matplotlib-users Mailing List for matplotlib (Page 2)

>>>>> "Robert" == Robert Cimrman <cim...@nt...> writes:

    Robert> John Hunter wrote:
    >>>>>>> "Robert" == Robert Cimrman <cim...@nt...> writes:
    Robert> Does it work for SciPy sparse matrices
    Robert> (e.g. scipy.sparse.csr_matrix)?  I don't think so... That
    Robert> is why I provided my solution. Otherwise, of course, I am
    Robert> well aware of spy, spy2 :-)
    >> I see -- can you post a simple example (with version nums for
    >> numpy/scipy/mpl) that exposes the bug.  Hopefully there will be
    >> an easy fix.

    Robert> Sure:

    Robert> matplotlib: 0.87.2-r2

    Robert> from SVN: numpy: Checked out revision 2433.  scipy:
    Robert> Checked out revision 1888.

    Robert> The problem is, that you cannot use 'where' for sparse
    Robert> matrices yet...

For spy we can use

    def spy(self, Z,  marker='s', markersize=10, **kwargs):
        """
        SPY(Z, **kwargs) plots the sparsity pattern of the matrix Z
        using plot markers.

        kwargs give the marker properties - see help(plot) for more
        information on marker properties

        The line handles are returned
        """
        if hasattr(Z, 'tocoo'):
            c = Z.tocoo()
            x = c.row
            y = c.col
            z = c.data
        else:            
            x,y,z = matplotlib.mlab.get_xyz_where(Z, Z>0)
        return self.plot(x+0.5,y+0.5, linestyle='None',
                         marker=marker,markersize=markersize, **kwargs)

you may want to plug this into your axes.py and test.

For spy2 it is a bit tricker, since it uses an image.  One option
would be to create a regular array of dimensions MxN and fill in the
nonempty cells, but this kind of defeats the purpose of using sparse
arrays.  Another is to use a special purpose RegularPolygonCollection,
and build rectangles at each non-zero pixel.  I like this option best
because it preserves sparsity and allows colormapping, etc..


JDH

John Hunter wrote:
>>>>>>"Robert" == Robert Cimrman <cim...@nt...> writes:
> 
> 
>     Robert> Does it work for SciPy sparse matrices
>     Robert> (e.g. scipy.sparse.csr_matrix)?  I don't think so... That
>     Robert> is why I provided my solution. Otherwise, of course, I am
>     Robert> well aware of spy, spy2 :-)
> 
> I see -- can you post a simple example (with version nums for
> numpy/scipy/mpl) that exposes the bug.  Hopefully there will be an
> easy fix.

Sure:

matplotlib: 0.87.2-r2

from SVN:
numpy: Checked out revision 2433.
scipy: Checked out revision 1888.

The problem is, that you cannot use 'where' for sparse matrices yet...

In [14]:import scipy.sparse as sp
In [15]:import scipy as nm
In [16]:a = nm.array( [[1, 0, 2, 0, 3]] )
In [17]:b = sp.csr_matrix( a )
In [18]:b
Out[18]:
<1x5 sparse matrix of type '<type 'float64scalar'>'
         with 3 stored elements (space for 3)
         in Compressed Sparse Row format>

In [19]:p b
   (0, 0)        1.0
   (0, 2)        2.0
   (0, 4)        3.0

In [20]:spy(b )
---------------------------------------------------------------------------
exceptions.TypeError                                 Traceback (most 
recent call last)

/home/eldaran/<console>

/usr/lib/python2.4/site-packages/matplotlib/pylab.py in spy(*args, **kwargs)
    2184         hold(h)
    2185     try:
-> 2186         ret =  gca().spy(*args, **kwargs)
    2187         draw_if_interactive()
    2188     except:

/usr/lib/python2.4/site-packages/matplotlib/axes.py in spy(self, Z, 
marker, markersize, **kwargs)
    3557         The line handles are returned
    3558         """
-> 3559         x,y,z = matplotlib.mlab.get_xyz_where(Z, Z>0)
    3560         return self.plot(x+0.5,y+0.5, linestyle='None',
    3561                          marker=marker,markersize=markersize, 
**kwargs)

/usr/lib/python2.4/site-packages/scipy/sparse/sparse.py in __cmp__(self, 
other)

TypeError: comparison of sparse matrices not implemented

In [23]:spy( a )
Out[23]:[<matplotlib.lines.Line2D instance at 0xb5709aac>]

>>>>> "Robert" == Robert Cimrman <cim...@nt...> writes:

    Robert> Does it work for SciPy sparse matrices
    Robert> (e.g. scipy.sparse.csr_matrix)?  I don't think so... That
    Robert> is why I provided my solution. Otherwise, of course, I am
    Robert> well aware of spy, spy2 :-)

I see -- can you post a simple example (with version nums for
numpy/scipy/mpl) that exposes the bug.  Hopefully there will be an
easy fix.

