matplotlib-users Mailing List for matplotlib

After trying hard to send a plot to my widget, I finally found a solution that works: get the figure from the widget!
In short, it is has simple as:


from pyqtgraph.Qt import QtGui, QtCore
import matplotlib
from matplotlib.backends.backend_qt4agg import FigureCanvasQTAgg as FigureCanvas
from matplotlib.backends.backend_qt4agg import NavigationToolbar2QTAgg as NavigationToolbar
from matplotlib.figure import Figure

class MatplotlibWidget(QtGui.QWidget):
    """
    Implements a Matplotlib figure inside a QWidget.
    Use getFigure() and redraw() to interact with matplotlib.

    Example::

        mw = MatplotlibWidget()
        subplot = mw.getFigure().add_subplot(111)
        subplot.plot(x,y)
        mw.draw()
    """

    def __init__(self, size=(5.0, 4.0), dpi=100):
        QtGui.QWidget.__init__(self)
        self.fig = Figure(size, dpi=dpi)
        self.canvas = FigureCanvas(self.fig)
        self.canvas.setParent(self)
        self.toolbar = NavigationToolbar(self.canvas, self)

        self.vbox = QtGui.QVBoxLayout()
        self.vbox.addWidget(self.toolbar)
        self.vbox.addWidget(self.canvas)

        self.setLayout(self.vbox)

    def getFigure(self):
        return self.fig

    def draw(self):
        self.canvas.draw()


A special thanks to the pyqtgraph maintainer
http://luke.campagnola.me/code/pyqtgraph/


________________________________
 From: Emmanuel Mayssat <ema...@ya...>
To: Emmanuel Mayssat <ema...@ya...>; Eric Firing <ef...@ha...>; "mat...@li..." <mat...@li...> 
Sent: Tuesday, April 3, 2012 4:36 PM
Subject: Re: [Matplotlib-users] assigning a plot object to a figure
 

Although this work, I lose the position of the axes.
How can I reposition the axes2 which was at 221 in fig2, to 221 in fig1 ?




________________________________
 From: Emmanuel Mayssat <ema...@ya...>
To: Eric Firing <ef...@ha...>; "mat...@li..." <mat...@li...> 
Sent: Monday, April 2, 2012 3:20 PM
Subject: Re: [Matplotlib-users] assigning a plot object to a figure
 

Well, not sure if that is supported, but I found a way !:
        [...]

        self.fig1 = Figure(figsize=(width, height), dpi=dpi)
        self.fig2 = Figure(figsize=(width, height), dpi=dpi)
        self.axes1 = self.fig1.add_subplot(223)
        self.axes2 = self.fig2.add_subplot(221)

        [...]
        self.axes2.plot([0,1])
(1)     self.axes2.set_figure(self.fig1)
(2)     self.fig1.add_axes(self.axes2)

Note that (1) and (2) are required in that order!
From the code above, if I show the fig1, then I see axes1 and axes2 on the figure....


Summary:
I created the axes2 independently of fig1

PS: I want to use my axes2 as an object to sent to my Qt widget (a blank figure/canvas)

--
Emmanuel


________________________________
 From: Eric Firing <ef...@ha...>
To: mat...@li... 
Sent: Sunday, April 1, 2012 1:36 PM
Subject: Re: [Matplotlib-users] assigning a plot object to a figure
 
On 03/30/2012 01:23 PM, Emmanuel Mayssat wrote:
> Hello all,
>
> I can create a figure
> and get the corresponding axes/plots
>
> for examples:
> 1/
> fig = Figure(figsize=(width, height), dpi=dpi)
> ax = gif.add_subplot(111)
>
> 2/
> fig, ax = plt.subplots()
>
> but I would like to create my plot independently from the figure and
> assign it to a figure
>
> I code I would like to do something like this
>
> my_ax = Axes(...)
> my_ax.plot(x,y)
>
>
> ax = my_ax
>
> or
>
> my_ax.set_figure(fig)
>
> How can I create the axes/plot object independently from the figure?

As far as I can see, you can't.  The figure is deeply embedded in the 
Axes object; that's why the figure is an argument in Axes.__init__().

> Better yet, how can I assign an axe to a
 figure?

I don't think you can.  It looks to me like Axes.set_figure should be 
clearly marked as a private method, not to be used except in 
Axes.__init__().

Short of a major refactoring of mpl, I think you will need to find some 
other way of accomplishing your ultimate aim.

