matplotlib-users Mailing List for matplotlib

I was having the same problem, but the recent xpdf thread pointed me to 
a/the solution. I was able to successfully run your script by outputting 
as eps instead of ps (which produced truncated output, as you were 
experiencing).
Matt
On 2007年5月10日, Johann Cohen-Tanugi wrote:
> Well, matplotlib starts with a canvas bigger than my screen (vertically only) 
> but in the end it seems to resize it so everything fits in. I can see the 
> whole drawing and the bottom toolbar. Again, saving in png or jpg works 
> perfectly.... and of course I checked that gv was not cutting the graph when 
> displaying it.
> I just tried to use GTK as a backend, and ipython tells me that there are 
> "non implemented" errors on my script....
> I attach it here.
> thanks!
> Johann
>
> Darren Dale wrote:
>> On Wednesday 09 May 2007 12:19:24 pm Johann Cohen-Tanugi wrote:
>> 
>>> I am creating a "big" drawing ( figure(figsize=(16,20)) ), and when I
>>> try to save it in eps/ps form, it mishandle the overall size and only
>>> save a portion of the drawing. Saving in png or jpg works fine though.
>>> Any idea?
>>> 
>> 
>> On my system, you cant create a figure that is larger than the monitor 
>> size, unless I use a non-gui backend like agg or ps. It doesnt matter 
>> whether I save an eps, png, or jpg. Also, make sure your postscript viewer 
>> is not truncating the page due to an inappropriate page size setting.
>> 
>> Darren
>>

Hi, list:
In my quest to create MPL plots for insertion into Word documents, I tried post-script (doesn't work well - Word doesn't like the ps file created by MPL), pdf (can't create more then 8 plots per job and can not import into word), png (works but each plot needs to be magified manually before saving), and now I am playing around with the svg format. Word doesn't recognize the svg format but I can open the file in gimp, select the portion of the chart I need, and do cut-and-paste over to Word. I can do the same with png format too but svg format works better if I have to do cut and paste and scale.
My question is this:
While the svg files are much smaller then the cooresponding png file (yes, and scalable), I am surprised that the quality of the plot doesn't seem to be that much better (seem like about the same as compared to a corresponding png figure). I would have thought that they should (see for example: http://luxor-xul.sourceforge.net/talk/jug-nov-2002/slides.html#svg-1). Is this because MPL is not taking advantage of the capabilities offered by svg?
 
I know that not every one works with Word but for good or for bad, that's what I have to work with in oder to submit my paper for publishing. Surely many of you have come across this kind of chanllenge. What have you been doing for inserting MPL created charts into Word?
Thanks,
--
John Henry

It is working great if I just use .eps
Thanks guys.,,
On Thu, May 10, 2007 at 03:22:20PM -0400, Darren Dale wrote:
> On Thursday 10 May 2007 02:39:51 pm Troels Kofoed Jacobsen wrote:
> > I want to make some good vector graphs to include in LaTeX.
> >
> > I use
> > backend : PS
> > ps.papersize : auto
> > ps.usedistiller : xpdf
> >
> > My problem is that when i export the graphs [savefig('test')] the
> > exported graph is in the middle of a large page, which isn't suitable
> > for inclusion in LaTeX.
> 
> ps.papersize: auto picks the smallest standard papersize that will accommodate 
> your figure.
> 
> > How can I make the page the same size as the graph?
> 
> Try saving an eps file instead.
> 
> > Please help me. Matplotlib makes the nicest graphs i know -- don't make
> > me use Matlab.
> 
> ...
-- 
Troels Kofoed Jacobsen
tkj...@gm...
tel: +45 20880798

I have updated the "Using Python for Interactive Data Analysis" 
tutorial to use numpy instead of numarray (finally!). There are 
further improvements I would like to make in its organization and 
formatting (in the process including suggestions others have made to 
that end), but I'd rather get this version out, which I believe 
addresses all the content changes needed to make it useful for numpy, 
without delaying it any further.
The tutorial, as well as other supporting material and information, 
can be obtained from:
http://www.scipy.org/wikis/topical_software/Tutorial
I'm sure errors remain; please let me know of any you find.
Perry

On Thursday 10 May 2007 02:39:51 pm Troels Kofoed Jacobsen wrote:
> I want to make some good vector graphs to include in LaTeX.
>
> I use
> backend : PS
> ps.papersize : auto
> ps.usedistiller : xpdf
>
> My problem is that when i export the graphs [savefig('test')] the
> exported graph is in the middle of a large page, which isn't suitable
> for inclusion in LaTeX.
ps.papersize: auto picks the smallest standard papersize that will accommodate 
your figure.
> How can I make the page the same size as the graph?
Try saving an eps file instead.
> Please help me. Matplotlib makes the nicest graphs i know -- don't make
> me use Matlab.
...

