Thread: [Matplotlib-users] Polar plot - problem with negative values for radius

Hi,

I am trying to display some complex values in a polar plot. Displaying
linear magnitude vs. angle - of course - works without any issues. But I'd
rather display the logarithmic magnitute vs. angle. Since the data for the
radius gets negative then, it'll be wrapped around / rotated by 180deg by
matplotlib.

How can I display negative values for the radius w/o having them rotated by
180deg?

Of course I can add an offset to the data before plotting it but since my
plot is interactive and I use the xdata/ydata of events I'd like them to
represent the real data, without an offset. I thought it might be possible
to achieve with a custom transformation but I don't actually get it to work.

-Stephan
-- 
View this message in context: http://old.nabble.com/Polar-plot---problem-with-negative-values-for-radius-tp30936638p30936638.html
Sent from the matplotlib - users mailing list archive at Nabble.com.

On Wed, Feb 16, 2011 at 9:16 AM, Stephan Markus <zw...@we...> wrote:

>
> Hi,
>
> I am trying to display some complex values in a polar plot. Displaying
> linear magnitude vs. angle - of course - works without any issues. But I'd
> rather display the logarithmic magnitute vs. angle. Since the data for the
> radius gets negative then, it'll be wrapped around / rotated by 180deg by
> matplotlib.
>
> How can I display negative values for the radius w/o having them rotated by
> 180deg?
>
> Of course I can add an offset to the data before plotting it but since my
> plot is interactive and I use the xdata/ydata of events I'd like them to
> represent the real data, without an offset. I thought it might be possible
> to achieve with a custom transformation but I don't actually get it to
> work.
>
> -Stephan
> --
> View this message in context:
> http://old.nabble.com/Polar-plot---problem-with-negative-values-for-radius-tp30936638p30936638.html
> Sent from the matplotlib - users mailing list archive at Nabble.com.
>
>
Stephan,

>From the polar axes object, you can call ax.set_rscale('log') and that will
automatically make the radial ticks scale and label the axes properly.  You
can then input the original data as you would without worry of negative
values.

Let us know how that works for you!

Ben Root

Ben,

I should have mentioned that I already tried that. When I set the rscale to
'log' the plot crashes when zooming or mpl cannot even create it.

Maybe some example code will help:
----------------
from numpy import arange, sin, pi, cos, ones
from matplotlib.backends.backend_tkagg import FigureCanvasTkAgg,
NavigationToolbar2TkAgg
from matplotlib.figure import Figure
import Tkinter as Tk

root = Tk.Tk()


f = Figure(figsize=(5,4), dpi=100)
ax = f.add_subplot(111, projection='polar')

t = arange(0.0,2*pi,0.01)
s1 = ones(len(t))*10

ax.set_rscale('log')
ax.plot(t,s1)


canvas = FigureCanvasTkAgg(f, master=root)
canvas.show()
canvas.get_tk_widget().pack(side=Tk.TOP, fill=Tk.BOTH, expand=1)
toolbar = NavigationToolbar2TkAgg(canvas, root)
toolbar.update()
canvas._tkcanvas.pack(side=Tk.TOP, fill=Tk.BOTH, expand=1)
Tk.mainloop() 
----------------

Using this code the plot will show up but you can see the 10^-1 and 10^-2
tick labels overlapping in the center. Ok, it shouldn't be a big deal to get
rid of them. The major problem is if you try to zoom out, the tick labels
move away from the center and if you try to zoom into the plot it eventually
throws an exception "ValueError: cannot convert float NaN to integer".
If you try to plot smaller values (e.g. replacing the line 's1 =
ones(len(t))*10' with 's1 = ones(len(t))') mpl also throws an ValueError
exception and does not even create the plot.

Let me know if you need full tracebacks.

Btw: I am using Matplotlib 1.0.1 and Python 2.6.0 on Windows 64.
-- 
View this message in context: http://old.nabble.com/Polar-plot---problem-with-negative-values-for-radius-tp30936638p30947935.html
Sent from the matplotlib - users mailing list archive at Nabble.com.

Small update:

I tried the very same code with MPL 1.0.1 and Python 2.5.0 on Linux 64 and
Python 2.5.4 on Win32 and it runs w/o throwing any exceptions there!

But: the behaviour is still not that what I expected. Still these issues are
remaining:
- the smallest magnitude (center magnitude in other words) is 0.1 and I'd
like to display way smaller values
- the smallest magnitude doesn't even change when zooming
- the grid lines and ticks do not show as I'd expect (see my last message)
when zooming


That's why I'd rather stick to a linear scale, doing my own logarithmic
conversion, limit my data at lowest value I need and just use an offset for
the scale.
-- 
View this message in context: http://old.nabble.com/Polar-plot---problem-with-negative-values-for-radius-tp30936638p30948265.html
Sent from the matplotlib - users mailing list archive at Nabble.com.

On Thu, Feb 17, 2011 at 4:12 AM, Stephan Markus <zw...@we...> wrote:

>
> Small update:
>
> I tried the very same code with MPL 1.0.1 and Python 2.5.0 on Linux 64 and
> Python 2.5.4 on Win32 and it runs w/o throwing any exceptions there!
>
> But: the behaviour is still not that what I expected. Still these issues
> are
> remaining:
> - the smallest magnitude (center magnitude in other words) is 0.1 and I'd
> like to display way smaller values
> - the smallest magnitude doesn't even change when zooming
> - the grid lines and ticks do not show as I'd expect (see my last message)
> when zooming
>
>
> That's why I'd rather stick to a linear scale, doing my own logarithmic
> conversion, limit my data at lowest value I need and just use an offset for
> the scale.
>


I see what you mean.  This problem is almost identical to another thread
going on where we can't seem to correctly do log scale for 3d plots.  The
tick locators are in the wrong positions (the major ticks should be evenly
spaced) and the error message is similar to one I have been encountering in
mplot3d.

I will look a little further into this and see if I can kill two birds with
one stone...

Ben Root

I am using a workaround now. But that is a hackery solution.

Before plotting my data I convert it to dBs and limit it to the lowest value
I want to display. Then I plot it using a regular polar plot with a custom
formatting function that sets the tick labels with respect to the data
offset.
Since I use a custom Navigation-Toolbar anyways it was no big deal to add
the few necessery lines of codes to handle the offset there, too.

The plot now looks exactly as I want it. But: I'd still prefer using a scale
that does all the work in the background.


-- 
View this message in context: http://old.nabble.com/Polar-plot---problem-with-negative-values-for-radius-tp30936638p30975519.html
Sent from the matplotlib - users mailing list archive at Nabble.com.