matplotlib-users Mailing List for matplotlib

On Thu, Jul 8, 2010 at 8:41 AM, Jeremy Conlin <jlc...@gm...> wrote:
> On Sun, Jul 4, 2010 at 8:38 PM, Benjamin Root <ben...@ou...> wrote:
>> Jeremy,
>>
>> The pcolor function can take a vmin and a vmax parameter if you wish to
>> control the colorscaling. In addition, you can use a special array
>> structure called a "masked array" to have pcolor ignore "special" values.
>> Assuming your data is 'vals':
>>
>> vals_masked = numpy.ma.masked_array(vals, vals == 0.0)
>>
>> Note that depending on your situation, doing an equality with with a
>> floating point value probably isn't very reliable, so be sure to test and
>> modify to suit your needs. 'vals_masked' can then be passed to pcolor
>> instead of vals.
>
> Yes, I think this is exactly what I need. Thanks!
>
To follow up with my response, I tried the above and it works nicely
with pyplot.pcolor. I would like to get a 3D version of this, like I
get using Axes3D.plot_surface. Is this just not implemented yet? I
am using 0.99.1.1. Has this been implemented in matplotlib 1.0?
Thanks,
Jeremy

Hi
I'm trying to plot several subplots. I have setup a scrollwidget and
viewport and I pack a canvas into a vbox in the viewport.
Problem is that when I scroll, either some of the subplots are missing,
or I get an error when I try to zoom on a graph that argument is not a
gdk.gtk.image (or something like that) but None.
I thought this was fixed in 1.0, but it isn't
Please advice!
Thanks in advance.
Preben

On Fri, Jul 9, 2010 at 10:49 AM, Johannes Röhrs <joh...@me...> wrote:
>
>
> Thanks a lot, this solutions seems to serve my purpose. A new method C.remove() would of course be even better.
>
> One could say the problem is solved, but why does there no method exist to update a contour plot as there is for many other plot routines, i.e.
> set_xdata/set_ydata for plot
> set_data for imshow or
> set_UVC for quiver and so on.
> set_array should be the corresponding method for contour plots, and if type C.get_array() I actually get the data array that I used to plot the countours!
>
> My purpose of this is to animate the contour plot, and I did read somewhere that updating the plot is much faster/more efficient than deleting and recreating the plot.
This is the case when setting up the initial book-keeping is a
significant portion of the time to make the plot. In this case, most
of the work is in generating the contours, so I don't think you'd get
much savings. (Granted, I haven't tried to verify these assumptions.)
Ryan
-- 
Ryan May
Graduate Research Assistant
School of Meteorology
University of Oklahoma

Thanks a lot, this solutions seems to serve my purpose. A new method C.remove() would of course be even better.
One could say the problem is solved, but why does there no method exist to update a contour plot as there is for many other plot routines, i.e.
set_xdata/set_ydata for plot
set_data for imshow or
set_UVC for quiver and so on.
set_array should be the corresponding method for contour plots, and if type C.get_array() I actually get the data array that I used to plot the countours!
My purpose of this is to animate the contour plot, and I did read somewhere that updating the plot is much faster/more efficient than deleting and recreating the plot.
----- Original Message -----
From: "Ryan May" <rm...@gm...>
To: "Johannes Röhrs" <joh...@me...>
Cc: mat...@li...
Sent: Friday, 9 July, 2010 5:11:37 PM
Subject: Re: [Matplotlib-users] update an existing contour plot with new data
On Fri, Jul 9, 2010 at 6:28 AM, Johannes Röhrs <joh...@me...> wrote:
> I have some troubles updating a contour plot. I reduced my code to a simple example to reproduce the problem:
>
> [code]
> from pylab *
> import scipy as sp
>
> x=sp.arange(0,2*sp.pi,0.1)
> X,Y=sp.meshgrid(x,x)
> f1=sp.sin(X)+sp.sin(Y)
> f2=sp.cos(X)+sp.cos(Y)
>
> figure()
> C=contourf(f1)
> show()
>
> C.set_array(f2)
> draw()
> [\code]
>
> What do I need to do to update an existing contour plot with new data?
The set_array() method (I think) only impacts the colormapping
information for contourf, and even then doesn't appear to update.
What you need to do is make a new contour plot and remove the old one,
especially if you need to change the underlying contoured data. This
should be as easy as C.remove(), but for some reason, this doesn't
exist (I'll go add it in a minute). So instead, you need to do the
following:
import matplotlib.pyplot as plt
import numpy as np
x = np.arange(0, 2 * np.pi, 0.1)
X,Y = np.meshgrid(x,x)
f1 = np.sin(X) + np.sin(Y)
f2 = np.cos(X) + np.cos(Y)
plt.figure()
C = plt.contourf(f1)
plt.show()
for coll in C.collections:
 plt.gca().collections.remove(coll)
C = plt.contourf(f2)
plt.draw()
Ryan
-- 
Ryan May
Graduate Research Assistant
School of Meteorology
University of Oklahoma

