matplotlib-users Mailing List for matplotlib

Hello everyone. I have a problem about the PS backend and the Draw Event.

I read the "Automatically make room for tick labels" section 
<http://matplotlib.sourceforge.net/faq/howto_faq.html#automatically-make-room-for-tick-labels> 
in the matplotlib Howto page. It is very useful to my project, and I 
would like to adopt it. I wrote a test program as follows:

> #!/usr/bin/env python
>
> import matplotlib
> matplotlib.use("PS")
>
> import pylab
>
> def on_draw(event):
>     fig = pylab.gcf()
>     cur_bottom = fig.subplotpars.bottom
>
>     if cur_bottom != 0.2:
>         fig.subplots_adjust(bottom=0.2)
>         fig.canvas.draw()
>
>     print("moved bottom")
>
> pylab.xlabel("x axis\nlabel\nis\nlong")
> pylab.gcf().canvas.mpl_connect('draw_event', on_draw)
> #pylab.subplots_adjust(bottom=0.2)
>
> pylab.savefig("test.eps")
which just lifts the bottom to 0.2. If I execute this program, "moved 
bottom" is displayed just once, but the figure is actually not moved. If 
I simple uncomment the line after mpl_connect, figure is lifted. 
However, if I change the backend from "PS" to "Agg" and generate 
"test.png" instead, "moved bottom" is displayed twice, and figure is moved.

So I think when using Agg to generate png image, the canvas draw() 
invokes the Draw Event again and shows the second "moved bottom". This 
makes sense. My question is, when using PS backend,
why does not the subplots_adjust work in on_draw()? And why does not the 
canvas draw() invokes the Draw Event? Is this is an known issue?

I cannot switch my backend to Agg, because Agg does not produce hatch, 
which is essential to my images.

Thanks in advance for you help!

It looks like you're all heading down the right path.  I'm happy to help 
(as the original author of the transforms code), but I'm going to be 
generally unavailable until Thursday or Friday of next week.

Cheers,
Mike

Eric Firing wrote:
> Cohen-Tanugi Johann wrote:
>> I tried to look at the code (axes.py I presume) in order to attempt a 
>> patch, but it defeated me, I do not have the  instructions to 
>> navigate through this code :)
>> Where is the actual transform of the error bars occurring?
>
> I think it is in the draw (or recache) method of the artists.
>
> The right way to handle this problem is not obvious to me.  Maybe with 
> a variant of the log scale that clips instead of masking invalid 
> points. And maybe instead of making a whole set of such variants, the 
> behavior--masking versus clipping, and the clip level if the 
> latter--should be handled by an rcParam.
>
> Eric
>
>> thanks,
>> Johann
>>
>> Michael Droettboom wrote:
>>> I have to say I don't really have a lot of experience with error 
>>> bars on log plots -- but the root cause here is that the lower bound 
>>> of the error bar goes negative, and as we all know, the log of a 
>>> negative number is undefined.  If you can suggest where the lower 
>>> bound should be drawn or provide third-party examples, I'm happy to 
>>> look into this further and resolve this "surprise".
>>>
>>> Mike
>>>
>>> Cohen-Tanugi Johann wrote:
>>>> yes exactly....
>>>> I should have provided a test case, thanks for following up!
>>>> Johann
>>>>
>>>> Matthias Michler wrote:
>>>>  
>>>>> Hello Johann,
>>>>>
>>>>> is the problem you are reporting the one I observe in the attached 
>>>>> picture? Namely some vertical and horizontal lines are missing 
>>>>> when using yscale="log". More precisely everything below y=1 seems 
>>>>> to be missing.
>>>>>
>>>>> The picture was generated with the code below and
>>>>> matplotlib.__version__ = '0.98.6svn'
>>>>> matplotlib.__revision__ = '$Revision: 6887 $'
>>>>>
>>>>> best regards Matthias
>>>>>
>>>>> ###############################
>>>>> import numpy as np
>>>>> import matplotlib.pyplot as plt
>>>>>
>>>>> plt.subplot(111, xscale="log", yscale="log")
>>>>> x = 10.0**np.linspace(0.0, 2.0, 20)
>>>>> y = x**2.0
>>>>> plt.errorbar(x, y, xerr=0.1*x, yerr=5.0+0.75*y)
>>>>> plt.show()
>>>>> ################################
>>>>> On Friday 27 March 2009 16:12:12 Cohen-Tanugi Johann wrote:
>>>>>     
>>>>>> Hello, what is the best way to get log log plots with error bars? I
>>>>>> tried putting log10() everywhere but as I was afraid results look 
>>>>>> ugly....
>>>>>> thanks,
>>>>>> johann

-- 
Michael Droettboom
Science Software Branch
Operations and Engineering Division
Space Telescope Science Institute
Operated by AURA for NASA

> image.set_clim(vmin=10)

This works great - thanks!

