matplotlib-users — Discussion related to using matplotlib

Re: [Matplotlib-users] figlegend and title

[John H. <jdh...@ac...>]

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

    Chris> By the way, I'm on a quest to make the OO interface to
    Chris> matplotlib fully functional and convenient. If you want to
    Chris> work on it, John would probably except a patch to make a
    Chris> figure.legend() method.

pylab.figlegend is a thin wrapper to matplotlib.Figure.legend; see
http://matplotlib.sf.net/matplotlib.figure.html#Figure-legend

<wink>

JDH

Re: [Matplotlib-users] figlegend and title

[Chris B. <Chr...@no...>]

Werner F. Bruhin wrote:
> However I can't get this to work within the wxcursor_demo.py, is 
> figlegend only available with "from pylab import *"?

I'm not sure about that one, but there are a number of convenience 
functions that are only in pylab. If so, you have two choices:

1) Use:
import pylab
pylab.figlegend(...)

2) you can look at the code in pylab.figlegend, and see what it does, 
and make your own version.

By the way, I'm on  a quest to make the OO interface to matplotlib fully 
functional and convenient. If you want to work on it, John would 
probably except a patch to make a figure.legend() method.

-Chris


-- 
Christopher Barker, Ph.D.
Oceanographer
                                     		
NOAA/OR&R/HAZMAT         (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chr...@no...

[Matplotlib-users] figlegend and title

[Werner F. B. <wer...@fr...>]

Hi,

I am just getting my feet wet with matplotlib.

I like plots with the legend outside the plotting area.  I found the 
demo which uses figlegend (thanks Andrea), this fits the bill - at least 
I think so.

However I can't get this to work within the wxcursor_demo.py, is 
figlegend only available with "from pylab import *"?

The other question is can I have sub-title lines, I saw that I can pass 
newline with title string but this uses the same font attributes, I 
would like a slightly smaller font and non bold etc for the second title 
line.

Looked through the archive on gmane, but couldn't find anything.

Appreciate any pointers on this.

See you
Werner

Re: [Matplotlib-users] Changing size of markers in legend??

[John H. <jdh...@ac...>]

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

    kristen> Hello John I was wondering how to change the size of a
    kristen> point marker in legend.  I only plot single points like:

    kristen> h1 = plot([1],[2],'ro') h2 = plot([3],[4],'go')
    kristen> legend((h1,h2),('1','2'), numpoints = 1)

    kristen> but the marker 'o' comes out quite small in the legend.
    kristen> Is there an easy way to enlarge it ??

See help(legend), specifically the markerscale kwarg

 markerscale = 0.6     # the relative size of legend markers vs. original

    kristen> By the way, why aren't things like fontsize,
    kristen> Frame=True/False etc legend kwargs ?? It would be much
    kristen> easier to change them this way compared to what you have
    kristen> shown in legend_demo.py. But, I guess there is a sensible
    kristen> reason for this.

You can pass in a kwarg 
   
  prop = FontProperties(size='smaller')  # the font properties

to control the font property.  See
http://matplotlib.sf.net/matplotlib.font_manager.html#FontProperties
for more information

As a general answer to your question, there are simply too many things
to customize to make them all kwargs.  We would have to add all the
properties of the frame, all the properties of the markers, all the
properties of the text, and these would be ambiguous since they all
derive from the base Artist class and hence properties like "visible"
which apply to all artists would be ambiguous.  matplotlib Artists
have a lot of configurable properties (eg for the frame alone)

In [5]: frame = leg.get_frame()
 
In [6]: get(frame)
    alpha = 1.0
    antialiased or aa = True
    clip_on = False
    edgecolor or ec = k
    facecolor or fc = w
    figure = <matplotlib.figure.Figure instance at 0x41a9ecac>
    fill = 1
    height = 0.0123998966675
    label =
    linewidth or lw = 1.0
    transform = <Affine object at 0x84b7cec>
    verts = ((0.89123573200992556, 0.96760010333247226), (0.89123573200992556, 0.97999999999999998), (0.97999999999999998, 0.97999999999999998), (0.97999999999999998, 0.96760010333247226))
    visible = True
    width = 0.0887642679901
    window_extent = <Bbox object at 0x84316ec>
    x = 0.89123573201
    y = 0.967600103332
    zorder = 1

and it would be tedious and difficult to maintain to expose all of
these in the kwarg interface.

