matplotlib-users Mailing List for matplotlib

Hi there,

I'm wondering whether there is an easy way to append an additional subplot
to an existing figure without losing the subplots already drawn.

Currently if I do something like
>>> subplot(211); plot(...); subplot(212); plot(...);

Then I get inconsistent drawing results if I try something like:
>>> subplot(313); plot(...);

Cheers,
Alex

On 06/27/2011 03:38 AM, Jonathan Slavin wrote:
> I tried the suggested clean-up but saw no difference in performance.  I
> left out a crucial piece of information, I think, in my earlier message.
> The delay in drawing occurs when I'm running the code from within
> ipython, invoked with the -pylab flag.  When I run it directly from the
> command line, I get no such delay.  I presume this is backend dependent.
> For my current purposes, just running it directly from the command line
> (i.e. something like: % python do_fits.py) works for me.  The ability to
> interactively examine variables, as one can when running within ipython,
> would be nicer, however.
>
> Jon
>
>>          On 06/24/2011 04:03 AM, Jonathan Slavin wrote:
>>          >  import matplotlib.pyplot as plt
>>          >  plt.ion()
>>          >  fig = plt.gcf()
>>          >  for obsid in obsids:
>>          >       <do fitting>
>>          >       plt.cla()
>>          >       fig = plt.gcf()
>>          >       ax = fig.add_axes([0.15,0.1,0.8,0.6])
>>          >       ax.plot(x,y)
>>          >       plt.draw()
>>          >       ans = raw_input('continue? ')
>>          >       if ans == 'n':
>>          >           break
>>
>>          The behavior may depend on mpl version and backend, but with
>>          1.0.1 or
>>          later, I think something like what you have will work with a
>>          little
>>          cleanup, e.g.:
>>
>>          import matplotlib.pyplot as plt
>>          import numpy as np
>>
>>          plt.ion()
>>          fig = plt.gcf()
>>          ax = fig.add_axes([0.15,0.1,0.8,0.6])
>>          for i in range(3):
>>               ax.cla()
>>               ax.plot(np.random.rand(10))
>>               plt.draw()
>>               raw_input("hit a key to proceed")


What happens if you replace the raw_input with the figure method 
waitforbuttonpress?  (Also available as a pyplot function.)

Eric


>>
>>
>>          Eric
>

> I don't know for sure if this is matplotlib's fault. ipython's fault or Mac OS
> X, but I ever since I upgraded to matplotlib 1.0.1 I have this problem that
> ipython will exit with segfault after I close a figure.
>

 >> What version of ipython are you using?

Hi Eric,

Thanks for your response.

I'm indeed using 0.10

Good to know a new iPython will be out.

Best
-Kaushik

I tried the suggested clean-up but saw no difference in performance.  I
left out a crucial piece of information, I think, in my earlier message.
The delay in drawing occurs when I'm running the code from within
ipython, invoked with the -pylab flag.  When I run it directly from the
command line, I get no such delay.  I presume this is backend dependent.
For my current purposes, just running it directly from the command line
(i.e. something like: % python do_fits.py) works for me.  The ability to
interactively examine variables, as one can when running within ipython,
would be nicer, however.

Jon

>         On 06/24/2011 04:03 AM, Jonathan Slavin wrote:
>         > import matplotlib.pyplot as plt
>         > plt.ion()
>         > fig = plt.gcf()
>         > for obsid in obsids:
>         >      <do fitting>
>         >      plt.cla()
>         >      fig = plt.gcf()
>         >      ax = fig.add_axes([0.15,0.1,0.8,0.6])
>         >      ax.plot(x,y)
>         >      plt.draw()
>         >      ans = raw_input('continue? ')
>         >      if ans == 'n':
>         >          break
>         
>         The behavior may depend on mpl version and backend, but with
>         1.0.1 or 
>         later, I think something like what you have will work with a
>         little 
>         cleanup, e.g.:
>         
>         import matplotlib.pyplot as plt
>         import numpy as np
>         
>         plt.ion()
>         fig = plt.gcf()
>         ax = fig.add_axes([0.15,0.1,0.8,0.6])
>         for i in range(3):
>              ax.cla()
>              ax.plot(np.random.rand(10))
>              plt.draw()
>              raw_input("hit a key to proceed")
>         
>         
>         Eric

Hi

I've made a lib to replace the default Matplotlib toolbar, it looks like
this:
http://i256.photobucket.com/albums/hh163/_pelya/mpl-gui.png

It replaces the pan/zoom button with wheels and handles, using new
subplot2grid API of matplotlib 1.0.1.
The reason behind that change is that people I'm currently writing my
application for can hardly handle
anything that doesn't look like their 50-years-old potentiostat. Also, I
needed some custom event code to
interact with the figure (typically to mark a part of the graph), and
enabling/disabling pan/zoom mode
each time I need to make a change to the figure is not convenient.