On 2007年5月10日, Troels Kofoed Jacobsen apparently wrote: 
> My problem is that when i export the graphs 
> [savefig('test')]
savefig('test.eps') always works fine for me...
Cheers,
Alan Isaac

I want to make some good vector graphs to include in LaTeX.
I use
backend : PS
ps.papersize : auto
ps.usedistiller : xpdf 
My problem is that when i export the graphs [savefig('test')] the 
exported graph is in the middle of a large page, which isn't suitable 
for inclusion in LaTeX.
How can I make the page the same size as the graph?
Please help me. Matplotlib makes the nicest graphs i know -- don't make 
me use Matlab.
Best Regards
T.
-- 
Troels Kofoed Jacobsen
tkj...@gm...
tel: +45 20880798

Good news. When the next "production" version will be released?
Le Jeudi 10 Mai 2007 15:47, John Hunter a =E9crit=A0:
> I recently added legend support for polar to matplotlib svn.
>
> JDH
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by DB2 Express
> Download DB2 Express C - the FREE version of DB2 express and take
> control of your XML. No limits. Just data. Click to get it now.
> http://sourceforge.net/powerbar/db2/
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
=2D-=20
Lionel Roubeyrie - lro...@li...
Chag=E9 d'=E9tudes et de maintenance
LIMAIR - la Surveillance de l'Air en Limousin
http://www.limair.asso.fr

Well, matplotlib starts with a canvas bigger than my screen (vertically 
only) but in the end it seems to resize it so everything fits in. I can 
see the whole drawing and the bottom toolbar. Again, saving in png or 
jpg works perfectly.... and of course I checked that gv was not cutting 
the graph when displaying it.
I just tried to use GTK as a backend, and ipython tells me that there 
are "non implemented" errors on my script....
I attach it here.
thanks!
Johann
Darren Dale wrote:
> On Wednesday 09 May 2007 12:19:24 pm Johann Cohen-Tanugi wrote:
> 
>> I am creating a "big" drawing ( figure(figsize=(16,20)) ), and when I
>> try to save it in eps/ps form, it mishandle the overall size and only
>> save a portion of the drawing. Saving in png or jpg works fine though.
>> Any idea?
>> 
>
> On my system, you cant create a figure that is larger than the monitor size, 
> unless I use a non-gui backend like agg or ps. It doesnt matter whether I 
> save an eps, png, or jpg. Also, make sure your postscript viewer is not 
> truncating the page due to an inappropriate page size setting.
>
> Darren
> 

Paul,
Paul Ray wrote:
>
> On May 10, 2007, at 11:05 AM, Werner F. Bruhin wrote:
>
>> Should you have a small matplotlib script which shows the problem 
>> I'll give it a try on my setup (XP or Vista with Python 2.5 and numpy.
>
> No need for a script. "import pylab" produces the error.
No problem here with this.
Just tried the example \matplotlib\examples\animation_blit_tk.py and 
that runs fine too (except it does not shut down nicely if has not done 
it's 1000 iterations:)).
> If you are using Windows XP, Python 2.5, numpy 1.0.2, and matplotlib 
> 0.90, then you have the same setup that is causing us the error. What 
> other packages do you have installed. My guess is that there is some 
> dependency we are not satisfying?
This was on Vista, Python 2.5, matplotlib 0.90, numpy 1.0.1. (can't try 
this config on XP, still have 0.82 installed on there at the moment)
If you look at _transforms.py _ns_transforms is imported for numpy. 
So, maybe your version of numpy is too far ahead?
>
> Thanks for your help,
You are welcome.
Werner

