matplotlib-users Mailing List for matplotlib (Page 2)

Alan G Isaac wrote:
> Since pylab's 'load' function has come up,
> it may be worth mentioning that it overrides scipy's load function.
> (And of course Numeric has its own 'load'.)

Nope, it's just Numeric and pylab that have a load() at top-level.

> It may be a bad habit,
> but I doubt I'm the only one who occasionally does
> from scipy import *
> from pylab import *

You're definitely *not* the only one.

> I would even claim that both encourage this ...
> especially for new users.  If that is correct,
> and if scipy+pylab is a common pair (as I
> believe it is), then perhaps ...

-- 
Robert Kern
rk...@uc...

"In the fields of hell where the grass grows high
  Are the graves of dreams allowed to die."
   -- Richard Harter

On Thu, 27 Jan 2005 Alan G Isaac apparently wrote:
> Since pylab's 'load' function has come up,
> it may be worth mentioning that it overrides scipy's load function.

Never mind: this is wrong as stated.
Sorry for the noise.

Alan Isaac

Since pylab's 'load' function has come up,
it may be worth mentioning that it overrides scipy's load function.
(And of course Numeric has its own 'load'.)

It may be a bad habit,
but I doubt I'm the only one who occasionally does
from scipy import *
from pylab import *

I would even claim that both encourage this ...
especially for new users.  If that is correct,
and if scipy+pylab is a common pair (as I
believe it is), then perhaps ...

fwiw,
Alan Isaac

             Hi,

I had a problem with the load function. I have a file with some data in 
two columns. I'm trying to use load:

x,y=load('toto.dat')

with toto something like:

1 3
3 4
5 6

I obtained the error message:

ValueError: too many values to unpack

I understanded why. It's because the array is not use by columns but by 
lines when you unpack the array from load to x and y so I don't have 
enough variable. To solve this problem I add the transpose function:

x,y=transpose(load('toto.dat'))

I don't know if I'm alone with this problem but if yes that will be a 
good idea to update the help of the function.

Thanks,

    Nicolas

>>>>> "Chris" == Chris Barker <Chr...@no...> writes:

    Chris> anyway, to cut my rant short, here is my vote for
    Chris> matplotlib development (not that I get a vote, but
    Chris> hopefully I'll have time to help out someday)

Hey, I live in Chicago -- we pioneered "vote early, vote often". 

    Chris> 2) Improve the OO interface to make it just as easy to use.

Do you have some suggestions which would make the OO interface easier
to use?  For concreteness, here is the archetypal OO script

  from matplotlib.backends.backend_svg import FigureCanvasSVG
  from matplotlib.figure import Figure
  fig = Figure()
  ax = fig.add_subplot(211)
  ax.plot([1,2,3])
  ax.set_xlabel('time')
  canvas = FigureCanvasSVG(fig)
  canvas.print_figure('myfile.svg')

Things that leap to my mind:

  * hide the complexity of having both a canvas and the figure from
    the user, allowing them to deal only with figures, or at least
    have the figure store a ref to its canvas and allow you to do

    fig.savefig('myfig')
 
    which can forward the call on to its canvas print_figure method.

    This would bring the OO interface closer to the pylab interface
    and would prevent you from having to remember which methods were
    fig methods and which methods were canvas methods.  We could also
    provide a pylab independent factory function that instantiates a
    figure and canvas with the relevant initialization, allowing you to
    do something like

    fig, canvas = figure_factory(*args, **kwargs)  # use the current backend

    but I'm not sure how advisable it is to hide this complexity
    because what you ultimately do with the objects depends on whether
    you are using a GUI backend or not.  But for pure image backends
    it would be possible to have a single OO interface which works
    across backends.

  * move the (few) remaining pylab only methods (axis, colorbar) into
    the OO framework for full compatibility

  * use properties/traits so you could do

     ax.xlabel = 'hi mom'

    while retaining the setters and getters for backwards
    compatibility

  * provide more pure OO examples in the examples directory

  * significant;y expand the OO section of the user's guide

Do you have additional ideas?  Or are these the issues your were thinking of?