On Fri, Jul 9, 2010 at 6:28 AM, Johannes Röhrs <joh...@me...> wrote:
> I have some troubles updating a contour plot. I reduced my code to a simple example to reproduce the problem:
>
> [code]
> from pylab *
> import scipy as sp
>
> x=sp.arange(0,2*sp.pi,0.1)
> X,Y=sp.meshgrid(x,x)
> f1=sp.sin(X)+sp.sin(Y)
> f2=sp.cos(X)+sp.cos(Y)
>
> figure()
> C=contourf(f1)
> show()
>
> C.set_array(f2)
> draw()
> [\code]
>
> What do I need to do to update an existing contour plot with new data?
The set_array() method (I think) only impacts the colormapping
information for contourf, and even then doesn't appear to update.
What you need to do is make a new contour plot and remove the old one,
especially if you need to change the underlying contoured data. This
should be as easy as C.remove(), but for some reason, this doesn't
exist (I'll go add it in a minute). So instead, you need to do the
following:
import matplotlib.pyplot as plt
import numpy as np
x = np.arange(0, 2 * np.pi, 0.1)
X,Y = np.meshgrid(x,x)
f1 = np.sin(X) + np.sin(Y)
f2 = np.cos(X) + np.cos(Y)
plt.figure()
C = plt.contourf(f1)
plt.show()
for coll in C.collections:
 plt.gca().collections.remove(coll)
C = plt.contourf(f2)
plt.draw()
Ryan
-- 
Ryan May
Graduate Research Assistant
School of Meteorology
University of Oklahoma

On 7/9/10 10:31 AM, per freem wrote:
> Also, I am not sure how to use alan's code.
>
> If I try:
>
> ec = empirical_cdf(my_data)
> plt.plot(ec)
>
> it doesn't actually look like a cdf
Make sure my_data is sorted first.
plt.plot(my_data, ec)
You probably want to use one of the "steps" linestyles; I'm not sure which one 
would be best. It probably doesn't matter much.
-- 
Robert Kern
"I have come to believe that the whole world is an enigma, a harmless enigma
 that is made terrible by our own mad attempt to interpret it as though it had
 an underlying truth."
 -- Umberto Eco

Also, I am not sure how to use alan's code.
If I try:
ec = empirical_cdf(my_data)
plt.plot(ec)
it doesn't actually look like a cdf
On Fri, Jul 9, 2010 at 10:17 AM, per freem <per...@gm...> wrote:
> How does Alan's code compare with using cumfreq and then plotting its
> result? Is the only difference that cumfreq bins the data?
>
> On Fri, Jul 9, 2010 at 10:12 AM, Robert Kern <rob...@gm...> wrote:
>> On 7/9/10 10:02 AM, per freem wrote:
>>> I'd like to clarify: I want the empirical cdf, but I want it to be
>>> normalized. There's a normed=True option to plt.hist but how can I do
>>> the equivalent for CDFs?
>>
>> There is no such thing as a normalized empirical CDF. Or rather, there is no
>> such thing as an unnormalized empirical CDF.
>>
>> Alan's code is good. Unless if you have a truly staggering number of points,
>> there is no reason to bin the data first.
>>
>> --
>> Robert Kern
>>
>> "I have come to believe that the whole world is an enigma, a harmless enigma
>> that is made terrible by our own mad attempt to interpret it as though it had
>> an underlying truth."
>>  -- Umberto Eco
>>
>>
>> ------------------------------------------------------------------------------
>> This SF.net email is sponsored by Sprint
>> What will you do first with EVO, the first 4G phone?
>> Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
>> _______________________________________________
>> Matplotlib-users mailing list
>> Mat...@li...
>> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>>
>

