matplotlib-users Mailing List for matplotlib

>>>>> "zunzun" == zunzun  <zu...@zu...> writes:

    zunzun> This is James Phillips of http://zunzun.com, I'm in the
    zunzun> process of switching the site graphics from DISLIN to
    zunzun> matplotlib.  I had some trouble pinning down why the Y
    zunzun> axis on my plots ws always in the wrong place, and thought
    zunzun> I'd pass along my findings to the mailing list.

    zunzun> If pylab is imported before vtk, everything works fine:

pylab starts the default backend GUI mainloop (if there is one) when
you call "show", and this may be conflicting with vtkpython.  The
default matplotlib GUI for the src distribution is GTKAgg and
vtkpython may start tk???  This is a wild guess.

For something like zunzun, a web app server, I suggest not using pylab
at all, but the matplotlib API, which avoids all the magical things
that pylab does that can just be a pain in an app server.

  http://matplotlib.sf.net/faq.html#OO

See for example

  http://matplotlib.sf.net/examples/agg_oo.py

Let me know if this helps.

If you really want to use pylab, make sure you set Agg or some other
image backend as your default backend in your matplotlibrc file
(http://matplotlib.sf.net/matplotlibrc) or do the following before
importing pylab

    import matplotlib
    matplotlib.use('Agg')
    import pylab

Hope this helps.  zunzun is amazing!!

JDH

John Hunter wrote:
>>>>>>"Steve" == Steve Schmerler <el...@gm...> writes:
> 
> 
>     Steve> Hi I had some issues with marker plots and automatic color
>     Steve> cycling.
>     Steve> gives only black crosses. It would be nice if someone could
>     Steve> please try this out. My guess is that the "o"s and "v"s do
>     Steve> have some kind of black margin and a (colored) filling area
>     Steve> which the "+"s and "x"s dont't have. If that's right then
>     Steve> anyone should see this behavior. Thanx for your help.
> 
> As noted by Christian, the reason for this discrepancy is that '+' is
> not filled and 'o' is.  filled marker colors are governed by their
> 'markeredgecolor' and 'markerfacecolor'.  line styles are goverened by
> 'color'.  Thus it is not entirely trivial to determine how these three
> colors should behave in automatic color cycling.  For example, for
> filled markers, you might want the face color to automatically cycle
> and the edge colors to remain black (or more precisely, their default
> rc value).  For non-filled markers, which do not have a facecolor, you
> might want the edge color to cycle.
> 
> Could you specify how you think this should behave?  My guess is that
> this is sufficiently complicated that different people will want
> different default behaviors.  As you posted earlier, it is pretty
> simple to force the kind of cycling you want in your own code.  But if
> you can spell out a clear policy of how *it should* work, and there
> seems to be a consensus on this, or at least no objection, I can take
> a stab at implementing it in the axes code.  You can take a look
> yourself if you want Axes._process_plot_var_args and
> Axes._process_plot_format.
> 
> JDH
> 
> 
> 

Hi

Well, since mpl changes the markeredgecolor when you request 'r+' and 
the markerfacecolor when you say 'ro' it would be the most natural way 
to keep this behavior when "turning on" automatic cycling by

	rcParams["lines.marker"] = <any_marker>
	rcParams["lines.linestyle"] = "None"

Some would even suggest to let the markeredgecolor == markerfacecolor 
(if there is one) by default for the sake of some "consistency" (i.e. 
remove the markeredgecolor completely, like in gnuplot). I personally 
don't care.

Btw, the issue here applies only to pure marker plots (i.e. nothing like 
'o-') right? What if you want automatic cycling there?


The ideal behavior would be: Say you want to plot 30 data sets with 
lines _and_ markers like '+-'. There would be a call like


setAutoMarkerAndColor()
for i in range(...):
	A = load(<dataset>) # or whatever
	plot(A[:,0], A[:,1]) # no formating at all

which would then behave like


def getFormatString(i):
     # works for len(c) = len(m) and all i >= 0
     c = ['b', 'g', 'r']
     # maybe without '.' and ',' because they are pretty small
     m = ['o', 'x', 'v']
     max = len(c)*len(m) - 1
     if (i > max):
         ii = i - max - 1
     else:
         ii = i
     mind = int(floor(ii / len(m)))
     cind = ii - mind * len(m)
     return c[cind] + m[mind]

		
for i in range(...):
	A = load(<dataset>)
	fs = getFormatString(i)
	plot(A[:,0], A[:,1], fs)


Actually, that's very much it, except that it only works if the numbers 
of colors and markers are the same.


cheers,
steve

-- 
Women are like cell phones. They like to be held and talked to, but push 
the wrong button, and you'll be disconnected.