JDH

John Hunter wrote:
>>>>>>"Nils" == Nils Wagner <nw...@ia...> writes:
> 
> 
>     Nils> Hi all, What is the state-of-the-art of making matrix
>     Nils> structure plots of sparse matrices with matplotlib ?
> 
> The irony of this question is that you asked the same question in
> 2004 and I implemented spy and spy2 in answer:
> 
>   http://sourceforge.net/mailarchive/message.php?msg_id=10020115  
>   http://sourceforge.net/mailarchive/message.php?msg_id=10020116

Does it work for SciPy sparse matrices (e.g. scipy.sparse.csr_matrix)? I 
don't think so... That is why I provided my solution. Otherwise, of 
course, I am well aware of spy, spy2 :-)

r.

>>>>> "Nils" == Nils Wagner <nw...@ia...> writes:

    Nils> Hi all, What is the state-of-the-art of making matrix
    Nils> structure plots of sparse matrices with matplotlib ?

The irony of this question is that you asked the same question in
2004 and I implemented spy and spy2 in answer:

  http://sourceforge.net/mailarchive/message.php?msg_id=10020115  
  http://sourceforge.net/mailarchive/message.php?msg_id=10020116


JDH

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
  <meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
  <title></title>
</head>
<body bgcolor="#ffffff" text="#000000">
Hi Nils,<br>
<br>
I think you're missing the simplicity of pylab if you don't use the
'spy' or 'spy2' approach. Here is what I had to do. (I needed to
install the suggested <a class="moz-txt-link-freetext" href="http://matplotlib.sourceforge.net/matplotlibrc">http://matplotlib.sourceforge.net/matplotlibrc</a>,
changing 'backend:TkAgg' and 'interactive:True')<br>
<br>
<img alt="[screenshot]"
 src="cid:par...@st..." height="394"
 width="515"><br>
<br>
Nils Wagner wrote:<br>
<blockquote cite="mid...@ia..." type="cite">
  <pre wrap=""><!---->Thank you for your feedback.
Finally I have used Robert Cimrman's code to generate a matrix structure
plot (structure.png).
  </pre>
</blockquote>
<br>
<pre class="moz-signature" cols="72">-- 
John Pye
Department of Mechanical and Manufacturing Engineering
University of New South Wales, Sydney, Australia
<a class="moz-txt-link-freetext" href="http://pye.dyndns.org/">http://pye.dyndns.org/</a></pre>
</body>
</html>

Hello!

I started using matblotlib few days ago, and like it.
I encountered the following problem:

I have a figure with two subplots. I use the toolbar for going back and
forth after changing axes' scales. I want to change something in the graphs
from time to time  -- like shifting some lines elsewhere (by editing lines'
properties). By playing with this I found a strange behaviour of the axes()
command:

I thought that calling simple axes('existing axes instance') should only
make the 'existing axes instance' current. But, at least for me, this
command also resets the possibility of using back & forward buttons. Is thi=
s
really wanted?

What I truly want to achieve is to be able to use those navigation buttons
independently on what is drawn in the axes -- only taking care of the
scales. I use the first subplot for visualizing input data, the second
subplot for visualizing fft of the input data. I have two additional
vertical lines in the former plot for selecting the range of the input data
to be then passed to fft. I want to be able to change scales of both graphs
at any time and do not discontinue this ability by adding new things to the
graphs...

1) is the described behaviour of axes() function intended?
2) what events have some influence on navigation buttons?
3) what shall I read in order to be able to exercise some control over the
buttons?

Thank you a lot! :)
Ruda Sykora

John Pye wrote:
> Hi Nils,
>
> You've got the 'spy' function, just like in MATLAB. Works a treat. For
> example, see under ' Viewing an incidence matrix' on the page:
> https://pse.cheme.cmu.edu/wiki/view/Ascend/ScreenShots
>
> In fact this example uses some tricks to perform colouring of the
> plot. All is explained in the matplotlib wiki.
>
> Cheers
> JP
>
> Nils Wagner wrote:
>> Hi all,
>>
>> What is the state-of-the-art of making matrix structure plots of sparse
>> matrices with matplotlib ?
>>  
>> Nils
>>
>> http://math.nist.gov/MatrixMarket/structureplots.html
Thank you for your feedback.
Finally I have used Robert Cimrman's code to generate a matrix structure
plot (structure.png).
However, the entries should lie on the main diagonal instead of the
antidiagonal.
So what is missing in the code below ? How do I change the orientation
of the plots ?

from pylab import show, plot, axis, xlabel, subplot, title
from scipy import *

K = io.mmread('k.mtx')
M = io.mmread('m.mtx')
print 'Reading finished'

rowk = K.row
colk = K.col
nItemk = len(colk)
nRowk, nColk = shape(K)

rowm = M.row
colm = M.col
nItemm = len(colm)
nRowm, nColm = shape(M)

subplot(211)
title('Stiffness matrix')
plot( colk + 0.5, rowk + 0.5, linestyle = 'None', marker = ',',
markersize = 0.5, markeredgewidth = 0.1 )
axis( [-0.5, nRowk+0.5, -0.5, nColk+0.5] )
axis( 'equal' )
xlabel( '%d x %d: %d nnz, %.2f%% fill' % (nRowk, nColk, nItemk, 100. *
nItemk / float( nRowk * nColk )) )

subplot(212)
title('Mass matrix')
plot( colm + 0.5, rowm + 0.5, linestyle = 'None', marker = ',',
markersize = 0.5, markeredgewidth = 0.1 )
axis( [-0.5, nRowm+0.5, -0.5, nColm+0.5] )
axis( 'equal' )
xlabel( '%d x %d: %d nnz, %.2f%% fill' % (nRowm, nColk, nItemm, 100. *
nItemm / float( nRowm * nColm )) )
show()

