Re: [Matplotlib-users] real time plotting

[Neal B. <ndb...@gm...>]

I go through a compute loop that takes maybe a few seconds per pass, then
plot a new point on the graph.  Do I have to?  No - I thought mpl was
supposed to do this and wanted to learn how.  If it really doesn't work
I'll do something else.

I don't think animation is correct here - I had the impression animation is
where my update would be run as a callback, with a main loop that calls me
periodically.  Could that fit the model I described, where a lengthy
computation produces a new value every few/10s of seconds?


On Mon, Mar 11, 2013 at 1:55 PM, David Hoese <dh...@gm...> wrote:

>  Someone may have to correct me, but I think this has to do with the Qt4
> event loop and it not being run properly. When you get into real time
> plotting it can get kind of tricky. In your case (I got the same results).
> I have made real-time PyQt4 GUIs before and have always used separate
> QThreads and Qt signals/slots to update the plot. I've never used GTK so
> I'm not sure why that worked vs Qt, I would think they would use similar
> principles but matplotlib does some magic behind the scenes sometimes. You
> can see different results if you comment out the while loop and import the
> module into your python/ipython interpreter. After doing this you'll see
> the figure pop up (you don't even need the fig.canvas.show() for this part
> if interactive mode is on. I went one step further and turned the while
> loop into a function:
>
> def one_iter(i):
>     # Contents of while loop
>
> Calling this in the interpreter shows the figure updating after each call,
> but running in a loop (even with sleep) won't show any updates until the
> loop is done. In my opinion you have a few choices that really depend on
> your programming comfort level:
>
> 1. Don't make a real-time plot.
>         Do you really need a real-time plot that updates from some
> external source?
> 2. Maybe you should look at the matplotlib animation functionality (
> http://matplotlib.org/api/animation_api.html). I like this tutorial:
> http://jakevdp.github.com/blog/2012/08/18/matplotlib-animation-tutorial/.
> This won't get you a real-time GUI exactly, but it can help if what you're
> doing isn't too complicated. It can also be nice for making videos of plot
> animations.
> 3. If you need a GUI with multiple plots and you need for future feature
> creep, I would research making PyQt4 GUIs, QThreads, Qt signals and slots,
> and putting matplotlib figures into a PyQt4 GUI. This is complex if you are
> not familiar with GUI programming and will take a while.
>
> Sorry I couldn't be of more help, but it really depends on what exactly
> you are doing.  Mainly, what do you mean by real-time? Do you really mean
> animation? Let me know what you come up with, I'm interested.
>
> -Dave
>
> P.S. Why use a while loop? You can do the same thing with:
>
>     for i in range(1000):
>         # Do stuff
>
>
> On 3/11/13 10:34 AM, Neal Becker wrote:
>
>  I added fig.canvas.show(). It still does nothing.
>
>  If I add
> mpl.use ('GTK'), now it seems to be doing realtime plotting.
>
>  import matplotlib as mpl
>
>  import matplotlib.pyplot as plt
> plt.ion()
> import numpy as np
> fig=plt.figure()
> plt.axis([0,1000,0,1])
>
>  i=0
> x=list()
> y=list()
>
>  fig.canvas.show()
> while i <1000:
>     temp_y=np.random.random()
>     x.append(i)
>     y.append(temp_y)
>     plt.scatter(i,temp_y)
>     i+=1
>     plt.draw()
>
>
>
> On Mon, Mar 11, 2013 at 10:35 AM, David Hoese <dh...@gm...> wrote:
>
>> Oops forgot to change the subject line.
>>
>> On 3/11/13 9:34 AM, David Hoese wrote:
>>
>>> You likely need to "show()" the canvas. I usually do this by calling
>>> "fig.canvas.show()" before the for loop.
>>> Since you are using a Qt4 backend the canvas used by the figure is a
>>> QWidget, the basic component of a Qt4 GUI. I don't know if there is a more
>>> matplotlib specific way of doing this, but when dealing with a larger
>>> system this is how I do it.
>>>
>>> I would also add a sleep ("from time import sleep") of a couple seconds
>>> for testing to make sure you are getting through the entire for loop before
>>> you can see it.
>>>
>>> Please CC in any replies, thanks.
>>>
>>> -Dave
>>>
>>>
>>> On 3/11/13 8:58 AM, ndb...@gm... wrote:
>>>
>>>> I want to update a plot in real time.  I did some goog search, and saw
>>>> various
>>>> answers.  Trouble is, they aren't working.
>>>>
>>>> Here's a typical example:
>>>>
>>>> import matplotlib.pyplot as plt
>>>> import numpy as np
>>>> fig=plt.figure()
>>>> plt.axis([0,1000,0,1])
>>>>
>>>> i=0
>>>> x=list()
>>>> y=list()
>>>>
>>>> while i <1000:
>>>>      temp_y=np.random.random()
>>>>      x.append(i)
>>>>      y.append(temp_y)
>>>>      plt.scatter(i,temp_y)
>>>>      i+=1
>>>>      plt.draw()
>>>>
>>>> If I run this, it draws nothing.
>>>>
>>>> This is my matplotlibrc:
>>>> backend : Qt4Agg
>>>> mathtext.fontset: stix
>>>>
>>>
>>>
>>
>
>