matplotlib-users Mailing List for matplotlib (Page 8)

Jeff Mangum wrote:
> 
> 
> 
> Robert Kern-2 wrote:
>> 
>> Jeff Mangum wrote:
>>> Hello,
>>> 
>>> I am having a problem installing matplotlib 0.93.3 from egg on Mac OSX 
>>> 10.5.5.  Even though I have numpy 1.2.1 installed in 
>>> /Library/Frameworks/..., the egg insists on using an older version of 
>>> numpy (1.0.4) in /opt/local/lib/python2.5/site-packages (which must have 
>>> been delivered with the OS). 
>> 
>> No, /opt/local is MacPorts territory.
>> 
>>> How can I tell the egg where to find the 
>>> proper version of numpy?  Thanks!
>> 
>> Are you sure you are using the same versions of Python to run and install
>> both 
>> of these?
>> 
>> -- 
>> Robert Kern
>> 
>> "I have come to believe that the whole world is an enigma, a harmless
>> enigma
>>   that is made terrible by our own mad attempt to interpret it as though
>> it had
>>   an underlying truth."
>>    -- Umberto Eco
>> 
>> 
>> -------------------------------------------------------------------------
>> This SF.Net email is sponsored by the Moblin Your Move Developer's
>> challenge
>> Build the coolest Linux based applications with Moblin SDK & win great
>> prizes
>> Grand prize is a trip for two to an Open Source event anywhere in the
>> world
>> http://moblin-contest.org/redirect.php?banner_id=100&url=/
>> _______________________________________________
>> Matplotlib-users mailing list
>> Mat...@li...
>> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>> 
>> 
> 
> Pretty sure that I am using the same versions of python to run and
> install.  Here is the trace when I try to install matplotlib:
> 
> torgo:Desktop jmangum$ easy_install
> ./matplotlib-0.98.3-py2.5-macosx-10.3.egg
> Processing matplotlib-0.98.3-py2.5-macosx-10.3.egg
> removing
> '/Library/Python/2.5/site-packages/matplotlib-0.98.3-py2.5-macosx-10.3.egg'
> (and everything under it)
> creating
> /Library/Python/2.5/site-packages/matplotlib-0.98.3-py2.5-macosx-10.3.egg
> Extracting matplotlib-0.98.3-py2.5-macosx-10.3.egg to
> /Library/Python/2.5/site-packages
> matplotlib 0.98.3 is already the active version in easy-install.pth
> 
> Installed
> /Library/Python/2.5/site-packages/matplotlib-0.98.3-py2.5-macosx-10.3.egg
> Processing dependencies for matplotlib==0.98.3
> Searching for matplotlib==0.98.3
> Reading http://pypi.python.org/simple/matplotlib/
> Reading http://matplotlib.sourceforge.net
> Reading
> https://sourceforge.net/project/showfiles.php?group_id=80706&package_id=278194
> Reading
> https://sourceforge.net/project/showfiles.php?group_id=80706&package_id=82474
> Reading http://sourceforge.net/project/showfiles.php?group_id=80706
> Best match: matplotlib 0.98.3
> Downloading
> http://downloads.sourceforge.net/matplotlib/matplotlib-0.98.3.tar.gz?modtime=1217773039&big_mirror=0
> Processing matplotlib-0.98.3.tar.gz
> Running matplotlib-0.98.3/setup.py -q bdist_egg --dist-dir
> /var/folders/P-/P-e5QX4vHmezGAqqg2iH1k++C86/-Tmp-/easy_install-wEpzrD/matplotlib-0.98.3/egg-dist-tmp-Op3Pcx
> ============================================================================
> BUILDING MATPLOTLIB
>             matplotlib: 0.98.3
>                 python: 2.5.1 (r251:54863, Apr 15 2008, 22:57:26)  [GCC
>                         4.0.1 (Apple Inc. build 5465)]
>               platform: darwin
> 
> REQUIRED DEPENDENCIES
>                         * numpy 1.1 or later is required; you have 1.0.1
> error: Setup script exited with 1
> 
> 
> Thanks for your help!
> 
> -- Jeff
> 
> 

Sorry, but indeed the easy_install is finding the "stock" Apple python
rather than the python.org version I recently installed:

============================================================================
BUILDING MATPLOTLIB
            matplotlib: 0.98.3
                python: 2.5.1 (r251:54863, Apr 15 2008, 22:57:26)  [GCC
                        4.0.1 (Apple Inc. build 5465)]
              platform: darwin

REQUIRED DEPENDENCIES
                        * numpy 1.1 or later is required; you have 1.0.1
error: Setup script exited with 1
torgo:Desktop jmangum$ which python
/Library/Frameworks/Python.framework/Versions/Current/bin/python
torgo:Desktop jmangum$ python
Python 2.5.2 (r252:60911, Feb 22 2008, 07:57:53) 
[GCC 4.0.1 (Apple Computer, Inc. build 5363)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> 

How can I instruct the matplotlib install to find the appropriate python
install?  Thanks!