Thomas

Thomas Robitaille wrote:
> Hello,
> 
> I am plotting a numpy array with
> 
> image = imshow(array,vmin=0,vmax=100)
> 
> If I want to change the array, I can then do
> 
> image.set_data(new_array)
> 
> and I can also update the colormap, etc. But I can't find any options  
> to update the vmin and vmax values. Am I missing something? What I am  
> looking for is something like this:
> 
> image = imshow(array,vmin=0,vmax=100)
> image.set_vmin = 10
> 
> but there is now such function.

image.set_clim(vmin=10)


You will have to call draw() because the OO interface does not do so 
automatically.  Or you can use the pyplot clim() function, which does 
call draw.

Eric

> 
> Thanks,
> 
> Thomas
> 
> ------------------------------------------------------------------------------
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Hello,

I am plotting a numpy array with

image = imshow(array,vmin=0,vmax=100)

If I want to change the array, I can then do

image.set_data(new_array)

and I can also update the colormap, etc. But I can't find any options  
to update the vmin and vmax values. Am I missing something? What I am  
looking for is something like this:

image = imshow(array,vmin=0,vmax=100)
image.set_vmin = 10

but there is now such function.

Thanks,

Thomas

I'm afraid that I'm still confused and there seems to be not much
thing I can help..

You're considering a circle, but you already have your function in a
cartesian coordinate. I'm not sure why you can't just plot in a
cartesian coordinate? (in other words, what is wrong with your
original script?)

There is an "set_image_extent" call in your original script (although
commented out). Maybe what you're trying to do is simply

 im = imshow(Z, cmap=cm.jet, extent=(-3, 3, -3, 3))

I'm not sure it would be relevant, but if you have your function or
data in  (r, theta) coordinate, one simple way is just to use the
polar axes with pcolormesh method.

 n_theta, n_r = 100, 50
 theta = np.linspace(0, 2*np.pi, n_theta+1)
 r = np.linspace(0., 1., n_r + 1)
 data = np.arange(0., n_theta*n_r).reshape(n_r, n_theta)
 ax=subplot(1,1,1, projection="polar", aspect=1.)
 ax.pcolormesh(theta, r, data)



-JJ

Hello,
I'll try to make myself clear with a simple example: consider a circle
in 2D Cartesian coordinates (x,y) (my domain).
On that circle, for each (x,y) coordinate, you want to plot the function

f(x,y)=x+y

How would you do that in pylab? perhaps I should have called this a
surface plot (as if I was plotting a temperature field on a tube
cross-section)
This is easy to do in pylab if the domain is e.g. a square,  but I do
not know how to handle the simple case I described above conveniently.
Many thanks

Lorenzo

2009/4/2 Jae-Joon Lee <lee...@gm...>:
> On Wed, Apr 1, 2009 at 6:50 AM, Lorenzo Isella <lor...@gm...> wrote:
>> Now, I would like to plot exactly the same function but on a circular
>> domain (circle of radius 1 centered at (0,0)).
>
> You have an image with x,y spanning from -1 to 1. How do you transform
> your x,y coordinate to r, theta? Or, x,y in your example  actually
> corresponds to r, theta in polar coordinate?
> In that case, is it simply that you want to your image centered at
> (0,0)? Use the extent keyword of imshow command.
>
>  http://matplotlib.sourceforge.net/api/pyplot_api.html#matplotlib.pyplot.imshow
>
> It is not clear what you want.
> Regards,
>
> -JJ
>



-- 
Life is what happens to you while you're busy making other plans.

Brian Granger-2 wrote:
> 
> Wow, it is really striking how much better the matplotlib version of
> this plot is compared to R.  Very nice!
> 
> Brian
> 

Brian, yes, I agree!
 