How does Alan's code compare with using cumfreq and then plotting its
result? Is the only difference that cumfreq bins the data?
On Fri, Jul 9, 2010 at 10:12 AM, Robert Kern <rob...@gm...> wrote:
> On 7/9/10 10:02 AM, per freem wrote:
>> I'd like to clarify: I want the empirical cdf, but I want it to be
>> normalized. There's a normed=True option to plt.hist but how can I do
>> the equivalent for CDFs?
>
> There is no such thing as a normalized empirical CDF. Or rather, there is no
> such thing as an unnormalized empirical CDF.
>
> Alan's code is good. Unless if you have a truly staggering number of points,
> there is no reason to bin the data first.
>
> --
> Robert Kern
>
> "I have come to believe that the whole world is an enigma, a harmless enigma
> that is made terrible by our own mad attempt to interpret it as though it had
> an underlying truth."
>  -- Umberto Eco
>
>
> ------------------------------------------------------------------------------
> This SF.net email is sponsored by Sprint
> What will you do first with EVO, the first 4G phone?
> Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

On 7/9/10 10:02 AM, per freem wrote:
> I'd like to clarify: I want the empirical cdf, but I want it to be
> normalized. There's a normed=True option to plt.hist but how can I do
> the equivalent for CDFs?
There is no such thing as a normalized empirical CDF. Or rather, there is no 
such thing as an unnormalized empirical CDF.
Alan's code is good. Unless if you have a truly staggering number of points, 
there is no reason to bin the data first.
-- 
Robert Kern
"I have come to believe that the whole world is an enigma, a harmless enigma
 that is made terrible by our own mad attempt to interpret it as though it had
 an underlying truth."
 -- Umberto Eco

I'd like to clarify: I want the empirical cdf, but I want it to be
normalized. There's a normed=True option to plt.hist but how can I do
the equivalent for CDFs?
On Fri, Jul 9, 2010 at 9:14 AM, Alan G Isaac <ala...@gm...> wrote:
> On 7/9/2010 12:02 AM, per freem wrote:
>> How can I plot the empirical CDF of an array of numbers in matplotlib
>> in Python?
>
>
> I recalled David Huard posted the below,
> which apparently was once in the sandbox...
> hth,
> Alan Isaac
>
> def empiricalcdf(data, method='Hazen'):
>   """Return the empirical cdf.
>
>   Methods available (here i goes from 1 to N)
>     Hazen:    (i-0.5)/N
>     Weibull:   i/(N+1)
>     Chegodayev: (i-.3)/(N+.4)
>     Cunnane:   (i-.4)/(N+.2)
>     Gringorten: (i-.44)/(N+.12)
>     California: (i-1)/N
>
>   :see:
> http://svn.scipy.org/svn/scipy/trunk/scipy/sandbox/dhuard/stats.py
>   :author: David Huard
>   """
>   i = np.argsort(np.argsort(data)) + 1.
>   nobs = len(data)
>   method = method.lower()
>   if method == 'hazen':
>     cdf = (i-0.5)/nobs
>   elif method == 'weibull':
>     cdf = i/(nobs+1.)
>   elif method == 'california':
>     cdf = (i-1.)/nobs
>   elif method == 'chegodayev':
>     cdf = (i-.3)/(nobs+.4)
>   elif method == 'cunnane':
>     cdf = (i-.4)/(nobs+.2)
>   elif method == 'gringorten':
>     cdf = (i-.44)/(nobs+.12)
>   else:
>     raise 'Unknown method. Choose among Weibull, Hazen, Chegodayev,
> Cunnane, Gringorten and California.'
>   return cdf
>
>
>
>
> ------------------------------------------------------------------------------
> This SF.net email is sponsored by Sprint
> What will you do first with EVO, the first 4G phone?
> Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