-- Jeff

-- 
View this message in context: http://www.nabble.com/matplotlib-egg-finds-wrong-version-of-numpy-tp20544369p20570792.html
Sent from the matplotlib - users mailing list archive at Nabble.com.

Robert Kern-2 wrote:
> 
> Jeff Mangum wrote:
>> Hello,
>> 
>> I am having a problem installing matplotlib 0.93.3 from egg on Mac OSX 
>> 10.5.5.  Even though I have numpy 1.2.1 installed in 
>> /Library/Frameworks/..., the egg insists on using an older version of 
>> numpy (1.0.4) in /opt/local/lib/python2.5/site-packages (which must have 
>> been delivered with the OS). 
> 
> No, /opt/local is MacPorts territory.
> 
>> How can I tell the egg where to find the 
>> proper version of numpy?  Thanks!
> 
> Are you sure you are using the same versions of Python to run and install
> both 
> of these?
> 
> -- 
> Robert Kern
> 
> "I have come to believe that the whole world is an enigma, a harmless
> enigma
>   that is made terrible by our own mad attempt to interpret it as though
> it had
>   an underlying truth."
>    -- Umberto Eco
> 
> 
> -------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's
> challenge
> Build the coolest Linux based applications with Moblin SDK & win great
> prizes
> Grand prize is a trip for two to an Open Source event anywhere in the
> world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
> 
> 

Pretty sure that I am using the same versions of python to run and install. 
Here is the trace when I try to install matplotlib:

torgo:Desktop jmangum$ easy_install
./matplotlib-0.98.3-py2.5-macosx-10.3.egg
Processing matplotlib-0.98.3-py2.5-macosx-10.3.egg
removing
'/Library/Python/2.5/site-packages/matplotlib-0.98.3-py2.5-macosx-10.3.egg'
(and everything under it)
creating
/Library/Python/2.5/site-packages/matplotlib-0.98.3-py2.5-macosx-10.3.egg
Extracting matplotlib-0.98.3-py2.5-macosx-10.3.egg to
/Library/Python/2.5/site-packages
matplotlib 0.98.3 is already the active version in easy-install.pth

Installed
/Library/Python/2.5/site-packages/matplotlib-0.98.3-py2.5-macosx-10.3.egg
Processing dependencies for matplotlib==0.98.3
Searching for matplotlib==0.98.3
Reading http://pypi.python.org/simple/matplotlib/
Reading http://matplotlib.sourceforge.net
Reading
https://sourceforge.net/project/showfiles.php?group_id=80706&package_id=278194
Reading
https://sourceforge.net/project/showfiles.php?group_id=80706&package_id=82474
Reading http://sourceforge.net/project/showfiles.php?group_id=80706
Best match: matplotlib 0.98.3
Downloading
http://downloads.sourceforge.net/matplotlib/matplotlib-0.98.3.tar.gz?modtime=1217773039&big_mirror=0
Processing matplotlib-0.98.3.tar.gz
Running matplotlib-0.98.3/setup.py -q bdist_egg --dist-dir
/var/folders/P-/P-e5QX4vHmezGAqqg2iH1k++C86/-Tmp-/easy_install-wEpzrD/matplotlib-0.98.3/egg-dist-tmp-Op3Pcx
============================================================================
BUILDING MATPLOTLIB
            matplotlib: 0.98.3
                python: 2.5.1 (r251:54863, Apr 15 2008, 22:57:26)  [GCC
                        4.0.1 (Apple Inc. build 5465)]
              platform: darwin

REQUIRED DEPENDENCIES
                        * numpy 1.1 or later is required; you have 1.0.1
error: Setup script exited with 1


Thanks for your help!

-- Jeff

-- 
View this message in context: http://www.nabble.com/matplotlib-egg-finds-wrong-version-of-numpy-tp20544369p20570681.html
Sent from the matplotlib - users mailing list archive at Nabble.com.

