matplotlib-users Mailing List for matplotlib

Hi Manuel,
On Fri, Apr 18, 2008 at 8:49 AM, Manuel Metz wrote:
>
> Andrea Gavana wrote:
> > Hi All,
> >
> > I was wondering about custom markers in the scatter method, and I
> > thought to ask here for some suggestions.
> > Basically, I have 3 variables to show, which are oil, gas and water
> > production. I would like to define the bubble size by the sum of these
> > three variables (or something akin), and then I would like to be able
> > to split the bubble marker in 3 sections (like having a small pie
> > chart in place of the marker), with each section area proportional to
> > the value of oil, gas and water production respectively. I know this
> > might sound not very clear, I attach a small picture of what I mean
> > (the picture shows the bubble divided into 2 sections, but the purpose
> > is the same).
> > I know about custom marker, but I am not so expert with matplotlib to
> > be able to implement it... could someone please share some suggestion
> > on how to do this?
> >
> > Thank you very much.
> >
> > Andrea.
> >
>
> Hi,
> as you already suggested, you have to do a little hand-work, but its not too
> hard. I attached an example, which you can use as a starting point...
Thank you for the sample, this is exactly wht I was looking for. Thank
you very much!
Andrea.
"Imagination Is The Only Weapon In The War Against Reality."
http://xoomer.alice.it/infinity77/

Fernando,
Your example works as you describe on recent matplotlib versions. I 
suspect you are using an old one. The preferred way of handling missing 
points in numpy, and therefore in matplotlib and pylab, however, is via 
masked arrays.
import pylab
import numpy as np
from numpy import ma
a = [1,2,3,4,5]
b = np.array([6,2,np.nan,1,9])
bm = ma.masked_where(np.isnan(b), b)
pylab.plot(a,bm)
pylab.show()
There are many other examples of masked array use in the examples 
directory of the matplotlib distribution.
Eric
Fernando Abilleira wrote:
> Dear sourceforge community,
> 
> I come from a Matlab environment so I am used to plotting matrices that 
> contain NaN elements. This is very useful because in some cases one 
> doesn't have data for the entire matrix. If one tries plotting the data, 
> the NaN elements won't be plotted.
> 
> Is there a similar element type or workaround I could use to get the 
> same effect?
> 
> In the following simple example:
> 
> a = [1,2,3,4,5]
> b = [6,2,NaN,1,9]
> mpylab.plot(a,b)
> 
> I would like to get two lines with a gap in between them at element [2].
> 
> Thanks for any help you can offer.
> 
> Regards,
> 
> Fernando

When I increase the xlabel fontsize, it overlaps with the tick label.
How can I adjust the distance between tick label and xlable (or ylabel)?
Please help me.
-- 
Yong-Duk Jin

Vsevolod Kovalenko wrote:
> Dear matplotlib
> 
> I have a newbie question (I am new to python, migrating from Matlab). Just
> like the subject suggests, I need to display an image X stored in a matrix
> (well, a 2D numpy array in fact) versus the coordinates specified by arrays
> x and y. Further, it would be nice if mouse position was returned in terms
> of those coordinates as well, but that's just for a nicety - I can map it
> myself. But main thing is I failed to find the way to get the image
> versuscoordinates in imshow() and figimage(). I also failed to find the
> answer in this forum.
> 
> Hence I ask for any suggestions.
There is not an *exact* clone of imagesc, but you can certainly get the 
same effect. Note that imshow has an origin kwarg that lets you specify 
the vertical orientation, and an extent kwarg that controls the mapping 
between pixels and your X and Y scales. If you want direct indexing 
into a colormap you can get it with the kwarg norm=mpl.colors.NoNorm(), 
provided your Z array has an integer dtype and includes values within 
the range of the colormap; the default colormaps all have 256 colors.
An alternative, if you want to specify the boundaries of the pixels 
rather than their centers (or if you have unevenly-spaced boundaries), 
is pcolor, or pcolormesh. If you have a recent version of mpl you can 
use the Axes method pcolorfast.
Offhand, though, it sounds like imshow() with suitable kwargs will do 
exactly what you want. See image_demo2.py in the mpl examples directory.
Eric