John Hunter-4 wrote:
> 
> Sorry forgot to answer your second question.  You can create multiple
> columns and rows of subplots using the subplot command.  Here is an
> example:
> 
>  
> http://matplotlib.sourceforge.net/examples/pylab_examples/line_styles.html
> 
> and here is the subplot tutorial
> 
>  
> http://matplotlib.sourceforge.net/users/pyplot_tutorial.html#working-with-multiple-figures-and-axes
> 
> When working with lots-o-subplots, you may want to adjust the spacing
> between them as well as around the edges with "subplots_adjust"
> 
>  
> http://matplotlib.sourceforge.net/api/figure_api.html#matplotlib.figure.Figure.
> subplots_adjust
> 
> There is also a pyplot command by the same name.
> 
> JDH
> 


John, thanks for the helpful links. Thinking about my data more and what I
am trying to visualize I now think that the spie charts are not appropriate,
as they are intended for merging two pie charts based on the same underlying
context (e.g. expected proportions vs actual proportions). That being said,
I think the polar bar chart example you sent plus the example of how to have
a gridded layout of plots will get me what I need. I realize these examples
were right there in the gallery, but it helps to have people suggest things
sometimes, so thanks again for responding as you did and not (the seemingly
typical forum) "RTFM".

If what I come up with is different/interesting enough I'll work on
submitting it as an example to the matplotlib gallery.

Josh 
-- 
View this message in context: http://www.nabble.com/spie-charts-in-matplotlib--tp22856064p22868864.html
Sent from the matplotlib - users mailing list archive at Nabble.com.

On Thu, Apr 2, 2009 at 3:11 PM, totallyunimodular <jh...@vn...> wrote:
>
> I am interested in making spie charts in matplotlib, as discussed
> http://www.nabble.com/forum/NewTopic.jtp?forum=2906 here  and
> http://addictedtor.free.fr/graphiques/RGraphGallery.php?graph=106 available
> in R . Additionally, I would like to have a grid of such charts within a
> single graphic frame, something along the lines of the
> http://addictedtor.free.fr/graphiques/RGraphGallery.php?graph=63 star chart
> demo  in the R gallery, but with spie charts instead of star charts.
>

Sorry forgot to answer your second question.  You can create multiple
columns and rows of subplots using the subplot command.  Here is an
example:

  http://matplotlib.sourceforge.net/examples/pylab_examples/line_styles.html

and here is the subplot tutorial

  http://matplotlib.sourceforge.net/users/pyplot_tutorial.html#working-with-multiple-figures-and-axes

When working with lots-o-subplots, you may want to adjust the spacing
between them as well as around the edges with "subplots_adjust"

  http://matplotlib.sourceforge.net/api/figure_api.html#matplotlib.figure.Figure.
subplots_adjust

There is also a pyplot command by the same name.

JDH

Cohen-Tanugi Johann wrote:
> I tried to look at the code (axes.py I presume) in order to attempt a 
> patch, but it defeated me, I do not have the  instructions to navigate 
> through this code :)
> Where is the actual transform of the error bars occurring?

I think it is in the draw (or recache) method of the artists.

The right way to handle this problem is not obvious to me.  Maybe with a 
variant of the log scale that clips instead of masking invalid points. 
And maybe instead of making a whole set of such variants, the 
behavior--masking versus clipping, and the clip level if the 
latter--should be handled by an rcParam.