Thank you Jae-Joon,

JL> get_frame() is deprecated (it is supposed to show you a
JL> DeprecationWarning). Axes class now has a "frame" property, so
JL> gca().frame.set_linewidth(2) should work.
JL> Currently get_frame() returns Axes.patch which is used to draw a
JL> axes background.

-- 
Christopher Brown, Ph.D.
Department of Speech and Hearing Science
Arizona State University

On Tue, Nov 18, 2008 at 9:20 AM, John Hunter <jd...@gm...> wrote:
>
>
> There are two related problems here: one easier, one harder.  We can
> start with the easier one.  The easy one is to have a "detachable
> axis".  Right now we have one on the right and one on the left for the
> x-axis.  We may as well generalize this to allow folks to have as many
> as they want and put them whereever they want (center, offset farther
> to the left, etc.).  It would be nice to support a different set of
> tick locators and formatters for each axis, eg you could have a left
> axis in dollars and a right axis in euros.  Note this is not the same
> thing as having a different scale, in the sense of the two_scales
> example, this is just two different formatting of the same scale.  The
> harder problem is supporting an arbitrary number of axis instances
> each with its own transform, etc.  Eg generalizing the two_scales
> example.  I suggest you punt on that one for now.
>

I spent some time on this a while ago. And I was able to partially
implement what John described above. I'm attaching my code, but note
that it is more like a proof of the concept, than a practical code.
What it can do now is to place an axis at some arbitrary position (can
be either an axes coordinate or a data coordinate).

In brief, the code is consisted of GridHelper and AxisLine. My
intention was an Axes object has a single GridHelper instance and
multiple AxisLine instances.
I wanted to make some framework which can work with any arbitrary
transformation (for example, see
http://www.atnf.csiro.au/people/mcalabre/WCS/PGSBOX/index.html), and
this affected some of my design. I only had a partial success on this,
and then swamped by other things.

So, I'm also willing to help Mike on this. And even though Mike start
with easy things, I wish the design is flexible enough to be extended
to support an arbitrary transformation  in a future.

-JJ

get_frame() is deprecated (it is supposed to show you a DeprecationWarning).
Axes class now has a "frame" property, so gca().frame.set_linewidth(2)
should work.
Currently get_frame() returns Axes.patch which is used to draw a axes
background.

-JJ





On Tue, Nov 18, 2008 at 12:24 PM, Christopher Brown <c-...@as...> wrote:
> Hi Users,
>
> I use the following to adjust the line width of the ticks and bounding box:
>
> for tl in plt.gca().get_xticklines() + plt.gca().get_yticklines():
>     tl.set_markeredgewidth(2)
> plt.gca().get_frame().set_linewidth(2)
>
> I'm not sure when it happened, but sometime relatively recently the
> get_frame().set_linewidth() command stopped working, and only the ticks
> are getting updated.
>
> In [5]: matplotlib.__version__
> Out[5]: '0.98.3'
>
> In [6]: os.name
> Out[6]: 'nt'
>
> Any ideas about what I'm doing wrong?
>
> --
> Christopher Brown, Ph.D.
> Department of Speech and Hearing Science
> Arizona State University
>
> -------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
> Build the coolest Linux based applications with Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

Hi,

A vertical line on the x axis of a semilogy plot will be in the correct 
position, but when saved with the save button of the toolbar it will be 
placed in an incorrect position, although savefig will do the right 
thing. Furthermore, zooming will not move the axvline correctly.

Matplotlib version: 0.98.3

Example script:
-----------------------------------------

import matplotlib.pyplot as pl
import numpy as np

x = np.linspace(0.0,1.0,100)

pl.semilogy(x,x**2)
pl.axvline(x=0.5,ls='--',color='k')
pl.show()
#pl.savefig("saved_image.png")
-----------------------------------------

image.png was saved using the toolbar, saved_image.png using savefig.

Regards,
João Silva

Hi Users,

I use the following to adjust the line width of the ticks and bounding box:

for tl in plt.gca().get_xticklines() + plt.gca().get_yticklines():
     tl.set_markeredgewidth(2)
plt.gca().get_frame().set_linewidth(2)

I'm not sure when it happened, but sometime relatively recently the 
get_frame().set_linewidth() command stopped working, and only the ticks 
are getting updated.

In [5]: matplotlib.__version__
Out[5]: '0.98.3'

In [6]: os.name
Out[6]: 'nt'

Any ideas about what I'm doing wrong?

