matplotlib-users Mailing List for matplotlib

Hi Everyone,
I am considering the best practices for producing high-quality plots for
publication. I would like to be able to use colour and and transparency
for screen viewing, but also want to ensure that the graphs print well.
The typical problem I run into is using colors for data curves. When 
printed
on laser printers, these curves become halftone grayscale which appear 
very
"blurry" compared with the pure black curves.
I would like to develop a set of practises that will allow me to specify
two types of colors - those for data that should be rendered in true 
black
(but perhaps with varying line-thickneses corresponding to the original
darkness) when printed, and those for shading, axis frames etc. in the
background where blurry halftones are acceptable.
Has this topic been discussed anywhere? I know of several threads 
discussing
general conversion to grayscale, but nothing discussion the issues of 
blurry
halftones when printing.
Thanks,
Michael.
Refs:
http://thread.gmane.org/gmane.comp.python.matplotlib.general/5479/focus=5484
http://thread.gmane.org/gmane.comp.python.matplotlib.general/28576/focus=28578

On 2012年09月23日 9:27 AM, Benjamin Root wrote:
>
>
> On Sunday, September 23, 2012, Giovanni Plantageneto wrote:
>
> Hi everybody,
> sorry, I guess the question is trivial, but I confess my matplotlib
> and python ignorance.
>
> I'm running some code written by someone else, and apparently some
> bits of the code are not compliant with newer versions of matplotlib.
> So, how can I rewrite the following, which give AttributError?
>
> > self.ax.get_figure().axes = []
>
> and
>
> > self.ax.get_figure().axes = [self.ax <http://self.ax>]
>
> Thanks a lot.
>
>
> Without context, it would be hard to say. What was the exception
> message? I bet it was a NoneType object being returned by get_figure(),
> which would mean that the Axes object was created without a figure,
> which is rarely done.
It looks to me like the code was trying to delete all axes but one from 
the figure. This probably worked when Figure.axes was a plain list, but 
for quite some time it has been a read-only list generated from an 
AxesStack instance.
The code will need rewriting based on an understanding of what it is 
trying to do, and how mpl works now. There is no shortcut. For this 
particular problem, you might be able to do something like this:
fig = self.ax.get_figure()
axlist = fig.axes
for ax in axlist:
 if not ax == self.ax:
 fig.delaxes(ax)
Eric
>
> Also, it looks like the code was trying to manage the hierarchy of
> objects itself (maybe the code was trying to detach an axes from one
> figure and transfer to another? Lots of bookkeeping code like this is
> not needed, but you may still have other issues lurking.
>
> Ben Root
>
>
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://ad.doubleclick.net/clk;258768047;13503038;j?
> http://info.appdynamics.com/FreeJavaPerformanceDownload.html
>
>
>
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

On 2012年09月23日 9:45 AM, Giovanni Plantageneto wrote:
> One of the suggestions I got works:
>
>> Maybe this:
>
>> self.ax.get_figure().clf()
>> self.ax.get_figure().add_axes(self.ax)
This seems a bit dangerous, because logically, even if does not 
presently do so, the clf() call could remove the figure reference from 
self.ax. If you want to go this route, then:
fig = self.ax.get_figure()
fig.clf()
fig.add_axes(self.ax)
Eric
>
>> But it looks really weird to me.
>
> If I understand it correctly, from matplotlib version 1.1.1 (?)
> statements as "self.ax.get_figure().axes = []" are not possible any
> more as axes are not lists anymore. Don't take my word for it, though.
>
> Thanks for the support.
>
>
> 2012年9月23日 Giovanni Plantageneto <g.p...@gm...>:
>> Hi everybody,
>> sorry, I guess the question is trivial, but I confess my matplotlib
>> and python ignorance.
>>
>> I'm running some code written by someone else, and apparently some
>> bits of the code are not compliant with newer versions of matplotlib.
>> So, how can I rewrite the following, which give AttributError?
>>
>>> self.ax.get_figure().axes = []
>>
>> and
>>
>>> self.ax.get_figure().axes = [self.ax]
>>
>> Thanks a lot.
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://ad.doubleclick.net/clk;258768047;13503038;j?
> http://info.appdynamics.com/FreeJavaPerformanceDownload.html
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

One of the suggestions I got works:
>Maybe this:
>self.ax.get_figure().clf()
>self.ax.get_figure().add_axes(self.ax)
>But it looks really weird to me.
If I understand it correctly, from matplotlib version 1.1.1 (?)
statements as "self.ax.get_figure().axes = []" are not possible any
more as axes are not lists anymore. Don't take my word for it, though.
Thanks for the support.
2012年9月23日 Giovanni Plantageneto <g.p...@gm...>:
> Hi everybody,
> sorry, I guess the question is trivial, but I confess my matplotlib
> and python ignorance.
>
> I'm running some code written by someone else, and apparently some
> bits of the code are not compliant with newer versions of matplotlib.
> So, how can I rewrite the following, which give AttributError?
>
>> self.ax.get_figure().axes = []
>
> and
>
>> self.ax.get_figure().axes = [self.ax]
>
> Thanks a lot.

