matplotlib-users Mailing List for matplotlib

The arrows worked quite well, but I can't figure out how to curve them 
in a natural way, so they don't overlap.  Also, how do I alter the 
breadth of the arrowhead, and to make it the same size along arrows of 
all lengths?

Cheers,
Mike

Jose Gomez-Dans wrote:
> Hi,
> 2009/11/29 Michael Cohen <mc...@ca... <mailto:mc...@ca...>>
> 
>     Hi all,
> 
>     I have a plot that has say 6 black X's, each separate, and 6 blue X's,
>     also separate, denoting where x's 1-6 have moved to (from black to
>     blue).
>     Currently each point is plotted with a separate plot function.
>     I would like to generate a plot where each black x and blue x pair has
>     an arrow pointing from one to the other.
> 
>  
> See this cookbook entry: <http://www.scipy.org/Cookbook/Matplotlib/Arrows>
> 
> What I usually do in these circumstances (and what is demonstrated in 
> the cookbook entry) is to plot the points, and then add an arrow patch 
> for each pair:
> arrow = pylab.Arrow ( x_from, y_from, distance_x, distance_y )
> ax = pylab.gca()
> ax.add_patch(arrow)
> 
> Hope that helps!
> Jose

Wayne Watson skrev:
> So far matplotlib pretty much leaves the user on his own. Yes, there are
> plenty of examples, but there seem to be no explanations about matters
> like figure or show. No tutor on any of this I guess. Apparently, they
> are leaving it up to one's MATLAB experience. It's been quite awhile
> since I used that. It still resides on my computer.
> 

I have mostly used my experience from matlab and other graphics 
packages, along with the examples and documentation available on the 
webpage, but there is now also a book about matplotlib.

http://sandrotosi.blogspot.com/2009/11/matplotlib-for-python-developers.html

/ johan

On 11/29/2009 12:10 PM, Eric Firing wrote:
> quiver(x[:-1], y[:-1], u, v, angles='xy', units='x', scale=1)
>

That works perfectly when there is change only in x,
but not when both coordinates change.  It seems like
I want a `units='xy'` option, to get the right
scaling along each axis, but it does not exist.

However I can "fake it" pretty well by plotting
first with lines and then on top of that doing
quiver(x[1:],y[1:],u,v, pivot='tip', angles='xy')

Thanks,
Alan

I'm on Win XP, and have no idea where the images went. I guess I'll have
to search my C-disk. I just began the search.

Looking at C:\Python25, I see no image files. Same with C:\Python25\Scripts

If I put some code up front as below, the image comes up on my screen,
but using a range of 0 to 8 and 0, 18 changes nothing. Further the
program hangs up.

So far matplotlib pretty much leaves the user on his own. Yes, there are
plenty of examples, but there seem to be no explanations about matters
like figure or show. No tutor on any of this I guess. Apparently, they
are leaving it up to one's MATLAB experience. It's been quite awhile
since I used that. It still resides on my computer.

Ah, the png files were found in the folder where the program is located.
I can clearly see the changes from your switches between range values.

=================inserted a few lines at the top and some print debug
stmts=====
import matplotlib.pyplot as plt
print "let's start the histogram"
plt.hist([1, 2, 3, 4, 5], range=(0, 18))
plt.show()
#raw_input("Hey")
fig = plt.figure()
plt.hist([1, 2, 3, 4, 5], range=(0, 20))
plt.show()
plt.savefig("hist_0_20.png")
plt.close(fig)

fig = plt.figure()
plt.hist([1, 2, 3, 4, 5], range=(0, 8))
plt.savefig("hist_0_8.png")
plt.close(fig)
print "finished"

Johan Grönqvist wrote:
> Wayne Watson skrev:
>   
>> Anyway, where, folder, does your program write the files? I'm not
>> familiar with figure, but apparently using it produces some "canvas"
>> that plt.hist places it's output on. One can than save fig to a file.
>> What happens if I don't use figure? I just put a copy of the line
>> plt.hist([1, 2, 3, 4, 5], range=(0, 20)) after import. When I execute
>> the program, I don't see a graphic appear. So doesn't matplotlib produce
>> graphic output aside from use of figure?
>>
>>     
>
> My intuition is exactly like that.
>
> After importing matplotlib.pyplt as plt, plt.figure creates an object 
> that acts as canvas, and then I plot various things (hist, is one 
> example), and when I am satisfied, I plt.savefig(path) or plt.show() the 
> figure. The figures are saved in the directory where the script is run 
> (its cwd, or current working directory, on linux).
>
> I have also used hist without having a figure, but that was because I 
> wanted the histogram data, i.e., the numbers of data points in the 
> different bins. That script used plt.hist to generate such data, and 
> later (after running fig = plt.figure) used plt.plot to plot parts of 
> the data in different ways.
>
> By the way, matplotlib.pyplot is one way of using matplotlib. There is 
> also the "object oriented interface", which I have never used.
>
>
>
>
>
> / johan
>
>
> ------------------------------------------------------------------------------
> Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
> trial. Simplify your report design, integration and deployment - and focus on 
> what you do best, core application coding. Discover what's new with
> Crystal Reports now.  http://p.sf.net/sfu/bobj-july
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>   

