matplotlib-users Mailing List for matplotlib

Well I did tried this but it didn't work out. It actually removes the 
black color from the color_cycle but the pie still prints it. Moreover I 
noticed that this color_cycle has 7 colors that repeats after the first 
7 so it won't do what I want. I will need to find another way to set the 
colors or just avoid the pie.

Thanks though for the help!

Regards
Grigoris

On 11/30/2011 06:09 PM, Tony Yu wrote:
> On Wed, Nov 30, 2011 at 10:34 AM, Grigoris Maravelias 
> <gr....@gm... <mailto:gr....@gm...>> wrote:
>
>     Hello list!
>
>     I have a question regarding the colors of the pie diagram of
>     matplotlib. When no colors are assigned then the pie function
>     automatically selects some colors, like the example image I have
>     attached. But in this case the black color covers the text. How
>     can we avoid this?Is there an easy (perhaps?) way to exclude a color?
>
>
> I don't really use pie charts, but I think it just uses the default 
> color cycle. This can be altered by changing the rcParams:
>
> >>> import matplotlib.pyplot as plt
> >>> plt.rcParams['axes.color_cycle'].remove('k')
>
> The color_cycle parameter is just a python list, so I use list.remove 
> to remove black (which is the letter 'k' since 'b' is blue). There are 
> other ways of setting rcParams, as detailed in the help files 
> <http://matplotlib.sourceforge.net/users/customizing.html> (Note that 
> `rc` and `rcParams` is in both matplotlib.pyplot and the main 
> matplotlib package).
>
> Best,
> -Tony

On Wed, Nov 30, 2011 at 11:42 AM, Jeffrey Blackburne <
jbl...@al...> wrote:

> Hi Steven,
>
> Try this:
>
> import numpy as np
> import numpy.random
> import matplotlib as mpl
> import matplotlib.pyplot as plt
>
> x = np.random.randn(1000)
> h, binedg = np.histogram(x, 10)
>
> wid = binedg[1:] - binedg[:-1]
> plt.bar(binedg[:-1], h/float(x.size), width=wid)
>
>
> On Nov 30, 2011, at 10:25 AM, Steven Boada wrote:
>
> > Hi Users,
> >
> > I'm looking to make a histogram that is normalized by the total number
> > of items shown in the histogram. For example:
> >
> > Let's say that I have an array 1000 items long. If I make a
> > histogram in
> > the normal way hist(x,10) then I get a histogram showing the total
> > number of items in each bin. What I want to do is take that total
> > number
> > in each bin and divide them by 1000 and then make the plot.
> >
> > So if one of my bins has 350 objects in it, then it would be
> > changed to
> > 0.35.
> >
> > Another way to say it would be that I want the height of the histogram
> > to represent the fraction of the total. I am pretty sure that this is
> > different than using the "normed=True" flag, but I couldn't find
> > anyone
> > talking about this when I searched.
> >
> > Thanks
> >
> > Steven
>
>
One option: You can plot the normal `hist` and then change the tick labels
appropriately. Here's some code for accomplishing that:

#~~~
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.ticker import FuncFormatter, MultipleLocator


N = 350
ytick_step = 0.05
data = np.random.normal(size=N)

def norm_num(x, pos):
    return '%g' % (x / float(N))

locator = MultipleLocator(N * ytick_step)
formatter = FuncFormatter(norm_num)

f, ax = plt.subplots()
ax.yaxis.set_major_formatter(formatter)
ax.yaxis.set_major_locator(locator)
ax.hist(data)

plt.show()
#~~~

Note that the formatter object is all you need to change to the desired
scale. But, that result will usually look ugly, because you'll get tick
labels with long, ugly floating point numbers. The locator object fixes
that issue.

