Matplotlib twinx with wx

[pJmp]

I am building a wxPython application that plots some data using matplotlib. I am embedding several axis within a figure by creating several repeatedly calling

ax1 =fig.add_axes([0.1, 0.7, 0.8, 0.2], **axprops)

where axprops is a dict(). This way the plots can share an X-axis. However, I wanted to have one plot have two sets of information. I am familiar with the two_scales.py example, but the same calls do not seem to work in the wx app. Only the original plot succeeds and formats the left axis, while nothing is displayed on the right axis. Is twinx() not supported in wx?

-pJmp

[hattawayd]

import pylab
ax2 = pylab.twinx()

[bartonc]

import pylab
ax2 = pylab.twinx()

Thanks, hattawayd, for the very useful information.

I hope you stick around,
Barton

[jlm699]

I am trying to accomplish the same thing but with no luck... here is my code:

class CanvasFrame(wx.Panel):
 
    def __init__(self, parent):
        wx.Panel.__init__(self, parent, -1)
 
        self.SetBackgroundColour(wx.NamedColor("WHITE"))
 
        self.figure = Figure()
        self.ax1 = self.figure.add_subplot(111)
        t = arange(0.0,3.0,0.01)
        s = sin(2*pi*t)
 
        self.ax1.plot(t,s)
        self.ax1.set_xlabel('t')
        self.ax1.set_ylabel('sin(t)')
 
        self.ax2 = pylab.twinx()
        s2 = pylab.exp(t)
 
        self.ax2.plot(t, s2, 'r.')
        self.ax2.set_ylabel('exp(t)')
        self.ax2.yaxis.tick_right()
 
        self.figure_canvas = FigureCanvas(self, -1, self.figure)
 
        # Note that event is a MplEvent
        self.figure_canvas.mpl_connect('motion_notify_event', self.UpdateStatusBar)
        self.figure_canvas.Bind(wx.EVT_ENTER_WINDOW, self.ChangeCursor)
 
        self.sizer = wx.BoxSizer(wx.VERTICAL)
        self.sizer.Add(self.figure_canvas, 1, wx.LEFT | wx.TOP | wx.GROW)
        self.SetSizer(self.sizer)
        self.Fit()
 
        self.toolbar = NavigationToolbar2Wx(self.figure_canvas)
        self.sizer.Add(self.toolbar, 0, wx.LEFT | wx.EXPAND)
        self.toolbar.Show()

This is displayed in a wx.Frame however it only displays the first axis (the sin wave one), and nothing for the second. I am assuming that my problem is that I am not associating the second axis with the figure, however with the horrifyingly lackluster matplotlib API/documentation I am at a complete loss of how to do this.

[jlm699]

I cannot figure this out and it's just killing me!
I tried using pylab's show() method as in the two_scales example and a window pops up with both y-axes displayed correctly, but then I close that window and my actual application appears with only a single y-axis. Here's my codes.

code: (python)

class CanvasFrame(wx.Panel):
    def __init__(self, parent):
        wx.Panel.__init__(self, parent, -1)
 
        self.SetBackgroundColour(wx.NamedColor("WHITE"))
 
        self.time_range = 100.0
        self.time_div = 0.5
        self.curr_time = 0.0
        self.y = [0.0]
        self.x = [0.0]
 
        self.tv = track_vars(self)
 
        self.sizer = wx.BoxSizer(wx.VERTICAL)
        self.sizer.Add(self.tv.canvas, 1, wx.LEFT | wx.TOP | wx.GROW)
        self.SetSizer(self.sizer)
        self.Fit()
 
        self.toolbar = NavigationToolbar2Wx(self.tv.canvas)
        self.sizer.Add(self.toolbar, 0, wx.LEFT | wx.EXPAND)
        self.toolbar.Show()
 
class MainWindow(wx.Frame):
    def __init__(self, *args, **kwargs):
        wx.Frame.__init__(self, *args, **kwargs)
 
        icon = wx.Icon(name = 'rDT.ico', type = wx.BITMAP_TYPE_ICO)
        self.Bind(wx.EVT_CLOSE, self.OnExit)
        self.SetIcon(icon)
        self.ds = None
 