Eric

> thanks,
> Johann
> 
> Michael Droettboom wrote:
>> I have to say I don't really have a lot of experience with error bars 
>> on log plots -- but the root cause here is that the lower bound of the 
>> error bar goes negative, and as we all know, the log of a negative 
>> number is undefined.  If you can suggest where the lower bound should 
>> be drawn or provide third-party examples, I'm happy to look into this 
>> further and resolve this "surprise".
>>
>> Mike
>>
>> Cohen-Tanugi Johann wrote:
>>> yes exactly....
>>> I should have provided a test case, thanks for following up!
>>> Johann
>>>
>>> Matthias Michler wrote:
>>>  
>>>> Hello Johann,
>>>>
>>>> is the problem you are reporting the one I observe in the attached 
>>>> picture? Namely some vertical and horizontal lines are missing when 
>>>> using yscale="log". More precisely everything below y=1 seems to be 
>>>> missing.
>>>>
>>>> The picture was generated with the code below and
>>>> matplotlib.__version__ = '0.98.6svn'
>>>> matplotlib.__revision__ = '$Revision: 6887 $'
>>>>
>>>> best regards Matthias
>>>>
>>>> ###############################
>>>> import numpy as np
>>>> import matplotlib.pyplot as plt
>>>>
>>>> plt.subplot(111, xscale="log", yscale="log")
>>>> x = 10.0**np.linspace(0.0, 2.0, 20)
>>>> y = x**2.0
>>>> plt.errorbar(x, y, xerr=0.1*x, yerr=5.0+0.75*y)
>>>> plt.show()
>>>> ################################
>>>> On Friday 27 March 2009 16:12:12 Cohen-Tanugi Johann wrote:
>>>>      
>>>>> Hello, what is the best way to get log log plots with error bars? I
>>>>> tried putting log10() everywhere but as I was afraid results look 
>>>>> ugly....
>>>>> thanks,
>>>>> johann

hello..... Anyone? I would very much love to see this fixed, and I am 
ready to help out, but I do not know how to browse through the code. 
Despite the fact that log(errors) should of course not be used, but 
rathter errors/values/log(10), Michael's point still remains : values- 
errors in log scale can be negative, so that the artist should just draw 
a bar until the lower limit of the vertical bar. I would say that this 
is the standard practice.
Sorry for my previous email beside the point.
Johann

Cohen-Tanugi Johann wrote:
> I tried to look at the code (axes.py I presume) in order to attempt a 
> patch, but it defeated me, I do not have the  instructions to navigate 
> through this code :)
> Where is the actual transform of the error bars occurring?
> thanks,
> Johann
>
> Michael Droettboom wrote:
>   
>> I have to say I don't really have a lot of experience with error bars 
>> on log plots -- but the root cause here is that the lower bound of the 
>> error bar goes negative, and as we all know, the log of a negative 
>> number is undefined.  If you can suggest where the lower bound should 
>> be drawn or provide third-party examples, I'm happy to look into this 
>> further and resolve this "surprise".
>>
>> Mike
>>
>> Cohen-Tanugi Johann wrote:
>>     
>>> yes exactly....
>>> I should have provided a test case, thanks for following up!
>>> Johann
>>>
>>> Matthias Michler wrote:
>>>  
>>>       
>>>> Hello Johann,
>>>>
>>>> is the problem you are reporting the one I observe in the attached 
>>>> picture? Namely some vertical and horizontal lines are missing when 
>>>> using yscale="log". More precisely everything below y=1 seems to be 
>>>> missing.
>>>>
>>>> The picture was generated with the code below and
>>>> matplotlib.__version__ = '0.98.6svn'
>>>> matplotlib.__revision__ = '$Revision: 6887 $'
>>>>
>>>> best regards Matthias
>>>>
>>>> ###############################
>>>> import numpy as np
>>>> import matplotlib.pyplot as plt
>>>>
>>>> plt.subplot(111, xscale="log", yscale="log")
>>>> x = 10.0**np.linspace(0.0, 2.0, 20)
>>>> y = x**2.0
>>>> plt.errorbar(x, y, xerr=0.1*x, yerr=5.0+0.75*y)
>>>> plt.show()
>>>> ################################
>>>> On Friday 27 March 2009 16:12:12 Cohen-Tanugi Johann wrote:
>>>>      
>>>>         
>>>>> Hello, what is the best way to get log log plots with error bars? I
>>>>> tried putting log10() everywhere but as I was afraid results look 
>>>>> ugly....
>>>>> thanks,
>>>>> johann
>>>>>    
>>>>> ------------------------------------------------------------------------ 
>>>>>
>>>>>
>>>>> ------------------------------------------------------------------------ 
>>>>>
>>>>>
>>>>> ------------------------------------------------------------------------------ 
>>>>>
>>>>>     
>>>>> ------------------------------------------------------------------------ 
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Matplotlib-users mailing list
>>>>> Mat...@li...
>>>>> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>>>>>           
>>>>>           
>>> ------------------------------------------------------------------------------ 
>>>
>>> _______________________________________________
>>> Matplotlib-users mailing list
>>> Mat...@li...
>>> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>>>   
>>>       
>
> ------------------------------------------------------------------------------
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>   

Wow, it is really striking how much better the matplotlib version of
this plot is compared to R.  Very nice!