Best,
-Tony

Hi Steven,

Try this:

import numpy as np
import numpy.random
import matplotlib as mpl
import matplotlib.pyplot as plt

x = np.random.randn(1000)
h, binedg = np.histogram(x, 10)

wid = binedg[1:] - binedg[:-1]
plt.bar(binedg[:-1], h/float(x.size), width=wid)


On Nov 30, 2011, at 10:25 AM, Steven Boada wrote:

> Hi Users,
>
> I'm looking to make a histogram that is normalized by the total number
> of items shown in the histogram. For example:
>
> Let's say that I have an array 1000 items long. If I make a  
> histogram in
> the normal way hist(x,10) then I get a histogram showing the total
> number of items in each bin. What I want to do is take that total  
> number
> in each bin and divide them by 1000 and then make the plot.
>
> So if one of my bins has 350 objects in it, then it would be  
> changed to
> 0.35.
>
> Another way to say it would be that I want the height of the histogram
> to represent the fraction of the total. I am pretty sure that this is
> different than using the "normed=True" flag, but I couldn't find  
> anyone
> talking about this when I searched.
>
> Thanks
>
> Steven
>
>
> ---------------------------------------------------------------------- 
> --------
> All the data continuously generated in your IT infrastructure
> contains a definitive record of customers, application performance,
> security threats, fraudulent activity, and more. Splunk takes this
> data and makes sense of it. IT sense. And common sense.
> http://p.sf.net/sfu/splunk-novd2d
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

On Wed, Nov 30, 2011 at 10:34 AM, Grigoris Maravelias <
gr....@gm...> wrote:

> Hello list!
>
> I have a question regarding the colors of the pie diagram of matplotlib.
> When no colors are assigned then the pie function automatically selects
> some colors, like the example image I have attached. But in this case the
> black color covers the text. How can we avoid this?Is there an easy
> (perhaps?) way to exclude a color?
>

I don't really use pie charts, but I think it just uses the default color
cycle. This can be altered by changing the rcParams:

>>> import matplotlib.pyplot as plt
>>> plt.rcParams['axes.color_cycle'].remove('k')

The color_cycle parameter is just a python list, so I use list.remove to
remove black (which is the letter 'k' since 'b' is blue). There are other
ways of setting rcParams, as detailed in the help
files<http://matplotlib.sourceforge.net/users/customizing.html>(Note
that `rc` and `rcParams` is in both matplotlib.pyplot and the main
matplotlib package).

Best,
-Tony

i use something like that:

from datetime import datetime, timedelta
import matplotlib.pyplot as pl
import matplotlib.ticker as ticker
import matplotlib.dates as mdates

dates1 = [datetime(2005,5,11)+n*timedelta(days=1) for n in range(500)]
dates2 = [datetime(...

ax1 = pl.subplot(2,1,1)
ax1.plot(dates1, y1, 'r')
ax1.fmt_xdata = mdates.DateFormatter('%Y-%m-%d')

ax2 = pl.subplot(2,1,2)
ax2.plot(dates2, ...

that's what you want?

regards, yoshi

Hello list!

I have a question regarding the colors of the pie diagram of matplotlib. 
When no colors are assigned then the pie function automatically selects 
some colors, like the example image I have attached. But in this case 
the black color covers the text. How can we avoid this?Is there an easy 
(perhaps?) way to exclude a color?

Of course there is not problem if I specifically select the colors but I 
do not know how many parts exist beforehand (and I would like to assign 
it automatically).

Thanks!

Regards

Grigoris

Hi Users,

I'm looking to make a histogram that is normalized by the total number 
of items shown in the histogram. For example:

Let's say that I have an array 1000 items long. If I make a histogram in 
the normal way hist(x,10) then I get a histogram showing the total 
number of items in each bin. What I want to do is take that total number 
in each bin and divide them by 1000 and then make the plot.

So if one of my bins has 350 objects in it, then it would be changed to 
0.35.

Another way to say it would be that I want the height of the histogram 
to represent the fraction of the total. I am pretty sure that this is 
different than using the "normed=True" flag, but I couldn't find anyone 
talking about this when I searched.