JDH

----- Original Message -----
From: John Hunter <jdh...@ac...>
> Eg, if you have set your dpi parameter correctly to reflect your
> monitor 
> 
>    from pylab import *
>    fig = figure(dpi=96)
>    plot([1,2,3])
>    text(1,2,'hi mom', fontsize=72)
>    show()
> 
> creates text that is an inch high, irrespective of the view port.  I
> think this is the behavior most people expect, at least those who are
> used to thinking about fontsizes in points.

Well, I'd rather set the fontsize in Pixels, but I suppose I can do that
by setting DPI to 72. However, you generally don't want to zoom text as
you zoom in and out, you are zooming in on teh DATA, no the PICTURE of
the data.

By the way, if you do want that, check out my FloatCanvas, in the
wxPython lib. You'll have to draw your own Axes, but if you use
ScaledText, you'll get the whole picture to zoom and scroll.

-Chris

>>>>> "Stephen" == Stephen Walton <ste...@cs...> writes:

    Stephen> Nils Wagner wrote:
    >> How can I add vertical/horizontal lines (x=constant, y=constant
    >> respectively) to an existing plot with matplotlib ?

    Stephen> I've always done something like:

    Stephen> plot(x,y) a=axis() plot(a[0:2],[50,50],'k')

    Stephen> for example, to get a line at y=50.  If there's something
    Stephen> shorter I'm open to suggestions.

if you want to plot a line in data coordinates, you can use vline

  plot([0,1,2],[3,4,5])
  vlines([1], [4,5], hold=True)

See also vlines.  This x and y location of this line will "move" with
the data when you pan and zoom.

If you want to plot a line in axes coords, us axvline, where x is in
data coords but y is now interpreted as a fraction of the axes width
and the x location will not move with pans and zooms.  Thus if you
want a vertical bar at x=1 that ranges from the top to the bottom, do

  axvline(1)  # ymin=0 (bottom) and ymax=1 (top) default

The x location moves with pan/zoom but the y data are fixed.

JDH

Hi folks,

I am trying to do something similar to:

from pylab import *
plot([1,2,3,4])
text(1,3,'a)')
savefig('test.eps')

I can't render the eps file because of the way that "a)" is recorded ["(a)" 
will render, so its just a parsing issue]. I think, in the eps file "(a)) 
show" should read "(a\)) show" for the image will render.

Changing line 353 in backend_ps.py to 

self._pswriter.write("(%s) show\n"%s.replace(')','\)').replace('(','\('))

would do the trick.

I'm using gentoo linux and matplotlib 0.71.
-- 

Darren

Nils Wagner wrote:

> Hi all,
>
> How can I add vertical/horizontal  lines (x=constant, y=constant 
> respectively) to an existing plot with matplotlib ?
> A small example would be appreciated.


See http://matplotlib.sourceforge.net/matplotlib.pylab.html#-axhline
and http://matplotlib.sourceforge.net/matplotlib.pylab.html#-axvline

>
> Is there a simple command to plot closed polygons, which are defined 
> by the coordinates  of the vertices ?
>
http://matplotlib.sourceforge.net/matplotlib.pylab.html#-fill

See axhspan_demo.py and fill_demo.py for example usage.

-Jeff

-- 
Jeffrey S. Whitaker         Phone  : (303)497-6313
Meteorologist               FAX    : (303)497-6449
NOAA/OAR/CDC  R/CDC1        Email  : Jef...@no...
325 Broadway                Web    : www.cdc.noaa.gov/~jsw
Boulder, CO, USA 80303-3328 Office : Skaggs Research Cntr 1D-124

I know if I modify the code, I can avoid these things. But out of my =
eccentric nature, all I am asking is how to do the zooming more nicely?

o The zooming of text, as you can test in <mathtext_demo.py>. I remember =
once I put some mathtext symbols, as I wished to see the subscript I =
tried zoom to see more sharply but I failed. I saved png and zoomed =
finally.
o The zooming of subplot, as you can test in <axes_demo.py>, well =
personally I assume if you zoom on a subplot you would expect it gets =
enlarged.
o The zooming of crosshaired frontend, as you can test in =
<wxcursor_demo.py>, this is minor problem, but when I zoom a random =
area, I see a vestigial image of the cross-hair which is a tiny bit =
unnatural.

And of course the speed of zooming...  but I see the pylab is redrawing =
the figure so this is understandable. I know this is a plotting module =
not an ACDsee or picasa, so forgive my eccentricity.

btw: along with this note
- how to insert a space in a bunch of mathtext as I wish to put spaces =
near '=3D'?
- could you stop the interactive console (at least make it configurable) =
after the figure is closed so that I don't need to put sys.exit().