-- 
Christopher Brown, Ph.D.
Department of Speech and Hearing Science
Arizona State University

Hello,

Thanks for updating the files API_CHANGES & MIGRATION.txt, I have now 
read them and fear that a problem I have is due to calling:

    f = figure()
    h = f.bbox.height()

Which should change to:

    f = figure()
    h = f.bbox.height

Is the expectation that I should write something like this instead:

    if f.bbox.height.hasattr("__call__"):
        h = f.bbox.height()
    else:
        h = f.bbox.height


It lacks something aesthetically. How are people handling this? Do you 
insist on users having a pre- or post- 0.98 version? Have I missed a 
trick? Perhaps some kind of monkeypatch that can be applied so that old 
versions would appear to look like the new versions?

Thanks,

Jon

While I though this would be fairly
easy, I've yet to find a decent solution.  I'm interested in using the
RangeSelector or standard zoom feature along with an autoscaling
function.  In the example below there are two gaussian profiles created
in a matrix.  I would like to have the behavior in which when I zoom in
that the Z scale is reordered to account for the new range of data.  Is
my only solution to take the limits returned by a RangeSelector event
and extract a sub-matix from the original data and replot each time I
zoom in or out or is there a more elegant solution?  Is there a way to also extend such functionality to contour, pcolor, etc.?



Cheers,



Brian




######paste in ipython###########################################


from numpy import *


from matplotlib.widgets import RectangleSelector


import pylab as P


def gaussian(height, center_x, center_y, width_x, width_y):


    """Returns a gaussian function with the given parameters"""


    width_x = float(width_x)


    width_y = float(width_y)


    return lambda x,y: height*exp(-(((center_x-x)/width_x)**2+((center_y-y)/width_y)**2)/2)


def line_select_callback(event1, event2):


    'event1 and event2 are the press and release events'


    x1, y1 = event1.xdata, event1.ydata


    x2, y2 = event2.xdata, event2.ydata


    print "(%3.2f, %3.2f) --> (%3.2f, %3.2f)"%(x1,y1,x2,y2)


    print " The button you used were: ",event1.button, event2.button


        


        


cla()   


# Create the gaussian data


Xin, Yin = mgrid[0:201, 0:201]


Xin2, Yin2 = mgrid[0:201, 0:201]


data = gaussian(20, 100, 100, 20, 40)(Xin, Yin) + random.random(Xin.shape)


data2 = gaussian(10, 10, 10, 2, 4)(Xin2, Yin2) + random.random(Xin.shape)


P.imshow(data+data2, cmap=cm.jet)


current_ax = gca()


LS = RectangleSelector(current_ax, line_select_callback, drawtype='box',useblit=True,\


                       minspanx=5,minspany=5,spancoords='pixels')




P.show()


 



      

John Hunter wrote:
> On Tue, Nov 18, 2008 at 9:41 AM, Hrafnkell Pálsson <hr...@hi...> wrote:
>   
>> Ok, I tried your last suggestion and sure enough it worked.
>> But it turns out to solve only half of my problem. I'd like to be able to
>> restore the background (using the Agg backend) and then use it further, i.e.
>> plot on it, without it disappearing.
>>
>> Elaborating on my real use case, what I want to do is to make a Basemap
>> instance, plot the coastlines and other things that are constant in time
>> (put marker on the maps for cities, watermark the figure, draw meridians and
>> parallels and so on) and save that as a background.
>> I would then restore this background and plot "the weather" (open and filled
>> contours, quivers and barbs) onto it. I typically have data for some dozens
>> of timesteps, so I'd be restoring the background many times. Each time I'd
>>     
>
> This should work fine as long as you make the figure frame and axes
> frame invisible.  Basically, you need to create the figure background,
> and the axes background with all the grid lines, tick labels,
> coastlines, etc, save it, and then for the new figure make sure all
> the stuff you have saved is turned off.  I am not a basemap user, but
> I know all this stuff is ultimately exposed.  Eg for a plain vanilla
> axes, you would do
>
>
>   fig1 = figure(frameon=False)
>   ax1 = fig1.add_subplot(111)
>
>
>   fig = figure(frameon=False)
>   ax = fig.add_subplot(111, frameon=false, xticks=[], yticks=[])
>   fig.canvas.restore_region(background)
>
>
> This will turn off the rendering of both the background rectangle as
> well as all the ticks and their associated labels, ticklines and
> gridlines.  Perhaps Jeff can advise you vis-a-vis the basemap api what
> the equivalent is
>
>
>   
Don't see why this wouldn't work fine with basemap, as long as you don't 
call the drawmapboundary method (which will set the axes frame on).