Eric

> --
> Emmanuel

------------------------------------------------------------------------------
This SF email is sponsosred by:
Try Windows Azure free for 90 days Click Here 
http://p.sf.net/sfu/sfd2d-msazure
_______________________________________________
Matplotlib-users mailing list
Mat...@li...
https://lists.sourceforge.net/lists/listinfo/matplotlib-users



------------------------------------------------------------------------------
Better than sec? Nothing is better than sec when it comes to
monitoring Big Data applications. Try Boundary one-second 
resolution app monitoring today. Free.
http://p.sf.net/sfu/Boundary-dev2dev
_______________________________________________
Matplotlib-users mailing list
Mat...@li...
https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Thanks again, Eric.


I had been trying, and finally got a good result.
My answer is installing PyQt4 for newer Python (v2.7.2) from sources.

Here is the successful steps:

1. Install SIP from source (sip-4.13.2.tar.gz) at default directory

2. Install QtSDK from .run file (Qt_SDK_Lin64_online_v1_2_en.run)
   at /usr/local/QtSDK
   * To install Qt from source, we have to get additional libraries.
     It was very bothersome, so I used official .run file.

3. Install PyQt4 from source (PyQt-x11-gpl-4.9.1.tar.gz)
   When I did ./configure, I set these options:
      python ./configure.py -g -q /usr/local/QtSDK/Desktop/Qt/4.8.0/gcc/bin/qmake
   If not setting -q option, pre-installed Qt may be selected,
   and we may get DBusPendingCall errors.


After above, I set backend and plotted line as below:

> import matplotlib
> matplotlib.use("Qt4Agg")
> import matplotlib.pyplot as plt
> plt.plot([1,2,4,3])
> plt.show()

Then I got plot window from PyQt4! :)


Magician


On 2012/04/01, at 19:42, mat...@li... wrote:

> Date: Sat, 31 Mar 2012 07:53:21 -1000
> From: Eric Firing <ef...@ha...>
> Subject: Re: [Matplotlib-users] Display problem on CentOS
> To: mat...@li...
> Message-ID: <4F7...@ha...>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> 
> On 03/31/2012 04:02 AM, Magician wrote:
>> Alexis&  Eric
>> 
>> 
>> Thanks for your advices.
>> I've been trying, but I still have some problems.
>> 
>> I tried matplotlib.matplotlib_fname() and I found my silly mistake.
>> Python answered '/usr/local/lib/python2.7/site-packages/matplotlib/mpl-data/matplotlibrc'
>> My RC file was named ".matplotlibrc".
>> When the prefixed dot removed, it works perfectly.
>> 
>> And I found to change the backends by using matplotlib.use() command.
>> I tried all of them, but nothing worked (with errors).
>> 
>> When using matplotlib.get_backend() on pre-installed python2.6,
>> it said 'GTKAgg' and matplotlib.pyplot.show() exactly worked.
>> So I tried to install PyGTK from source code, but it's fairly complicated.
>> 
>> Is there an easy way to install backends for additionally installed Python?
>> My python is v2.7.2.
> 
> Can you find CentOS packages for any of the gui toolkits for your python 
> version? pygtk, pyqt4, tkinter, wxpython--any of them?  If so, you 
> should be just about set.  Only pygtk and tkinter would even require 
> rebuilding mpl; pyqt4 and wxpython don't require any mpl extension code.
> 
> Eric
> 
>> 
>> 
>> Magician
>> 
>> 
>> On 2012/03/27, at 23:04, Alexis Praga wrote:
>> 
>>> You can check you are editing the correct configuration file with (in
>>> the Python shell) :
>>>>>> import matplotlib
>>>>>> matplotlib.matplotlib_fname()
>>> 
>>> You can also try other backends. For a list, see :
>>> http://matplotlib.sourceforge.net/faq/usage_faq.html#what-is-a-backend
>>> 
>>> On Tue, Mar 27, 2012 at 2:46 PM, Magician<f_m...@ma...>  wrote:
>>>> Thank you, Alexis.
>>>> 
>>>> I try to install PyQt4 and set Qt4Agg just now, but nothing displayed.
>>>> Ummm...what's wrong...??
>>>> 
>>>> 
>>>> Magician
>>>> 
>>>> 
>>>> On 2012/03/27, at 17:05, Alexis Praga wrote:
>>>> 
>>>>> I had the same problem on Debian.
>>>>> Editing the matplotlibrc (should be installed somewhere in your
>>>>> systeme) and changing the "backend" variable to Qt4Agg did the trick.
>>>>> 
>>>>> 
>>>>> 
>>>>> On Tue, Mar 27, 2012 at 12:14 AM, Magician<f_m...@ma...>  wrote:
>>>>>> Hi.
>>>>>> 
>>>>>> I want to install Matplotlib from source code on CentOS.
>>>>>> I've been using Matplotlib for a year.
>>>>>> But this is the first time for me to install CentOS by myself.
>>>>>> 
>>>>>> I installed CentOS 6.2 in basic install option.
>>>>>> Next, I installed NumPy and Matplotlib.
>>>>>> .matplotlibrc isn't set.
>>>>>> 
>>>>>> It looks successfully installed, but when I type show() command,
>>>>>> nothing appears.
>>>>>> I can export PNG image by using savefig() command, so maybe
>>>>>> I'm using invalid backend.
>>>>>> 
>>>>>> How can I display plots with show() command?
>>>>>> 
>>>>>> 
>>>>>> Magician

