matplotlib-users Mailing List for matplotlib

Thank you for your fast reply and suggestion.   I downloaded the GNU tar
ball and looked at it.  Unfortunately due to my own limitations, I need
a win32 installer.
I'll have to bide my time I guess.

RDY

-----Original Message-----
From: Christoph Gohlke [mailto:cg...@uc...] 
Sent: Thursday, December 23, 2010 2:47 PM
To: mat...@li...
Subject: Re: [Matplotlib-users] Python 3



On 12/23/2010 1:01 PM, Robert Young wrote:
> Hi, I have used Matplotlib extensively now for 2 years with python
2.x.
> I recently needed to move to python 3.1 which was greatly facilitated
by
> numpy and scipy being ported to python 3. I was lucky in that all I
have
> to change is many print statements. All on a Windows OS.
>
> But my progress is severely limited by having no port of Matplotlib to
> python 3. I am definitely a user so have contributed twice to
Matplotlib
> development.
>
> Plea: If the stars align properly, I would be so grateful for a port
of
> matplotlib to python 3.
>
> Thanks for hearing me.
>

Did you try the py3k branch at 
<http://matplotlib.svn.sourceforge.net/viewvc/matplotlib/branches/py3k/>
? It 
does work for simple plots.

--
Christoph

------------------------------------------------------------------------
------
Learn how Oracle Real Application Clusters (RAC) One Node allows
customers
to consolidate database storage, standardize their database environment,
and, 
should the need arise, upgrade to a full multi-node Oracle RAC database 
without downtime or disruption
http://p.sf.net/sfu/oracle-sfdevnl
_______________________________________________
Matplotlib-users mailing list
Mat...@li...
https://lists.sourceforge.net/lists/listinfo/matplotlib-users

On 12/23/2010 1:01 PM, Robert Young wrote:
> Hi, I have used Matplotlib extensively now for 2 years with python 2.x.
> I recently needed to move to python 3.1 which was greatly facilitated by
> numpy and scipy being ported to python 3. I was lucky in that all I have
> to change is many print statements. All on a Windows OS.
>
> But my progress is severely limited by having no port of Matplotlib to
> python 3. I am definitely a user so have contributed twice to Matplotlib
> development.
>
> Plea: If the stars align properly, I would be so grateful for a port of
> matplotlib to python 3.
>
> Thanks for hearing me.
>

Did you try the py3k branch at 
<http://matplotlib.svn.sourceforge.net/viewvc/matplotlib/branches/py3k/>? It 
does work for simple plots.

--
Christoph

Hi,  I have used Matplotlib extensively now for 2 years with python 2.x.
I recently needed to move to python 3.1 which was greatly facilitated by
numpy and scipy being ported to python 3.  I was lucky in that all I
have to change is many print statements.  All on a Windows OS.

 

But my progress is severely limited by having no port of Matplotlib to
python 3.  I am definitely a user so have contributed twice to
Matplotlib development.

 

Plea:  If the stars align properly, I would be so grateful for a port of
matplotlib to python 3.

 

Thanks for hearing me.

 

 

Hey guys,

Thank you so much for your clear answers which have been very helpful!

Pawel

-----Wiadomość oryginalna-----
Od: Paul Ivanov [mailto:piv...@gm...] 
Wysłano: Thursday, December 23, 2010 3:03 PM
Do: mat...@li...
DW: Pawel Janowski
Temat: Re: ODP: [Matplotlib-users] starting with pplots

Pawel Janowski, on 2010-12-23 10:09,  wrote:
> Hi Pavel,
> 
> Thanks for your help. Matplotlib seems to be a really cool tool. Your 
> response almost answered my question. What I want is for the 3D plot 
> to be 2D. I mean the z-axis can only take on 5 discreet values so I 
> don't want to visualize 3 dimensions but just two with the data points 
> colored five different colors depending on the z value.

Pawel,

(I'm replying back to the list, so that others may benefit - hello there,
search engine visitors from the future!)

In that case, you can either follow Goya's suggestion - if you only want to
draw points. scatter will actually rescale the color values you give to
whatever colormap you're using - and for your case, with just five z values
in range(1,6), I found a slight tweak to the 'hsv' colormap does the trick.

  from numpy.random import rand, randint
  import matplotlib.pyplot as plt
  x,y = rand(2,100)
  z = randint(1,6,100)
  plt.scatter(x,y,c=z, vmin=-1, cmap=plt.get_cmap('hsv'))

You can see the built-in colormaps here:
http://matplotlib.sourceforge.net/examples/pylab_examples/show_colormaps.htm
l

and as Goyo showed, it's pretty easy to make a new one.

