matplotlib-users Mailing List for matplotlib

All,
Where do bugs and document corrections/suggestions get sent?
David

Hi -- According to what I've read in the docs, the following code should
render a graph with the Y-Axis NOT in scientific-mode. However this isn't
the case. What am I failing to understand?
import matplotlib.pylab as pylab
pylab.figure()
yfmt = pylab.gca().yaxis.get_major_formatter()
yfmt.set_scientific(False)
pylab.draw()
pylab.plot([110100,110102,110100,110103])
pylab.show()
...I'm running on Ubuntu Intrepid. matplotlib.__version__ reports 0.98.3 
Any thoughts?
Thanks!
-Dave
-- 
View this message in context: http://old.nabble.com/set_scientific%28False%29-doesn%27t-do-anything.-tp27631188p27631188.html
Sent from the matplotlib - users mailing list archive at Nabble.com.

On Wed, Feb 17, 2010 at 9:28 AM, John Hunter <jd...@gm...> wrote:
> On Wed, Feb 17, 2010 at 9:06 AM, Matthias Michler
> <Mat...@gm...> wrote:
>
> > thanks a lot for taking the time to go through this patch. I tried to
> > incorporate you remarks and attached a new patch.
> > The most difficult task is about the
> > documentation 'mpl/doc/users/navigation_toolbar.rst', because I'm not
> sure
> > what is needed and I not familiar with the used programming language.
>
> Hey mattias,
>
>
Thanks John adding this feature into the code-base. Also Matthias thanks for
your additions and keeping the subject hot :)
> I just committed this to svn HEAD. I tested the L key on WXAgg, TkAgg
> and Qt4Agg and it is working on all but WXAgg. Some wx expert may
> want to look and make sure that the shifted keys are getting properly
> recognized on wx.
>
L is working for me both for Qt4Agg and WXAgg.
For some reason I can't make "f" key toggling the fullscreen. It does
nothing -no error messages and no life sign. Any confirmation?
Lastly, I know "s" behaviour is also being discussed in another thread, and
I have seen with WXAgg it raises a not implemented error. I think "s" could
simply bring the save dialogue instead of showing those messages. What do
you think?
> As for the docs in mpl/doc/users/navigation_toolbar.rst, the language
> is sphinx and restructured text
>
> http://sphinx.pocoo.org/rest.html
>
> It looks like you got it right by following the pattern. This is the
> document used to generate the mpl html and pdf documentation, eg
>
> http://matplotlib.sourceforge.net/users/navigation_toolbar.html
>
> Thanks again!
> JDH
>
I see that you also updated the documentation. I was waiting for someone to
take a look at the code before making changes there. Thanks again.
>
>
> ------------------------------------------------------------------------------
> SOLARIS 10 is the OS for Data Centers - provides features such as DTrace,
> Predictive Self Healing and Award Winning ZFS. Get Solaris 10 NOW
> http://p.sf.net/sfu/solaris-dev2dev
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
-- 
Gökhan

Jan Strube wrote:
> Dear matplotters,
> encouraged from the excellent response times to my last problem, I am 
> trying to explore more features of matplotlib.
> 
> My current problem is with hexbin.
> I have been using numpy.histogram2d and imshow so far for 2d histograms, 
> but I must admit that hexbin looks quite pretty.
> 
> In order to emphasize small differences over a large scale, I am using 
> hexbin with the bins='log' option.
> The result is very nice, but when I add a pyplot.colorbar(), I would 
> like to have the original number of entries on the ticks, not their log10.
> 
> Ideally, I'd change the yaxis of the colorbar to a logscale, and use 
> numpy.power(10, yaxis_values) on it.
> I think colorbar().ax.set_yscale('log') should do what I need for the 
> first part, but how do I get the original number of entries back, rather 
> than their log10.
> I could cheat and simply replace each ticklabel with 10^ticklabel, but I 
> don't even know how to do that.
Offhand, I suspect this would be the best solution at present, if the 
only problem now is that you want the ticklabels written differently. 
Try using the "format" kwarg of the colorbar, supplying either a format 
string or a formatter.
Something like this (untested):
from matplotlib.ticker import LogFormatter
class LogFormatterHB(LogFormatter):
 def __call__(self, v, pos=None):
 vv = self._base ** v
 return LogFormatter.__call__(self, vv, pos)
...
cbar = colorbar( ... , format=LogFormatterHB())
Eric