JDH

[Matplotlib-users] Changing size of markers in legend??

[kristen k. <co...@ya...>]

Hello John

I was wondering how to change the size of a point
marker in legend.
I only plot single points like:

h1 = plot([1],[2],'ro')
h2 = plot([3],[4],'go')
legend((h1,h2),('1','2'), numpoints = 1)

but the marker 'o' comes out quite small in the
legend.
Is there an easy way to enlarge it ??

By the way, why aren't things like fontsize,
Frame=True/False etc legend kwargs ?? It would be much
easier to change them this way compared to what you
have shown in legend_demo.py. But, I guess there is a
sensible reason for this.

Cheers
Kristen



		
__________________________________ 
Do you Yahoo!? 
Yahoo! Personals - Better first dates. More second dates. 
http://personals.yahoo.com

[Matplotlib-users] Re: Legend

[John H. <jdh...@ac...>]

>>>>> "Travis" == Travis Brady <td...@fa...> writes:

    Travis> Hi John, I've finally gotten to the point where I'm making
    Travis> really nice plots on the fly in a bit of code I wrote that
    Travis> I call from within the analysis part of my code and I must
    Travis> say, Matplotlib rules.

Great -- glad it's starting to work for you.

In the future, could you post to the list instead of me directly.
That way the responses show up in the archives, and others can provide
help as well

    Travis> But there are a few things that I always find difficult.
    Travis> Most of the time the legend is a good bit larger than I'd
    Travis> like and I can't seem to figure out how to change that,
    Travis> though I'm thinking it probably involves something like
    Travis> what you do in legend_demo.py in the examples.  Is this
    Travis> true, getting the text and then setting its fontsize? and
    Travis> then setting the width and height of the frame?

Yes, you can control the size of the text following the example from
examples/legend_demo.py.  There are a few other parameters to tweak as
well, eg the following kwargs to the legend command

  numpoints = 4         # the number of points in the legend line
  fontprop = ...        # the font property
  pad = 0.2             # the fractional whitespace inside the legend border
  markerscale = 0.6     # the relative size of legend markers vs. original
  labelsep = 0.005      # the vertical space between the legend entries
  handlelen = 0.05      # the length of the legend lines
  handletextsep = 0.02  # the space between the legend line and legend text
  axespad = 0.02        # the border between the axes and legend edge


Setting the width and height of the frame directly probably won't
help, as this is dynamically resized at drawing time to accommodate
the artists contained inside it.

    Travis> Also, I'm making a plot of some signed errors and for some
    Travis> reason the y axis seems broken as the negative values are
    Travis> being plotting above the positive ones, any idea why this
    Travis> might be?

I haven't seen anything like this before.  An example which shows the
bug would help, as would the version info provided when you run your
script with --verbose-helpful.

JDH

Re: [Matplotlib-users] Plot dimensions

[Tim L. <ti...@cs...>]

And to answer my own question, I need to do:

plot(range(75))
axis([0, 74, 0, 74])
show()

Sorry for the noise.

Tim

On Fri, 08 Apr 2005, Tim Leslie <ti...@cs...> wrote...

> Hi All,
> 
> I've got a nice simple question for you. If I do:
> 
> plot(range(75))
> show()
> 
> The figure shows up as expected, but the x and y axes both run from 0 to
> 80. Is there some way to force them to only include the required range of
> points, eg 0 to 75? I had a play around with the axes() and figure()
> functions but I wasn't able to get them to do what I wanted. Not sure if I
> was maybe barking up the wrong tree.
> 
> Cheers,
> 
> Tim Leslie
> 
> 
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from real users.
> Discover which products truly live up to the hype. Start reading now.
> http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
> 
`-

[Matplotlib-users] Plot dimensions

[Tim L. <ti...@cs...>]

Hi All,

I've got a nice simple question for you. If I do:

plot(range(75))
show()

The figure shows up as expected, but the x and y axes both run from 0 to
80. Is there some way to force them to only include the required range of
points, eg 0 to 75? I had a play around with the axes() and figure()
functions but I wasn't able to get them to do what I wanted. Not sure if I
was maybe barking up the wrong tree.

