matplotlib-users Mailing List for matplotlib

Eric Firing wrote:
>
>>
>> Thanks Eric - I just wanted to make sure I wasn't missing some 
>> obvious way to do it.
>
>
>
> Jeff,
>
> A possible trick would be to make a very skinny subplot (using custom 
> sizing and positioning) and use contourf itself to make the colorbar 
> in it. I think this is actually the way I am doing it in Matlab. (A 
> colleague and I greatly customized the Matlab colorbar for this sort 
> of thing a long time ago.)
>
> Eric
Eric/Phil: Here is a hack that produces colorbars with discrete contour 
intervals for use with contourf. It's in the form of a patch for 
pylab.py and figure.py (based on 0.83.2). When pylab.colorbar is 
called, it checks to see if the current image was created by contourf, 
and if so calls figure.colobar_contourf (which uses contourf instead of 
imshow to make the colorbar). Here's a test script:
"""test script for contourf colorbar"""
from pylab import *
def func3(x,y):
 return (1- x/2 + x**5 + y**3)*exp(-x**2-y**2)
dx, dy = 0.05, 0.05
X,Y = meshgrid(arange(-3.0,3.0001,dx),arange(-3.0,3.0001,dy))
Z = func3(X, Y)
levels = linspace(-1.2,1.2,13)
l,c = contour (X, Y, Z, levels, linewidths=0.5, colors='k')
clabel(c,l,fmt='%4.2f')
l,c = contourf(X, Y, Z, levels, cmap=cm.jet)
colorbar(tickfmt='%4.2f',orientation='horizontal') # horiz colorbar
colorbar(tickfmt='%4.2f',orientation='vertical') # vert colorbar
axis([-3,3,-3,3])
show()
This doesn't address the other two of Phil's requests (triangles at the 
ends, and constant intervals on the colorbar for non-constant contour 
levels), but I think it's a start.
-Jeff
-- 
Jeffrey S. Whitaker Phone : (303)497-6313
Meteorologist FAX : (303)497-6449
NOAA/OAR/CDC R/CDC1 Email : Jef...@no...
325 Broadway Office : Skaggs Research Cntr 1D-124
Boulder, CO, USA 80303-3328 Web : http://tinyurl.com/5telg

Hi All,
I'm striving to understand how I ca stuff a file object (e.g. StringIO) to
print_figure for an Agg backend. backend_agg.py states : "If filename is a
fileobject, write png to file object (thus you can, for example, write the png
to stdout".
I tried :
 canvas = FigureCanvasAgg(fig)
 data = StringIO.StringIO()
 canvas.print_figure(data, dpi=150)
 return data.get_value()
But that returns an error from backend_agg.py line 383 (matplotlib v. - 0.76) :
StringIO instance has no attribute 'rfind'.
I've decided to use tempfile, but that leads to something hefty and I hate
writing to disk :
 tempplotfilename = tempfile.mkstemp(suffix='.png')
 canvas.print_figure(tempplotfilename[1], dpi=150)
 data = os.read(tempplotfilename[0],os.fstat(tempplotfilename[0]).st_size)
# self.REQUEST.RESPONSE.setHeader('Content-Length',
os.fstat(tempplotfilename[0]).st_size)
# self.REQUEST.RESPONSE.setHeader('Pragma', 'no-cache')
 self.REQUEST.RESPONSE.setHeader('Content-Type', 'image/png')
# os.close(tempplotfilename[0])
# os.unlink(tempplotfilename[1])
 return data
Can this be done ?
Thank you,
Yves Moisan

On 2005年8月23日, T) Sascha GL apparently wrote: 
> Never mind... I found the solution. 
This list is more useful if the solution is actually posted.
Cheers,
Alan Isaac

>>>>> "Sascha" == Sascha GL <Sas...@gm...> writes:
 Sascha> Never mind... I found the solution. Thanks. Sascha
And the answer, for the archives.... ?
JDH

Never mind... I found the solution. Thanks.
Sascha
-- 
5 GB Mailbox, 50 FreeSMS http://www.gmx.net/de/go/promail
+++ GMX - die erste Adresse für Mail, Message, More +++

I realize that this is probably not the appropriate place to ask this
question but it may be related to using matplotlib's pylab interface. The
issue is that when I print an array, I get only the first 300 elements and
then " + XX more elements". Is there a way to get rid of this behavior? I'd
guess so because when using Numeric directly, the complete array is printed.
Thanks in advance for your help.
Sascha
-- 
5 GB Mailbox, 50 FreeSMS http://www.gmx.net/de/go/promail
+++ GMX - die erste Adresse für Mail, Message, More +++