Thanks

Steven

I have a short script to plot 20 years of river flow data. I can use
the plot_date command to create a plot, using this snippet:
f = figure()
ax1 = f.add_subplot(111)
ax1.plot_date(dates0,y1,'g', label='observed', xdate=True,visible=True)
ax1.plot_date(dates1,y2,'r', label='simulated', xdate=True,visible=True)
years    = YearLocator(1, month=6, day=30)   # every year
months   = MonthLocator(1)  # every month
ax1.set_xlim(date2num(datetime.date(1990,1,1)),date2num(datetime.date(1999,12,31)))
ax1.xaxis.set_major_locator(years)
ax1.xaxis.set_minor_locator(months)
labels = ax1.get_xticklabels()
setp(labels, fontsize=8,visible=True)

The problem is with the x-axis (time axis) labels when I add a second
subplot, to add the next time segment. I change the above to ax1 =
f.add_subplot(211), and then:
ax2 = f.add_subplot(212)
ax2.plot_date(dates0,y1,'g')  #plots the time series
ax2.plot_date(dates1,y2,'r')  #need to call twice, unlike plot,
plot_date takes one set
ax2.set_xlim(date2num(datetime.date(2000,1,1)),date2num(datetime.date(2009,12,31)))
ax2.xaxis.set_major_locator(years)
ax2.xaxis.set_minor_locator(months)
setp(labels, fontsize=8,visible=True)

The x-axis labels only appear for the last subplot. I'm guessing that
plot_date assumes that more than one subplot must share a time axis.
There must be a simple way to stop plot_date from doing this, if this
is indeed the problem. Any guidance would be appreciated.

Thanks,
Ed

On Mon, Nov 28, 2011 at 10:49 PM, Jae-Joon Lee <lee...@gm...> wrote:
> This is a bug. In the current implementation, "annotate" has a
> side-effect that modifies the arrowprops dictionary.

For a future reference, this should now be fixed in the v1.1.x branch
which also has been merged into the master branch.

https://github.com/matplotlib/matplotlib/commit/b3a2ab77c89fdb3ab860edeb1a781b5307347070

Regards,

-JJ

On 11/29/2011 01:29 PM, Nat Echols wrote:
> I'm plotting values that cover a very small range with a relatively
> large base, e.g.
>
> 375.0001
> 375.00025
> 375.0002
> ...
>
> In practice, the data series hovers at a single value for several
> hundred elements in a row, then fluctuates slightly.  Initially
> matplotlib does what I expect, and the Y-axis ticks are labeled 373,
> 374, 376, 376, etc.  Once the small fluctuations are plotted, however,
> it switches to showing relative values, with the absolute reference
> point above the plot, as in the attached image.  This is needlessly
> confusing - is there a way to prevent such behavior?
>
> thanks,
> Nat

If you are using mpl 1.0 or later, try

pyplot.ticklabel_format(useOffset=False)

It is also available as an Axes method.

Eric
>
>
>
> ------------------------------------------------------------------------------
> All the data continuously generated in your IT infrastructure
> contains a definitive record of customers, application performance,
> security threats, fraudulent activity, and more. Splunk takes this
> data and makes sense of it. IT sense. And common sense.
> http://p.sf.net/sfu/splunk-novd2d
>
>
>
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

I'm plotting values that cover a very small range with a relatively
large base, e.g.

375.0001
375.00025
375.0002
...

In practice, the data series hovers at a single value for several
hundred elements in a row, then fluctuates slightly.  Initially
matplotlib does what I expect, and the Y-axis ticks are labeled 373,
374, 376, 376, etc.  Once the small fluctuations are plotted, however,
it switches to showing relative values, with the absolute reference
point above the plot, as in the attached image.  This is needlessly
confusing - is there a way to prevent such behavior?

thanks,
Nat

I would like to suggest the following fixes for the
mpl_toolkits.basemap module to improve its treatment of celestial
(rather than geographic) coordinates.