Thank you (smile).

Bao=20

Nils Wagner wrote:

> How can I add vertical/horizontal  lines (x=constant, y=constant 
> respectively) to an existing plot with matplotlib ? 

I've always done something like:

plot(x,y)
a=axis()
plot(a[0:2],[50,50],'k')

for example, to get a line at y=50.  If there's something shorter I'm 
open to suggestions.

On Wednesday 26 January 2005 10:32 am, Perry Greenfield wrote:
> On Jan 26, 2005, at 10:14 AM, John Hunter wrote:
> >>>>>> "Haibao" == Haibao Tang <ba...@ug...> writes:
> >
> >     Haibao> Hi folks, I've noticed that zooming on the text won't make
> >     Haibao> the text any larger, and some other instances like
> >     Haibao> sub-axes share the problem too. And it is not the default
> >     Haibao> zoom behavior people usually expect. Any solution to that?
> >
> > Well, I suspect we'd have to poll people to know whether this is the
> > behavior they usually expect....
>
> Exactly. I'd guess that most people would not want the text to zoom. I
> don't. If one takes zooming as a literal zoom of everything in the
> field, yes, I can understand that is what some might expect. But what
> is really most useful? I doubt that literal zooming is in the great
> majority of cases.

I also would expect the default behavior to not zoom the text.

-- 

Darren

Hi all,

How can I add vertical/horizontal  lines (x=constant, y=constant 
respectively) to an existing plot with matplotlib ?
A small example would be appreciated.

Is there a simple command to plot closed polygons, which are defined by 
the coordinates  of the vertices ?
 
Thanks in advance.

Nils

 

On Jan 26, 2005, at 10:14 AM, John Hunter wrote:

>>>>>> "Haibao" == Haibao Tang <ba...@ug...> writes:
>
>     Haibao> Hi folks, I've noticed that zooming on the text won't make
>     Haibao> the text any larger, and some other instances like
>     Haibao> sub-axes share the problem too. And it is not the default
>     Haibao> zoom behavior people usually expect. Any solution to that?
>
> Well, I suspect we'd have to poll people to know whether this is the
> behavior they usually expect....
>
Exactly. I'd guess that most people would not want the text to zoom. I 
don't. If one takes zooming as a literal zoom of everything in the 
field, yes, I can understand that is what some might expect. But what 
is really most useful? I doubt that literal zooming is in the great 
majority of cases.

Perry

On Wed, 2005-01-26 at 10:09, John Hunter wrote:
> >>>>> "Hans" == Hans Fangohr <H.F...@so...> writes:
> 
>     Hans> I didn't make myself very clear, or I have misunderstood the
>     Hans> meaning of "interactive". I presumed that 'interactive'
>     Hans> means: I issue the plot command and the figure pops up
>     Hans> immediatly, etc.
> 
>     Hans> Not interactive (in my understanding) means that I issue all
>     Hans> the matplotlib commands I'd like to use and at the end I can
>     Hans> use show() to display the picture.
> 
> This partially but not completely correct, but it is admittedly
> confusing.  Make sure you have read
> http://matplotlib.sourceforge.net/interactive.html and
> http://matplotlib.sourceforge.net/faq.html#SHOW .  As emphasized in
> the FAQ
> 
>   IMPORTANT: show should called at most once per script and it should
>   be the last line of your script. At that point, the GUI takes
>   control of the interpreter. If you want to force a figure draw, use
>   draw instead.
> 
> If you are issuing plot commands from idle, you need interactive :
> True, you should not use show, and if you want finer control of when
> the plot pops up and it drawn, use the commands ion, ioff and draw, as
> explained on the interactive.html link.
> 
> Hope this helps -- if you still encounter problems please let us know
> because it is always possible there is a bug ....

I'm really busy this morning (like what else is new for any of us),  but
I can confirm that there is a bug in 0.71.  The root cause is that an
"extra" mainloop is now run by matplotlib in "Idle -n" which already has
a mainloop of its own.

