matplotlib-users Mailing List for matplotlib

fred a écrit :

I get anoter issue now.

How can I know if the displayed image has been zoomed or not
(I mean not on testing xlim & ylim) ?

TIA

-- 
http://scipy.org/FredericPetit

On 6/18/07, Itai Arad <it...@gm...> wrote:

> I am trying to produce a very simple EPS figure, but the EPS file is
> different than what I see on the screen in the GTKAgg window, or when
> I save the figure as a PNG file. It looks like the coordinate system
> in the PS case is different, and so the objects in the figure are
> placed in different locations.

There appears to be a problem with Agg -- not sure about PS yet.  We
recently added support for "true ellipses" using arcs rather than
polygons, and it appears we have messed something up.  Charlie, take a
look at this case -- depending on the y-zoom the x extent of the
ellipse is different:

from pylab import figure, show
from matplotlib.patches import Ellipse

xy = 0.38,0.52

props = dict(xy=xy, width=1e-1, height=3e-1,
             facecolor='red',
             angle=30.0, linewidth=2, fill=True, alpha=0.5)
e1 = Ellipse(**props)
e2 = Ellipse(**props)

fig = figure()
ax1 = fig.add_subplot(211)
ax1.add_artist(e1)
ax1.set_xlim(-.1, .7)
ax1.set_ylim(0.35, .7)

ax2 = fig.add_subplot(212, sharex=ax1)
ax2.add_artist(e2)
ax2.set_ylim(0.4, .6)

show()


Any ideas?

JDH

On 6/17/07, Duncan Child <dun...@gm...> wrote:
> We would like to provide a Matplotlib-like plotting function in a Java
> application we are building.  We can't require our users to install
> native code so we can't use Matplotlib.
>
> We have a simple implementation that would be sufficient for v1.0 of
> our application that uses jython and jfreechart.
>
> To re-implement the whole of Matplotlib would require a collaborative
> effort and would take a while. You have to start somewhere though so I
> thought I should post here to ask for advice.
>
> Does anyone have any comments or suggestions?

I would think Java would have some pretty sophisticated plotting
libraries and if you are committed to Java I would try and find/extend
a native Java library.  Porting mpl is probably not an option -- we
use a fair amount of C++ with python bindings to do rendering and
since you can't install client side code this approach probably won't
work.  One could use the SVG output combined with a Java SVG
renderer....

Alternatively, if you have a client/server mode, you could generate
your figures with matplotlib server side and send them to the client.

JDH

fred a écrit :
> Sorry for the noise.

Thanks anyway, Jouni, I fixed my issue thanks to you ;-)

You showed me that the get_?lim methods were right,
so I followed my feeling further.

Cheers,

-- 
http://scipy.org/FredericPetit

fred a écrit :
> Maybe I'm doing something wrong ?
Yes :-)

You only need to set extent to the const xmin/xmax & ymin/ymax
and not those you get with get_xlim & get_ylim.

Sorry for the noise.

Cheers,

-- 
http://scipy.org/FredericPetit

Jouni K. Seppänen a écrit :
> Try:
>
> imshow(rand(50,50))
> # zoom with mouse
> xlim=getp(gca(), 'xlim'); ylim = getp(gca(), 'ylim')
> imshow(rand(50,50))
> setp(gca(), 'xlim', xlim); setp(gca(), 'ylim', ylim)
>   
Well..., I forgot to mention something, sorry ;-)
I already apply this trick: I use get_{x,y}lim and set_{x,y}lim.
Which works almost fine.

In fact, I use a grid, which is defined by xmin, xmax, dx & ymin, ymax, dy.
I don't want do display on the axes the number of points, but the real world
axis dimension.

If xmin=ymin=0 and xmax=ymax=500 with dx=dy=1,
the trick does work fine.

Now, if I set xmax=ymax=1000 but dx=dy=2 (so I have the same number of 
points)
I have to use imshow extent option, because of the use of _xlim & _ylim.

Please see http://fredantispam.free.fr/a.png without extent option,
http://fredantispam.free.fr/b.png with extent option.

But in this case, after zoomed,

see http://fredantispam.free.fr/c.png

the trick does not work anymore: x & y limits are right,
but the whole data is displayed by imshow.

see http://fredantispam.free.fr/d.png.

Maybe I'm doing something wrong ?

TIA.