This is James Phillips of http://zunzun.com, I'm in
the process of switching the site graphics from DISLIN
to matplotlib.  I had some trouble pinning down why the
Y axis on my plots ws always in the wrong place, and
thought I'd pass along my findings to the mailing list.

If pylab is imported before vtk, everything works fine:


import pylab, vtkpython
pylab.ylabel('Frequency\n', multialignment='center', rotation=90)
n, bins, patches = pylab.hist([1,1,1,2,2,3,4,5,5,5,8,8,8,8], 5)
pylab.show()


If however vtk is imported first:


import vtkpython, pylab
pylab.ylabel('Frequency\n', multialignment='center', rotation=90)
n, bins, patches = pylab.hist([1,1,1,2,2,3,4,5,5,5,8,8,8,8], 5)
pylab.show()


then the Y axis label is positioned incorrectly on the
plots.

     James Phillips
     http://zunzun.com

On Wednesday 03 August 2005 11:57 am, kristen kaasbjerg wrote:
> The problem only occurs when you save the file in ps
> or eps format. If instead png is used the error
> doesn't appear, at least not at my machine!

I think I understand this. The problem is that the ghostscript command is 
different on linux and windows. On Linux it is "gs", and on windows its 
something like gs32win. Would you try editing your backend_ps.py file to read 
the following, (but make sure gs32win is really the name of the executable), 
note there are subtle changes throughout the block:

command = 'latex -interaction=nonstopmode "%s"' % texfile 
stdin, stdout, stderr = os.popen3(command)
verbose.report(stdout.read(), 'debug-annoying')
verbose.report(stderr.read(), 'helpful')
command = 'dvips -R -T %fin,%fin -o "%s" "%s"' % (pw, ph, psfile, dvifile)
stdin, stdout, stderr = os.popen3(command)
verbose.report(stdout.read(), 'debug-annoying')
verbose.report(stderr.read(), 'helpful')
os.remove(epsfile)
if ext.startswith('.ep'):
    dpi = rcParams['ps.distiller.res']
    command = 'gs32win -dBATCH -dNOPAUSE -dSAFER -r%d \
                     -sDEVICE=epswrite -dLanguageLevel=2 -dEPSFitPage \
                     -sOutputFile="%s" "%s"'% (dpi, epsfile, psfile)    
    stdin, stdout, stderr = os.popen3(command)
    verbose.report(stdout.read(), 'debug-annoying')
    verbose.report(stderr.read(), 'helpful')
    shutil.move(epsfile, outfile)
else: shutil.move(psfile, outfile)

At some point I'll come up with a way to automatically call the right command 
for ghostscript, but I am really busy at the moment...

-- 

Darren

The problem only occurs when you save the file in ps
or eps format. If instead png is used the error
doesn't appear, at least not at my machine!

Kristen

--- Darren Dale <dd...@co...> wrote:

> On Wednesday 03 August 2005 04:35 am, Alex Rada
> wrote:
> > Hi all,
> >
> > following
>
http://www.scipy.org/wikis/topical_software/UsingTex,
> and in
> > particular the following example:
> >
> > /from matplotlib import rc
> > from matplotlib.numerix import arange, cos, pi
> > from pylab import figure, axes, plot, xlabel,
> ylabel, title, \
> >      grid, savefig, show
> >
> > rc('text', usetex=True)
> > figure(1)
> > ax = axes([0.15, 0.1, 0.8, 0.7])
> > t = arange(0.0, 1.0+0.01, 0.01)
> > s = cos(2*2*pi*t)+2
> > plot(t, s)
> >
> > xlabel(r'\bf{time (s)}')
> > ylabel(r'\it{voltage (mV)}',fontsize=16)
> > title(r"\TeX\ is Number
> >
>
$\displaystyle\sum_{n=1}^\infty\frac{-e^{i\pi}}{2^n}$!",
> >       fontsize=16, color='r')
> > savefig('tex_demo.eps')
> >
> > show()
> > /
> > I obtain the following error:
> >
> > [....]
> > File
>
"/usr/lib/python2.3/site-packages/matplotlib/texmanager.py",
> line
> > 296, in get_rgba
> >     X = readpng(pngfile)
> > RuntimeError: _image_module::readpng could not
> open PNG file
> >
>
/home/alex/.matplotlib/tex.cache/3d522deaf85577451c01974654b36ad3_270.png
> > for reading
> >
> > What does it means!?? I'm using matplotlib 0.83 on
> a linux box with
> > tetex-3.0 and ghostscript-8.51 installed...
> 
> Did you read the "common hangups" section at the end
> of the wiki? You have to 
> make sure that latex, and dvipng can be found on
> your PATH.
> 
> -- 
> 
> Darren
> 
> 
>
-------------------------------------------------------
> SF.Net email is sponsored by: Discover Easy Linux
> Migration Strategies
> from IBM. Find simple to follow Roadmaps,
> straightforward articles,
> informative Webcasts and more! Get everything you
> need to get up to
> speed, fast.
>
http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
>
https://lists.sourceforge.net/lists/listinfo/matplotlib-users
> 