The first one, posted at
https://github.com/mollyswanson/basemap/commit/23db4bbebf4d7fe6ca202b5dad50b6a2054dd685
changes the call function in basemap's init.py to correctly transform
lat/lon values into xy map coordinates in the case of a cyclic or
polycyclic projection with lon_0 not equal to 0.

The second, posted at
https://github.com/mollyswanson/basemap/commit/35470b51523e9429d26cefc911dca843264581b9
changes one line in the drawparallels. This line is to avoid drawing
lines between points on the parallel that span the whole map. However,
the old version uses a fixed value for the distance between the points
rather than scaling it to the radius of the sphere used in the
projection, so if you use a non-default radius (such as 180/pi, so
your x-y values are in degrees on the sky instead of meters on the
earth) it won't work. This fix scales the cutoff value to the radius
of the projection sphere.

The following example illustrates the issues that are addressed here:

from mpl_toolkits.basemap import Basemap
import matplotlib.pyplot as plt
figure(1)
#make a basemap centered on longitude of 90
m=Basemap(celestial=False,lon_0=90,projection='hammer')
#draw map boundary and grid
m.drawmapboundary()
m.drawparallels(np.arange(-90.,91.,30.),labels=[1,0,0,0])
m.drawmeridians(np.arange(-90.,271.,30.),labels=[0,0,0,0])
#define a test polygon - a triangle with corners at
[lon,lat]=[90,30],[120,60],[120,30]
polygon=array([[90,30],[120,60],[120,30],[90,30]])
#convert to map coordinates
polyxy=m(polygon[:,0],polygon[:,1])
plt.plot(polyxy[0],polyxy[1])
plt.savefig('basemap1.png')
figure(2)
#make a celestial basemap centered on longitude of 90
m=Basemap(celestial=True,lon_0=90,projection='hammer',rsphere=180./pi)
#draw map boundary and grid
m.drawmapboundary()
m.drawparallels(np.arange(-90.,91.,30.),labels=[1,0,0,0])
m.drawmeridians(np.arange(-90.,271.,30.),labels=[0,0,0,0])
#define a test polygon - a triangle with corners at
[lon,lat]=[90,30],[120,60],[120,30]
polygon=array([[90,30],[120,60],[120,30],[90,30]])
#convert to map coordinates
polyxy=m(polygon[:,0],polygon[:,1])
plt.plot(polyxy[0],polyxy[1])
plt.savefig('celestial_basemap1.png')

Thank you!
Molly Swanson

I have a short script to plot 20 years of river flow data. I can use
the plot_date command to create a plot, using this snippet:
f = figure()
ax1 = f.add_subplot(111)
ax1.plot_date(dates0,y1,'g', label='observed', xdate=True,visible=True)
ax1.plot_date(dates1,y2,'r', label='simulated', xdate=True,visible=True)
years    = YearLocator(1, month=6, day=30)   # every year
months   = MonthLocator(1)  # every month
ax1.set_xlim(date2num(datetime.date(1990,1,1)),date2num(datetime.date(1999,12,31)))
ax1.xaxis.set_major_locator(years)
ax1.xaxis.set_minor_locator(months)
labels = ax1.get_xticklabels()
setp(labels, fontsize=8,visible=True)

The problem is with the x-axis (time axis) labels when I add a second
subplot, to add the next time segment. I change the above to ax1 =
f.add_subplot(211), and then:
ax2 = f.add_subplot(212)
ax2.plot_date(dates0,y1,'g')  #plots the time series
ax2.plot_date(dates1,y2,'r')  #need to call twice, unlike plot,
plot_date takes one set
ax2.set_xlim(date2num(datetime.date(2000,1,1)),date2num(datetime.date(2009,12,31)))
ax2.xaxis.set_major_locator(years)
ax2.xaxis.set_minor_locator(months)
setp(labels, fontsize=8,visible=True)

The x-axis labels only appear for the last subplot. I'm guessing that
plot_date assumes that more than one subplot must share a time axis.
There must be a simple way to stop plot_date from doing this, if this
is indeed the problem. Any guidance would be appreciated.