I reuse a Basemap instance to reduce processing time. What I do is
basically:
1. Remember all items I added to the map
2. Remove all items I don't want to keep
I use a list for every category of items (contoursets, clabels, texts,
...), as the way to remove them is slightly different.
Then I remove them from the map axes:
for contourset in contoursets_to_remove:
 for coll in contourset.collections:
 if coll in map_axes.collections:
 map_axes.collections.remove(coll)
for label in clabels_to_remove:
 if label in map_axes.artists:
 map_axes.artists.remove(label)
for txt in texts_to_remove:
 if txt in map_axes.texts:
 map_axes.texts.remove(txt)
Works well for me, but I would not be suprised if there's an more
matplotlib-like method...
While writing these lines, I start to wonder if it wouldn't be more
elegant to remember the items you want to keep (coastlines,...), then do
map_axes.cla() an then add your items to the map axes again?! For my
application, this seems to be the better way, as I'm deleting more
elements than reusing elements.
Shouldn't be too hard to rewrite my code, maybe I can tell you if it
works in a few days.
Cheers
Simon
Jesper Larsen schrieb:
> On Tuesday 08 May 2007 20:55, Jeff Whitaker wrote:
>> Jesper: You might consider just deleting the figure elements that you
>> don't want to re-use (instead of figuring out everything you do want to
>> re-use).
> 
> I have an axes instance on which I have plotted my map decorations 
> (coastlines, meridians and so on). I would like to reuse that axes. On the 
> same axes* I would like to make filled contour plot and potentially a quiver 
> plot. These should be discarded after saving the plot to a file.
> 
> I know I have to discard part of the "collection" variable in the axes 
> instance. But it seems like other things are changed as well. Is it safe to 
> ignore these and simply do something like this:
> 
> nsave = len(ax.collections)
> 
> ...do and save plot...
> 
> ax.collections = ax.collections[0:nsave]
> 
> or would you recommend doing something else?
> 
> Regards,
> Jesper
> 
> *) I have tried to make an extra axes for the contour plot itself but it seems 
> like matplotlib does not create an extra axes when it is the same size as an 
> existing axes.
> 
> -------------------------------------------------------------------------
> This SF.net email is sponsored by DB2 Express
> Download DB2 Express C - the FREE version of DB2 express and take
> control of your XML. No limits. Just data. Click to get it now.
> http://sourceforge.net/powerbar/db2/
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

On May 10, 2007, at 11:05 AM, Werner F. Bruhin wrote:
> Should you have a small matplotlib script which shows the problem 
> I'll give it a try on my setup (XP or Vista with Python 2.5 and numpy.
No need for a script. "import pylab" produces the error.
If you are using Windows XP, Python 2.5, numpy 1.0.2, and matplotlib 
0.90, then you have the same setup that is causing us the error. 
What other packages do you have installed. My guess is that there is 
some dependency we are not satisfying?
Thanks for your help,
-- Paul

Hi Paul,
Paul Ray wrote:
>
> On May 10, 2007, at 10:34 AM, Werner F. Bruhin wrote:
>
>> What backend are you using?
>>
>> If you use wxPython then as of 0.87.? something you need to either 
>> use wxPython 2.6 Unicode or rename the following file:
>
> I'm not 100% sure. We certainly didn't intentionally use wxPython, 
> since I don't even know what it is.
It is a Python wrapping of wxWidgets, much nicer then Tk in my view.
> The only packages we installed were python, numpy, and matplotlib. 
> I think his backend is TkAgg, but I can't check since it was on my 
> colleagues computer and he is on travel today. (All the rest of us us 
> python/numpy/matplotlib on Macs or Linux and everything is fine with 
> 0.90)
If you or he uses TkAgg then I have no idea what might cause the problem.
Should you have a small matplotlib script which shows the problem I'll 
give it a try on my setup (XP or Vista with Python 2.5 and numpy.
>
> I know there are a bunch of .pyd files in his matplotlib directory. 
> Are those Windows DLL versions of packages? As far as I could tell 
> the relevant one "_ns_transforms.pyd" is there, so I'm not sure why it 
> is having trouble finding it (though I'm working from memory here).
Someone more technical has to explain that, but I believe you right in 
saying that the .pyd files are a kind of a DLL on Windows.
Werner