Nils Wagner wrote:
> Hi all,
> 
> What is the state-of-the-art of making matrix structure plots of sparse
> matrices with matplotlib ?
>  
> Nils
> 
> http://math.nist.gov/MatrixMarket/structureplots.html


I use this simple code:

row, col are arrays of indices of nonzero matrix entries nItem long, 
nRow, nCol is the matrix shape

plot( col + 0.5, row + 0.5, linestyle = 'None', marker = ',', markersize 
= 0.5, markeredgewidth = 0.1 )
axis( [-0.5, nRow+0.5, -0.5, nCol+0.5] )
axis( 'equal' )
xlabel( '%d x %d: %d nnz, %.2f%% fill' % (nRow, nCol, nItem, 100. * 
nItem / float( nRow * nCol )) )

In SciPy, to get row, col, convert your matrix to the 'COO' format via m 
= m.tocoo(), then m.row, m.col are your indices.

r.

Hi Nils,

You've got the 'spy' function, just like in MATLAB. Works a treat. For 
example, see under ' Viewing an incidence matrix' on the page:
https://pse.cheme.cmu.edu/wiki/view/Ascend/ScreenShots

In fact this example uses some tricks to perform colouring of the plot. 
All is explained in the matplotlib wiki.

Cheers
JP

Nils Wagner wrote:
> Hi all,
>
> What is the state-of-the-art of making matrix structure plots of sparse
> matrices with matplotlib ?
>  
> Nils
>
> http://math.nist.gov/MatrixMarket/structureplots.html
-- 
John Pye
Department of Mechanical and Manufacturing Engineering
University of New South Wales, Sydney, Australia
http://pye.dyndns.org/

Hi all,

What is the state-of-the-art of making matrix structure plots of sparse
matrices with matplotlib ?
 
Nils

http://math.nist.gov/MatrixMarket/structureplots.html

I'm working on a little tool using callbacks, so I have it set up so that
I can interactively add special points in the plot window. I would also
like to delete points. I tried writing a white '+' on top of my original '+'
but it looks like the dithering causes the original point to only fade a 
bit, and not disappear. I tried overwriting circles ('o'), and the interiors
disappeared, but I'm left with the border.

Any suggestions on how to do this? I thought it was going to be easy!

I guess I could just save all the commands and do a redraw, but I'd like
to avoid that if I can.

-- 
-----------------------------------------------------------------------
| Alan K. Jackson            | To see a World in a Grain of Sand      |
| al...@aj...          | And a Heaven in a Wild Flower,         |
| www.ajackson.org           | Hold Infinity in the palm of your hand |
| Houston, Texas             | And Eternity in an hour. - Blake       |
-----------------------------------------------------------------------

Glen,

The change you ran into is mine.  The relevant parameter is in this 
method of the base Locator class:

     def nonsingular(self, vmin, vmax, expander=0.001, tiny=1e-6):
         if vmax < vmin:
             vmin, vmax = vmax, vmin
         if vmax - vmin <= max(abs(vmin), abs(vmax)) * tiny:
             if vmin==0.0:
                 vmin -= 1
                 vmax += 1
             else:
                 vmin -= expander*abs(vmin)
                 vmax += expander*abs(vmax)
         return vmin, vmax

This method gets called in several places.  Some of them supply the 
expander kwarg, but I don't think any supply the tiny kwarg, so the 
default, which I arbitrarily set to 1e-6, is used.  I dimly recall 
finding that using a *much* smaller value could cause trouble elsewhere.

Rather than patching the code you send out, I suggest you consider using 
a custom Locator, which would allow you to set this parameter to 
whatever works for you, and maybe put in additional tweaks for your 
unusual use case.  You could subclass whatever Locator you are using 
now, e.g.

class myLocator(MaxNLocator):
     def nonsingular(self, vmin, vmax, expander=0.001, tiny=1e-12):
         return Locator.nonsingular(self, vmin, vmax,
                                    expander=expander, tiny=tiny)
and then

ax.xaxis.set_major_locator(myLocator())

This is untested, but I think something along these lines is all you 
would need.  Everything except nonsingular would be inherited, and 
nonsingular would be called with your new default value for tiny.

As an alternative approach, can you simply use as your x-coordinate time 
deltas from a suitable base time, so that vmin and vmax would not be so 
huge compared to their difference?  My assumption while working with the 
ticker.py code was that this would be the logical solution to this sort 
of problem.

Longer term, it would be easy to modify the Locator classes so that the 
default values of tiny and expander would appear as a kwargs in 
__init__.  Whether this would be a valuable change or mere clutter, I 
don't know; I will listen for more advice.