Cheers,

Tim Leslie

Re: [Matplotlib-users] Candlestick Issues

[Daisy F. <eli...@ya...>]

John,

Thank you for clarifying things for me!  As far as the
replacement function goes, well -- yours is right on
the money.  It's a simple moving average calculation
and it gets the job done.  I just kept getting a
different result when comparing to an actual financial
chart since the calculation was based on open prices. 
I just added a calculation for the Exponential Moving
Average (based on
http://www.stockcharts.com/education/IndicatorAnalysis/indic_movingAvg.html):

def ema(s, n):
    """
    returns an n period exponential moving average for
the time series s

    s is a list ordered from oldest (index 0) to most
recent (index -1)
    n is an integer

        returns a numeric array of the exponential
moving average
    """
    s = array(s)
    ema = []
    j = 1
    #get n sma first and calculate the next n period
ema
    sma = sum(s[:n]) / n
    multiplier = 2 / float(1 + n)
    ema.append(sma)
    #EMA(current) = ( (Price(current) - EMA(prev) ) x
Multiplier) + EMA(prev)
    ema.append(( (s[n] - sma) * multiplier) + sma)
    #now calculate the rest of the values
    for i in s[n+1:]:
        tmp = ( (i - ema[j]) * multiplier) + ema[j]
        j = j + 1
        ema.append(tmp)
    return ema

I hope you can find it useful in your examples.

--- John Hunter <jdh...@ac...> wrote:

> OK, I see what is going on.  The lines are being
> plotted over the
> rectangles, so even if the rectangles are
> transparent, you still see
> the lines.  There are two candlestick functions in
> matplotlib
> candlestick and candlestick2.  They have slightly
> different call
> signatures and a different implementation under the
> hood.  candlestick
> creates a bunch of separate lines and rectangles,
> candlestick2 uses
> collections (see the help for the
> matplotlib.collections module).
> 
> You can control the z-ordering on the plot by
> setting the zorder
> property (see examples/zorder_demo.py).  For
> candlestick (see
> examples/candlestick_demo.py) you would do
> 
>   lines, patches = candlestick(ax, quotes,
> width=0.6)
>   set(lines, zorder=0.9*patches[0].zorder)
> 
> for candlestick2 you would do (untested)
>   
>  linecol, rectcol = candlestick2(blah) 
>  z = rectcol.get_zorder()
>  linecol.set_zorder(0.9*z)
> 


> Argg, that's embarrassing.  Good thing mpl is
> distributed with no
> warranties....  No telling how many billions this
> bug has cost the
> wall street barons already!
> 
> In matplotlib/finance.py in the candlestick2
> function, find this code
> 
>     colord = { True : colorup,
>                False : colordown,
>                }
>     colors = [colord[open>=close] for open, close in
> zip(opens, closes) if open!=-1 and close !=-1]
> 
> 
> That should read
> 
>     colors = [colord[close>=open] for open, close in
> zip(opens, closes) if open!=-1 and close !=-1]
> 
> 
> right?  I believe this is already correct in
> candlestick, so this is a
> candlestick2 specific bug.
> 
> 
 
> OK, if you submit a replacement function that better
> matches actual
> plots, I will be happy to include it.
> 
> Thanks for the report!
> JDH
> 
> 
>
-------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT
> Products from real users.
> Discover which products truly live up to the hype.
> Start reading now.
>
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
>
https://lists.sourceforge.net/lists/listinfo/matplotlib-users
> 


		
__________________________________ 
Do you Yahoo!? 
Take Yahoo! Mail with you! Get it on your mobile phone. 
http://mobile.yahoo.com/maildemo 

Re: [Matplotlib-users] imshow axis directions

[Stephen W. <ste...@cs...>]

Robert Kern wrote:

> Exuberant ctags
>
> http://ctags.sourceforge.net/
>
Which, as it turns out, is the version of ctags on Fedora Core 3.  I did 
a 'ctags -R' in the matplotlib root directory and am now happily 
browsing the source using nedit.

Thanks, Robert!