> On Tue, Feb 16, 2010 at 9:53 PM, John Hunter <jd...@gm...> wrote:
> > I think it means that Darren, who did most of the heavy lifting for
> > these features, was getting tired of the endless line of additional
> > things TeX users wanted to do and the difficulties supporting these
> > across all platforms (eg MikTex on Windows) and drew a line in the
> > sand saying "I'll add this stuff but if you report problems I may not
> > stay up all night trying to fix them".
> From: Darren Dale [mailto:dsd...@gm...]
> Sent: Wednesday, February 17, 2010 4:55 AM
> That is a fair explanation. Adding arbitrary lines to the latex
> preamble adds an incredible amount of flexibility if you are a user,
> and it provides plenty of rope with which to hang yourself if you are
> a developer (answering questions/problems related to people writing
> bad python code, misconfigured latex installations, and bad latex
> code!). At the time, we had finally shaped the latex backend into
> something that worked across platforms (and different versions of
> latex installations), plus I was also finishing up grad school and
> starting a new job, so I had reservations about even adding this
> feature to the library. But people really wanted it so we compromised.
Compromised or not, it's a great feature and I'm *really* happy it made it was included! Thanks for all of the effort.
-Paul H.

Dear matplotters,
encouraged from the excellent response times to my last problem, I am trying
to explore more features of matplotlib.
My current problem is with hexbin.
I have been using numpy.histogram2d and imshow so far for 2d histograms, but
I must admit that hexbin looks quite pretty.
In order to emphasize small differences over a large scale, I am using
hexbin with the bins='log' option.
The result is very nice, but when I add a pyplot.colorbar(), I would like to
have the original number of entries on the ticks, not their log10.
Ideally, I'd change the yaxis of the colorbar to a logscale, and use
numpy.power(10, yaxis_values) on it.
I think colorbar().ax.set_yscale('log') should do what I need for the first
part, but how do I get the original number of entries back, rather than
their log10.
I could cheat and simply replace each ticklabel with 10^ticklabel, but I
don't even know how to do that.
Thanks for any pointers,
Cheers,
 Jan

On Tue, Feb 16, 2010 at 2:03 PM, Jae-Joon Lee <lee...@gm...> wrote:
> If what you want is to have more padding for the major tick labels, I
> recommend you to use
>
> rcParams['xtick.major.pad'] = 20
>
> If you don't like to change the global setting, you may set the
> ticklabel padding for an specific axis. Try
>
> for tck in ax.xaxis.get_major_ticks():
>  tck.set_pad(20)
>  tck.label1 = tck._get_text1()
>
> Regards,
>
> -JJ
>
Wow. This is really useful and perhaps, should be placed in a wiki
somewhere. Is this a common enough task to warrant a method?
 ax.xaxis.repad_ticks(20)
Being unfamiliar with the Tick class, I wouldn't have expected that it
would have been necessary to reassign label1 (after setting the pad
size).

On Wed, Feb 17, 2010 at 9:06 AM, Matthias Michler
<Mat...@gm...> wrote:
> thanks a lot for taking the time to go through this patch. I tried to
> incorporate you remarks and attached a new patch.
> The most difficult task is about the
> documentation 'mpl/doc/users/navigation_toolbar.rst', because I'm not sure
> what is needed and I not familiar with the used programming language.
Hey mattias,
I just committed this to svn HEAD. I tested the L key on WXAgg, TkAgg
and Qt4Agg and it is working on all but WXAgg. Some wx expert may
want to look and make sure that the shifted keys are getting properly
recognized on wx.
As for the docs in mpl/doc/users/navigation_toolbar.rst, the language
is sphinx and restructured text
 http://sphinx.pocoo.org/rest.html
It looks like you got it right by following the pattern. This is the
document used to generate the mpl html and pdf documentation, eg
 http://matplotlib.sourceforge.net/users/navigation_toolbar.html
Thanks again!
JDH