On May 10, 2007, at 10:34 AM, Werner F. Bruhin wrote:
> What backend are you using?
>
> If you use wxPython then as of 0.87.? something you need to either 
> use wxPython 2.6 Unicode or rename the following file:
I'm not 100% sure. We certainly didn't intentionally use wxPython, 
since I don't even know what it is. The only packages we installed 
were python, numpy, and matplotlib. I think his backend is TkAgg, 
but I can't check since it was on my colleagues computer and he is on 
travel today. (All the rest of us us python/numpy/matplotlib on Macs 
or Linux and everything is fine with 0.90)
I know there are a bunch of .pyd files in his matplotlib directory. 
Are those Windows DLL versions of packages? As far as I could tell 
the relevant one "_ns_transforms.pyd" is there, so I'm not sure why 
it is having trouble finding it (though I'm working from memory here).
Cheers,
-- Paul

On 5/10/07, Nils Wagner <nw...@ia...> wrote:
> Hi all,
>
> How can I increase the distance between the title and the spy plot ?
>
> I have used
>
> figure(1)
> subplot(121)
> spy(M_bc)
> title(r'$M_{red}$')
> subplot(122)
> title(r'$K_{red}$')
> spy(K_bc)
>
You can set the "y" attribute of the title, either with
title('my title', y=1.05)
or
t = title('my title')
t.set_y(1.05)
JDH

Hi Paul,
Paul Ray wrote:
> Hi,
>
> A colleague of mine uses Python on Windows. When he upgraded to 
> matplotlib 0.90 (from 0.82, I think) he started getting an import 
> error when importing pylab. We tried many combinations and the only 
> thing that fixed it was downgrading back to 0.82.
>
> We tried:
> python2.4 and python2.5
> numpy 1.0.1 and 1.0.2
> matplotlib 0.90 and 0.87.7
> 
What backend are you using?
If you use wxPython then as of 0.87.? something you need to either use 
wxPython 2.6 Unicode or rename the following file:
matplot\backends\_wxagg.pyd
to
matplot\backends\_wxagg DO NOT USE.pyd (or something similar).
The 0.87 version was compiled against wxPython 2.6 to take accelarate some functions, by renaming the above file you basically deactive this accelaration code.
I believe the next version of matplotlib will get rid of this dependency.
Werner

Hi,
A colleague of mine uses Python on Windows. When he upgraded to 
matplotlib 0.90 (from 0.82, I think) he started getting an import 
error when importing pylab. We tried many combinations and the only 
thing that fixed it was downgrading back to 0.82.
We tried:
python2.4 and python2.5
numpy 1.0.1 and 1.0.2
matplotlib 0.90 and 0.87.7
We got the same error with all of those combinations. The error 
message was an "ImportError: DLL load failed", when doing "from 
matplotlib._ns_transforms import *". It is the same problem that 
others reported earlier:
	http://www.mail-archive.com/mat...@li.../ 
msg02841.html
All of our installs were from the .exe Windows binaries, not source 
compilations, and in all cases import numpy, numpy.test(1), and 
import matplotlib produced no errors.
Note that we have not installed either Numeric or numarray, but we 
tried adding Numeric and it didn't help.
Has anyone found a solution? We'd like to upgrade to a matplotlib 
later that 0.87.5 to get matplotlib.patches.Ellipse().
Thanks,
-- Paul

On Tuesday 08 May 2007 20:55, Jeff Whitaker wrote:
> Jesper: You might consider just deleting the figure elements that you
> don't want to re-use (instead of figuring out everything you do want to
> re-use).
I have an axes instance on which I have plotted my map decorations 
(coastlines, meridians and so on). I would like to reuse that axes. On the 
same axes* I would like to make filled contour plot and potentially a quiver 
plot. These should be discarded after saving the plot to a file.
I know I have to discard part of the "collection" variable in the axes 
instance. But it seems like other things are changed as well. Is it safe to 
ignore these and simply do something like this:
nsave = len(ax.collections)
...do and save plot...
ax.collections = ax.collections[0:nsave]
or would you recommend doing something else?
Regards,
Jesper
*) I have tried to make an extra axes for the contour plot itself but it seems 
like matplotlib does not create an extra axes when it is the same size as an 
existing axes.