Steve

Re: [Matplotlib-users] imshow axis directions

[Robert K. <rk...@uc...>]

Stephen Walton wrote:

> I'm sure the code is fine, John, and it's just me.  Specifically, if I 
> want to see how matplotlib is doing something, I can't just trace 
> function calls like I'm used to doing with Fortran and C, because there 
> are methods too and I have to find the class which defines those methods 
> to see what's happening.  Is there anything like ctags for Python?

Exuberant ctags

http://ctags.sourceforge.net/

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

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

Re: [Matplotlib-users] imshow axis directions

[Stephen W. <ste...@cs...>]

John Hunter wrote:

>>>>>>"Stephen" == Stephen Walton <ste...@cs...> writes:
>>>>>>            
>>>>>>
>Does changing the rc param
>
>  image.origin : upper             # lower | upper
>
>to lower help?
>  
>
Yes, it makes the image the right way around.  I was simply confused 
because I expected the pixel at (x,y) in the imshow display to be the 
value of img[x,y].  It appears not to be.  I hasten to add MATLAB's 
worse, as the pixel at (x,y) is actually the value of img[y,x];  MATLAB 
displays with the first array coordinate increasing downward and the 
second from left to right.

>    Stephen> I've tried to look at the source, really, but I feel like
>    Stephen> I'm *years* away from being enough of a Python hacker to
>    Stephen> understand matplotlib.
>
>Hmm, the code's not that bad is it :-)
>  
>
I'm sure the code is fine, John, and it's just me.  Specifically, if I 
want to see how matplotlib is doing something, I can't just trace 
function calls like I'm used to doing with Fortran and C, because there 
are methods too and I have to find the class which defines those methods 
to see what's happening.  Is there anything like ctags for Python?

Steve

[Matplotlib-users] feature request

[Rich D. <dr...@in...>]

Some ability to specify empty space between subplots would be very 
convenient.  I'm not sure if this could best be done at figure creation or 
at subplot creation with an additional argument that says how much space 
to leave above or below.

The motivation is that creating subplots with separate titles is a very
common desire, and doing so now results in title text that invariably
overlaps the plot above (or if the font is reduced enough, it may fit but
looks squished in there).

Using manual axes creation is another approach that has been suggested, 
but this is a pain and seems to obviate the convenience of subplot().

If there is a better approach than creating custom axes, I'd love to hear
about it.

Thanks to all the pylab developers!

Rich

Re: [Matplotlib-users] imshow axis directions

[John H. <jdh...@ac...>]

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

    Stephen> Hi, I was just comparing images displayed with imshow()
    Stephen> with what I get from IRAF's display command and ds9.  The
    Stephen> latter puts the first axis increasing horizontally and
    Stephen> the second axis increasing vertically, as if the image
    Stephen> array were addressed as img[ix,iy], where ix and iy are
    Stephen> integer x and y coordinates.  The axes imshow() produces
    Stephen> are labeled as if this is true for it as well, but in
    Stephen> fact the image is upside down with respect to what I see
    Stephen> with IRAF.  Even odder, to see the same display I see in
    Stephen> IRAF I have to do imshow(img[::-1,:]), as if I'm
    Stephen> reversing the direction of the first axis.


Does changing the rc param

  image.origin : upper             # lower | upper

to lower help?


    Stephen> I've tried to look at the source, really, but I feel like
    Stephen> I'm *years* away from being enough of a Python hacker to
    Stephen> understand matplotlib.

Hmm, the code's not that bad is it :-)

JDH

Re: [Matplotlib-users] imshow axis directions

[Perry G. <pe...@st...>]

check out the .matplotlibrc image.origin configuration parameter. The 
default is to have pixel 0,0 on the upper left which is opposite of 
what us weird astronomers think is the natural locationl

Perry

On Apr 7, 2005, at 2:08 PM, Stephen Walton wrote:

> Hi,
>
> I was just comparing images displayed with imshow() with what I get 
> from IRAF's display command and ds9.  The latter puts the first axis 
> increasing horizontally and the second axis increasing vertically, as 
> if the image array were addressed as img[ix,iy], where ix and iy are 
> integer x and y coordinates.  The axes imshow() produces are labeled 
> as if this is true for it as well, but in fact the image is upside 
> down with respect to what I see with IRAF.  Even odder, to see the 
> same display I see in IRAF I have to do imshow(img[::-1,:]), as if I'm 
> reversing the direction of the first axis.
>
> I've tried to look at the source, really, but I feel like I'm *years* 
> away from being enough of a Python hacker to understand matplotlib.
>
> Steve
>
>
>
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from real 
> users.
> Discover which products truly live up to the hype. Start reading now.
> http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