On Sunday, September 23, 2012, Giovanni Plantageneto wrote:
> Hi everybody,
> sorry, I guess the question is trivial, but I confess my matplotlib
> and python ignorance.
>
> I'm running some code written by someone else, and apparently some
> bits of the code are not compliant with newer versions of matplotlib.
> So, how can I rewrite the following, which give AttributError?
>
> > self.ax.get_figure().axes = []
>
> and
>
> > self.ax.get_figure().axes = [self.ax]
>
> Thanks a lot.
>
>
Without context, it would be hard to say. What was the exception message?
 I bet it was a NoneType object being returned by get_figure(), which would
mean that the Axes object was created without a figure, which is rarely
done.
Also, it looks like the code was trying to manage the hierarchy of objects
itself (maybe the code was trying to detach an axes from one figure and
transfer to another? Lots of bookkeeping code like this is not needed, but
you may still have other issues lurking.
Ben Root

Hi everybody,
sorry, I guess the question is trivial, but I confess my matplotlib
and python ignorance.
I'm running some code written by someone else, and apparently some
bits of the code are not compliant with newer versions of matplotlib.
So, how can I rewrite the following, which give AttributError?
> self.ax.get_figure().axes = []
and
> self.ax.get_figure().axes = [self.ax]
Thanks a lot.

Hey Ben,
Here is the code I am using to generate this plot, in addition to an
example input. It's basicaIlly a list of lists, where each inner list is a
time series. Plotting a 3D time series as a 3D surface rectangle is
probably a pretty common problem. It would be awesome if I didn't have to
switch to using rpy + ggplot or even worse, mlabwrap. Thank for any ideas
that anyone has!
def ThreeDSurfacePlot(list_of_lists):
 plt.clf()
 fig = plt.figure()
 list_of_lists = Smooth(list_of_lists)
 xs = numpy.arange(0,list_of_lists.shape[1],1)
 zs = numpy.arange(0,len(list_of_lists),1)
 ax = p3.Axes3D(fig)
 X = numpy.meshgrid(xs,zs)[0]
 Y = numpy.meshgrid(xs,zs)[1]
 Z = list_of_lists
 ax.plot_surface(X, Y, Z, cmap=cm.jet, cstride=1, rstride=1)
 ax.set_xlabel("Sequence Elements")
 ax.set_ylabel("Trial")
 ax.set_zlabel("Inter-Key-Interval")
 ax.set_xscale
 # Set up x ticks
 tick_locs_x = range(20)
 tick_lbls_x = ['d', 'j', 'k', 'f', 'j', 'd', 'f', 'k', 'd', 'f', 'k',
'j', 'd', 'f', 'k', 'f', 'j', 'd', 'k', 'j']
 plt.xticks(tick_locs_x, tick_lbls_x)
 # Set up y (or z?) ticks
 tick_locs_y = range(len(list_of_lists))
 tick_lbls_y = []
 for day in range(len(data["keylog"])):
 day_trial_ctr = 0
 for trial in range(len(data["keylog"][day])):
 if len(data["keylog"][day][trial]) == 20:
 if day_trial_ctr % 20 == 0:
 tick_lbls_y.append("Day:" + str(day) + ", Trial: " +
str(day_trial_ctr))
 else:
 tick_lbls_y.append("")
 day_trial_ctr += 1
 plt.yticks(tick_locs_y, tick_lbls_y, fontsize=10)
 ax.auto_scale_xyz([0,20],[0,50],[0,1])
 ax.view_init(20,45)
 ax.axis('tight')
 plt.savefig(subject + "_3d_surface.png")
 if show: plt.show()
>>> print accurateseries[:3]
[[ 0.5 0.5 0.49699092 0.5 0.68226504 0.48422813
 0.42276716 0.46813011 0.42340088 0.40479589 0.41090202 0.31301808
 0.30782294 0.27784109 0.36982799 0.48932219 0.38784313 0.33056998
 0.40356588 0.32964206]
 [ 0.5 0.57195497 0.30683708 0.46926498 0.44043994 0.46917915
 0.32043695 0.41017413 0.40825605 0.28631306 0.40151811 0.31961489
 0.35328102 0.22550416 0.36752486 0.55106211 0.39073801 0.38961005
 0.36436582 0.34787703]
 [ 0.4480989 0.49201202 0.25450802 0.39503598 0.32998705 0.33187294
 0.35646415 0.36470699 0.3162992 0.28596401 0.39307094 0.4239881
 0.32525587 0.30294204 0.38540196 0.296211 0.35584903 0.31555796
 0.35734415 0.36554003]]
On Thu, Sep 20, 2012 at 7:46 PM, Benjamin Root <ben...@ou...> wrote:
>
>
> On Thursday, September 20, 2012, Brian J Mingus wrote:
>
>> Hi all,
>>
>> I have managed to create a 3d plot with uneven aspect ratio via
>> auto_scale_xyz but I haven't yet figured out how to fix the grid. If you
>> could give me a pointer I would appreciate it.
>>
>> http://imagebin.org/index.php?mode=image&id=229196
>>
>> Thanks,
>>
>>
> Brian,
>
> Just today, I submitted a PR that involved extensive fixes to autoscaling
> in mplot3d. I don't have the PR number on me, but it should be easy to
> find on the github page. While I don't think it will fix much for you, I
> would be interested to know if it breaks your code.
>
> As for your issues, it is hard to help out without a code sample to see
> how you got to where you are. Did you happen to use my experimental
> daspect branch?
>
> Cheers!
> Ben Root
>
>>
>>
-- 
Brian Mingus
Graduate student
Computational Cognitive Neuroscience Lab
University of Colorado at Boulder
http://grey.colorado.edu/mingus
1-720-587-9482