Hi,
I have the following problem: I do a wavelet decomposition on a
signal, sampled at 1 kHz. When I take 1024 datapoints I get
coefficients cD1, cD2, cD3, ... , cD10, cA10, where the 512 cD1
coefficients correspond to the frequencies 250-500 Hz, the next 256
cD2 coefficients are from the 125-250Hz band and so on.
I would now like to do a color coded plot of these coefficients,
x-axis is time, y-axis is frequency, with colors for each coefficient
in it's own area, i.e. the first cD1 must be from 0 to 2 ms and from
250 to 500 Hz, the first cD2 is from 0 to 4 ms, filling 125 to 250 Hz
and so on...
How can I do this in python? I could use imshow of course, but how is
the easiest way to prepare one 1024 x 250? array that hold the
coefficients just like this? Thank you for your help.
Thorsten

Hi,
I am currently learning metplotlib to make beautiful figures for a LaTeX
thesis.
I will need to have graphs with a break in the x-axis, comparable to this
one
http://www.originlab.com/www/Products/images/Date_Axis_plot_with_XY_break_500px.png
As I found out, there is no native support for this. So I fiddled around a
lot and came up with the following solutions.
Please take this as a starting point if you happen to have the same
intensions with matplotlib.
Also I am happy for any input from more experienced users.
The code:
#@@@@@@@@@@@@@@@@@@@@@@@@@ begin
from matplotlib import rc
import numpy as np
import matplotlib.pyplot as plt
y = np.loadtxt('532_0.1_10.txt')
 