Eric

Glen W. Mabey wrote:
> Glen W. Mabey wrote:
> 
>>I'm in the process of migrating some software I've written from a
>>machine with MPL 0.86.2 on it to another with 0.87.2 .
>>
>>The app is Qt-based with MPL embedded as a widget.  The plot itself uses
>>dates on the x-axis.
>>
>>It appears that the behavior of autoscale_view() changed quite a bit
>>somewhere between these two versions.  
>>
>>In my application I create a patches.Rectangle that is 100 seconds wide:
>>
>>    start_time_num = matplotlib.dates.date2num( start_time )
>>    lower_left = [ matplotlib.dates.date2num( start_time ), start_y ]
>>    width = matplotlib.dates.date2num( start_time + datetime.timedelta( seconds=10 ) ) - start_time_num
>>    r = matplotlib.patches.Rectangle( lower_left, width, height )
>>
>>With the 0.86.2 version, the total span of the x-axis after an
>>autoscale_view() call is about 12 seconds.  For the 0.87.2 version
>>however, the span is more than 200 years. (see attached .png files)
>>
>>I don't see this type of behavior in the pylab environment, yet I'm
>>having a hard time creating a concise example that demonstrates the
>>behavior.  
>>
>>I can keep working on extracting an snippet, but I thought I'd ask
>>if this is some intended behavioral change, and if so, if there is a
>>workaround.
> 
> 
> Okay, I figured out what the critical parameter is: if the width of the
> Rectangle is less than one-millionth of the x value, then I get the 
> centuries of timespan on the x-axis phenomenon.  
> 
> (That's probably blindingly obvious in the source code ... somewhere.)
> 
> So, first of all, this behavior has nothing to do with the OO interface,
> nor the Qt backend.
> 
> This seems like reasonable behavior for autoscaling in general.
> Certainly you can't zoom in nicely around a patch with zero width;
> you have to draw the line somewhere.  (ha ha, no pun intended)
> 
> It just so happens that for my application, I need to plot Rectangles
> that represent events as a function of time and frequency.  As such, my
> x-values (as produced by date2num()) are on the order of 732421, yet 
> I need to display patches that are like 10 seconds wide, which turns 
> out to be 0.000115740695037, after a date2num() conversion.
> 
> This behavior is exhibited in the following lines of code.  A delta of 
> 65000 seconds is sufficient to get the scaling right.
> 
> 
> #########
> 
> 
> import matplotlib
> import matplotlib.dates
> from matplotlib.pylab import *
> 
> import datetime
> 
> ax = subplot(111)
> 
> start_time = datetime.datetime(2006, 4, 21)
> start_time_num = matplotlib.dates.date2num( start_time )
> 
> lower_left = [ start_time_num, 1.345680002 ]
> width = matplotlib.dates.date2num( start_time + datetime.timedelta( seconds=100 ) ) - start_time_num
> print 'width = ', width
> 
> ax.add_patch( matplotlib.patches.Rectangle( lower_left, width, 2 ) )
> 
> ax.autoscale_view()
> ax.xaxis.set_major_formatter( matplotlib.dates.DateFormatter('%Y') )
> 
> print ax.get_xlim()
> show()
> 
> 
> ##########
> 
> So, this seems like a fairly narrow case where the threshold in use
> causes a problem.  I can't imagine another instance where 1e-6 would not
> be prefectly fine.
> 
> One solution would be to just patch the source for my deployment of this
> software ...
> 
> I hesitate to suggest that this level ought to be globally tunable by
> the user, although an additional parameter that could be passed to
> autoscale_view() wouldn't be as bad.  Still that seems like clutter.
> 
> What say ye?
> 
> Thank you,
> Glen Mabey
> 
> 
> -------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Glen W. Mabey wrote:
> I'm in the process of migrating some software I've written from a
> machine with MPL 0.86.2 on it to another with 0.87.2 .
> 
> The app is Qt-based with MPL embedded as a widget.  The plot itself uses
> dates on the x-axis.
> 
> It appears that the behavior of autoscale_view() changed quite a bit
> somewhere between these two versions.  
> 
> In my application I create a patches.Rectangle that is 100 seconds wide:
> 
>     start_time_num = matplotlib.dates.date2num( start_time )
>     lower_left = [ matplotlib.dates.date2num( start_time ), start_y ]
>     width = matplotlib.dates.date2num( start_time + datetime.timedelta( seconds=10 ) ) - start_time_num
>     r = matplotlib.patches.Rectangle( lower_left, width, height )
> 
> With the 0.86.2 version, the total span of the x-axis after an
> autoscale_view() call is about 12 seconds.  For the 0.87.2 version
> however, the span is more than 200 years. (see attached .png files)
> 
> I don't see this type of behavior in the pylab environment, yet I'm
> having a hard time creating a concise example that demonstrates the
> behavior.  
> 
> I can keep working on extracting an snippet, but I thought I'd ask
> if this is some intended behavioral change, and if so, if there is a
> workaround.

Okay, I figured out what the critical parameter is: if the width of the
Rectangle is less than one-millionth of the x value, then I get the 
centuries of timespan on the x-axis phenomenon.  