-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-113
Boulder, CO, USA 80303-3328 Web    : http://tinyurl.com/5telg

On Tue, Nov 18, 2008 at 9:41 AM, Hrafnkell Pálsson <hr...@hi...> wrote:
>
> Ok, I tried your last suggestion and sure enough it worked.
> But it turns out to solve only half of my problem. I'd like to be able to
> restore the background (using the Agg backend) and then use it further, i.e.
> plot on it, without it disappearing.
>
> Elaborating on my real use case, what I want to do is to make a Basemap
> instance, plot the coastlines and other things that are constant in time
> (put marker on the maps for cities, watermark the figure, draw meridians and
> parallels and so on) and save that as a background.
> I would then restore this background and plot "the weather" (open and filled
> contours, quivers and barbs) onto it. I typically have data for some dozens
> of timesteps, so I'd be restoring the background many times. Each time I'd

This should work fine as long as you make the figure frame and axes
frame invisible.  Basically, you need to create the figure background,
and the axes background with all the grid lines, tick labels,
coastlines, etc, save it, and then for the new figure make sure all
the stuff you have saved is turned off.  I am not a basemap user, but
I know all this stuff is ultimately exposed.  Eg for a plain vanilla
axes, you would do


  fig1 = figure(frameon=False)
  ax1 = fig1.add_subplot(111)


  fig = figure(frameon=False)
  ax = fig.add_subplot(111, frameon=false, xticks=[], yticks=[])
  fig.canvas.restore_region(background)


This will turn off the rendering of both the background rectangle as
well as all the ticks and their associated labels, ticklines and
gridlines.  Perhaps Jeff can advise you vis-a-vis the basemap api what
the equivalent is

Ok, I tried your last suggestion and sure enough it worked.
But it turns out to solve only half of my problem. I'd like to be able to
restore the background (using the Agg backend) and then use it further, i.e.
plot on it, without it disappearing.

Elaborating on my real use case, what I want to do is to make a Basemap
instance, plot the coastlines and other things that are constant in time
(put marker on the maps for cities, watermark the figure, draw meridians and
parallels and so on) and save that as a background.
I would then restore this background and plot "the weather" (open and filled
contours, quivers and barbs) onto it. I typically have data for some dozens
of timesteps, so I'd be restoring the background many times. Each time I'd
plot a particular weather (corresponding to a particular timestep) onto the
background I'd save that figure, so in the end I have some dozens of
figures.
Sliding the slider beneath the pictures at  http://www.belgingur.is
http://www.belgingur.is  should give a good idea of what I mean (notice that
the website is also in English).

A somewhat simplified example of what I'd like to do (but it catches the
essence of it) follows.
The problem that arises is that when I try to plot onto the background it
disappears and only what I tried to plot onto it remains. So figure2.png
contains the coastline of Iceland while figure3.png contains only the
quivers, the coastline has disappeared.

import matplotlib
matplotlib.use('Agg')

import matplotlib.pyplot as plt
import numpy
from mpl_toolkits.basemap import Basemap

plt.close('all')

basemap = Basemap(llcrnrlat=62.8, 
                  llcrnrlon=-24.8, 
                  urcrnrlat=66.7,
                  urcrnrlon=-12.4,
                  lat_0=65.0, 
                  lon_0=-19.5)

# make sure we have all the same properties on all figs
figprops = dict(figsize=(8,6), dpi=100, facecolor='white')
fig1 = plt.figure(1, **figprops)

basemap.drawcoastlines()

fig1.canvas.draw()
background = fig1.canvas.copy_from_bbox(fig1.bbox)

# turn the frame off or it will overwrite the background
fig2 = plt.figure(2, frameon=False, **figprops)
fig2.canvas.restore_region(background)
fig2.savefig('figure2.png', dpi=100)

# turn the frame off or it will overwrite the background
fig3 = plt.figure(3, frameon=False, **figprops)
fig3.canvas.restore_region(background)
# create a lon-lat grid for plotting quivers
n, m = 10, 5
latitudes = numpy.resize(numpy.linspace(63.5, 65.0, n), (n, m))
longitudes = numpy.transpose(numpy.resize(numpy.linspace(-22, -14, n),(n,
m)))
x, y = basemap(longitudes, latitudes)
# create the u and v components of the quivers
u = numpy.resize([5], (n, m))
v = numpy.resize([5], (n, m))
# since keyword ax is not supplied the current axes instance is used, i.e.
the axes of fig3
basemap.quiver(x, y, u, v)
fig3.savefig('figure3.png', dpi=100)

Hrafnkell
-- 
View this message in context: http://www.nabble.com/Save-a-plot-background-tp20519596p20562028.html
Sent from the matplotlib - users mailing list archive at Nabble.com.