fig_width_pt = 400 # Get this from LaTeX using \showthe\columnwidth
inches_per_pt = 1.0/72.27 # Convert pt to inches
golden_mean = (np.sqrt(5)-1.0)/2.0 # Aesthetic ratio
fig_width = fig_width_pt*inches_per_pt # width in inches
fig_height =fig_width*golden_mean # height in inches
fig_size = [fig_width,fig_height]
rc('font',**{'family':'serif','serif':['Computer Modern Roman']})
params = {'backend': 'ps',
 'axes.labelsize': 10,
 'text.fontsize': 10,
 'legend.fontsize': 8,
 'xtick.labelsize': 8,
 'ytick.labelsize': 8,
 'text.usetex': True,
 'figure.figsize': fig_size,
 'axes.unicode_minus': True}
matplotlib.rcParams.update(params)
# setup figure with specified size
fig = plt.figure(figsize=fig_size)
width = 0.44 # width of the two subplots
spacing = 0.025/2 # spacing between subplots
if 1: # left subplot
 ax1 = plt.axes([0.08,0.15,width,0.8])
 for loc, spine in ax1.spines.iteritems():
 if loc in ['right']:
 spine.set_color('none')
 else:
 spine.set_position(('outward',0))
 plt.plot(zip(*y)[0], zip(*y)[1])
 ax1.set_xlim(740,820)
 plt.xticks([750,800])
if 1: #right subplot with shared y-axis
 ax2 = plt.axes([0.08+width+spacing,0.15,width,0.8], sharey=ax1)
 for loc, spine in ax2.spines.iteritems():
 if loc in ['left', 'right']:
 spine.set_color('none')
 else:
 spine.set_position(('outward',0))
 plt.plot(zip(*y)[0], zip(*y)[1])
 ax2.set_xlim(930, 1010)
 ax2.set_xticks([950,1000])
 ax2.set_yticks([])
 ax2.set_ylim(0, 11000) # after second plot
if 1: # axes labels and break lines
 ax3 = plt.axes([0.08,0.15,2*width+0.025/2,0.8]) # change position to
adjust labels
 for loc, spine in ax3.spines.iteritems():
 if loc in ['left', 'bottom', 'top']:
 spine.set_color('none')
 else:
 spine.set_position(('outward',0))
 ax3.set_yticks([])
 ax3.set_xticks([])
 ax3.set_xlabel(r'Raman shift (cm-1)', labelpad = 25) 
 # ax3.xaxis.set_label_coords(0.5,-0.06) #alternative to labelpad
 ax3.set_ylabel('Intensity (arb. unit)', labelpad = 10)
 ax3.patch.set_facecolor('None') # set background transparent. Must for
.eps!
 shift = spacing/2
 # the four small lines
 line =
matplotlib.lines.Line2D([0.5-0.005-shift,0.5+0.005-shift],[-0.013,0.013],
linewidth=1, color='black', clip_on=False)
 ax3.add_line(line) 
 line =
matplotlib.lines.Line2D([0.5-0.005+shift,0.5+0.005+shift],[-0.013,0.013],
linewidth=1, color='black', clip_on=False)
 ax3.add_line(line) 
 line =
matplotlib.lines.Line2D([0.5-0.005-shift,0.5+0.005-shift],[1+-0.013,1+0.013],
linewidth=1, color='black', clip_on=False)
 ax3.add_line(line) 
 line =
matplotlib.lines.Line2D([0.5-0.005+shift,0.5+0.005+shift],[1+-0.013,1+0.013],
linewidth=1, color='black', clip_on=False)
 ax3.add_line(line) 
plt.show()
#plt.savefig('breaktest.png')
#@@@@@@@@@@@@@@@@@@@@@@@@@@@@ end
And here some explanations:
I decided not to use subplots, but axes instead. I feel this gives more
control, even it involves arithmetics.
So there is one axes (ax1) for the left figure (no right axis)
then another axes (ax2) for the right figure (no left and right axis, sharey
with left figure)
And in the end a transparent axes (ax3) around both (only right axis), which
is used to plot the axis labels and the four small breaklines.
Looks better than Origin ;)
EPS output: http://old.nabble.com/file/p27624781/image.eps image.eps 
PNG output: http://old.nabble.com/file/p27624781/matplotlib_axis_break.png
matplotlib_axis_break.png 
Raw Data: http://old.nabble.com/file/p27624781/532_0.1_10.txt
532_0.1_10.txt 
-- 
View this message in context: http://old.nabble.com/Break-in-X-Axis-tp27624781p27624781.html
Sent from the matplotlib - users mailing list archive at Nabble.com.

On Wed, Feb 17, 2010 at 8:08 AM, Matthias Michler
<Mat...@gm...> wrote:
> Hi list, Hi Gökhan,
>
> I once more would like to say that I like the 2 new features introduced by
> Gökhan (key 'k' for xscaling and the generalized handling of the key-mapping,
> which allows the user to choose its prefered key for a certain task) and I'd
> like to see this in matplotlib.
>
> I attached a patch (against todays svn) of my final version, which is a
> slightly modified version of Gökhans patch.
>
> If anybody could test it and comments on this I'd be happy. Otherwise I will
> place the patch on the patch tracker in (let's say) a week.
>
> Thanks in advance for any help in order to bring this into matplotlib.
Hi -- these look like useful changes. In most of matplotlib rc we
have a namespace for the rc keys, so I prefer something like
 keymap.fullscreen : f # toggling
 keymap.home : h, r, home # home or reset mnemonic
Also, could you include in your patch a diff against
 mpl/doc/users/navigation_toolbar.rst
documenting the changes.
Also, I'm not sure you need
+ # to get a common standard we move all keys for each action into a list
+ for key in [fullscreen_keys, home_keys, back_keys, forward_keys,
+ pan_keys, zoom_keys, save_keys, grid_keys,
+ toggle_xscale_keys, toggle_yscale_keys, all]:
+ if type(key) is not list:
+ key = list(key)
+
If you setup your validate func properly in rcsetup, you can guarantee
that each of these keymap.* entries will be a list of strings, even
those changes interactively from the prompt. So I suggest changing
each of these key entries to have a validate_stringlist validator and
then you won't need this check.
Thanks,
JDH

