matplotlib-users Mailing List for matplotlib

Hi,

when I wrote a script like:


import numarray
image = numarray.ones((30,30))

from pylab import *
matshow(image)
show()

I obtain an error message:

Traceback (most recent call last):
  File "test.py", line 6, in ?
    matshow(image)
  File "/usr/lib/python2.4/site-packages/matplotlib/pylab.py", line 
1647, in matshow
    w,h = figaspect(arr)
  File "/usr/lib/python2.4/site-packages/matplotlib/figure.py", line 
480, in figaspect
    nr,nc = arr.shape[:2]
AttributeError: 'module' object has no attribute 'shape'


if I'm call pylab before to create the numarray array it's ok:

from pylab import *
import numarray
image = numarray.ones((30,30))

matshow(image)
show()

just to let you know.

N.

Dear all,

Perhaps this idea appears strange to some, but in my field (atmospheric
turbulence) it is a common problem: I want to plot data with a log-axis (say
the x-axis) with both positive and negative numbers for x. This implies that I
want to zoom in on small values of |x|. The way to do this, is to define a
'gap' around zero in which no data exist, or are ignored. So if my x-data would
range
from -10 to -0.01 and from 0.01 to 10, the x-axis would look like:

|-------|-------|-------|-------|-------|------|
-10     -1     -0.1  +/-0.01   0.1      1      10

There are few (if any) plotting programs that can do this, but it would make
life a lot easier for me. By now I have hacked my own pylab script to do this,
but it has many limitations. To do it properly, it should be done on a somewhat
lower level in the code, I suppose. The idea is to split the data into either 2
(semilogx and semilogy) or 4 quadrants (loglog) and to plot the data in each
quadrant seperately. If the lower limit of the x-axis (or y-axis) is taken
positive, a normal semilogx (or semilogy) plot is recovered.

More people that need/like this? Any volunteers who know what they are doing (in
terms of low-level pylab coding)?

Regards,

Arnold

-- 
------------------------------------------------------------------------
Arnold F. Moene                          NEW tel: +31 (0)317 482604
Meteorology and Air Quality Group            fax: +31 (0)317 482811
Wageningen University                   e-mail: Arnold.Moene at wur.nl
Duivendaal 2                             url: http://www.met.wau.nl
6701 AP  Wageningen
The Netherlands
------------------------------------------------------------------------
Openoffice.org - Freedom at work
Firefox        - The browser you can trust (www.mozilla.org)
------------------------------------------------------------------------

>>>>> "David" == David Fugate <df...@uc...> writes:

    John> add_base_flags(module) before the call to
    John> ext_modules.append(module) in both the Numeric and numarray
    John> sections.  Ditto for the build_contour function in
    John> setupext.py.  

    David> BTW A new identical problem appeared which was easily fixed
    David> by doing the same thing to the build_contour function.

Which is why I said "Ditto for the build_contour function in
setupext.py" <wink>

Glad it helped -- thanks for the report and persevering!

JDH

John Hunter wrote:
> No, it looks like we are doing something wrong.  I can't believe it
> has not been reported in the umpteen mpl releases that have had this
> problem....
> 
> build_transforms needs to call
> 
>   add_base_flags(module)
> 
> before the call to 
> 
>         ext_modules.append(module)
> 
> in both the Numeric and numarray sections.  Ditto for the
> build_contour function in setupext.py.
> 
> Hope this helps!  Let me know...
> 
> JDH

Yes, this fixed it! Thank you.

David

BTW
A new identical problem appeared which was easily fixed by doing the 
same thing to the build_contour function.

-- 
There are 10 types of people in the world. Those that understand binary 
and those that don't.

>>>>> "kristen" == kristen kaasbjerg <co...@ya...> writes:

    kristen> Hi again Working with legend I've encountered another
    kristen> problem.  Changing the fontsize in a legend seems to be a
    kristen> little harder than first assumed. Is there an easy way to
    kristen> do this??

http://matplotlib.sf.net/examples/legend_demo.py shows you how to
customize the legend text font size.  The examples directory is really
an indispensable tool in learning matplotlib.  If you are using the
source distribution, the examples directory is included.  If you are
using a binary distribution, a zip file is found here
http://matplotlib.sourceforge.net/matplotlib_examples_0.72.zip

The relevant code fragment from legend_demo.py is

  ltext  = leg.get_texts()      # all the text.Text instance in the legend
  set(ltext, fontsize='small')  # the legend text fontsize

JDH

kristen kaasbjerg wrote:
> Hi again
> Working with legend I've encountered another problem.
> Changing the fontsize in a legend seems to be a little
> harder than first assumed. Is there an easy way to do
> this??

 From the mailing list a couple of days ago...

You need to pass in a FontProperties instance that specifies the size you want:

prop = FontProperties(size="x-small')

size    - Either an absolute value of xx-small, x-small, small,
           medium, large, x-large, xx-large; or a relative value
           of smaller or larger; or an absolute font size, e.g. 12;
           or scalable

i.e. lgnd = ax.legend((lines, labels, prop = FontProperties(size="x-small'), 
..other_params_as_required)

Robert

PS This looks like something to add to my 'Getting Started' document....

Hi again
Working with legend I've encountered another problem.
Changing the fontsize in a legend seems to be a little
harder than first assumed. Is there an easy way to do
this??

Kristen



	
		
__________________________________ 
Celebrate Yahoo!'s 10th Birthday! 
Yahoo! Netrospective: 100 Moments of the Web 
http://birthday.yahoo.com/netrospective/

>>>>> "David" == David Fugate <df...@uc...> writes:

    David> Just looking at the output it appears as if the changes
    David> made to basedir (i.e., 'linux2') in setupext.py are not
    David> having any sort of effect (hence the error message about
    David> numarray/arrayobject.h not existing). Is there something
    David> blatantly wrong I'm doing? Any help would be greatly
    David> appreciated.

No, it looks like we are doing something wrong.  I can't believe it
has not been reported in the umpteen mpl releases that have had this
problem....

build_transforms needs to call

  add_base_flags(module)

before the call to 

        ext_modules.append(module)

in both the Numeric and numarray sections.  Ditto for the
build_contour function in setupext.py.

Hope this helps!  Let me know...

JDH