Here's the sources:
http://elchemgraphview.svn.sourceforge.net/viewvc/elchemgraphview/

Currently it supports only subplot() API, similar to the Matplotlib API with
the same name,
also it requires quite a lot of a boilerplate code - see simpletest.py or
test.py for example.
As a bonus you can add your own buttons to that new toolbar, however my
designer skills
are rather poor, so it looks like MS Paint drawing.
If anyone is interested I can extend it to the point where you need only to
write
"import vintage" to get the new controls - it will replace several
matplotlib functions like pyplot.figure(),
pyplot.subplot() and pyplot.connect() with it's own handlers, right inside
pyplot module namespace.

On Sunday, June 26, 2011, Warren Weckesser
<war...@en...> wrote:
>
>
> On Sun, Jun 26, 2011 at 7:44 PM, Benjamin Root <ben...@ou...> wrote:
>
>
> On Sunday, June 26, 2011, Carl Karsten <ca...@pe...> wrote:
>> http://matplotlib.sourceforge.net/examples/api/radar_chart.html
>>
>> "Exception occurred rendering plot."
>>
>
> Without more information, we can't help you.  What version of
> matplotlib are you using? On what OS?  How did you install it?  Do the
> tests pass?  And which backend?
>
>
> That error is what shows up on the web page when you follow the link.
>
> Warren
>
>

Ah, indeed it is.  I apologize for misunderstanding, what is odd is
that the demo didn't work, but the mpl logo rendered fine.

Who was it that uploaded the recent rebuild of the docs?

Ben Root

On Sun, Jun 26, 2011 at 7:44 PM, Benjamin Root <ben...@ou...> wrote:

> On Sunday, June 26, 2011, Carl Karsten <ca...@pe...> wrote:
> > http://matplotlib.sourceforge.net/examples/api/radar_chart.html
> >
> > "Exception occurred rendering plot."
> >
>
> Without more information, we can't help you.  What version of
> matplotlib are you using? On what OS?  How did you install it?  Do the
> tests pass?  And which backend?
>


That error is what shows up on the web page when you follow the link.

Warren



> Ben Root
>
>
> ------------------------------------------------------------------------------
> All of the data generated in your IT infrastructure is seriously valuable.
> Why? It contains a definitive record of 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-d2d-c2
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

I am trying to graph bandwidth over time.   I am trying to get a grip
on the details of how my ISP throttles high amounts of traffic.  like
when I upload a 300 mb video.  I am collecting data during the upload,
but I am having trouble understanding the raw numbers.   Pretty sure a
graph will show me what I expect to see, and maybe help me get the
exact numbers I am looking for: bandwidth drops to X after Y
something.

I am guessing this will easy for someone familiar with matplotlib.  I
have been looking at docs and example, cant figure it out.  Thus my
cry for help post.

my data looks like this:

uploadlog.csv: seconds from midnight, timestamp, bytes sent
(accumulative over a 300mb upload)

Here is something that calcs the deltas that I think I need.




#!/usr/bin/python

import datetime
import csv
import StringIO
import pprint

log="""56521.93324, 2011-06-24 15:42:01.933240, 0
56521.933569, 2011-06-24 15:42:01.933569, 1292
56521.933722, 2011-06-24 15:42:01.933722, 1488
56522.022575, 2011-06-24 15:42:02.022575, 16488
56522.023069, 2011-06-24 15:42:02.023069, 31488
56522.03704, 2011-06-24 15:42:02.037040, 46488
56522.079995, 2011-06-24 15:42:02.079995, 61488
56522.080119, 2011-06-24 15:42:02.080119, 76488
56522.116328, 2011-06-24 15:42:02.116328, 91488"""


# reader = csv.reader(open('uploadlog.csv', 'rb'))
reader = csv.reader(StringIO.StringIO(log))

i=0
dat = []
last_sec, last_bytes = 0,0
for row in reader:
    sec=float(row[0])
    bytes_sent = int(row[2])
    if last_sec:
        duration=sec - last_sec
        chunk = bytes_sent - last_bytes
        bps = chunk/duration
        dat.append( [chunk,duration,bps] )

    last_sec = sec
    last_bytes = bytes_sent

pprint.pprint(dat)

"""
[[1292, 0.00032900000223889947, 3927051.645008286],
 [196, 0.00015300000086426735, 1281045.7443976079],
 [15000, 0.08885300000110874, 168818.16033012758],
 [15000, 0.00049399999988963827, 30364372.476419158],
 [15000, 0.013971000000310596, 1073652.5660057638],
 [15000, 0.042954999997164123, 349202.65396322421],
 [15000, 0.00012399999832268804, 120967743.57177937],
 [15000, 0.036208999998052604, 414261.6476789398]]
"""


-- 
Carl K

On Sunday, June 26, 2011, Carl Karsten <ca...@pe...> wrote:
> http://matplotlib.sourceforge.net/examples/api/radar_chart.html
>
> "Exception occurred rendering plot."
>

Without more information, we can't help you.  What version of
matplotlib are you using? On what OS?  How did you install it?  Do the
tests pass?  And which backend?

Ben Root