On Wed, Feb 17, 2010 at 7:42 AM, Wolfgang Kerzendorf
<wke...@go...> wrote:
> Hello,
>
> I would like to build a bit of an interactive fitter with matplotlib and ipython (in pylab environment). I would like to have a a function, which takes x and y as input, then plots these and fits a line to it (just numpy polyfit). if I click a point it will be removed from the fit pool and the line will be refitted (optionally after pressing 'f'). when I'm done I can press 'q' or close the window and the function will come to an end and spit out the fitting parameter.
> I tried this a year or two ago and I had terrible problems with getting stopping the event loop and waiting for the interactive part to finish and then finish the function. I'm running os 10.6 and use the wx backend (or mac os x, if that's easier). Can you point me to an example or give me a crude overview of how to do that in the right way. Is that understandable?
Take a look at
 http://matplotlib.sourceforge.net/examples/event_handling/poly_editor.html
You can use the 'i' and 'd' keys to insert and delete vertexes, can
click and drag them to move them.
See also the event handling tutorial at
 http://matplotlib.sourceforge.net/users/event_handling.html
JDH

On Wed, Feb 17, 2010 at 7:41 AM, Michiel de Hoon <mjl...@ya...> wrote:
> An inconsistency in the definition of save_figure between different backends is causing this problem.
>
> The GTK backends use
>  def save_figure(self, button):
>
> but the tkagg, qt, qt4, and macosx backends use
>  def save_figure(self):
>
> so without the second argument. The line that is causing the error is
>
>      self.canvas.toolbar.save_figure(self.canvas.toolbar)
>
> in backend_bases.py. This assumes that the save_figure method is defined as in the GTK backends.
>
> As far as I can tell, the GTK backend has the second argument because that is what pygtk passes when save_figure is called as a callback. The second argument is not actually used inside the method.
>
> So I would suggest the following:
>
> In backend_bases.py, change the offending line to
>
>      self.canvas.toolbar.save_figure()
>
> and the backend_gtk, change the definition of the save_figure method to
>
>  def save_figure(self, button-None):
>
> Any objections, anybody?
The base class signature is
 def save_figure(self, *args):
 'save the current figure'
 raise NotImplementedError
But I think the problem is the line
 self.canvas.toolbar.save_figure(self.canvas.toolbar)
it shouldn't be passing the toolbar in, but should just read
 self.canvas.toolbar.save_figure()
We could make both changes -- make sure all the signatures of the
derived classes comply with
 def save_figure(self, *args):
and remove the self.canvas.toolbar argument from the save_figure call.
Michiel, do you want to take the lead on this?
JDH
No?

Hello,
I would like to build a bit of an interactive fitter with matplotlib and ipython (in pylab environment). I would like to have a a function, which takes x and y as input, then plots these and fits a line to it (just numpy polyfit). if I click a point it will be removed from the fit pool and the line will be refitted (optionally after pressing 'f'). when I'm done I can press 'q' or close the window and the function will come to an end and spit out the fitting parameter. 
I tried this a year or two ago and I had terrible problems with getting stopping the event loop and waiting for the interactive part to finish and then finish the function. I'm running os 10.6 and use the wx backend (or mac os x, if that's easier). Can you point me to an example or give me a crude overview of how to do that in the right way. Is that understandable?
Cheers
 Wolfgang