[Matplotlib-users] imshow axis directions

[Stephen W. <ste...@cs...>]

Hi,

I was just comparing images displayed with imshow() with what I get from 
IRAF's display command and ds9.  The latter puts the first axis 
increasing horizontally and the second axis increasing vertically, as if 
the image array were addressed as img[ix,iy], where ix and iy are 
integer x and y coordinates.  The axes imshow() produces are labeled as 
if this is true for it as well, but in fact the image is upside down 
with respect to what I see with IRAF.  Even odder, to see the same 
display I see in IRAF I have to do imshow(img[::-1,:]), as if I'm 
reversing the direction of the first axis.

I've tried to look at the source, really, but I feel like I'm *years* 
away from being enough of a Python hacker to understand matplotlib.

Steve

[Matplotlib-users] problems with py2exe

[<oli...@ma...>]

Hi,

I have some problems compiling an .exe-file for my application. I use
wxPython together with matplotlib and tried to make an .exe-file with both
packages for the first time.

Windows XP
Python 2.4
wxPython 2.5.3.1
matplotlib 0.74

I checked the FAQ at the matplotlib homepage and the py2exe-examples and
did the following:



from distutils.core import setup
import glob
import py2exe

data = glob.glob(r'C:\Python24\Lib\site-packages\matplotlib\*')
data.append(r'C:\Python24\Lib\site-packages\matplotlib\.matplotlibrc')

setup(
    console = ["PanelCheck.py"],
    data_files = [("matplotlibdata", data)],
    )


After py2exe is finished I find both folders 'dist' and 'build' as ususal,
however I can't finde an .exe-file of my application PanelCheck.py in the
dist-folder.
I assume that 'window' is what I want to use in the code above, but when I
do that I get the following error:

--(just a small part of the console output)--
error: can't copy 'C:\Python24\Lib\site-packages\matplotlib\backends':
doesn't exist or not a regular file


What am I doing wrong? Help is really appreciated!

Oliver

[Matplotlib-users] py2exe Problems...

[<and...@ti...>]

Hello NG,

    this morning I have done the really BIG mistake of uninstalling the
following things:

- Numeric 23.5
- ctypes 0.9.2
- numarray 1.1.1
- Matplotlib 0.72

And I have installed the following:

- Numeric 23.8
- ctypes 0.9.6
- numarray 1.2.3
- Matplotlib 0.74

Using py2exe 0.5.4, now I get some errors on not-found modules, that I pu=
t
at the end of the mail. I have never encountered these problems with olde=
r
versions of the aforementioned tools. Please note that, a part the VTK th=
ings,
almost all the missing modules are from numerix/backends. If I try to sta=
rt
the application, I get a Visual C++ runtime error (and not the usual MYEX=
E.exe.log
message from py2exe, so I do not know how to track the error). 
I am on a Windows 2000 machine.
Deinstalling the new tools and re-installing the old ones didn't help any=
more...

Does anyone have a suggestion?