-- 
           Wayne Watson (Watson Adventures, Prop., Nevada City, CA)

             (121.015 Deg. W, 39.262 Deg. N) GMT-8 hr std. time)
              Obz Site:  39° 15' 7" N, 121° 2' 32" W, 2700 feet

                   350 350 350 350 350 350 350 350 350 350
                     Make the number famous. See 350.org
            The major event has passed, but keep the number alive.

                    Web Page: <www.speckledwithstars.net/>

Hi!

>> How can I put the bottom axis on top (or on top AND on bottom) for a barh plot?

> It's somewhat counter-intuitive, but it can be done.
> You have to create some "twin" axes with the "twiny" option,
> then make the plot on the twin axes so it will use the
> top axis. The bottom axis still have to be adjusted manually
> to make it match the top one and remove the labels.
> 
> See this example:
> 

Thanks for your example.  With your help I was able to finally create this with matplotlib,
  http://www.hep.wisc.edu/cms/comp/cmsprod/diskUserUsage.png
and this is my final script,
  http://www.hep.wisc.edu/cms/comp/cmsprod/plotDiskUse.py



One thing that surprised me was to shrink the font for the names required changing ax1 even though the barh plot is on ax2.  See these lines:

----------
  ax2.barh(pos, listUse, color=listColors, height=1.0, align='center')
     ...
  for tick in ax1.yaxis.get_major_ticks():
    tick.label1.set_fontsize(8)
----------

Surprising.
Mike

Wayne Watson skrev:
> Anyway, where, folder, does your program write the files? I'm not
> familiar with figure, but apparently using it produces some "canvas"
> that plt.hist places it's output on. One can than save fig to a file.
> What happens if I don't use figure? I just put a copy of the line
> plt.hist([1, 2, 3, 4, 5], range=(0, 20)) after import. When I execute
> the program, I don't see a graphic appear. So doesn't matplotlib produce
> graphic output aside from use of figure?
> 

My intuition is exactly like that.

After importing matplotlib.pyplt as plt, plt.figure creates an object 
that acts as canvas, and then I plot various things (hist, is one 
example), and when I am satisfied, I plt.savefig(path) or plt.show() the 
figure. The figures are saved in the directory where the script is run 
(its cwd, or current working directory, on linux).

I have also used hist without having a figure, but that was because I 
wanted the histogram data, i.e., the numbers of data points in the 
different bins. That script used plt.hist to generate such data, and 
later (after running fig = plt.figure) used plt.plot to plot parts of 
the data in different ways.

By the way, matplotlib.pyplot is one way of using matplotlib. There is 
also the "object oriented interface", which I have never used.





/ johan

Alan G Isaac wrote:
> I have a collection of 2d points xy.
> Instead of simply plotting the points,
> I'd like to plot a collection of arrows
> that goes from each point to the next.
> The "from" is easy of course;
> the "to" is the problem.
> 
> I think I should be able to use quiver,
> but I'm not getting it quite right.
> E.g.,
> 
> x,y = zip(*xy)
> u,v = np.diff(x), np.diff(y)
> quiver(x[:-1], y[:-1], u, v, angles='xy')
> 
> looks like a start but does not get there.
> I guessed that `scale=1` would complete
> the picture, but that just showed that I
> misunderstood `scale`.
> 
> Suggestions?

Try this:

quiver(x[:-1], y[:-1], u, v, angles='xy', units='x', scale=1)


Eric

Hi, this is an odd result. I copied your code and tried to execute it
with Python 2.5. It came back with a socket area, and a reference to my
firewall. Very strange. Ah, I happen to have a comm port open to another
computer. Apparently Python didn't like that.

Anyway, where, folder, does your program write the files? I'm not
familiar with figure, but apparently using it produces some "canvas"
that plt.hist places it's output on. One can than save fig to a file.
What happens if I don't use figure? I just put a copy of the line
plt.hist([1, 2, 3, 4, 5], range=(0, 20)) after import. When I execute
the program, I don't see a graphic appear. So doesn't matplotlib produce
graphic output aside from use of figure?