If you want more control, such as changing the shape of the marker, not just
the color, or if there's some order to your points that you want to also see
(for example, draw lines between points of the same z value) - you can use a
boolean mask. 

  from numpy.random import rand, randint
  import matplotlib.pyplot as plt
  x,y = rand(2,100)
  z = randint(1,6,100)
  for i,c,m in zip(range(1,6),'rgbmk', 'odp*s'):
      mask = z==i
      plt.plot(x[mask],y[mask], color=c, marker=m)

hope that helps,
--
Paul Ivanov
314 address only used for lists,  off-list direct email at:
http://pirsquared.org | GPG/PGP key id: 0x0F3E28F7 

Pawel Janowski, on 2010-12-23 10:09,  wrote:
> Hi Pavel,
> 
> Thanks for your help. Matplotlib seems to be a really cool tool. Your
> response almost answered my question. What I want is for the 3D plot to be
> 2D. I mean the z-axis can only take on 5 discreet values so I don't want to
> visualize 3 dimensions but just two with the data points colored five
> different colors depending on the z value. 

Pawel,

(I'm replying back to the list, so that others may benefit -
hello there, search engine visitors from the future!)

In that case, you can either follow Goya's suggestion - if you
only want to draw points. scatter will actually rescale the color
values you give to whatever colormap you're using - and for your
case, with just five z values in range(1,6), I found a slight
tweak to the 'hsv' colormap does the trick.

  from numpy.random import rand, randint
  import matplotlib.pyplot as plt
  x,y = rand(2,100)
  z = randint(1,6,100)
  plt.scatter(x,y,c=z, vmin=-1, cmap=plt.get_cmap('hsv'))

You can see the built-in colormaps here:
http://matplotlib.sourceforge.net/examples/pylab_examples/show_colormaps.html

and as Goyo showed, it's pretty easy to make a new one.

If you want more control, such as changing the shape of the
marker, not just the color, or if there's some order to your points
that you want to also see (for example, draw lines between points
of the same z value) - you can use a boolean mask. 

  from numpy.random import rand, randint
  import matplotlib.pyplot as plt
  x,y = rand(2,100)
  z = randint(1,6,100)
  for i,c,m in zip(range(1,6),'rgbmk', 'odp*s'):
      mask = z==i
      plt.plot(x[mask],y[mask], color=c, marker=m)

hope that helps,
-- 
Paul Ivanov
314 address only used for lists,  off-list direct email at:
http://pirsquared.org | GPG/PGP key id: 0x0F3E28F7 

Václav Šmilauer, on 2010-12-23 14:51,  wrote:
> Hi there,
> 
> when I use twinx() to have y1 and y2 axes and set ticklabel_format
> style to 'sci' on the y2 axis, ticks on the y2 are properly numbered,
> but the "1e-5" that is supposed to be atop y2 appears on the top of
> y1 instead. When both y1 and y2 use the exponents, they overwrite
> each other -- a minimal example (result attached in pdf):
> 
> import pylab
> 
> pylab.plot([0,1e-2,2e-2,3e-2],[1e3,5e3,6.1e3,1e3],'g-')
> 
> pylab.ticklabel_format(style='sci',scilimits=(0,0),axis='both')  # this is not necessary to show the bug
> 
> pylab.twinx()
> 
> pylab.plot([1e-2,2e-2,3e-2,4e-2],[2e-5,3e-5,0,-1e-5],'r-')
> 
> pylab.ticklabel_format(style='sci',scilimits=(0,0),axis='both')   # makes 1e-5 appear on the left instead of on the right
> 
> pylab.show()
> 
Hi Václav,

thanks for the bug report. As a temporary workaround - use 

plt.gca().yaxis.set_offset_position('right')

Committers: the patch attached fixes this problem. I thought that
there might be a similar problem for twiny() - but ax.xaxis does
not appear to have .set_offset_position() method.
 
> I've had this issue with versions .99, 1.0.0, running on Linux
> (Ubuntu, versions 9.04 through to 10.10).

me too, and I kept forgetting to report it.