An inconsistency in the definition of save_figure between different backends is causing this problem.
The GTK backends use
 def save_figure(self, button):
but the tkagg, qt, qt4, and macosx backends use
 def save_figure(self):
so without the second argument. The line that is causing the error is
 self.canvas.toolbar.save_figure(self.canvas.toolbar)
in backend_bases.py. This assumes that the save_figure method is defined as in the GTK backends.
As far as I can tell, the GTK backend has the second argument because that is what pygtk passes when save_figure is called as a callback. The second argument is not actually used inside the method.
So I would suggest the following:
In backend_bases.py, change the offending line to
 self.canvas.toolbar.save_figure()
and the backend_gtk, change the definition of the save_figure method to
 def save_figure(self, button-None):
Any objections, anybody?
--Michiel
--- On Sun, 2/14/10, David Arnold <dwa...@su...> wrote:
> From: David Arnold <dwa...@su...>
> Subject: [Matplotlib-users] Easy come easy go
> To: mat...@li...
> Date: Sunday, February 14, 2010, 11:36 PM
> All,
> 
> This example: http://matplotlib.sourceforge.net/examples/event_handling/keypress_demo.html
> 
> Raises this exception o my Macbook when the key 's' is
> pressed:
> 
> The debugged program raised the exception unhandled
> TypeError
> "save_figure() takes exactly 1 argument (2 given)"
> File:
> /Library/Frameworks/Python.framework/Versions/6.0.0/lib/python2.6/site-packages/matplotlib/backend_bases.py,
> Line: 1703
> 
> David
> ------------------------------------------------------------------------------
> SOLARIS 10 is the OS for Data Centers - provides features
> such as DTrace,
> Predictive Self Healing and Award Winning ZFS. Get Solaris
> 10 NOW
> http://p.sf.net/sfu/solaris-dev2dev
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
> 
 

That actually did it. I thought I had tried this several times and kept
getting errors. It would not take the color names like red, blue etc. But
when I changed it as suggested it worked! 
Thank you! 
Jae-Joon Lee wrote:
> 
> On Tue, Feb 16, 2010 at 2:15 PM, duckman <tdu...@cr...>
> wrote:
>> I am trying to make the table at the bottom and the lines the same color.
>> I
>> took the code from one of the online examples and modified it to do most
>> of
>> what I want but cannot get the color in the table working properly! Can
>> someone please help? Thank you
>>
> 
> Do you want the background of the row labels to match your line colors?
> The answer seems to be so trivial and I'm not sure if this is what you
> want.
> 
> the_table = table(cellText=cellText,
> rowLabels=rowLabels, rowColours=linecolor,
> colLabels=colLabels,
> loc='bottom')
> 
> -JJ
> 
> ------------------------------------------------------------------------------
> SOLARIS 10 is the OS for Data Centers - provides features such as DTrace,
> Predictive Self Healing and Award Winning ZFS. Get Solaris 10 NOW
> http://p.sf.net/sfu/solaris-dev2dev
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
> 
> 
-- 
View this message in context: http://old.nabble.com/Color-in-table-help-please%21--I-have-already-searched-previous-posts-tp27613553p27623776.html
Sent from the matplotlib - users mailing list archive at Nabble.com.

On Tue, Feb 16, 2010 at 9:53 PM, John Hunter <jd...@gm...> wrote:
> On Tue, Feb 16, 2010 at 3:57 PM, Ernest Adrogué <ead...@gm...> wrote:
>> 16/02/10 @ 16:20 (-0500), thus spake Filipe Pires Alvarenga Fernandes:
>>> "\\" works for titles and label, but not for DateFormatter, but \vspace did
>>> the trick!
>>>
>>> Thanks again for the help.
>>>
>>> ps: I'm new to python, but maybe there is a way to mix Latex and unicode?
>>
>> Yes, the inputenc package from latex lets you use unicode.
>> Try adding this to your script:
>> plt.rc('text.latex', preamble='\usepackage[utf]{inputenc}')
>>
>> Note that this setting is not officially supported, whatever
>> that means :)
>
> I think it means that Darren, who did most of the heavy lifting for
> these features, was getting tired of the endless line of additional
> things TeX users wanted to do and the difficulties supporting these
> across all platforms (eg MikTex on Windows) and drew a line in the
> sand saying "I'll add this stuff but if you report problems I may not
> stay up all night trying to fix them".
That is a fair explanation. Adding arbitrary lines to the latex
preamble adds an incredible amount of flexibility if you are a user,
and it provides plenty of rope with which to hang yourself if you are
a developer (answering questions/problems related to people writing
bad python code, misconfigured latex installations, and bad latex
code!). At the time, we had finally shaped the latex backend into
something that worked across platforms (and different versions of
latex installations), plus I was also finishing up grad school and
starting a new job, so I had reservations about even adding this
feature to the library. But people really wanted it so we compromised.
Darren

