2011/9/30 <petr_aleksandrov@mail.ru>
How to make a plot of a discontinuous 1d function without line at a function jump. An example is shown on a picture. This was plotted by:
import numpy
from matplotlib import pyplot
x = numpy.linspace(-1.0, 1.0, 100)
y = numpy.sign(x) * numpy.cos(x)
axes = pyplot.gca()
axes.set_ylim(-1.1, 1.1)
pyplot.plot(x, y)

If you happen to know where the discontinuity is, then you could do two separate plot commands. Automatically, it is harder. Maybe something like this would help:

threshold = 1000.0
# Use gradient instead of diff because it returns an array of the same shape
dydx = numpy.gradient(y) / numpy.gradient(x)
idx = (numpy.abs(dydx) < threshold)
axes.plot(x[idx], y[idx])

Although, that gets rid of valid data points. Another trick would be to rebuild the x and y arrays, but with NaNs placed between datapoints that would form the discontinuity. matplotlib automatically treats NaNs as a indicator to break a line.

Ben Root