Thanks,
Ed

Hello all,

I am struggling to improve the layout of these subplots.  You can see that the x tick labels on the left subplot are crunched.  I simply want to stretch the x axis a bit in order to spread out the x ticks.  Do you have any suggestions?

Thanks!

Neil
-----

Here is my code:

P.figure()
P.subplots_adjust(wspace=0.4)

P.subplot(121)
P.hist([el_mean_diff,ln_mean_diff],bins=[-.7,-.6,-.5,-.4,-.3,-.2,-.1,0,.1,.2,.3,.4,.5,.6,.7],color=['red','blue'],align='mid')
P.title('a) mean')
P.xlabel('Average difference at each grid point (m/s) ')
P.ylabel('Number of grid points')
P.tick_params(axis='both',direction='out',top='off',right='off',length=2,labelsize='10')
P.grid(True)

P.subplot(122)
P.hist([el_stdv_diff,ln_stdv_diff],bins=[-.5,-.4,-.3,-.2,-.1,0,.1,.2,.3,.4,.5],color=['red','blue'],align='mid')
P.title('b) standard deviation')
P.xlabel('Average difference at each grid point (m/s) ')
P.ylabel('Number of grid points')
P.tick_params(axis='both',direction='out',top='off',right='off',length=2,labelsize='10')
P.grid(True)

P.show()
----

On Tue, Nov 29, 2011 at 4:16 AM, Fabien Lafont <laf...@gm...> wrote:
> Hello everyone,
>
> I don't understand how works TimerBase.
>
>
>
> >From matplotlib import backend_bases
>
> def write(x):
>     print x
>
> backend_bases.TimerBase._timer_start
> backend_bases.TimerBase(1000,write(2))

TimerBase is a do-nothing skeleton class that provides the common
infrastructure for other backends to implement a timer that works with
them (just like the rest of backend_bases). For example, the gtk
backend uses this as a starting point for its own timer class. You
really shouldn't be instantiating TimerBase yourself as it won't do
anything.

> It returns only "2" one time. Why it doesn't return 2 every second?

The only reason you actually see anything at all is because you call
write yourself when you do:

write(2)

The timer never actually does anything. The proper call is to separate
the function and its arguments, since as the docs say, it takes a
"list of (func, args) tuples that will be
called upon timer events":

TimerBase(1000, [(write, 2)])

However, the proper way to create a timer, which will intergrate
properly with the figure event loop, is shown in the example:

http://matplotlib.sourceforge.net/examples/event_handling/timers.html

In your case:

timer = fig.canvas.new_timer(interval=1000)
timer.add_callback(write, 2)
timer.start()

Ryan

-- 
Ryan May
Graduate Research Assistant
School of Meteorology
University of Oklahoma

Hello everyone,

I don't understand how works TimerBase.



>From matplotlib import backend_bases

def write(x):
     print x

backend_bases.TimerBase._timer_start
backend_bases.TimerBase(1000,write(2))





It returns only "2" one time. Why it doesn't return 2 every second?



Thx in advance,

Fabien

> This is a bug. In the current implementation, "annotate" has a
> side-effect that modifies the arrowprops dictionary.
> As a workaround, you may do,
>
> arrowprops = dict(arrowstyle='-', relpos=(0, 1))
> plt.annotate('Good relpos', (3, 3), xytext = (3, 2),
>
             annotation_clip=False, arrowprops=arrowprops.copy())
>
>
Works for me. Thanks a lot!