Cheers,

-- 
http://scipy.org/FredericPetit

fred <fr...@gm...> writes:

> imshow(rand(50,50))
> then zoom with the mouse and redo
> imshow(rand(50,50))
> the zoom factor is not kept, ie xrange/yrange are still (0,50).
>
> How can I keep the zoom factor ?

Try:

imshow(rand(50,50))
# zoom with mouse
xlim=getp(gca(), 'xlim'); ylim = getp(gca(), 'ylim')
imshow(rand(50,50))
setp(gca(), 'xlim', xlim); setp(gca(), 'ylim', ylim)

-- 
Jouni K. Seppänen
http://www.iki.fi/jks

Hi,

I want to keep the image zoomed to the same factor for all my images.

If I try

ipython -pylab
imshow(rand(50,50))

then zoom with the mouse and redo

imshow(rand(50,50))

the zoom factor is not kept, ie xrange/yrange are still (0,50).

How can I keep the zoom factor ?

TIA.


Cheers,

-- 
http://scipy.org/FredericPetit

Hello everybody and a bit thanks to the matplotlib team. Its great stuff.

I would like to place the minor ticks for days, lower than the major ticks
for hours, so that they are visible. Right now they overlap each other. I
would like to achieve a result like

11:00    12:00    01:00
                |
           Monday

The vertical line is not important, I just need days to be lower than hours.

The pythonic API could certainly use more documentation. Pylab is good but
only for working in the same fashion as matlab.

Hi,

I am using the matplotlib library that comes with Ubuntu 7.04. This is
version 0.87.7 with ESP ghostscript 8.15.4.

I am trying to produce a very simple EPS figure, but the EPS file is
different than what I see on the screen in the GTKAgg window, or when
I save the figure as a PNG file. It looks like the coordinate system
in the PS case is different, and so the objects in the figure are
placed in different locations.

The code that I am using is:

import matplotlib
from pylab import *
from matplotlib.patches import Ellipse
from matplotlib.text import Text


xy1=(0.38,0.52)
xy2=(0.5,0.5)

e1 = Ellipse(xy=xy1, width=1e-1, height=3e-1, \
     angle=30.0, linewidth=2, fill=False)

e2 = Ellipse(xy=xy2, width=1e-1, height=3e-1, \
     angle=-30, linewidth=2, fill=False)


ax=gca()
ax.add_artist(e1)
ax.add_artist(e2)

axis('off')

savefig('ellipse.eps')
savefig('ellipse.png')

If you run it and view the two files ellipse.eps and ellipse.png -
you'll notice they look different. In the PNG file the ellipses
intersect, but in the EPS file they don't.

What's wrong here?

Thanks,
Itai.

Hello everybody,

I am going through a real epic trying to build matplotlib on cygwin. Here is 
my environment:
freetype-2.3.2 matplotlib-0.90.0 numpy-1.0.2 python-2.5-1 rebase 2.4.3-1 
gcc-3.4.4-3 cygwin_NT-5.1
Here is what I tried to do:

$ python setup.py build
running build
running build_py
running build_ext
building 'matplotlib._isnan' extension
gcc -fno-strict-aliasing -DNDEBUG -g -O3 -Wall -Wstrict-prototypes -I/usr/includ
e -I. -I/usr/include/python2.5 -c src/_isnan.c -o 
build/temp.cygwin-1.5.24-i686-
2.5/src/_isnan.o
gcc: fork: Resource temporarily unavailable
error: command 'gcc' failed with exit status 1

Previous postings on various groups pointed to this being a manifestation of 
the dreaded rebase problem. I ran rebaseall from the ash shell and gave it 
another shot:

$ python setup.py build
running build
running build_py
running build_ext
building 'matplotlib._isnan' extension
gcc -shared -Wl,--enable-auto-image-base 
build/temp.cygwin-1.5.24-i686-2.5/src/_
isnan.o -L/usr/local/lib -L/usr/lib -L/usr/lib/python2.5/config -lpython2.5  
-o b
uild/lib.cygwin-1.5.24-i686-2.5/matplotlib/_isnan.dll
gcc: fork: Resource temporarily unavailable
error: command 'gcc' failed with exit status 1

Interestingly when I run the gcc command string from the dollar prompt, the 
.o file builds without a problem. Any Ideas?
Regards,

//Pete Harris