On Tue, Feb 16, 2010 at 2:15 PM, duckman <tdu...@cr...> wrote:
> I am trying to make the table at the bottom and the lines the same color. I
> took the code from one of the online examples and modified it to do most of
> what I want but cannot get the color in the table working properly! Can
> someone please help? Thank you
>
Do you want the background of the row labels to match your line colors?
The answer seems to be so trivial and I'm not sure if this is what you want.
the_table = table(cellText=cellText,
 rowLabels=rowLabels, rowColours=linecolor,
 colLabels=colLabels,
 loc='bottom')
-JJ

On Tue, Feb 16, 2010 at 3:57 PM, Ernest Adrogué <ead...@gm...> wrote:
> 16/02/10 @ 16:20 (-0500), thus spake Filipe Pires Alvarenga Fernandes:
>> "\\" works for titles and label, but not for DateFormatter, but \vspace did
>> the trick!
>>
>> Thanks again for the help.
>>
>> ps: I'm new to python, but maybe there is a way to mix Latex and unicode?
>
> Yes, the inputenc package from latex lets you use unicode.
> Try adding this to your script:
> plt.rc('text.latex', preamble='\usepackage[utf]{inputenc}')
>
> Note that this setting is not officially supported, whatever
> that means :)
I think it means that Darren, who did most of the heavy lifting for
these features, was getting tired of the endless line of additional
things TeX users wanted to do and the difficulties supporting these
across all platforms (eg MikTex on Windows) and drew a line in the
sand saying "I'll add this stuff but if you report problems I may not
stay up all night trying to fix them".

On Mon, Feb 15, 2010 at 8:28 PM, T J <tj...@gm...> wrote:
> On Mon, Feb 15, 2010 at 5:22 PM, John Hunter <jd...@gm...> wrote:
>> Very nice and thorough work. I think this should be included, but
>> I'll wait to hear from other developers before committing. Could you
>> confirm that the unit tests pass?
>>
>>>>> import matplotlib
>>>>> matplotlib.test()
>>
>
> Confirmed on rev 8133:
>
> Ran 124 tests in 341.585s
>
> FAILED (KNOWNFAIL=2, errors=2)
> and the errors were something to do with hexbin extents and the figimage method.
>
Great -- I committed this patch in r8138
>> I think the markerangle would also be a useful contribution, though it
>> would render some of the markers redundant (eg triangle left, right,
>> etc, would all just be triangles with different angles...)
>>
>
> That was a concern I had as well, but I suppose > ^ v < (etc) could
> just be considered shortcuts to particular angles. Presumably, we
> would not be removing them. Correct? Also, is the standard to have
> the angle specified in degrees? So what is more useful: markerangle
> or markerdeg?
We would definitely be leaving these as shortcuts and for backward
compatibility. And yes the standard is to use degrees -- for
consistency with the text "rotation" property, we may want
markerrotation specified in degrees.
>
> The other difference is that when one specifies fillstyle='left', then
> it would only apply to the marker at 0 degrees. Whereas, marker='v',
> fillstyle='left', markerangle=0 would correspond to marker='^',
> fillstyle='right', markerangle=180  (or something like that).
You can think about what the right way to do this is. My first
inclination is that that left, right, etc, apply to the unrotated
marker, and then you apply the rotation. So 'd' with markerrotation=0
and fillstyle='left' would be identical to 'd' with fillstyle='right'
and markerrotation=180. But any convention you want to apply would
probably be fine as long as it is documented. Note I am not sure this
is a terribly useful feature, but it might be marginally useful and it
seems like something that could be implemented unobtrusively. So
don't kill yourself on it.
Thanks again for the nice work.
JDH