While I though this would be fairly
easy, I've yet to find a decent solution.  I'm interested in using the
RangeSelector or standard zoom feature along with an autoscaling
function.  In the example below there are two gaussian profiles created
in a matrix.  I would like to have the behavior in which when I zoom in
that the Z scale is reordered to account for the new range of data.  Is
my only solution to take the limits returned by a RangeSelector event
and extract a sub-matix from the original data and replot each time I
zoom in or out or is there a more elegant solution?  Is there a way to also extend such functionality to contour, pcolor, etc.?



Cheers,



Brian




######paste in ipython###########################################


from numpy import *


from matplotlib.widgets import RectangleSelector


import pylab as P


def gaussian(height, center_x, center_y, width_x, width_y):


    """Returns a gaussian function with the given parameters"""


    width_x = float(width_x)


    width_y = float(width_y)


    return lambda x,y: height*exp(-(((center_x-x)/width_x)**2+((center_y-y)/width_y)**2)/2)


def line_select_callback(event1, event2):


    'event1 and event2 are the press and release events'


    x1, y1 = event1.xdata, event1.ydata


    x2, y2 = event2.xdata, event2.ydata


    print "(%3.2f, %3.2f) --> (%3.2f, %3.2f)"%(x1,y1,x2,y2)


    print " The button you used were: ",event1.button, event2.button


        


        


cla()   


# Create the gaussian data


Xin, Yin = mgrid[0:201, 0:201]


Xin2, Yin2 = mgrid[0:201, 0:201]


data = gaussian(20, 100, 100, 20, 40)(Xin, Yin) + random.random(Xin.shape)


data2 = gaussian(10, 10, 10, 2, 4)(Xin2, Yin2) + random.random(Xin.shape)


P.imshow(data+data2, cmap=cm.jet)


current_ax = gca()


LS = RectangleSelector(current_ax, line_select_callback, drawtype='box',useblit=True,\


                       minspanx=5,minspany=5,spancoords='pixels')




P.show()


 



      

Hi,

It seems that the Legend command doesnt function properly when using
errorbar to plot. It includes extra lines for the errorbars themselves.

I found out this is an old problem that have been fixed before??

http://osdir.com/ml/python.matplotlib.devel/2006-05/msg00038.html


Regards,
Soren

Hi,

Has anyone else noticed that using the pan tool is very slow when the data
is plotted on a double log scale, or single log scale for that matter.. it's
smooth as silk when the plots are in lin-lin.

Is there a way to make it faster or is it a bug?

Regards,
Soren

On Tue, Nov 18, 2008 at 4:25 AM, Mike Hansen <mh...@gm...> wrote:

> What would you recommend as the best way going about this?  I'm
> willing to put some work in on this.  There was someone back in July
> who posted some code moving in this direction, but it didn't feel like
> it was the right way to do things.  He didn't get any response from
> the list.

That was Tony's work, and it is a shame we didn't give him more
feedback.  It is definitely an important enhancement.  Sometimes the
things that are harder get marked with a "star" in my inbox, and the
things that are easier get tackled right away, and the harder things
get pushed down the queue.  Which is why I encourage folks to make
gentle reminders: the silence is usually not lack of interest, but
lack of time.

You should start by taking a look at where Tony left off.  I will
review his code too, and when we have both had a chance to digest it,
we can discuss a plan of attack.
 http://www.nabble.com/Modifications-to-axis.py-to-allow-more-flexibility-in-positioning-axis-lines%2C-ticks%2C-etc.-td18573959.html#a18573959

Tony may be interested in contributing too.  Reading over his email,
one of the things that stumped him was trying to get this done right
and maintain backwards compatibility.  That is admirable, and we
should try to achieve that where possible, but I don't think it is a
requirement.  This is a major feature enhancement, and is one of the
things that have kept me from releasing mpl as 1.0, so if we need some
breakage to get there we can live with it.  Not too many people tweak
the axis layout, in part because there is not much to tweak.  Also,
most use the Axes front end helper methods (eg ax.get_xticklabels())
which will insulate them from a reorg at the Axis layer.