Good day!
Could You please tell me how can I get axises autoscaling in the 
animated plot example. I've take an example and have modifyed it 
slightly so the second line in plot gets out of bounding box during 
animation. What I need is autoscaling of axises during animation. Please 
point mee what I have to do.
Thanks a lot!
Hereafter a modifyed example:
--------------------------------------------------------------------
# For detailed comments on animation and the techniqes used here, see
# the wiki entry http://www.scipy.org/Cookbook/Matplotlib/Animations
import sys
from matplotlib.figure import Figure
from matplotlib.backends.backend_qt4agg import FigureCanvasQTAgg as 
FigureCanvas
from PyQt4 import QtGui
ITERS = 100
import numpy as np
import time
class BlitQT(FigureCanvas):
 def __init__(self):
 FigureCanvas.__init__(self, Figure())
 self.ax = self.figure.add_subplot(111)
 self.ax.grid()
 self.draw()
 self.old_size = self.ax.bbox.width, self.ax.bbox.height
 self.ax_background = self.copy_from_bbox(self.ax.bbox)
 self.cnt = 0
 self.x = np.arange(0,2*np.pi,0.01)
 self.sin_line, = self.ax.plot(self.x, np.sin(self.x), 
animated=True)
 self.cos_line, = self.ax.plot(self.x, np.cos(self.x), 
animated=True)
 self.draw()
 self.tstart = time.time()
 self.startTimer(10)
 def timerEvent(self, evt):
 current_size = self.ax.bbox.width, self.ax.bbox.height
 if self.old_size != current_size:
 self.old_size = current_size
 self.ax.clear()
 self.ax.grid()
 self.draw()
 self.ax_background = self.copy_from_bbox(self.ax.bbox)
 self.restore_region(self.ax_background, bbox=self.ax.bbox)
 # update the data
 self.sin_line.set_ydata(np.sin(self.x+self.cnt/10.0))
 self.cos_line.set_ydata((self.x+self.cnt)/50.0)
 # just draw the animated artist
 self.ax.draw_artist(self.sin_line)
 self.ax.draw_artist(self.cos_line)
 # just redraw the axes rectangle
 self.blit(self.ax.bbox)
 if self.cnt == 0:
 # TODO: this shouldn't be necessary, but if it is excluded the
 # canvas outside the axes is not initially painted.
 self.draw()
 if self.cnt==ITERS:
 # print the timing info and quit
 print 'FPS:' , ITERS/(time.time()-self.tstart)
 sys.exit()
 else:
 self.cnt += 1
app = QtGui.QApplication(sys.argv)
widget = BlitQT()
widget.show()
sys.exit(app.exec_())

On 7/9/2010 12:02 AM, per freem wrote:
> How can I plot the empirical CDF of an array of numbers in matplotlib
> in Python?
I recalled David Huard posted the below,
which apparently was once in the sandbox...
hth,
Alan Isaac
def empiricalcdf(data, method='Hazen'):
 """Return the empirical cdf.
 Methods available (here i goes from 1 to N)
 Hazen: (i-0.5)/N
 Weibull: i/(N+1)
 Chegodayev: (i-.3)/(N+.4)
 Cunnane: (i-.4)/(N+.2)
 Gringorten: (i-.44)/(N+.12)
 California: (i-1)/N
 :see:
http://svn.scipy.org/svn/scipy/trunk/scipy/sandbox/dhuard/stats.py
 :author: David Huard
 """
 i = np.argsort(np.argsort(data)) + 1.
 nobs = len(data)
 method = method.lower()
 if method == 'hazen':
 cdf = (i-0.5)/nobs
 elif method == 'weibull':
 cdf = i/(nobs+1.)
 elif method == 'california':
 cdf = (i-1.)/nobs
 elif method == 'chegodayev':
 cdf = (i-.3)/(nobs+.4)
 elif method == 'cunnane':
 cdf = (i-.4)/(nobs+.2)
 elif method == 'gringorten':
 cdf = (i-.44)/(nobs+.12)
 else:
 raise 'Unknown method. Choose among Weibull, Hazen, Chegodayev,
Cunnane, Gringorten and California.'
 return cdf

