.use("GTK")
 >>> from pylab import *

Traceback (most recent call last):
   File "<pyshell#2>", line 1, in <module>
     from pylab import *
   File "C:\Python26\lib\site-packages\pylab.py", line 1, in <module>
     from matplotlib.pylab import *
   File "C:\Python26\lib\site-packages\matplotlib\pylab.py", line 253, 
in <module>
     from matplotlib.pyplot import *
   File "C:\Python26\lib\site-packages\matplotlib\pyplot.py", line 75, 
in <module>
     new_figure_manager, draw_if_interactive, show = pylab_setup()
   File "C:\Python26\lib\site-packages\matplotlib\backends\__init__.py", 
line 25, in pylab_setup
     globals(),locals(),[backend_name])
   File 
"C:\Python26\lib\site-packages\matplotlib\backends\backend_gtk.py", line 
25, in <module>
     from matplotlib.backends.backend_gdk import RendererGDK, 
FigureCanvasGDK
   File 
"C:\Python26\lib\site-packages\matplotlib\backends\backend_gdk.py", line 
29, in <module>
     from matplotlib.backends._backend_gdk import pixbuf_get_pixels_array
ImportError: No module named _backend_gdk
 >>>

Here is my configuration:
Python: 2.6.2 (r262:71605, Apr 14 2009, 22:40:02) [MSC v.1500 32 bit 
(Intel)] (as given by sys.version)
pygtk: 2.12.1
gtk: 2.16.1
matplotlib: 0.98.5.3 (installed using windows installer from sourceforge)

It seems to me that the matplotlib version provided was no built with 
GTK support (since there is no _backend_gdk.pyd present), am I right or 
did I miss something?

Based on this assumption, I tried to build matplotlib from source... 
Everything went just fine until mingw tried to link this nasty 
_backend_gdk.pyd.
The following command:
C:\MinGW\bin\gcc.exe -mno-cygwin -shared -s 
build\temp.win32-2.6\Release\src\backend_gdk.o 
build\temp.win32-2.6\Release\src\_backend_gdk.def -Lwin32_static\lib 
-LC:/GTK/bin -LC:/GTK/lib -Lwin32_static\lib -LC:\Python26\libs 
-LC:\Python26\PCbuild -lpython26 -lmsvcr90 -o 
build\lib.win32-2.6\matplotlib\backends\_backend_gdk.pyd

actually ended in:
build\temp.win32-2.6\Release\src\backend_gdk.o:backend_gdk.c:(.text+0x18d): 
undefined reference to `gdk_pixbuf_get_type'
build\temp.win32-2.6\Release\src\backend_gdk.o:backend_gdk.c:(.text+0x19f): 
undefined reference to `g_type_check_instance_cast'
build\temp.win32-2.6\Release\src\backend_gdk.o:backend_gdk.c:(.text+0x1a9): 
undefined reference to `gdk_pixbuf_get_height'
build\temp.win32-2.6\Release\src\backend_gdk.o:backend_gdk.c:(.text+0x1b4): 
undefined reference to `gdk_pixbuf_get_width'
build\temp.win32-2.6\Release\src\backend_gdk.o:backend_gdk.c:(.text+0x1bf): 
undefined reference to `gdk_pixbuf_get_has_alpha'
build\temp.win32-2.6\Release\src\backend_gdk.o:backend_gdk.c:(.text+0x1d2): 
undefined reference to `gdk_pixbuf_get_pixels'
build\temp.win32-2.6\Release\src\backend_gdk.o:backend_gdk.c:(.text+0x232): 
undefined reference to `gdk_pixbuf_get_rowstride'
collect2: ld returned 1 exit status
error: command 'gcc' failed with exit status 1

I really have no clue on this one... I must have forgotten something but 
I don't get what!

Has someone got an idea?

Thanks,

Séb