There are two related problems here: one easier, one harder.  We can
start with the easier one.  The easy one is to have a "detachable
axis".  Right now we have one on the right and one on the left for the
x-axis.  We may as well generalize this to allow folks to have as many
as they want and put them whereever they want (center, offset farther
to the left, etc.).  It would be nice to support a different set of
tick locators and formatters for each axis, eg you could have a left
axis in dollars and a right axis in euros.  Note this is not the same
thing as having a different scale, in the sense of the two_scales
example, this is just two different formatting of the same scale.  The
harder problem is supporting an arbitrary number of axis instances
each with its own transform, etc.  Eg generalizing the two_scales
example.  I suggest you punt on that one for now.

If you really want to keep this easy, you can forget all about the
locators and formatters and just focus on allowing the user to place
the axis lines wherever they want, using the standard Axis locator and
formatter for all the axis lines.  That is probably the place to start
-- we can always try to beef it up later.  That part should be fairly
easy.  Another enhancement that would be nice would be to support a
TextCollection for the tick labels.  Drawing the tick labels is
actually a pretty big bottleneck in the whole plot rendering time, and
since the properties are naturally homogeneous (eg the rotation, the
fontsize) we could probably save some time using a collection.

JDH

On Tue, Nov 18, 2008 at 3:41 AM, Benjamin Bardiaux <bar...@gm...> wrote:
> Dear matplotlib users,
>
> I'l looking for the files
>
> http://matplotlib.sf.net/API_CHANGES
> http://matplotlib.sourceforge.net/MIGRATION.txt
>
> that seem to have disappear from the new sf website.

Thanks -- I've re-added these and put links to them in the "need help"
sidebar at http://matplotlib.sf.net.  You may need to refresh your
browser to see the new links.  We should include these in the rest
docs proper, so they are exposed to the sphinx search tool, but I just
went with the quick and easy for now.

JDH

Hi Eric,

> From: Eric Firing <efiring@ha...> - 2008-11-18 00:08
> Unfortunately, the present mpl architecture does not lend itself to this
> sort of flexibility with respect to axis locations. It has been on the
> wish list for a long time, and it can be done manually, but to make it
> an easy-to-use mpl option will take some surgery and reconstruction on
> mpl's internals.

What would you recommend as the best way going about this?  I'm
willing to put some work in on this.  There was someone back in July
who posted some code moving in this direction, but it didn't feel like
it was the right way to do things.  He didn't get any response from
the list.

--Mike

Dear matplotlib users,

I'l looking for the files

http://matplotlib.sf.net/API_CHANGES
http://matplotlib.sourceforge.net/MIGRATION.txt

that seem to have disappear from the new sf website.

Many thanks !

Benjamin

On Mon, Nov 17, 2008 at 8:05 AM, Hrafnkell Pálsson <hr...@hi...> wrote:
>
> Hi
>
> I tried you suggestions but it didn't work out for me.
> In the following code I try to save the axes and the grid from figure1 into
> buffer and then restore it on figure2 but figure2.png turns out to be of an
> empty canvas.
>


OK, ever since Jae-Joon clued me into to the importance of the savefig
call, I realized there is a fairly easy solution for your problem.
The problem you are having is that savefig is redrawing the "figure
frame" which is essentially just filling the rectangular background of
the figure canvas.  This is obscuring the background you are trying to
restore.  SO just turn the frame off, and your background will be
revealed in its place.  Here is the example code:

import matplotlib
matplotlib.use('Agg')

import matplotlib.pyplot as plt

plt.close('all')

# make sure we have all the same properties on all figs
figprops = dict(figsize=(8,6), dpi=100, facecolor='white')

fig1 = plt.figure(1, **figprops)
ax1 = fig1.add_subplot(111)
ax1.grid()
fig1.canvas.draw()
background = fig1.canvas.copy_from_bbox(fig1.bbox)
fig1.savefig('figure1.png', dpi=100)


# turn the frame off or it will overwrite the background
fig2 = plt.figure(2, frameon=False, **figprops)
fig2.canvas.restore_region(background)
fig2.savefig('figure2.png', dpi=100)

 
 

Matplotlibbers,

While I though this would be fairly
easy, I've yet to find a decent solution.  I'm interested in using the
RangeSelector or standard zoom feature along with an autoscaling
function.  In the example below there are two gaussian profiles created
in a matrix.  I would like to have the behavior in which when I zoom in
that the Z scale is reordered to account for the new range of data.  Is
my only solution to take the limits returned by a RangeSelector event
and extract a sub-matix from the original data and replot each time I
zoom in or out or is there a more elegant solution?  Is there a way to also extend such functionality to contour, pcolor, etc.?