> PS. what's wrong with the sf.net bugzilla? I was not able to post the
> issue there --

this I do not know.

best,
-- 
Paul Ivanov
314 address only used for lists,  off-list direct email at:
http://pirsquared.org | GPG/PGP key id: 0x0F3E28F7 

Hi there,

when I use twinx() to have y1 and y2 axes and set ticklabel_format style 
to 'sci' on the y2 axis, ticks on the y2 are properly numbered, but the 
"1e-5" that is supposed to be atop y2 appears on the top of y1 instead. 
When both y1 and y2 use the exponents, they overwrite each other -- a 
minimal example (result attached in pdf):

import pylab

pylab.plot([0,1e-2,2e-2,3e-2],[1e3,5e3,6.1e3,1e3],'g-')

pylab.ticklabel_format(style='sci',scilimits=(0,0),axis='both')  # this is not necessary to show the bug

pylab.twinx()

pylab.plot([1e-2,2e-2,3e-2,4e-2],[2e-5,3e-5,0,-1e-5],'r-')

pylab.ticklabel_format(style='sci',scilimits=(0,0),axis='both')   # makes 1e-5 appear on the left instead of on the right

pylab.show()


I've had this issue with versions .99, 1.0.0, running on Linux (Ubuntu, 
versions 9.04 through to 10.10).

Cheers, Václav

PS. what's wrong with the sf.net bugzilla? I was not able to post the 
issue there --

Pawel, on 2010-12-18 20:04,  wrote:
> Hi,
> 
> I am a new user of matplotlib so maybe my question is elementary, but 
> have not been able to find an answer to my problem in the archive.
> 
> I would like to make a 2D plot of colored points of 3D data (clusters). 
> My data looks like this:
> 
> 11837.2120    -0.0858    2.0000
> 23975.2120    -0.0672    2.0000
> 37609.2120    -0.0306    2.0000
> 53263.9800    -0.0690    2.0000
> 72106.6760    0.2708     1.0000
> 92674.6760    -0.0129    3.0000
> 116758.676    -0.1245    3.0000
> ...
> 
> So I need to plot the first and second column as points on the x-y axis 
> and color the points according to the numbers in the third column (which 
> are integers ranging from 1 to5).
> 
> I'd appreciate any help. I realize something so typical should be 
> somewhere in the documentation but I was not able to find it.

Hi Paul,
welcome to matplotlib! So you need to read in those columns
somehow (as numpy arrays, or lists), but once you've got that,
it's just a matter of initiating a 3d plot, and calling scatter
with the three columns.

take a look at this example and it's source code:
http://matplotlib.sourceforge.net/mpl_toolkits/mplot3d/tutorial.html#scatter-plots

for your purposes, the code will be something like:

  from mpl_toolkits.mplot3d import Axes3D
  import matplotlib.pyplot as plt
  fig = plt.figure()
  ax = fig.add_subplot(111, projection='3d')
  ax.scatter(x,y,z)

best,
-- 
Paul Ivanov
314 address only used for lists,  off-list direct email at:
http://pirsquared.org | GPG/PGP key id: 0x0F3E28F7 

2010/12/19 Pawel <pa...@gm...>:
> Hi,
>
> I am a new user of matplotlib so maybe my question is elementary, but
> have not been able to find an answer to my problem in the archive.
>
> I would like to make a 2D plot of colored points of 3D data (clusters).
> My data looks like this:
>
> 11837.2120    -0.0858    2.0000
> 23975.2120    -0.0672    2.0000
> 37609.2120    -0.0306    2.0000
> 53263.9800    -0.0690    2.0000
> 72106.6760    0.2708     1.0000
> 92674.6760    -0.0129    3.0000
> 116758.676    -0.1245    3.0000
> ...
>
> So I need to plot the first and second column as points on the x-y axis
> and color the points according to the numbers in the third column (which
> are integers ranging from 1 to5).
>
> I'd appreciate any help. I realize something so typical should be
> somewhere in the documentation but I was not able to find it.

Try this:

import numpy as np
import matplotlib.pyplot as plt
from matplotlib.colors import ListedColormap
x, y, z = np.loadtxt('data.txt', unpack=True)
cmap = ListedColormap(['b', 'g', 'r', 'c', 'm'])
plt.scatter(x, y, c=z, cmap=cmap, vmin=1, vmax=5)
plt.show()

You'll need to use a single space as column delimiter in your data
file or deal with more loadtxt arguments.

If your z data were color specifications you could just use
plt.scatter(x, y, c=z) as stated in the scatter docstring. Converting
arbitrary data to color specifications is the non trivial issue here.
You can write your own code to do this or use colormaps.


Goyo