Just for completness I will attach the figure this time.
Phil
On Mon, Aug 22, 2005 at 02:02:57PM -1000, Eric Firing wrote:
> Phil Rasch wrote:
> >On Mon, Aug 22, 2005 at 12:55:16PM -1000, Eric Firing wrote:
> >
> >>(from phil:)
> >
> >.... stuff deleted ......
> >
> >>Phil, Jeff,
> >>
> >>Yes, the present colorbar is designed for use with images, not with 
> >>contourf, and fixing it (or making a new one) is something I have wanted 
> >>to do for quite a while. I just haven't gotten to it yet. Prompted by 
> >>your two emails, maybe I can at least take a close look during the next 
> >>week to see what it would take. It might be easy--probably is. No 
> >>promises yet, though.
> >>
> >>Eric
> >
> >
> >Thanks Eric,
> >
> >I also believe it should be easy to do. But it is better done by
> >somebody more familiar with matplotlib and python than I. I have
> >written such codes for other languages but need more practice before
> >I could do it elegantly in python.
> >
> >As long as you are digging around in there, can I have you think about
> >another couple features? 
> >
> >1) The best colorbars in my mind have "triangles" at the endpoints
> >that indicate the color for the region higher than the highest
> >contour, and lower than the lowest contours. This allows one to label
> >only the meaningful boundaries and not specify how much above or below
> >those regions.
> >
> >2) I frequently need to set contour intervals (the filled region
> >boundaries) to be approximately logarithmic. But I dont want to have
> >these filled regions occupy a fraction of the colorbar proportional to
> >their fraction of the total interval. I want each region to be equal
> >area on the colorbar. 
> >
> >An example of these features can be seen in the attached PNG figure
> >that I created in Yorick with a colorbar code I wrote. I havent tried
> >for beauty, but these figure are OK for for working plots. You can see
> >the point for the unequal contour intervals in the difference plot at
> >the bottom.
> >
> >For the codes I wrote this necessitated supplying arguments to the
> >colorbar function like....
> >
> >colorbar(levs, colors)
> >where levels was an N element array, and colors was an N+1 element
> >list containing color info for each filled region. 
> >
> >Thanks for listening.
> >
> >Phil
> >
-- 
Phil Rasch, Climate Modeling Section, National Center for Atmospheric Research
Mail --> P.O. Box 3000, Boulder CO 80307 
Shipping --> 1850 Table Mesa Dr, Boulder, CO 80305
email: pj...@uc..., Web: http://www.cgd.ucar.edu/cms/pjr Phone: 303-497-1368, FAX: 303-497-1324 

Phil,
> 
> As long as you are digging around in there, can I have you think about
> another couple features? 
OK.
> 
> An example of these features can be seen in the attached PNG figure
I think you forgot to attach the figure--I didn't see it. Your 
description is pretty clear, though.
Eric

Phil Rasch wrote:
> On Mon, Aug 22, 2005 at 12:55:16PM -1000, Eric Firing wrote:
> 
>>(from phil:)
> 
> .... stuff deleted ......
> 
>>Phil, Jeff,
>>
>>Yes, the present colorbar is designed for use with images, not with 
>>contourf, and fixing it (or making a new one) is something I have wanted 
>>to do for quite a while. I just haven't gotten to it yet. Prompted by 
>>your two emails, maybe I can at least take a close look during the next 
>>week to see what it would take. It might be easy--probably is. No 
>>promises yet, though.
>>
>>Eric
> 
> 
> Thanks Eric,
> 
> I also believe it should be easy to do. But it is better done by
> somebody more familiar with matplotlib and python than I. I have
> written such codes for other languages but need more practice before
> I could do it elegantly in python.
> 
> As long as you are digging around in there, can I have you think about
> another couple features? 
> 
> 1) The best colorbars in my mind have "triangles" at the endpoints
> that indicate the color for the region higher than the highest
> contour, and lower than the lowest contours. This allows one to label
> only the meaningful boundaries and not specify how much above or below
> those regions.
> 
> 2) I frequently need to set contour intervals (the filled region
> boundaries) to be approximately logarithmic. But I dont want to have
> these filled regions occupy a fraction of the colorbar proportional to
> their fraction of the total interval. I want each region to be equal
> area on the colorbar. 
> 
> An example of these features can be seen in the attached PNG figure
> that I created in Yorick with a colorbar code I wrote. I havent tried
> for beauty, but these figure are OK for for working plots. You can see
> the point for the unequal contour intervals in the difference plot at
> the bottom.
> 
> For the codes I wrote this necessitated supplying arguments to the
> colorbar function like....
> 
> colorbar(levs, colors)
> where levels was an N element array, and colors was an N+1 element
> list containing color info for each filled region. 
> 
> Thanks for listening.
> 
> Phil
>