['backends.draw_if_interactive', 'backends.new_figure_manager', 'backends=
.show',
 'cephes', 'dl', 'libvtkCommonPython', 'libvtkFilteringPython', 'libvtkGr=
aphicsP
ython', 'libvtkHybridPython', 'libvtkIOPython', 'libvtkImagingPython', 'l=
ibvtkPa
rallelPython', 'libvtkPatentedPython', 'libvtkRenderingPython', 'numerix.=
ArrayTy
pe', 'numerix.Complex', 'numerix.Complex32', 'numerix.Complex64', 'numeri=
x.Float
', 'numerix.Float32', 'numerix.Float64', 'numerix.Int', 'numerix.Int16',
'numeri
x.Int32', 'numerix.Int8', 'numerix.Matrix', 'numerix.UInt16', 'numerix.UI=
nt32',
'numerix.UInt8', 'numerix.absolute', 'numerix.add', 'numerix.allclose',
'numerix
.alltrue', 'numerix.arange', 'numerix.arccos', 'numerix.arccosh', 'numeri=
x.arcsi
n', 'numerix.arcsinh', 'numerix.arctan', 'numerix.arctan2', 'numerix.arct=
anh',
'
numerix.argmax', 'numerix.argmin', 'numerix.argsort', 'numerix.around',
'numerix
.array', 'numerix.arrayrange', 'numerix.asarray', 'numerix.asum', 'numeri=
x.bitwi
se_and', 'numerix.bitwise_or', 'numerix.bitwise_xor', 'numerix.ceil', 'nu=
merix.c
hoose', 'numerix.clip', 'numerix.compress', 'numerix.concatenate', 'numer=
ix.conj
ugate', 'numerix.convolve', 'numerix.cos', 'numerix.cosh', 'numerix.cross=
_correl
ate', 'numerix.cumproduct', 'numerix.cumsum', 'numerix.diagonal', 'numeri=
x.divid
e', 'numerix.dot', 'numerix.equal', 'numerix.exp', 'numerix.fabs', 'numer=
ix.fft.
fft', 'numerix.floor', 'numerix.fmod', 'numerix.fromfunction', 'numerix.f=
romstri
ng', 'numerix.greater', 'numerix.greater_equal', 'numerix.hypot', 'numeri=
x.ident
ity', 'numerix.indices', 'numerix.innerproduct', 'numerix.less', 'numerix=
.less_e
qual', 'numerix.log', 'numerix.log10', 'numerix.logical_and', 'numerix.lo=
gical_n
ot', 'numerix.logical_or', 'numerix.logical_xor', 'numerix.matrixmultiply=
',
'num
erix.maximum', 'numerix.minimum', 'numerix.mlab.amax', 'numerix.mlab.amin=
',
'num
erix.mlab.cov', 'numerix.mlab.diff', 'numerix.mlab.flipud', 'numerix.mlab=
.hannin
g', 'numerix.mlab.rand', 'numerix.mlab.std', 'numerix.mlab.svd', 'numerix=
.multip
ly', 'numerix.negative', 'numerix.nonzero', 'numerix.not_equal', 'numerix=
.nx',
'
numerix.ones', 'numerix.outerproduct', 'numerix.pi', 'numerix.power', 'nu=
merix.p
roduct', 'numerix.put', 'numerix.putmask', 'numerix.rank', 'numerix.ravel=
',
'num
erix.repeat', 'numerix.reshape', 'numerix.resize', 'numerix.searchsorted'=
,
'nume
rix.shape', 'numerix.sin', 'numerix.sinh', 'numerix.size', 'numerix.somet=
rue',
'
numerix.sort', 'numerix.sqrt', 'numerix.subtract', 'numerix.swapaxes', 'n=
umerix.
take', 'numerix.tan', 'numerix.tanh', 'numerix.trace', 'numerix.transpose=
',
'num
erix.where', 'numerix.which', 'numerix.zeros', 'vtkParallelPython', 'matp=
lotlib.
numerix.absolute', 'matplotlib.numerix.equal', 'numarray.Complex', 'numar=
ray.Com
plex32', 'numarray.Complex64', 'numarray.Float', 'numarray.Float32', 'num=
array.F
loat64', 'numarray.Int', 'numarray.Int16', 'numarray.Int32', 'numarray.In=
t8',
'n
umarray.NumArray', 'numarray.UInt16', 'numarray.UInt32', 'numarray.UInt8'=
,
'numa
rray._dotblas', 'numarray.asarray', 'numarray.dot', 'numarray.fromlist',
'numarr
ay.shape', 'numarray.typecode', 'numarray.zeros', 'wx.BitmapFromImage',
'wx.Empt
yIcon']

Re: [Matplotlib-users] Using RendererAgg with windows

[John H. <jdh...@ac...>]

>>>>> "jim" == jim  <jl...@yv...> writes:

    jim> John: What I need to do is create drawings of windows (as in
    jim> plug holes in houses and let in light) with grid and other
    jim> options on the fly.

    jim> So what I need are the graphic primitives -- canvas, lines,
    jim> fills ...  The output needs to be a graphic file.