I think the key to solving this class of problem is to detect the
presence of "freebe" mainloops to avoid running two mainloops while
ensuring that we run at least one.  To do this,  I'm hoping matplotlib
can ask Tkinter if a mainloop is running or not.  The only alternative
that occurs to me is to detect the presence of particular shells and
maintain knowledge about the "freebe mainloop" status of each.  

This morning, I tried to detect IDLE using "'idlelib.__main__' in
sys.modules.keys()" and was able to fix the current bug by suppressing
the mainloop and assuming that "Idle -n" is running and not "Idle".  It
would be much better to detect the mainloop directly though.

Regards,
Todd

>>>>> "Haibao" == Haibao Tang <ba...@ug...> writes:

    Haibao> Hi folks, I've noticed that zooming on the text won't make
    Haibao> the text any larger, and some other instances like
    Haibao> sub-axes share the problem too. And it is not the default
    Haibao> zoom behavior people usually expect. Any solution to that?

Well, I suspect we'd have to poll people to know whether this is the
behavior they usually expect....

Text locations are in data coordinates, but text sizes are physical
sizes, ie 12 points equal 12/72 inches, approximately...  Physical
sizes don't change when you zoom in because the physical size of the
font will not change when you change the view limits of the data
coordinate system.

Eg, if you have set your dpi parameter correctly to reflect your
monitor 

    from pylab import *
    fig = figure(dpi=96)
    plot([1,2,3])
    text(1,2,'hi mom', fontsize=72)
    show()

creates text that is an inch high, irrespective of the view port.  I
think this is the behavior most people expect, at least those who are
used to thinking about fontsizes in points.

JDH

>>>>> "Hans" == Hans Fangohr <H.F...@so...> writes:

    Hans> I didn't make myself very clear, or I have misunderstood the
    Hans> meaning of "interactive". I presumed that 'interactive'
    Hans> means: I issue the plot command and the figure pops up
    Hans> immediatly, etc.

    Hans> Not interactive (in my understanding) means that I issue all
    Hans> the matplotlib commands I'd like to use and at the end I can
    Hans> use show() to display the picture.

This partially but not completely correct, but it is admittedly
confusing.  Make sure you have read
http://matplotlib.sourceforge.net/interactive.html and
http://matplotlib.sourceforge.net/faq.html#SHOW .  As emphasized in
the FAQ

  IMPORTANT: show should called at most once per script and it should
  be the last line of your script. At that point, the GUI takes
  control of the interpreter. If you want to force a figure draw, use
  draw instead.

If you are issuing plot commands from idle, you need interactive :
True, you should not use show, and if you want finer control of when
the plot pops up and it drawn, use the commands ion, ioff and draw, as
explained on the interactive.html link.

Hope this helps -- if you still encounter problems please let us know
because it is always possible there is a bug ....

JDH

>>>>> "Humufr" == Humufr  <hu...@ya...> writes:

    Humufr>        Hi John, I saw that you add in the load function
    Humufr> the possibility to have some line comment with the
    Humufr> character '%'. I would like to change this function and
    Humufr> generalize a little bit this. Not everyone is using the
    Humufr> character '%' but some other like '#' or '!'. So I add
    Humufr> this possibilty to the function. I don't know if you are
    Humufr> agree with it but in case I join the change below.

OK, that's a good idea.  I added it to pylab.py.  In the future,
please add changes to the docstring as well when you change the
function, so that folks who do help(load) will get some information on
what the comments kwarg does.  Thanks...

    Humufr> Another things, I think there are a small bug for the
    Humufr> Tkagg backend. When I close the windows instead to come
    Humufr> back to my shell I'm arriving in a python shell. I tried
    Humufr> (when I had my laptop) with the Gtkagg backend and I
    Humufr> didn't notice this problem.

Not technically a bug, as it was intentional, but perhaps an
annoyance.  In order for python to work with idle, we need this line     

  os.environ['PYTHONINSPECT'] = '1'

in the show function of backends/backend_tkagg.py.  If you comment out
this line the shell will not be raised. 

Todd, any objection to making this an rc parameter, so that people can
optionally have this feature?  Something like

tk.window_focus   : False    # Maintain shell focus for TkAgg
tk.pythoninspect  : False    # True for interactive idle

JSH

John,

>>>>>> "Hans" == Hans Fangohr <H.F...@so...> writes:
>
>    Hans> Hm. Have upgraded to 0.71. Now I don't get the error message
>    Hans> anymore, however, once I close the figure window, the IDLE
>    Hans> (-n) prompt doesn't respond anymore. (Same scenario as
>    Hans> described in last email.)
>
>    Hans> Ideas?
> Well, if you are using matplotlib interactively from within IDLE< then
> interactive : True should be set in your rc file, no?