Brian

On Thu, Apr 2, 2009 at 7:21 PM, John Hunter <jd...@gm...> wrote:
> On Thu, Apr 2, 2009 at 3:11 PM, totallyunimodular <jh...@vn...> wrote:
>>
>> I am interested in making spie charts in matplotlib, as discussed
>> http://www.nabble.com/forum/NewTopic.jtp?forum=2906 here  and
>> http://addictedtor.free.fr/graphiques/RGraphGallery.php?graph=106 available
>> in R . Additionally, I would like to have a grid of such charts within a
>> single graphic frame, something along the lines of the
>> http://addictedtor.free.fr/graphiques/RGraphGallery.php?graph=63 star chart
>> demo  in the R gallery, but with spie charts instead of star charts.
>
>
> Is this what you are looking for:
>
> http://matplotlib.sourceforge.net/examples/pylab_examples/polar_bar.html
>
> JDH
>
> ------------------------------------------------------------------------------
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

Christopher Barker wrote:

block quote
R. Haynie wrote:
 
block quote
I am using VirtualEnv to have multiple virtual environments that have
their own site-packages folder.  So each environment is its own sandbox.
    
block quote end

another reason why core python really needs to have an "official"
versioning system - either something like virtualenv, or (my
preference), package versioning built in.
Anyway, in the mean time, two ideas to try:
1) "easxy_install matplotlib" run with the easy_install in your virtual
env -- hopefully, it will install a binary Windows egg -- does one exist?
 
block quote end

Tried this towards the end of 2008 a few times, did not fair well at all
in Windows.

block quote
2) try manually moving the matplotlib package from the system
site-packages to your virtualenv.
Ok, this is what I tried today:
 
block quote end

Uninstalled any python package I had installed with a windows installer,
uninstalled all of python, and cleared the cache folder, deleted
python25 folder.
Installed python 2.5 with windows installer.
Took note of what was in site-packages.
Ran NumPy windows installer, and moved all new files and folders out of
Site-Packages.  Moved them to a folder I named c:\site-packages_NumPy.
Did the same for MatPlotLib windows installer.  Moved them to a folder I
named c:\site-packages_MatPlotLib.
Got real anal here and uninstalled numpy and matplotlib using the
windows uninstallers.  Then uninstalled Python and cleared the cache
folder and deleted python25.

Now this is where I thought I could go and make some nice virtual
environments and have some safe sandboxes.  But of course not.
Installed python, easy_install, virtualenv, pywin32.
Created new virtual environment and activated it.  Copied the files from
the numpy and matplotlib site-packages folders into the new virtual
environment site-packages folder.
Tryed a simple plot, and got an error that pylab module did not exist.
So, I figured that the windows installers did more than met the eye.
They probably install some DLLs etc.
I then installed numpy and matplotlib using the windows installers.
Tried my plot test from within the virtual environment and it worked.

I created a new virtual environment, installed nothing, tried the plot
test from within that environment and the test worked.  It should not
have worked.

Looks like I can only have 1 numpy and 1 matplotlib on a windows box at
a time, virtual environments or not.


block quote
Let us know how it works out...
 
block quote end

I sure hope someone reads this and finds something glaringly wrong with
my process.

On Thu, Apr 2, 2009 at 3:11 PM, totallyunimodular <jh...@vn...> wrote:
>
> I am interested in making spie charts in matplotlib, as discussed
> http://www.nabble.com/forum/NewTopic.jtp?forum=2906 here  and
> http://addictedtor.free.fr/graphiques/RGraphGallery.php?graph=106 available
> in R . Additionally, I would like to have a grid of such charts within a
> single graphic frame, something along the lines of the
> http://addictedtor.free.fr/graphiques/RGraphGallery.php?graph=63 star chart
> demo  in the R gallery, but with spie charts instead of star charts.


Is this what you are looking for:

http://matplotlib.sourceforge.net/examples/pylab_examples/polar_bar.html

JDH