		
____________________________________________________
Start your day with Yahoo! - make it your home page 
http://www.yahoo.com/r/hs 
 

Could you post the exact sequence of commands that you're using and what 
you have interactive mode set to?

I tried this w/ regular python and a version of python that we deliver to 
our customers that supports interactive mode w/ qt and it seems to work 
fine for me.

----------
[non-interactive mode]
import pylab as p
p.figure( 1 )
p.plot( [ 1, 2 ] )
p.show()
[ kill the window ]
p.figure( 1 )
p.plot( [ 3, 4 ] )
p.show()
[ new window pops up fine ]

----------
[interactive mode]
import pylab as p
p.figure( 1 )
[ window pops up ]
p.plot( [ 1, 2 ] )
[ kill the window ]
p.figure( 1 )
[ window pops up ]


At 12:24 AM 8/3/2005, Eugen Wintersberger wrote:
>Hi there.
>I have a problem with the QtAgg backend. It is not that serious but a
>bit nasty. After opening a figure with
>
> > figure(1)
>
>I have no problems to plot there. But after closing the window it is
>no longer possible to open a figure with ID 1 again. Ipython gives no
>error message, everything looks ok. But if one tries to plot something
>in this window nothing appears on the screen.
>I use Ipython 0.6.15 with matplotlib 0.8 on a Debian 3.1 system.
>Does anyone of you have an idea what I'm doing wrong or is this a known
>bug/error.
>
>cu
>  Eugen
>
>
>-------------------------------------------------------
>SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
>from IBM. Find simple to follow Roadmaps, straightforward articles,
>informative Webcasts and more! Get everything you need to get up to
>speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click
>_______________________________________________
>Matplotlib-users mailing list
>Mat...@li...
>https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Ted Drain     Jet Propulsion Laboratory   ted...@jp...    

On Wednesday 03 August 2005 04:35 am, Alex Rada wrote:
> Hi all,
>
> following http://www.scipy.org/wikis/topical_software/UsingTex, and in
> particular the following example:
>
> /from matplotlib import rc
> from matplotlib.numerix import arange, cos, pi
> from pylab import figure, axes, plot, xlabel, ylabel, title, \
>      grid, savefig, show
>
> rc('text', usetex=True)
> figure(1)
> ax = axes([0.15, 0.1, 0.8, 0.7])
> t = arange(0.0, 1.0+0.01, 0.01)
> s = cos(2*2*pi*t)+2
> plot(t, s)
>
> xlabel(r'\bf{time (s)}')
> ylabel(r'\it{voltage (mV)}',fontsize=16)
> title(r"\TeX\ is Number
> $\displaystyle\sum_{n=1}^\infty\frac{-e^{i\pi}}{2^n}$!",
>       fontsize=16, color='r')
> savefig('tex_demo.eps')
>
> show()
> /
> I obtain the following error:
>
> [....]
> File "/usr/lib/python2.3/site-packages/matplotlib/texmanager.py", line
> 296, in get_rgba
>     X = readpng(pngfile)
> RuntimeError: _image_module::readpng could not open PNG file
> /home/alex/.matplotlib/tex.cache/3d522deaf85577451c01974654b36ad3_270.png
> for reading
>
> What does it means!?? I'm using matplotlib 0.83 on a linux box with
> tetex-3.0 and ghostscript-8.51 installed...

Did you read the "common hangups" section at the end of the wiki? You have to 
make sure that latex, and dvipng can be found on your PATH.

-- 

Darren

>>>>> "Steve" == Steve Schmerler <el...@gm...> writes:

    Steve> Hi I had some issues with marker plots and automatic color
    Steve> cycling.
    Steve> gives only black crosses. It would be nice if someone could
    Steve> please try this out. My guess is that the "o"s and "v"s do
    Steve> have some kind of black margin and a (colored) filling area
    Steve> which the "+"s and "x"s dont't have. If that's right then
    Steve> anyone should see this behavior. Thanx for your help.