I didn't make myself very clear, or I have misunderstood the meaning of 
"interactive". I presumed that 'interactive' means: I issue the plot 
command and the figure pops up immediatly, etc.

Not interactive (in my understanding) means that I issue all the 
matplotlib commands I'd like to use and at the end I can use show() to 
display the picture.

I presumed that when I close the figure windows (by clicking on the right 
symbol on the window), I could carry on using the IDLE session (in this 
example).

Here is the bit that I didn't expect and assumed would be some kind of 
bug:

I need to kill IDLE (when clicking on the [X] symbol in the upper right 
hand corner of the IDLE -n window in MS XP, I get the message :" The 
program is still running! Do you want to kill it?"). In summary, I can't 
carry on using the session once I have used the show() command.

Thanks,

Hans

>>>>> "Hans" == Hans Fangohr <H.F...@so...> writes:

    Hans> Hm. Have upgraded to 0.71. Now I don't get the error message
    Hans> anymore, however, once I close the figure window, the IDLE
    Hans> (-n) prompt doesn't respond anymore. (Same scenario as
    Hans> described in last email.)

    Hans> Ideas?
Well, if you are using matplotlib interactively from within IDLE< then
interactive : True should be set in your rc file, no?

JDH

Hi John,

>    Hans> The figure windows is interacting (despite the error
>    Hans> message). Once I close the figure window, I can't get back
>    Hans> to the Python prompt; i.e. the shell doesn't respond any
>    Hans> more.
>
>    Hans> Do we expect matplotlib (0.70.1) to work in non-interactive
>    Hans> mode with "IDLE -n"?
>
> This is an unrelated bug fixed in 0.71.
>
> Thanks for the report!

Hm. Have upgraded to 0.71. Now I don't get the error message anymore, 
however, once I close the figure window, the IDLE (-n) prompt doesn't 
respond anymore. (Same scenario as described in last email.)

Ideas?


> JDH
>
>
>
>
>
>
>
>

-------------------------------------------------
Dr Hans Fangohr

Computational Engineering & Design Research Group
School of Engineering Sciences
University of Southampton
Southampton, SO17 1BJ
United Kingdom

Location: Building 25, Room 1027
phone : +44 (0) 23 8059 8345
fax   : +44 (0) 23 8059 7082
email : fa...@so...
-------------------------------------------------

       Hi John,

I saw that you add in the load function the possibility to have some 
line comment with the character '%'. I would like to change this 
function and generalize a little bit this. Not everyone is using the 
character '%' but some other like '#' or '!'. So I add this possibilty 
to the function. I don't know if you are agree with it but in case I 
join the change below.
Another things, I think there are a small bug for the Tkagg backend. 
When I close the windows instead to come back to my shell I'm arriving 
in a python shell. I tried (when I had my laptop) with the Gtkagg 
backend and I didn't notice this problem.

Thanks.

    Nicolas



def load(fname,comments='%'):
    """
    Load ASCII data from fname into an array and return the array.

    The data must be regular, same number of values in every row

    fname can be a filename or a file handle

    matfile data is not currently supported, but see
    Nigel Wade's matfile ftp://ion.le.ac.uk/matfile/matfile.tar.gz

    Example usage:

    x,y = load('test.dat')  # data in two columns

    X = load('test.dat')    # a matrix of data

    x = load('test.dat')    # a single column of data

    """

    if is_string_like(fname):
        fh = file(fname)
    elif hasattr(fname, 'seek'):
        fh = fname
    else:
        raise ValueError('fname must be a string or file handle')
   
    X = []
    numCols = None
    for line in fh:
        line = line[:line.find(comments)].strip()
        if not len(line): continue
        row = [float(val) for val in line.split()]
        thisLen = len(row)
        if numCols is not None and thisLen != numCols:
            raise ValueError('All rows must have the same number of 
columns')
        X.append(row)

    X = array(X)
    r,c = X.shape
    if r==1 or c==1:
        X.shape = max([r,c]),
    return X

>>>>> "Hans" == Hans Fangohr <H.F...@so...> writes:

    Hans> The figure windows is interacting (despite the error
    Hans> message). Once I close the figure window, I can't get back
    Hans> to the Python prompt; i.e. the shell doesn't respond any
    Hans> more.