On Fri, Jul 9, 2010 at 7:25 AM, Karianne Holhjem
<kar...@as...> wrote:
> Regarding numpy - what you say is intersting. I couldn't find any such
> problems in my google-searches. I am running version 1.2.1:
> [karianneholhjem:/] karianne% python -c 'import numpy; print numpy.__version__'
> 1.2.1
Can you try upgrading numpy to the latest released version? This is
likely your problem. I would rm -rf the old numpy in your
site-packages directory and upgrade to 1.4.1
https://sourceforge.net/projects/numpy/files/NumPy/1.4.1/numpy-1.4.1-py2.5-python.org.dmg/download
Are you using python.org python or Apple python -- it appears the
installer above is for python.org python
JDH

Hi,
I changed the script to what you suggested and this is the output:
[karianneholhjem:~] karianne% python bla.py --verbose-helpful
$HOME=/Users/karianne
CONFIGDIR=/Users/karianne/.matplotlib
matplotlib data path /Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/matplotlib/mpl-data
loaded rc file /Users/karianne/.matplotlib/matplotlibrc
matplotlib version 1.0.0
verbose.level helpful
interactive is False
units is False
platform is darwin
Bus error
I get the same output when commenting out either line 2 or 3. 
(/Users/karianne/.matplotlib/matplotlibrc is just a link to 
/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/matplotlib/mpl-data/matplotlibrc 
as was in my previous output.)
Regarding numpy - what you say is intersting. I couldn't find any such 
problems in my google-searches. I am running version 1.2.1:
[karianneholhjem:/] karianne% python -c 'import numpy; print numpy.__version__'
1.2.1
I couldn't find the version requirements in the README file so I probably 
found the wrong README
(/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/numpy/README.txt). 
It did provide a way of testing numpy, I don't know if it's interesting 
but here is the output:
[karianneholhjem:/] karianne% python -c 'import numpy; numpy.test()'
Running unit tests for numpy
NumPy version 1.2.1
NumPy is installed in 
/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/numpy
Python version 2.5.2 (r252:60911, Feb 22 2008, 07:57:53) [GCC 4.0.1 (Apple 
Computer, Inc. build 5363)]
nose version 0.11.3
...........................................................................................................................................................................................................................................................................................................................................................................................................................................................K................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................
----------------------------------------------------------------------
Ran 1740 tests in 8.691s
OK (KNOWNFAIL=1)
Is there any more information I can provide to give you clues as of why 
pyplot and pylab aren't working? Thank you for your help so far, I did a 
course on python/numpy/matplotlib and I would really like to use 
matplotlib as it is a really powerful plotting tool!
cheers,
Karianne
On Thu, 8 Jul 2010, John Hunter wrote:
> On Thu, Jul 8, 2010 at 5:50 AM, Karianne Holhjem
> <kar...@as...> wrote:
> > Hi,
> >
> > I'm having trouble getting pyplot and pylab to work on my Mac v.10.4.11
> > (Tiger). I've tried searching in both google and different macusers
> > forums, but haven't found an answer to my problems. If I have overlooked a
> > webpage please send me a link to the solution.
> >
> > To download matplotlib I have downloaded the dmg package from the official
> > matplotlib page
> > http://sourceforge.net/projects/matplotlib/files/matplotlib/matplotlib-1.0/http://sourceforge.net/projects/matplotlib/files/matplotlib/matplotlib-1.0/
> >
> > and it seems to install fine. However, I cannot use neither pyplot nor
> > pylab:
> 
> 
> You're on the right track with the debugging information you are
> trying to provide. One problem is in your script bla.py. You do
> 
> import matplotlib as mpl
> import mpl.pyplot
> import mpl.pylab
> 
> but you need to do
> 
> import matplotlib as mpl
> import matplotlib.pyplot
> import matplotlib.pylab
> 
> This won't fix your segfault, but it may help you get get better
> debugging information.
> 
> My first guess is a numpy version conflict -- what version are you
> running? You can check the version requirements of the OSX installer
> in the README that is provides. Many recent versions of numpy are not
> ABI compatible, unfortunately.
> 
> JDH
> 