As noted by Christian, the reason for this discrepancy is that '+' is
not filled and 'o' is.  filled marker colors are governed by their
'markeredgecolor' and 'markerfacecolor'.  line styles are goverened by
'color'.  Thus it is not entirely trivial to determine how these three
colors should behave in automatic color cycling.  For example, for
filled markers, you might want the face color to automatically cycle
and the edge colors to remain black (or more precisely, their default
rc value).  For non-filled markers, which do not have a facecolor, you
might want the edge color to cycle.

Could you specify how you think this should behave?  My guess is that
this is sufficiently complicated that different people will want
different default behaviors.  As you posted earlier, it is pretty
simple to force the kind of cycling you want in your own code.  But if
you can spell out a clear policy of how *it should* work, and there
seems to be a consensus on this, or at least no objection, I can take
a stab at implementing it in the axes code.  You can take a look
yourself if you want Axes._process_plot_var_args and
Axes._process_plot_format.

JDH

>>>>> "Alex" == Alex Rada <ale...@ya...> writes:

    Alex> What does it means!?? I'm using matplotlib 0.83 on a linux
    Alex> box with tetex-3.0 and ghostscript-8.51 installed...

What version of dvipng do you have installed?  I recommend 1.6 or
later.  We need to make the texmanager smarter about raising errors
when it doesn't find the software it needs.

JDH

Hi all,

following http://www.scipy.org/wikis/topical_software/UsingTex, and in 
particular the following example:

/from matplotlib import rc
from matplotlib.numerix import arange, cos, pi
from pylab import figure, axes, plot, xlabel, ylabel, title, \
     grid, savefig, show

rc('text', usetex=True)
figure(1)
ax = axes([0.15, 0.1, 0.8, 0.7])
t = arange(0.0, 1.0+0.01, 0.01)
s = cos(2*2*pi*t)+2
plot(t, s)

xlabel(r'\bf{time (s)}')
ylabel(r'\it{voltage (mV)}',fontsize=16)
title(r"\TeX\ is Number 
$\displaystyle\sum_{n=1}^\infty\frac{-e^{i\pi}}{2^n}$!",
      fontsize=16, color='r')
savefig('tex_demo.eps')

show()
/
I obtain the following error:

[....]
File "/usr/lib/python2.3/site-packages/matplotlib/texmanager.py", line 
296, in get_rgba
    X = readpng(pngfile)
RuntimeError: _image_module::readpng could not open PNG file 
/home/alex/.matplotlib/tex.cache/3d522deaf85577451c01974654b36ad3_270.png 
for reading

What does it means!?? I'm using matplotlib 0.83 on a linux box with 
tetex-3.0 and ghostscript-8.51 installed...

Thanks for the Help,

Alex.

Steve Schmerler wrote:
> this out. My guess is that the "o"s and "v"s do have some kind of black 
> margin and a (colored) filling area which the "+"s and "x"s dont't have. 
> If that's right then anyone should see this behavior. Thanx for your help.

You're right. The marker's margin color is set by

rcParams['lines.markeredgecolor']=  'green'

and you even can have markers without edge by setting

rcParams['lines.markeredgewidth']=  0.0

but that doesn't work for '+' markers as they appear to consist only of the 
edge. As a result, they won't show up.

Christian

Hi there. 
I have a problem with the QtAgg backend. It is not that serious but a
bit nasty. After opening a figure with

> figure(1)

I have no problems to plot there. But after closing the window it is 
no longer possible to open a figure with ID 1 again. Ipython gives no
error message, everything looks ok. But if one tries to plot something
in this window nothing appears on the screen. 
I use Ipython 0.6.15 with matplotlib 0.8 on a Debian 3.1 system. 
Does anyone of you have an idea what I'm doing wrong or is this a known 
bug/error. 

cu 
 Eugen

Hi

I had some issues with marker plots and automatic color cycling.

John suggested

    	rcParams["lines.marker"] = "o"
    	rcParams["lines.linestyle"] = "None"
    	for i in range(4):
		plot(rand(5), rand(5))

which produces nice colorful dots :) "v" also works.

But changing the marker to '+' (or 'x')

	rcParams["lines.marker"] = "+"

gives only black crosses. It would be nice if someone could please try 
this out. My guess is that the "o"s and "v"s do have some kind of black 
margin and a (colored) filling area which the "+"s and "x"s dont't have. 
If that's right then anyone should see this behavior. Thanx for your help.

cheers,
steve