        # Set-up a Status Bar divided into two fields
        self.sb = self.CreateStatusBar(2)
 
        # Set up splitter window
        self.splitwin = SplitterWindow(self)
 
        # Set up the texual dataviewer for the left pane
        self.lc = CheckListCtrl(self.splitwin)
        self.lc.InsertColumn(0, "Variable Name")
        self.lc.InsertColumn(1, "Value")
        self.currentItem = 0
        self.Bind(wx.EVT_LIST_ITEM_SELECTED, self.OnItemSelected, self.lc)
 
        # Set up the visual dataviewer for the right pane
        self.graph = CanvasFrame(self.splitwin)
 
        # Now we build our menus
        menu = self.MakeMenu()
        self.SetMenuBar(menu)
 
        # Now initialize the split window for program start-up appearance
        self.splitwin.SetMinimumPaneSize(275)
        self.splitwin.SplitVertically(self.lc, self.graph, 275)
 
        # Setup our Heartbeat and Collection timers
        Global.heartbeat_timer_id = wx.NewId()
        Global.collection_timer_id = wx.NewId()
        Global.heartbeat_timer = wx.Timer(self, Global.heartbeat_timer_id)
        Global.collection_timer = wx.Timer(self, Global.collection_timer_id)
        wx.EVT_TIMER(self, Global.heartbeat_timer_id, self.OnHeartbeat)
        wx.EVT_TIMER(self, Global.collection_timer_id, self.OnCollection)
 
        return None
 
class track_vars:
    def __init__(self, parent):
        self.figure = Figure()
        self.ax = self.figure.add_subplot(111)
        self.ax.set_xlabel('Time')
        self.ax.set_ylabel('Testing data collection')
        self.ax2 = pylab.twinx()
        self.ax2.set_ylabel('Second y-axis')
        self.ax2.yaxis.tick_right()
        pylab.show()
        self.canvas = FigureCanvas(parent, -1, self.figure)

Does anyone have ANY idea why this is not working embedded in a wxPython application?! Anyone? Anyone? Bueller?... Bueller?

[jlm699]

Ok, well I figured it out finally... and to any of those poor souls who in the future are trying to accomplish anything similar I basically had to just do it manually (kind of a jerry-rig) and scrap twinx. Here's my code:

import matplotlib
from matplotlib.figure import Figure
from matplotlib.backends.backend_wxagg import FigureCanvasWxAgg as FigureCanvas
...
        self.figure = Figure()
        self.ax1 = self.figure.add_subplot(111)
        self.ax1.set_ylabel('Value')
        self.ax1.yaxis.tick_left() # w/o this the ticks were echoed on right side
        self.ax2 = self.figure.add_subplot(111, sharex=self.ax1, frameon=False)
        self.ax2.set_xlabel('Samples')
        self.ax2.yaxis.tick_right()
        self.canvas = FigureCanvas(parent, -1, self.figure)
        self.canvas.realize()
        self.canvas.draw()
 

I know, it's very simple and should have been more obvious to me earlier, but oh well. The one thing that I did not bother tackling was figuring out how to place the ylabel on the right side of the graph for ax2, as using ax.set_ylabel placed it right below the label for ax1.

If anyone has any suggestions I'd appreciate them, but I'm not really concerned about putting a label on the right side as it's rather redundant anyway.

[jougs]

The one thing that I did not bother tackling was figuring out how to place the ylabel on the right side of the graph for ax2, as using ax.set_ylabel placed it right below the label for ax1.

Does anyone have a solution to the problem of the position of the second ylabel? My data is such that I really need two different labels, but all my attempts have failed so far.

Thanks!

Greetings,
Jougs!