You probably want to be using matplotlib primitives.  Assuming you
have a matplotlib.axes.Axes (or Subplot) instance stored as "ax", The
primitives are

  matplotlib.lines.Line2D           - add with ax.add_line 
  matplotlib.patches.Rectangle      - add with ax.add_patch
  matplotlib.patches.Polygon        - add with ax.add_patch
  matplotlib.patches.RegularPolygon - add with ax.add_patch
  matplotlib.patches.Circle         - add with ax.add_patch
  matplotlib.patches.Text           - add with ax.add_artist

Ie, there are not too many mpl primitives, and using these will
insulate you from changes in the mpl backend (renderer) api.  The
backend API is meant only for mpl developers.  The classes referred to
above are all part of the matplotlib Artist hierarchy.

    jim> I will go back and study the docs, but a suggestion of what
    jim> area to use would be most welcome.  Or, of course, a
    jim> different module than matplotlib if that would be more
    jim> appropriate.

To reiterate, there is one base class matplotlib.artist.Artist that
all the objects that render into the figure derive from.  From this,
there are just a few derived classes to be aware of: Line2D, Patch,
Text, and Collection.  From these there are a few more derived classes (eg
Patch and Collection have some specialized derived classes, Line2D and
Text do not as of yet).  Other Artists (including Figure, Axes,
Legend, Table and so on) are simply composites of these primitive
types.  See the following class docs for more info:

  http://matplotlib.sourceforge.net/matplotlib.artist.html
  http://matplotlib.sourceforge.net/matplotlib.lines.html
  http://matplotlib.sourceforge.net/matplotlib.patches.html
  http://matplotlib.sourceforge.net/matplotlib.text.html
  http://matplotlib.sourceforge.net/matplotlib.collections.html

In short, you should concentrate on building the primitive types you
need rather than calling the renderer methods directly.  The
primitives will call the right renderer methods as necessary (as they
do in matplotlib.lines, for example).

Hope this helps,
JDH

Re: [Matplotlib-users] Using RendererAgg with windows

[jim <jl...@yv...>]

John:

What I need to do is create drawings of windows (as in plug holes in houses 
and let in light) with grid and other options on the fly.

So what I need are the graphic primitives -- canvas, lines, fills ...
The output needs to be a graphic file.  

I will go back and study the docs, but a suggestion of what area to use would 
be most welcome.  Or, of course, a different module than matplotlib if that 
would be more appropriate.

Thanks,
Jim

On Wednesday 06 April 2005 08:13 pm, you wrote:
> >>>>> "jim" == jim  <jl...@yv...> writes:
> 
>     jim> Running this script gives the results below #! /usr/bin/env
>     jim> python
> 
>     jim> from matplotlib.backends.backend_agg import RendererAgg from
>     jim> matplotlib.transforms import Value
> 
> I know there are examples in the pdf user's guide discussing how to
> use the backend renderer directly, and these are meant mostly to be
> helpful to matplotlib developers.  Unfortunately, the guide has lagged
> behind the current development state.  The backend renderer API is in
> a transition state right now as we try to introduce some new methods
> to solve some old problems.  In particular, the draw_lines method that
> you are experiencing problems with has recently had a change in its
> call signature.
> 
> These changes have been discussed at some length recently on the
> matplotlib-devel list.  If you are interested, you might want to
> browse the matplotlib-devel archives and/or join the mailing list; see
> for example
> 
http://sourceforge.net/mailarchive/forum.php?thread_id=6938045&forum_id=36187 .
> 
> Thanks for reporting this discrepancy -- in the near term the backend
> API shouldn't be used on the user side.  If there is something you
> want to do but can't in the current API let me know.
> 
> As for the platform specific differences you report, my first guess is
> that the matplotlib versions may not be the same.  I wouldn't expect
> platform specific differences in the backend API for the same version.
> 
> JDH
> 
> 
> 
> 

Re: [Matplotlib-users] Using RendererAgg with windows

[John H. <jdh...@ac...>]

>>>>> "jim" == jim  <jl...@yv...> writes:

    jim> Running this script gives the results below #! /usr/bin/env
    jim> python

    jim> from matplotlib.backends.backend_agg import RendererAgg from
    jim> matplotlib.transforms import Value

I know there are examples in the pdf user's guide discussing how to
use the backend renderer directly, and these are meant mostly to be
helpful to matplotlib developers.  Unfortunately, the guide has lagged
behind the current development state.  The backend renderer API is in
a transition state right now as we try to introduce some new methods
to solve some old problems.  In particular, the draw_lines method that
you are experiencing problems with has recently had a change in its
call signature.