I think I've confused hist as part of numpy and histogram as part of
matplotlib. A visit to the numpy and matplotlib web sites shows I am
confused.

Let's see. I want to know how to use a histogram in Python, so there is
a histogram function and a hist. So a histogram is for histograms. What
fun word play. :-) Maybe numpy should have called their function ahistogram!


Johan Grönqvist wrote:
> Wayne Watson skrev:
>   
>> See Subject. I don't seem able to produce a simple example of using 
>> histogram that uses range. I tried a variety of ranges, range=(0,22), 
>> range=(0, 50.2), ... and I see no difference between any of the x values 
>> scale. Can someone provide an example that shows how it works?
>>
>>     
>
> The script below works for me. It produces two files with different 
> limits on the x-axis and corresponding histograms.
>
>
> ------------
> import matplotlib.pyplot as plt
>
> fig = plt.figure()
> plt.hist([1, 2, 3, 4, 5], range=(0, 20))
> plt.savefig("hist_0_20.png")
> plt.close(fig)
>
> fig = plt.figure()
> plt.hist([1, 2, 3, 4, 5], range=(0, 8))
> plt.savefig("hist_0_8.png")
> plt.close(fig)
> -------------
>
> Hope it helps
>
> / johan
>
>
> ------------------------------------------------------------------------------
> Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
> trial. Simplify your report design, integration and deployment - and focus on 
> what you do best, core application coding. Discover what's new with
> Crystal Reports now.  http://p.sf.net/sfu/bobj-july
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>   

-- 
           Wayne Watson (Watson Adventures, Prop., Nevada City, CA)

             (121.015 Deg. W, 39.262 Deg. N) GMT-8 hr std. time)
              Obz Site:  39° 15' 7" N, 121° 2' 32" W, 2700 feet

                   350 350 350 350 350 350 350 350 350 350
                     Make the number famous. See 350.org
            The major event has passed, but keep the number alive.

                    Web Page: <www.speckledwithstars.net/>

I have a collection of 2d points xy.
Instead of simply plotting the points,
I'd like to plot a collection of arrows
that goes from each point to the next.
The "from" is easy of course;
the "to" is the problem.

I think I should be able to use quiver,
but I'm not getting it quite right.
E.g.,

x,y = zip(*xy)
u,v = np.diff(x), np.diff(y)
quiver(x[:-1], y[:-1], u, v, angles='xy')

looks like a start but does not get there.
I guessed that `scale=1` would complete
the picture, but that just showed that I
misunderstood `scale`.

Suggestions?

Alan Isaac

Hi,
2009/11/29 Michael Cohen <mc...@ca...>

> Hi all,
>
> I have a plot that has say 6 black X's, each separate, and 6 blue X's,
> also separate, denoting where x's 1-6 have moved to (from black to blue).
> Currently each point is plotted with a separate plot function.
> I would like to generate a plot where each black x and blue x pair has
> an arrow pointing from one to the other.
>

See this cookbook entry: <http://www.scipy.org/Cookbook/Matplotlib/Arrows>

What I usually do in these circumstances (and what is demonstrated in the
cookbook entry) is to plot the points, and then add an arrow patch for each
pair:
arrow = pylab.Arrow ( x_from, y_from, distance_x, distance_y )
ax = pylab.gca()
ax.add_patch(arrow)

Hope that helps!
Jose

Hi,
I have a strong preference to working with the keyboard instead of the mouse.
Today I was working on a script to label a sequence of images (mark position and
scale of a moving object). The default keyboard shortcuts are getting in my way,
many keys are already used and moreover they are spread all over the keyboard.
Looking at the doc, I could't find an easy way to customize or completely
disable them. If someone can suggest a (not too coonvoluted) solution, I'd love
to hear it.

Cheers,

Jorge

Wayne Watson skrev:
> See Subject. I don't seem able to produce a simple example of using 
> histogram that uses range. I tried a variety of ranges, range=(0,22), 
> range=(0, 50.2), ... and I see no difference between any of the x values 
> scale. Can someone provide an example that shows how it works?
> 

The script below works for me. It produces two files with different 
limits on the x-axis and corresponding histograms.