(That's probably blindingly obvious in the source code ... somewhere.)

So, first of all, this behavior has nothing to do with the OO interface,
nor the Qt backend.

This seems like reasonable behavior for autoscaling in general.
Certainly you can't zoom in nicely around a patch with zero width;
you have to draw the line somewhere.  (ha ha, no pun intended)

It just so happens that for my application, I need to plot Rectangles
that represent events as a function of time and frequency.  As such, my
x-values (as produced by date2num()) are on the order of 732421, yet 
I need to display patches that are like 10 seconds wide, which turns 
out to be 0.000115740695037, after a date2num() conversion.

This behavior is exhibited in the following lines of code.  A delta of 
65000 seconds is sufficient to get the scaling right.


#########


import matplotlib
import matplotlib.dates
from matplotlib.pylab import *

import datetime

ax = subplot(111)

start_time = datetime.datetime(2006, 4, 21)
start_time_num = matplotlib.dates.date2num( start_time )

lower_left = [ start_time_num, 1.345680002 ]
width = matplotlib.dates.date2num( start_time + datetime.timedelta( seconds=100 ) ) - start_time_num
print 'width = ', width

ax.add_patch( matplotlib.patches.Rectangle( lower_left, width, 2 ) )

ax.autoscale_view()
ax.xaxis.set_major_formatter( matplotlib.dates.DateFormatter('%Y') )

print ax.get_xlim()
show()


##########

So, this seems like a fairly narrow case where the threshold in use
causes a problem.  I can't imagine another instance where 1e-6 would not
be prefectly fine.

One solution would be to just patch the source for my deployment of this
software ...

I hesitate to suggest that this level ought to be globally tunable by
the user, although an additional parameter that could be passed to
autoscale_view() wouldn't be as bad.  Still that seems like clutter.

What say ye?

Thank you,
Glen Mabey

Glen W. Mabey wrote:
> On Wed, Apr 26, 2006 at 02:32:39PM -0600, Jeff Whitaker wrote:
>   
>> Glen:  That works for me - a small window pops up and I see the 
>> continents and the topo field.  I'm using the latest svn matplotlib and 
>> basemap.
>>     
>
> Jeff:
>
> Could you tell me what versions of Qt and Python you're using?  I'm on
> Python 2.4.3 with Qt 3.3.3, on AMD x86-64.
>
> I'll give it a try on Qt 3.3.6 here shortly ...
>
> Glen
>   

Glen:  Python 2.4.3, qt 3.3.5, pyqt 3.15 on macos x ppc (using x11, not 
aqua).

-Jeff

-- 
Jeffrey S. Whitaker         Phone  : (303)497-6313
Meteorologist               FAX    : (303)497-6449
NOAA/OAR/PSD  R/PSD1        Email  : Jef...@no...
325 Broadway                Office : Skaggs Research Cntr 1D-124
Boulder, CO, USA 80303-3328 Web    : http://tinyurl.com/5telg

On Wed, Apr 26, 2006 at 02:32:39PM -0600, Jeff Whitaker wrote:
> Glen:  That works for me - a small window pops up and I see the 
> continents and the topo field.  I'm using the latest svn matplotlib and 
> basemap.

Jeff:

Could you tell me what versions of Qt and Python you're using?  I'm on
Python 2.4.3 with Qt 3.3.3, on AMD x86-64.

I'll give it a try on Qt 3.3.6 here shortly ...

Glen

Glen W. Mabey wrote:
> On Wed, Apr 26, 2006 at 01:43:30PM -0600, Jeff Whitaker wrote:
>   
>> Glen W. Mabey wrote:
>>     
>>> Hello,
>>>
>>> I'm trying to follow the basemap examples in using basemap as a
>>> matplotlib widget within a Qt-based application.  Taking the example
>>> code
>>>    simpletest_oo.py
>>> and modifying it to target QtAgg instead, I only get the lat/long lines,
>>> as in the attached .png.  The original example does draw all of the
>>> continent lines and country boundaries correctly on my system.
>>>
>>> I'm using matplotlib and basemap svn of 25 Apr 2006; the same results
>>> were produced with 0.87.2 / 0.8.2 .
>>>
>>> If anyone can see something obvious that I'm leaving out, I would really
>>> appreciate a hint as to why the continents, countries, etc. are not
>>> getting drawn.
>>>
>>> Thank you!
>>> Glen Mabey
>>>
>>>  
>>>       
>> Glen:  I can't run your example, since I don't have all the dependencies 
>> (such as mplwidget and justaplot_designer).
>>     
>
> Oh, whoops.  I thought I had everything in there.  Here it is, in a
> single file:
>
>
>   
....(snipped)....
> I think I'm following fairly closely the suggestions posted to this
> mailing list regarding the use of OO matplotlib with basemap: passing
> the 'ax' to Basemap().
>
> Yet, it seems that the call to drawmapboundary() resets the entire axis.
> If it is commented out, then everything else works.  With it, only the
> boundary, parallels, and meridians are shown.
>
> I'm continuing to look into this ...  Thanks for your reply.
>
> Glen
>   

Glen:  That works for me - a small window pops up and I see the 
continents and the topo field.  I'm using the latest svn matplotlib and 
basemap.