These changes have been discussed at some length recently on the
matplotlib-devel list.  If you are interested, you might want to
browse the matplotlib-devel archives and/or join the mailing list; see
for example
http://sourceforge.net/mailarchive/forum.php?thread_id=6938045&forum_id=36187 .

Thanks for reporting this discrepancy -- in the near term the backend
API shouldn't be used on the user side.  If there is something you
want to do but can't in the current API let me know.

As for the platform specific differences you report, my first guess is
that the matplotlib versions may not be the same.  I wouldn't expect
platform specific differences in the backend API for the same version.

JDH

[Matplotlib-users] Using RendererAgg with windows

[jim <jl...@yv...>]

Running this script gives the results below

#! /usr/bin/env python

from matplotlib.backends.backend_agg import RendererAgg
from matplotlib.transforms import Value

dpi   = Value(100.0)

o = RendererAgg(100,200, dpi)
gc = o.new_gc()
gc.set_foreground('k')
face = (1,1,1)
   
gc.set_linewidth(1)
o.draw_rectangle(gc, face, 0,0, 100,200)

o.draw_lines(gc, (1,100,1), (1,100,200))

o._renderer.write_png("test.png")

#*** Results from running the above ****************
# J:\>python test4.py
# Traceback (most recent call last):
#   File "test4.py", line 16, in ?
#     o.draw_lines(gc, x, y)
# IndexError: Unexpected SeqBase<T> length.
#
# J:\>
#****************************************************

Running in on linux-gentoo 2004 runs just fine.  Gives the expected output 
file.

My windows setup is win2k, python2.4
I loaded numeric and numarray.  Gives the same results in either case.

I have Borland5.5 installed have build an extension with it.  Other than that 
have several other nonstock modules installed.

Running it on win98 with python 2.4 gives a different error, but much the same 
effect.

Thanks in advance
Jim Hurlburt
Yakima, WA

Re: [Matplotlib-users] Looking for memory tips for interactive use

[Fernando P. <Fer...@co...>]

John Hunter wrote:
>>>>>>"Fernando" == Fernando Perez <Fer...@co...> writes:
> 
>     Fernando> Oh, I don't think it's necessarily a leak.  Simply a ton
>     Fernando> of unreachable stuff lying around, which a call to
>     Fernando> gc.collect() may help with.
> 
> I hope you're right -- matplotlib uses a lot of cyclic references.
> But if he's using the pylab interface to manage the figure windows,
> the destroy event of the window manager already triggers a call to
> gc.collect in the _pylab_helpers module.

Never mind then.  If gc.collect is already called reasonably often, an extra 
manual call will do zilch.  Sorry.

Best,

f

Re: [Matplotlib-users] Looking for memory tips for interactive use

[John H. <jdh...@ac...>]

>>>>> "Fernando" == Fernando Perez <Fer...@co...> writes:
    Fernando> Oh, I don't think it's necessarily a leak.  Simply a ton
    Fernando> of unreachable stuff lying around, which a call to
    Fernando> gc.collect() may help with.

I hope you're right -- matplotlib uses a lot of cyclic references.
But if he's using the pylab interface to manage the figure windows,
the destroy event of the window manager already triggers a call to
gc.collect in the _pylab_helpers module.

JDH

Re: [Matplotlib-users] Looking for memory tips for interactive use

[Fernando P. <Fer...@co...>]

John Hunter wrote:
>>>>>>"Fernando" == Fernando Perez <Fer...@co...> writes:
> 
> 
>     Fernando> You can try to do:
> 
>     Fernando> import gc gc.collect()
> 
>     Fernando> This will force the garbage collector to kick in, which
>     Fernando> sometimes may help.
> 
> I'm not seeing any leak on linux with my standard memory leak test, eg
> 
>   unit/memleak_hawaii3.py
> 
> in CVS.

Oh, I don't think it's necessarily a leak.  Simply a ton of unreachable stuff 
lying around, which a call to gc.collect() may help with.

Best,

f