    Hans> Do we expect matplotlib (0.70.1) to work in non-interactive
    Hans> mode with "IDLE -n"?

This is an unrelated bug fixed in 0.71.

Thanks for the report!
JDH

Hi John,

I am coming back to our earlier discussion ...

>>>>>>> "Hans" == Hans Fangohr <H.F...@so...> writes:
>> 
>>> Hm. Done that. A figure window pops up immediately but is
>>> not being updated correctly: a part of the graph is visible,
>>> the other part of the figure (which was underneath another
>>> window initially) appears white. The titel of the figure
>>> says "Figure 1 (Not responding)".  IDLE, however, responds
>>> happily ;-) T
>> 
>> Todd, many moons ago you wrote me and said I had inadvertently screwed
>> up something in tkagg show that broke the idle -n behavior.  If I
>> recall correctly, we never reverted, since I didn't understand exactly
>> what to do.  Perhaps you could revisit the current tkagg versus the
>> old one to make sure it is right.
>> 
>> Hans, does it make a difference if you uncomment
>> 
>> #os.environ['PYTHONINSPECT'] = '1'
>> 
>> in the show method of
>> site-packages/matplotlib/backends/backend_tkagg.py? You should set
>> 'interactive : True' as Todd suggested.  Do not use show, as it is not
>> designed for interactive use.  I suggest carefully testing all 4
>> combinations: with and without the PYTHONINSPECTmline, and with and
>> without -n.
>
> Setting 'interactive=True' in .matplotlibrc
>
> IDLE           #os.environ['PYTHONINSPECT'] = '1'        -> fail
> IDLE -n        #os.environ['PYTHONINSPECT'] = '1'        -> fail
> IDLE           os.environ['PYTHONINSPECT'] = '1'         -> fail
> IDLE -n        os.environ['PYTHONINSPECT'] = '1'         -> success
>
> So that's good.

This doesn't work in non-interactive mode. What I observe is this:

- start Idle with -n (the "==== No Subprocess ====" message appears).
- at prompt issue commands:
>>>import pylab
>>>pylab.plot(range(10))
>>>pylab.show()
- figure windows shows up
- python reports this error message:

>>> pylab.show()
Exception in Tkinter callback
Traceback (most recent call last):
   File "C:\Python23\lib\lib-tk\Tkinter.py", line 1345, in __call__
     return self.func(*args)
   File 
"C:\Python23\lib\site-packages\matplotlib\backends\backend_tkagg.py",
line 215, in key_release
     FigureCanvasBase.key_release_event(self, key)
   File "C:\Python23\Lib\site-packages\matplotlib\backend_bases.py",
line 677, in key_release_event
     event = KeyEvent('key_release_event', self, key, self._lastx, 
self._lasty)
   File "C:\Python23\Lib\site-packages\matplotlib\backend_bases.py",
line 640, in __init__
     LocationEvent.__init__(self, name, canvas, x, y)
   File "C:\Python23\Lib\site-packages\matplotlib\backend_bases.py",
line 566, in __init__
     if a.in_axes(self.x, self.y):
   File "C:\Python23\Lib\site-packages\matplotlib\axes.py", line 1544, in 
in_axes
     return self.bbox.contains(xwin, ywin)
TypeError: float() argument must be a string or a number


The figure windows is interacting (despite the error message). Once I 
close the figure window, I can't get back to the Python prompt; i.e. the 
shell doesn't respond any more.

Do we expect matplotlib (0.70.1) to work in non-interactive mode with 
"IDLE -n"?

Thanks,

Hans

As a protanope, and knowing that lots of folks with interest in graphics 
frequent this newsgroup, I thought I'd call everyone's attention to the 
article entitled "The End of the Rainbow? Color Schemes for Improved 
Data Graphics" which appeared in the 5 October 2004 issue of the 
American Geophysical Union's weekly newsmagazine EOS.  A Web site with 
the sample color schemes and a copy of the article is at 
http://geography.uoregon.edu/datagraphics/