-Jeff

-- 
Jeffrey S. Whitaker         Phone  : (303)497-6313
Meteorologist               FAX    : (303)497-6449
NOAA/OAR/PSD  R/PSD1        Email  : Jef...@no...
325 Broadway                Office : Skaggs Research Cntr 1D-124
Boulder, CO, USA 80303-3328 Web    : http://tinyurl.com/5telg

>>>>> "Tom" == Tom Denniston <tom...@al...> writes:
    Tom> I was able to work around the error simply by doing the
    Tom> obvious, making var = 0 if max(abs(vmin), abs(vmax)).  I
    Tom> think someone who understood the code better could probably
    Tom> come up with a better solution.

Hey Tom, this bug is already fixed in matplotlib svn.  Thanks for the
report.

JDH

The following code in matplotlib.ticker has a bug:

730     def scale_range(vmin, vmax, n =3D 1, threshold=3D100):
731         dv =3D abs(vmax - vmin)
732         meanv =3D 0.5*(vmax+vmin)
733 B->     var =3D dv/max(abs(vmin), abs(vmax))
734         if var < 1e-12:
735             return 1.0, 0.0
736         if abs(meanv)/dv < threshold:
737             offset =3D 0
738         elif meanv > 0:

When both dmin and dmax are 0 it results in 0 division.  I have been
unable to write a snippet of code to reproduce the problem
unfortunately.

I was able to work around the error simply by doing the obvious,
making var =3D 0 if max(abs(vmin), abs(vmax)).  I think someone who
understood the code better could probably come up with a better
solution.

--Tom

On Wed, Apr 26, 2006 at 01:43:30PM -0600, Jeff Whitaker wrote:
> Glen W. Mabey wrote:
> >Hello,
> >
> >I'm trying to follow the basemap examples in using basemap as a
> >matplotlib widget within a Qt-based application.  Taking the example
> >code
> >    simpletest_oo.py
> >and modifying it to target QtAgg instead, I only get the lat/long lines,
> >as in the attached .png.  The original example does draw all of the
> >continent lines and country boundaries correctly on my system.
> >
> >I'm using matplotlib and basemap svn of 25 Apr 2006; the same results
> >were produced with 0.87.2 / 0.8.2 .
> >
> >If anyone can see something obvious that I'm leaving out, I would really
> >appreciate a hint as to why the continents, countries, etc. are not
> >getting drawn.
> >
> >Thank you!
> >Glen Mabey
> >
> >  
> 
> Glen:  I can't run your example, since I don't have all the dependencies 
> (such as mplwidget and justaplot_designer).

Oh, whoops.  I thought I had everything in there.  Here it is, in a
single file:



#################
import matplotlib
matplotlib.use('QtAgg')

from matplotlib.backends.backend_qtagg import FigureCanvasQTAgg as FigureCanvas

class MatplotlibWidget( FigureCanvas ):
    """Ultimately, this is a QWidget (as well as a FigureCanvasAgg, etc.)."""
    def __init__( self, parent=None, name=None ):
        self.init_parent = parent
        self.init_name = name

        if self.init_parent:
            # slightly more robust behavior here would be to check to make sure parent is a QWidget ...
            bgc = self.init_parent.backgroundBrush().color()
            bgcolor = float(bgc.red())/255.0, float(bgc.green())/255.0, float(bgc.blue())/255.0
            #bgcolor = "#%02X%02X%02X" % (bgc.red(), bgc.green(), bgc.blue())    Equivalent to above line

        self.fig = Figure( dpi=100, facecolor=bgcolor, edgecolor=bgcolor )
        self.axes = self.fig.add_subplot(111)

        # We want the axes cleared every time plot() is called
        self.axes.hold(False)

        # Apparently the FigureCanvas class has neither a name nor a label attribute.
        FigureCanvas.__init__(self, self.fig)
        self.reparent( self.init_parent, qt.QPoint(0, 0))
        self.axes.set_label( self.init_name )

        self.setMinimumSize( 100, 100 )
        self.setSizePolicy( qt.QSizePolicy.Expanding, qt.QSizePolicy.Expanding )
        self.updateGeometry()

import qt

class PlotWin( qt.QDialog ):
    def __init__( self, parent=None, name=None, modal=0, fl=0 ):
        qt.QDialog.__init__(self,parent,name,modal,fl)