On Wednesday, April 4, 2012, Nikolaus Rath wrote:

> Hello,
>
> I'm plotting a rotating phase:
>
> # ipython --pylab
> x1 = linspace(0, 2*pi, 30)
> x = concatenate((x1,x1,x1,x1))
> plot(x)
>
> The resulting plot has ugly vertical lines whenever x wraps from 2*pi
> back to zero.
>
> Does someone have a nice, general way to get to get rid of such lines?
>
> (the actual data is of course less predictable and regular than the
> example above).
>
>
> Best,
>
>   -Nikolaus
>
>
Put [np.nan] between each complete rotation. Nans in a plot breaks the line.

Ben Root

Forwarding an email that I sent directly to Nikolaus. (I think every other
mailing list that I used defaults to something like "Reply to list" or
"Reply to all".)

Warren

---------- Forwarded message ----------
From: Warren Weckesser <war...@en...>
Date: Wed, Apr 4, 2012 at 11:18 AM
Subject: Re: [Matplotlib-users] How to remove vertical lines when plotting
rotating phase
To: Nikolaus Rath <Nik...@ra...>




On Wed, Apr 4, 2012 at 10:48 AM, Warren Weckesser <
war...@en...> wrote:

>
>
> On Wed, Apr 4, 2012 at 10:30 AM, Nikolaus Rath <Nik...@ra...> wrote:
>
>> Hello,
>>
>> I'm plotting a rotating phase:
>>
>> # ipython --pylab
>> x1 = linspace(0, 2*pi, 30)
>> x = concatenate((x1,x1,x1,x1))
>> plot(x)
>>
>> The resulting plot has ugly vertical lines whenever x wraps from 2*pi
>> back to zero.
>>
>> Does someone have a nice, general way to get to get rid of such lines?
>>
>> (the actual data is of course less predictable and regular than the
>> example above).
>>
>>
>>
>
> You can use numpy.unwrap, e.g.
>
> plot(unwrap(x))
>
>
> Warren
>
>

You might not want the "unwrapped" effect.  You can figure out where the
"big" jumps occur, and plot the data in pieces with something like this
example. It doesn't plot lines that jump by more than pi:

-----
import numpy as np
import matplotlib.pyplot as plt

t = np.linspace(0, 10, 200)
y = np.mod(2 * t * np.sin(0.5 * t), 2 * np.pi)

jumps = np.r_[0, np.where(np.abs(np.diff(y)) > np.pi)[0] + 1, y.size]

for k in range(jumps.size-1):
    start, end = jumps[k:k + 2]
    plt.plot(t[start:end], y[start:end], 'b')
plt.show()
-----

I don't know if something like this already exists in matplotlib, or if
there is some other technique that would be simpler.

Warren

Hello,

I'm plotting a rotating phase:

# ipython --pylab
x1 = linspace(0, 2*pi, 30)
x = concatenate((x1,x1,x1,x1))
plot(x)

The resulting plot has ugly vertical lines whenever x wraps from 2*pi
back to zero.

Does someone have a nice, general way to get to get rid of such lines?

(the actual data is of course less predictable and regular than the
example above).


Best,

   -Nikolaus

-- 
 »Time flies like an arrow, fruit flies like a Banana.«

  PGP fingerprint: 5B93 61F8 4EA2 E279 ABF6  02CF A9AD B7F8 AE4E 425C

I create my figure in my (non-interactive) script like so:

import matplotlib.pyplot as plt
fig = plt.figure()
ax = fig.add_subplot(111)
ax.plot(...)

Then I want to set the spacing of y grid to 50 units (axis units). Do I 
need a method of the matplotlib.axis.Axis, or matplotlib.axes.Axes (or 
are these the same??), or something else?

Ben.