Hi,
I have some troubles updating a contour plot. I reduced my code to a simple example to reproduce the problem: 
[code]
from pylab *
import scipy as sp
x=sp.arange(0,2*sp.pi,0.1)
X,Y=sp.meshgrid(x,x)
f1=sp.sin(X)+sp.sin(Y)
f2=sp.cos(X)+sp.cos(Y)
figure()
C=contourf(f1)
show()
C.set_array(f2)
draw()
[\code]
The problem is that C.set_array(f2) does not show any effect, not even after I call draw(). Shouldn't the array f2 be displayed after that? In comparison, the following code using imshow instead of contour works well:
[code]
figure()
I=imshow(f1)
show()
I.set_data(f2)
draw()
[\code]
What do I need to do to update an existing contour plot with new data?
Greetings
Johannes

Hi,
I posted this to stackoverflow
(http://stackoverflow.com/questions/3190798/scale-legend-box-border-dashed-and-dotted-lines-when-the-figure-size-is-changed),
but didn't get any answer, so here goes again:
I'm trying to use matplotlib to prepare some figures for publication.
In order to make the font sizes match the text of the manuscript I'm
trying to create the figure in the final size to begin with, so that I
avoid scaling the figure when inserting it into the manuscript.
The problem I'm having is that as the figure is then pretty small, I
can scale font sizes, axis sizes, line widths etc., but what I've been
unable to figure out is how to scale dashed or dotted lines, as well
as the thickness of the legend border box. For a simplified and
somewhat exaggerated example, consider
#!/usr/bin/python
small = True
from matplotlib import use
use('pdf')
from matplotlib import rc
rc('ps', usedistiller='xpdf')
rc('text', usetex=True)
if small:
 figsize = (1.0, 0.5)
 rc('font', size=2)
 rc('axes', labelsize=2, linewidth=0.2)
 rc('legend', fontsize=2, handlelength=10)
 rc('xtick', labelsize=2)
 rc('ytick', labelsize=2)
 rc('lines', lw=0.2, mew=0.2)
 rc('grid', linewidth=0.2)
else:
 figsize = (8,8)
import numpy as np
x = np.arange(0, 10, 0.001)
y = np.sin(x)
import matplotlib.pyplot as plt
f = plt.figure(figsize=figsize)
a = f.add_subplot(111)
a.plot(x, y, '--', label='foo bar')
a.legend()
f.savefig('mplt.pdf')
If you change the first executable line to small = False you can see
how it should look in "normal" size. Compared to the normal size, the
small plot suffers from a legend box with too thick borders, and the
dashed line is too coarse, i.e. too long dashes and too long distance
between the dashes.
So my question is, is there a way to fix these two problems?
The matplotlib version I'm using is 0.99.1.2 on ubuntu 10.04 amd64.
-- 
Janne Blomqvist

Dear Matplotlib developers,
first of all my congratulations with the excellent 1.0 release, great work!
I currently use a custom compiled 64 bit version of Python 2.6.
But I would like to switch to the prebuild binaries for Python 2.7 as
soon as possible.
Are there any plans for supplying matplotlib 1.x dmg installers for
Python 2.7 (including 64 bit support on OSX)?
Kind regards,
Pim Schellart
P.S. I quickly tried compiling matplotlib from source against Python
2.7 which worked (after setting PREFIX=/usr/local and PYVERSION=2.7 in
make.osx and compiling with sudo make -f make.osx fetch deps mpl_build
mpl_install && sudo python setup.py install) but failed with a
segfault on import.

How can I plot the empirical CDF of an array of numbers in matplotlib
in Python? I'm looking for the cdf analog of pylab's "hist" function.
One thing I can think of is:
from scipy.stats import cumfreq
a = array([...]) # my array of numbers
num_bins = 20
b = cumfreq(a, num_bins)
plt.plot(b)
Is that correct though? Is there an easier/better way?
thanks.