        Form1Layout = qt.QVBoxLayout(self,11,6,"Form1Layout")

        self.matplotlibWidget = MatplotlibWidget(self,"matplotlibWidget")
        Form1Layout.addWidget(self.matplotlibWidget)

        self.resize(qt.QSize(322,222).expandedTo(self.minimumSizeHint()))
        self.clearWState(qt.Qt.WState_Polished)



from matplotlib.toolkits.basemap import Basemap
from matplotlib.figure import Figure
from matplotlib.mlab import meshgrid
import matplotlib.numerix as nx
import matplotlib.cm as cm

from matplotlib.mlab import load


# read in topo data (on a regular lat/lon grid)
# longitudes go from 20 to 380.
etopo = nx.array(load('etopo20data.gz'),'d')
lons = nx.array(load('etopo20lons.gz'),'d')
lats = nx.array(load('etopo20lats.gz'),'d')
# create figure.
plot_window = PlotWin()
# create axes instance, leaving room for colorbar at bottom.
ax = plot_window.matplotlibWidget.axes
# create Basemap instance for Robinson projection.
# set 'ax' keyword so pylab won't be imported.
m = Basemap(projection='robin',lon_0=0.5*(lons[0]+lons[-1]),ax=ax)
# reset figure size to have same aspect ratio as map.
# fig will be 8 inches wide.
# (don't use createfigure, since that imports pylab).
#fig.set_figsize_inches((8,m.aspect*8.))
# make filled contour plot.
x, y = m(*meshgrid(lons, lats))
cs = m.contourf(x,y,etopo,30,cmap=cm.jet)
# draw coastlines.
m.drawcoastlines()
# draw a line around the map region.
m.drawmapboundary()
# draw parallels and meridians.
m.drawparallels(nx.arange(-60.,90.,30.),labels=[1,0,0,0],fontsize=10)
m.drawmeridians(nx.arange(0.,420.,60.),labels=[0,0,0,1],fontsize=10)
# add a title.
ax.set_title('Robinson Projection')
# add a colorbar.
#cax = fig.add_axes([0.125, 0.05, 0.75, 0.05],frameon=False)
#fig.colorbar(cs, cax=cax, tickfmt='%d', orientation='horizontal',clabels=cs.levels[::3]) 
# save image (width 800 pixels with dpi=100 and fig width 8 inches).
plot_window.matplotlibWidget.draw()
plot_window.show()
plot_window.exec_loop()


#####################


If I was more familiar with some of the other GUI backends, I would try
a similar setup with them.

I think I'm following fairly closely the suggestions posted to this
mailing list regarding the use of OO matplotlib with basemap: passing
the 'ax' to Basemap().

Yet, it seems that the call to drawmapboundary() resets the entire axis.
If it is commented out, then everything else works.  With it, only the
boundary, parallels, and meridians are shown.

I'm continuing to look into this ...  Thanks for your reply.

Glen

Glen W. Mabey wrote:
> Hello,
>
> I'm trying to follow the basemap examples in using basemap as a
> matplotlib widget within a Qt-based application.  Taking the example
> code
>     simpletest_oo.py
> and modifying it to target QtAgg instead, I only get the lat/long lines,
> as in the attached .png.  The original example does draw all of the
> continent lines and country boundaries correctly on my system.
>
> I'm using matplotlib and basemap svn of 25 Apr 2006; the same results
> were produced with 0.87.2 / 0.8.2 .
>
> If anyone can see something obvious that I'm leaving out, I would really
> appreciate a hint as to why the continents, countries, etc. are not
> getting drawn.
>
> Thank you!
> Glen Mabey
>
>   

Glen:  I can't run your example, since I don't have all the dependencies 
(such as mplwidget and justaplot_designer).  However,  I can run the 
simpletest.py example using QtAgg and the continents display as 
expected, so I don't think it is a Qt issue.  But, you  probably already 
knew that - sorry I can't be of more help.

-Jeff

-- 
Jeffrey S. Whitaker         Phone  : (303)497-6313
Meteorologist               FAX    : (303)497-6449
NOAA/OAR/PSD  R/PSD1        Email  : Jef...@no...
325 Broadway                Office : Skaggs Research Cntr 1D-124
Boulder, CO, USA 80303-3328 Web    : http://tinyurl.com/5telg

Charlie Moad wrote:
> On 4/26/06, Christopher Barker <Chr...@no...> wrote:
>> Does it support all three numerix packages, TK, gtk, and wx?
> 
> - Yes to all the numerix.
> - Yes to Tk.
> - I never bother with gtk on mac, and I thought we discussed that
> primarily fink/darwin ports people would be gtk users.  They have a
> separate python.

yes, we did, sorry to write that so quick.

> - It should work fine with wx, but I didn't compile the _wxagg
> extension since there is not a universal wxpython yet.

There is a test version now. I'll try what you have done, and then maybe
try to build with the new Universal wxPython.

> The funny thing is that some of the dependencies on OSX appear to be
> PPC only.  Zlib is PPC only.

Do you mean that there is no zlib on intel Macs? Or that the zlib that
you have is a PPC only binary? If the later, then that makes sense. It's
a dynamic lib, so it's runtime linked. At runtime, you only need the
native one. Presumably Intel macs have an intel native zlib installed.

>  I guess they all work due to Rosetta,

Rosetta is pretty cool, but I don't think it supports mixing intel and
PPC libs in one running app.

>  I am glad to hear at least one
> person, Markus, had success with it.

If it ain't broke, don't fix it!

-thanks for all your work.

-Chris



-- 
Christopher Barker, Ph.D.
Oceanographer
                                     		
NOAA/OR&R/HAZMAT         (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chr...@no...