On 5/10/07, Lionel Roubeyrie <lro...@li...> wrote:
> It's PyQt3, I hope using PyQt4 in a future release. I'm also going to make
> some changes in the windfreq part, don't sure to keep missing values is good
> for the computation. For the moment, my major problem is the legend, because
> polar axes don't have one, so I must use figlegend, and it's not very
> useable.
I recently added legend support for polar to matplotlib svn.
JDH

Hi all,
How can I increase the distance between the title and the spy plot ?
I have used
figure(1)
subplot(121)
spy(M_bc)
title(r'$M_{red}$')
subplot(122)
title(r'$K_{red}$')
spy(K_bc)
for the attached figure.
Nils
 

Hi,
> I noticed on your Web Lab website that you suggest to download the 
> Enthouhht edition. Although I suggest that to my students for class 
> as well, we are currently running into some problems as the latest 
> Enthought edition still includes an old numpy (version less than 
> 1.0). Some important changes were made just before the 1.0 release, 
> so please be aware of that.
>
> Now that we are on the topic, does anybody have good enough 
> connections with Enthought to convince them to update the 
> distribution? I can toss-in a box of donuts if that would help,
I believe they are working on it, but they're shifting their approach 
to easily install the individual packages of the distribution 
individually/together/whatever by using eggs. It's good for the long 
term, but I guess unfortunate for people looking for the newest/ 
easiest install in the interim.
Still, though, I think I've heard people still having some success by 
just installing the latest enthought edition. After that, you can 
pull down numpy/scipy/matplotlib from svn and install each over the 
old version in enthought.
Lastly, Stef Mientki has apparently whipped up a "Portable SciPy" 
that is meant to be a one-click installer on windows if doing the 
above is unsatisfactory. Here's his announcement from the SciPy 
mailing list.
"""
Portable SciPy, is an easy installer of SciPy for M$ windows users.
For this moment, you can find the description page, with all links here
http://oase.uci.kun.nl/~mientki/data_www/pic/jalcc/python/ 
portable_scipy.html
For future use, it's advised to always use my redirector page
 http://pic.flappie.nl/
The simple method described here,
can be used to create any set of Python packages + other programs,
with just a few lines of code (example available).
have fun,
and let me hear what you think of it.
""""
Lastly, I can't really vouch for any of these things since I don't 
use Windows, hopefully they work easily enough.
-steve

On Wednesday 09 May 2007 12:19:24 pm Johann Cohen-Tanugi wrote:
> I am creating a "big" drawing ( figure(figsize=(16,20)) ), and when I
> try to save it in eps/ps form, it mishandle the overall size and only
> save a portion of the drawing. Saving in png or jpg works fine though.
> Any idea?
On my system, you cant create a figure that is larger than the monitor size, 
unless I use a non-gui backend like agg or ps. It doesnt matter whether I 
save an eps, png, or jpg. Also, make sure your postscript viewer is not 
truncating the page due to an inappropriate page size setting.
Darren

I noticed on your Web Lab website that you suggest to download the Enthouhht
edition. Although I suggest that to my students for class as well, we are
currently running into some problems as the latest Enthought edition still
includes an old numpy (version less than 1.0). Some important changes were
made just before the 1.0 release, so please be aware of that.
Now that we are on the topic, does anybody have good enough connections with
Enthought to convince them to update the distribution? I can toss-in a box
of donuts if that would help,
Mark

It's PyQt3, I hope using PyQt4 in a future release. I'm also going to make=
=20
some changes in the windfreq part, don't sure to keep missing values is goo=
d=20
for the computation. For the moment, my major problem is the legend, becaus=
e=20
polar axes don't have one, so I must use figlegend, and it's not very=20
useable.
Thanks for your compliments
Le Mercredi 09 Mai 2007 20:06, Christopher Barker a =E9crit=A0:
> Lionel Roubeyrie wrote:
> > PS: I'm working on a graphical interface for windrose, don't know when I
> > can finalyse it, maybe for the next month. I set a screenshot here.
>
> pretty cool. Is that pyGTK or wxPython?
>
> Keep us apprised, this looks nice, and I think I could use it (though I
> need wxPython)
>
> -Chris
=2D-=20
Lionel Roubeyrie - lro...@li...
Chag=E9 d'=E9tudes et de maintenance
LIMAIR - la Surveillance de l'Air en Limousin
http://www.limair.asso.fr

hello,
I am creating a "big" drawing ( figure(figsize=(16,20)) ), and when I 
try to save it in eps/ps form, it mishandle the overall size and only 
save a portion of the drawing. Saving in png or jpg works fine though.
Any idea?
Thanks,
Johann