------------
import matplotlib.pyplot as plt

fig = plt.figure()
plt.hist([1, 2, 3, 4, 5], range=(0, 20))
plt.savefig("hist_0_20.png")
plt.close(fig)

fig = plt.figure()
plt.hist([1, 2, 3, 4, 5], range=(0, 8))
plt.savefig("hist_0_8.png")
plt.close(fig)
-------------

Hope it helps

/ johan

Hi Michael,

Take a look at the quiver demo
<http://matplotlib.sourceforge.net/examples/pylab_examples/quiver_demo.html>

and the annotation2 demo
<http://matplotlib.sourceforge.net/examples/pylab_examples/quiver_demo.html>

More generally, have a look through the examples and gallery pages
<http://matplotlib.sf.net/examples/index.html>
<http://matplotlib.sf.net/gallery.html>

Gary R.

Michael Cohen wrote:
> Hi all,
> 
> I have a plot that has say 6 black X's, each separate, and 6 blue X's, 
> also separate, denoting where x's 1-6 have moved to (from black to blue).
> Currently each point is plotted with a separate plot function.
> I would like to generate a plot where each black x and blue x pair has 
> an arrow pointing from one to the other.
> 
> Currently I plot them like this:
> 
> x1black = value
> y1black = value
> plot([x1black],[y1black],'kx',markersize=10,markeredgewidth=2)
> x1blue = value
> y1blue = value
> plot([x1blue],[y1blue],'bx',markersize=10,markeredgewidth=2)
> 
> If I plotted,
> plot([x1black,x1blue],[y1black,y1blue])
> I could make the line between them into an arrow, but I wouldn't be able 
> to make one blue and the other black.
> 
> Also, I'd like to be able to curve my arrows to make them less confusing 
> in case they intersect too much.
> 
> 
> Can anyone point me to the right functions?
> 
> Cheers
> Michael

Hi all,

I have a plot that has say 6 black X's, each separate, and 6 blue X's, 
also separate, denoting where x's 1-6 have moved to (from black to blue).
Currently each point is plotted with a separate plot function.
I would like to generate a plot where each black x and blue x pair has 
an arrow pointing from one to the other.

Currently I plot them like this:

x1black = value
y1black = value
plot([x1black],[y1black],'kx',markersize=10,markeredgewidth=2)
x1blue = value
y1blue = value
plot([x1blue],[y1blue],'bx',markersize=10,markeredgewidth=2)

If I plotted,
plot([x1black,x1blue],[y1black,y1blue])
I could make the line between them into an arrow, but I wouldn't be able 
to make one blue and the other black.

Also, I'd like to be able to curve my arrows to make them less confusing 
in case they intersect too much.


Can anyone point me to the right functions?

Cheers
Michael

Hi All, 
I'm trying to create a set of strip charts so that I can see the relationships between a large number of time series.  I'd like to label the y-axis with a name on the left side and a value/percentile on the right hand side.  I can get the names on the left axis, and it looks very nice (yay Matplotlib!).  When I try to the code below I don't get the output I'd like.  None of the explicit tick labels that I'm setting make it to the display, and also the 3 charts on the page aren't kept separate (left tick labels from chart 132 hit the right tick labels from chart 131).  Any suggestions very appreciated. 

import numpy as np
import matplotlib.pyplot as plt

nameN = 25
obsN = 50

names = ["foo%d" % x for x in range(nameN)]
vols = np.random.uniform(0.2, 0.45, nameN)


labels = ['ValA', 'ValB', 'ValC']
d1 = np.random.normal(0,1, obsN*nameN).reshape(obsN, -1)*vols
d2 = np.random.normal(0,1, obsN*nameN).reshape(obsN, -1)*vols
d3 = np.random.normal(0,1, obsN*nameN).reshape(obsN, -1)*vols


f = 0.45
fig = plt.figure()
for label, dset, ax_id in zip(labels, (d1, d2, d3), (131, 132, 133)):
    
    ax = fig.add_subplot(ax_id)

    for i in range(len(names)):
        
        mx = np.max(dset[:,i])
        mn = np.min(dset[:,i])
        y = i + dset[:,i]*2*f/(mx - mn) - f*(mx + mn)/(mx - mn) + 1
        y_last = y[-1]*np.ones_like(y)
        x = np.arange(obsN)
        ax.fill_between(x, y_last, y)

    ax.set_ylim((0, nameN + 1))
    ax.set_yticks(range(1, nameN + 1))
    ax.set_title(label, fontsize=10)

    for tk, nm in zip(ax.yaxis.get_major_ticks(), list(reversed(names))):
        tk.tick1On = False
        tk.tick2On = False
        tk.label1On = True
        tk.label2On = True
        tk.set_label1(nm)
        tk.set_label2("%.1f\n(%.1f tile)" % (36.2, 98.6))
    
    #ytickNames = ax.set_yticklabels(list(reversed(names)), rotation = 45, fontsize = 8)
    
    
plt.show()