Hello,

I'm trying to follow the basemap examples in using basemap as a
matplotlib widget within a Qt-based application.  Taking the example
code
    simpletest_oo.py
and modifying it to target QtAgg instead, I only get the lat/long lines,
as in the attached .png.  The original example does draw all of the
continent lines and country boundaries correctly on my system.

I'm using matplotlib and basemap svn of 25 Apr 2006; the same results
were produced with 0.87.2 / 0.8.2 .

If anyone can see something obvious that I'm leaving out, I would really
appreciate a hint as to why the continents, countries, etc. are not
getting drawn.

Thank you!
Glen Mabey



########  justaplot_designer.py

from qt import *
from mplwidget import *

class Form1(QDialog):
    def __init__(self,parent = None,name = None,modal = 0,fl = 0):
        QDialog.__init__(self,parent,name,modal,fl)

        Form1Layout = QVBoxLayout(self,11,6,"Form1Layout")

        self.matplotlibWidget = MatplotlibWidget(self,"matplotlibWidget")
        Form1Layout.addWidget(self.matplotlibWidget)

        self.resize(QSize(322,222).expandedTo(self.minimumSizeHint()))
        self.clearWState(Qt.WState_Polished)




########  basemap_simpletest_oo_qt.py

#####################################
# pylab-free version of simpletest.py
#####################################

# set backend to QtAgg.
import matplotlib
matplotlib.use('QtAgg')
from matplotlib.backends.backend_qt import FigureCanvasQT as FigureCanvas

from matplotlib.toolkits.basemap import Basemap
from matplotlib.figure import Figure
from matplotlib.mlab import meshgrid
import matplotlib.numerix as nx
import matplotlib.cm as cm

from matplotlib.mlab import load

import justaplot_designer

class PlotWin( justaplot_designer.Form1 ):
    def __init__(self,parent = None,name = None,modal = 0,fl = 0):
        justaplot_designer.Form1.__init__( self, parent, name, modal, fl )

# read in topo data (on a regular lat/lon grid)
# longitudes go from 20 to 380.
etopo = nx.array(load('etopo20data.gz'),'d')
lons = nx.array(load('etopo20lons.gz'),'d')
lats = nx.array(load('etopo20lats.gz'),'d')
# create figure.
plot_window = PlotWin()
# create axes instance, leaving room for colorbar at bottom.
ax = plot_window.matplotlibWidget.axes
# create Basemap instance for Robinson projection.
# set 'ax' keyword so pylab won't be imported.
m = Basemap(projection='robin',lon_0=0.5*(lons[0]+lons[-1]),ax=ax)
# reset figure size to have same aspect ratio as map.
# fig will be 8 inches wide.
# (don't use createfigure, since that imports pylab).
#fig.set_figsize_inches((8,m.aspect*8.))
# make filled contour plot.
x, y = m(*meshgrid(lons, lats))
cs = m.contourf(x,y,etopo,30,cmap=cm.jet)
# draw coastlines.
m.drawcoastlines()
# draw a line around the map region.
m.drawmapboundary()
# draw parallels and meridians.
m.drawparallels(nx.arange(-60.,90.,30.),labels=[1,0,0,0],fontsize=10)
m.drawmeridians(nx.arange(0.,420.,60.),labels=[0,0,0,1],fontsize=10)
# add a title.
ax.set_title('Robinson Projection')
# add a colorbar.
#cax = fig.add_axes([0.125, 0.05, 0.75, 0.05],frameon=False)
#fig.colorbar(cs, cax=cax, tickfmt='%d', orientation='horizontal',clabels=cs.levels[::3]) 
# save image (width 800 pixels with dpi=100 and fig width 8 inches).
plot_window.matplotlibWidget.draw()
plot_window.show()
plot_window.exec_loop()

Hi all,

I only tested on my G4, so I am glad to hear at least one
> person, Markus, had success with it.



I ran some additional tests on it. All the animation stuff in the examples
runs just fine and very decent. The pylab mode of ipython also works out of
the box with this build.

Greetings,

Markus

On 4/26/06, Christopher Barker <Chr...@no...> wrote:
> Charlie Moad wrote:
> > Hey all (primarily Chris),
> >      I took the time to make a universal build that statically links
> > universal libpng and freetype builds.  It was built against the latest
> > pyarrays.  I don't know if anyone else got around to this yet.
>
> I know I hadn't. Thanks!
>
> Does it support all three numerix packages, TK, gtk, and wx?

- Yes to all the numerix.
- Yes to Tk.
- I never bother with gtk on mac, and I thought we discussed that
primarily fink/darwin ports people would be gtk users.  They have a
separate python.
- It should work fine with wx, but I didn't compile the _wxagg
extension since there is not a universal wxpython yet.

The funny thing is that some of the dependencies on OSX appear to be
PPC only.  Zlib is PPC only.  I didn't look into the tcl/tk framework.
 I guess they all work due to Rosetta, but I am still unclear on all
the details.  For example, it appears I successfully linked a
universal libpng to that libz which is not universal.  In any case,
everything provided by my build is universal, so that's the best we
can do.  I only tested on my G4, so I am glad to hear at least one
person, Markus, had success with it.

- Charlie