>
> > plt.annotate('No ha/va', (5, 5), xytext = (5, 4),
> > arrowprops=dict(arrowstyle='-'),
> >              ha='left', va='top')
> >
>
> ha and va controls the location of the text relative to the xytext,
> and I believe it does work as expected. It has nothing to do with the
> starting point of the arrow, which should be controlled by the relpos
> parameter.
>

Thanks for clarifying.

Regards,

Markus

Hello,

Are there examples of extending imshow or further customizing any of
the UI components of matpotlib?  Or anything you can point me at to
get started?  I have been able to connect events to imshow.  An
example of the type of thing I might want to do is to print the R,G,B
value in a status bar when I hover over an x y coordinate of an image.

-- 
Austin Godber

As related to another question(s) I've posted, can someone help with this
custom formatter?  This is for use in a FunctionFormatter to be used on the
y axis to format the ticks (in particular, to remove them when not
wanted).  Example:

def CustomFormatter(self,y,i):
        if y < 0:
            return ''

This says if the value of y < 0, put nothing on the axis tick.

QUESTION:  How can I implement the following two sorts of rules in
Matplotlib's (OO) API?

def CustomFormatter(self,y,i):
        if y falls in the bottom 50 pixels' worth of height of this plot:
            return ''

or

def CustomFormatter(self,y,i):
        if y falls in the bottom 10% of the height of this plot:
            return ''

Thanks,
Che

Hello,

 

I am fairly new to Matplotlib and appreciate your help with my question:


 

I am looking to generate a time series trend plots and include as my
xtick labels a selective list of strings. For ex: Month/Year label for
data corresponding to the 1st day of each month. I may be plotting 365
days of data but want to show only 12 xtick labels. 

 

Would it be possible to specify grid lines for every 7 days of data for
instance? This makes the grid line locations different from the xtick
label locations. 

 

I am sure there is support for such a feature in Matplotlib. I did look
through the docs/forums but could not find much help.

Any suggestions will be much appreciated. 

 

Thanks,
Shankar

> 
> From: "Marianne C." <mar...@gm...>
> Date: November 24, 2011 6:48:34 AM PST
> To: mat...@li...
> Subject: [Matplotlib-users] Removing ticks and frame (imshow)
> 
> 
> Hi all,
> 
> My name is Marianne, I am a beginner user of matplotlib.
> I am using imshow in pyplot.  I am desperate to get rid of
> the ticks on both x and y axes (see attached picture).  I
> do not need the black box around the data either.  Should
> I use imshow in axes.Axes instead, to be able to call
> set_ticks_position("none")?
> Thank you for your help,
> Marianne
> 
> Here is the code so far:
> 
> import numpy
> from matplotlib import pyplot
> 
> q=numpy.loadtxt('field.txt')
> 
> myfield = pyplot.imshow(q,aspect=1)
> myfield.set_clim(vmin=0, vmax=0.6)
> 
> pyplot.colorbar()
> 
> pyplot.savefig('field_1.eps')
> 
> <field_1.pdf>

Marianne,

Try

myfield.get_axes().axis('off')

-Sterling

On Thu, Nov 24, 2011 at 9:48 AM, Marianne C. <mar...@gm...> wrote:

> Hi all,
>
> My name is Marianne, I am a beginner user of matplotlib.
> I am using imshow in pyplot.  I am desperate to get rid of
> the ticks on both x and y axes (see attached picture).  I
> do not need the black box around the data either.  Should
> I use imshow in axes.Axes instead, to be able to call
>
> set_ticks_position("none")?
>
> Thank you for your help,
> Marianne
>
> Here is the code so far:
>
> import numpy
> from matplotlib import pyplot
>
> q=numpy.loadtxt('field.txt')
>
> myfield = pyplot.imshow(q,aspect=1)
> myfield.set_clim(vmin=0, vmax=0.6)
>
> pyplot.colorbar()
>
> pyplot.savefig('field_1.eps')
>


There are a number of ways to accomplish this, but the one I use is to make
the x and y axes invisible (gets rid of the ticks) and also make the spines
invisible (gets rid of the lines). I just throw these changes into a
utility function (`clear_frame` below) and put that in a module that's on
my python path so that it's easily reusable.