Cheers,



Brian




######paste in ipython###########################################


from numpy import *


from matplotlib.widgets import RectangleSelector


import pylab as P


def gaussian(height, center_x, center_y, width_x, width_y):


    """Returns a gaussian function with the given parameters"""


    width_x = float(width_x)


    width_y = float(width_y)


    return lambda x,y: height*exp(-(((center_x-x)/width_x)**2+((center_y-y)/width_y)**2)/2)


def line_select_callback(event1, event2):


    'event1 and event2 are the press and release events'


    x1, y1 = event1.xdata, event1.ydata


    x2, y2 = event2.xdata, event2.ydata


    print "(%3.2f, %3.2f) --> (%3.2f, %3.2f)"%(x1,y1,x2,y2)


    print " The button you used were: ",event1.button, event2.button


        


        


cla()   


# Create the gaussian data


Xin, Yin = mgrid[0:201, 0:201]


Xin2, Yin2 = mgrid[0:201, 0:201]


data = gaussian(20, 100, 100, 20, 40)(Xin, Yin) + random.random(Xin.shape)


data2 = gaussian(10, 10, 10, 2, 4)(Xin2, Yin2) + random.random(Xin.shape)


P.imshow(data+data2, cmap=cm.jet)


current_ax = gca()


LS = RectangleSelector(current_ax, line_select_callback, drawtype='box',useblit=True,\


                       minspanx=5,minspany=5,spancoords='pixels')




P.show()


 



      

On Mon, Nov 17, 2008 at 1:50 PM, Jae-Joon Lee <lee...@gm...> wrote:
> I think the savefig() command calls draw() internally, doesn't it?

Ahh yes.  The copy_from_bbox / restore_region was conceived as an aid
to animation and I don't think I've used it in the context of savefig
where none of the artists are involved in the restored region.  Ie, in
a typical animation use case, you draw the axes, background and ticks,
store the region, and then animate a line on top of it.  When you go
to savefig, the axes background is redrawn  because the associated
artists already reside in the figure, along with the line in its final
state.

It should be fairly easy to expose the agg BufferRegion object in the
artist API, so you could grab a buffer region from one canvas, bless
it as an artist, and add it to the figure or another figure, with the
x and y location  attrs exposed so it could be easily moved.  In this
case you would get the zorder for free.  This would help some, and
would certainly address the case at hand, but the really big win would
probably be something more like the chaco model, where each zorder is
rendered onto a different rendering buffer, with a fairly
straight-forward way to just re-render certain planes.  This would
certainly be harder.

Could you elaborate a bit on your use case, which will help me in
thinking about how this should be done?

JDH

jas...@cr... wrote:
> There is some work going on in Sage (http://www.sagemath.org) to more
> fully use matplotlib on a higher level than is currently done.  One
> issue that is coming up is using matplotlib to make a plot with axis
> lines crossing in the center of the plot, like is shown in the plot at
> http://sagemath.org/tour-education.html
> 
> Is this easy to do in matplotlib (preferably with the stock axes
> object)?  It seems that there is not an example illustrating an axes of 
> this form and I couldn't find any mention in the documentation of such a 
> feature.

Unfortunately, the present mpl architecture does not lend itself to this 
sort of flexibility with respect to axis locations.  It has been on the 
wish list for a long time, and it can be done manually, but to make it 
an easy-to-use mpl option will take some surgery and reconstruction on 
mpl's internals.

Eric

There is some work going on in Sage (http://www.sagemath.org) to more
fully use matplotlib on a higher level than is currently done.  One
issue that is coming up is using matplotlib to make a plot with axis
lines crossing in the center of the plot, like is shown in the plot at
http://sagemath.org/tour-education.html

Is this easy to do in matplotlib (preferably with the stock axes
object)?  It seems that there is not an example illustrating an axes of 
this form and I couldn't find any mention in the documentation of such a 
feature.

Thanks,

Jason

Jeff Mangum wrote:
> Hello,
> 
> I am having a problem installing matplotlib 0.93.3 from egg on Mac OSX 
> 10.5.5.  Even though I have numpy 1.2.1 installed in 
> /Library/Frameworks/..., the egg insists on using an older version of 
> numpy (1.0.4) in /opt/local/lib/python2.5/site-packages (which must have 
> been delivered with the OS). 

No, /opt/local is MacPorts territory.

> How can I tell the egg where to find the 
> proper version of numpy?  Thanks!

Are you sure you are using the same versions of Python to run and install both 
of these?

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
  that is made terrible by our own mad attempt to interpret it as though it had
  an underlying truth."
   -- Umberto Eco