Hope that helps,
-Tony

#~~~
import numpy as np
import matplotlib.pyplot as plt

def clear_frame(ax=None):
    if ax is None:
        ax = plt.gca()
    ax.xaxis.set_visible(False)
    ax.yaxis.set_visible(False)
    for spine in ax.spines.itervalues():
        spine.set_visible(False)

img = np.random.random((100,100))
plt.imshow(img)
clear_frame()

plt.colorbar()

plt.show()
#~~~

On 11/22/11 12:51 PM, Dave Xia wrote:
> Hi,
>
> I am a new user of matplotlib Basemap. I tried to draw  the latitude 
> with interval 0.1 degree using drawparallels, but failed.  I wonder if 
> the drawparallels can draw latitude lines with small interval instead 
> of integer values. (drawmeridian can work with small interval 0.1).
>
> Does anyone have any idea to fix this issue?
>
> Thanks!
>
> Dave
Confirmed.  This is now fixed in github master.  It's a one-liner, so 
you can manually patch your source if you prefer.

Just change line 2067 in lib/mpl_toolkits/basemap/__init__.py from

if t is not None: linecolls[int(lat)][1].append(t)

to

if t is not None: linecolls[lat][1].append(t)

Regards,

Jeff

On Mon, Nov 28, 2011 at 9:39 PM, Markus Baden <mar...@gm...> wrote:
> anno_args = {
>     'annotation_clip': False,
>     'arrowprops': dict(arrowstyle='-', relpos=(0, 1)),
>     }
> plt.annotate('Good relpos', (3, 3), xytext = (3, 2), **anno_args)
> plt.annotate('Bad relpos', (6, 6), xytext = (6, 5), **anno_args)

This is a bug. In the current implementation, "annotate" has a
side-effect that modifies the arrowprops dictionary.
As a workaround, you may do,

arrowprops = dict(arrowstyle='-', relpos=(0, 1))
plt.annotate('Good relpos', (3, 3), xytext = (3, 2),
             annotation_clip=False, arrowprops=arrowprops.copy())


> plt.annotate('No ha/va', (5, 5), xytext = (5, 4),
> arrowprops=dict(arrowstyle='-'),
>              ha='left', va='top')
>

ha and va controls the location of the text relative to the xytext,
and I believe it does work as expected. It has nothing to do with the
starting point of the arrow, which should be controlled by the relpos
parameter.

Regards,

-JJ

Hi list,

I'm trying to annotate points on a graph by drawing a simple line from the
point on the axis to the top left corner of the text. I can't figure out,
how to use pyplot.annotate so that it turns of the arrow head and I can use
horizontalalignment (ha) and verticalalignment (va). When I use
arrowstyle='-' in the arrowprops dictionary ha and va don't work. Instead I
use relpos=(0, 1) in the arrowprops dictionary, which works, but only when
I call the annotate function the first time. Below is a minimal example.

I'm using mpl version 1.0. as part of EPD 7.1 on Mac OS X 10.5.

Any hints on how to achieve my goal would be greatly appreciated!

Best regards,

Markus

---

The following code reproduces my problem

import numpy as np
import matplotlib.pyplot as plt

data = np.linspace(1,10)
plt.plot(data, data)
anno_args = {
    'annotation_clip': False,
    'arrowprops': dict(arrowstyle='-', relpos=(0, 1)),
    }
plt.annotate('Good relpos', (3, 3), xytext = (3, 2), **anno_args)
plt.annotate('Bad relpos', (6, 6), xytext = (6, 5), **anno_args)
plt.annotate('No ha/va', (5, 5), xytext = (5, 4),
arrowprops=dict(arrowstyle='-'),
             ha='left', va='top')
plt.show()