Preventing plot joining when values "wrap" in matplotlib plots
list, matplotlib, plot, python
Solution
Here is a generator function that finds the contiguous regions of 'wrapped' data:
import numpy as np
def unlink_wrap(dat, lims=[-np.pi, np.pi], thresh = 0.95):
"""
Iterate over contiguous regions of `dat` (i.e. where it does not
jump from near one limit to the other).
This function returns an iterator object that yields slice
objects, which index the contiguous portions of `dat`.
This function implicitly assumes that all points in `dat` fall
within `lims`.
"""
jump = np.nonzero(np.abs(np.diff(dat)) > ((lims[1] - lims[0]) * thresh))[0]
lasti = 0
for ind in jump:
yield slice(lasti, ind + 1)
lasti = ind + 1
yield slice(lasti, len(dat))
An example usage would be,
x = np.arange(0, 100, .1)
y = x.copy()
lims = [0, 24]
x = (x % lims[1])
fig, ax = matplotlib.pyplot.subplots()
for slc in unlink_wrap(x, lims):
ax.plot(x[slc], y[slc], 'b-', linewidth=2)
ax.plot(x, y, 'r-', zorder=-10)
ax.set_xlim(lims)
Which gives the figure below. Note that the blue lines (which utilize `unlink_wrap`) are broken and the standard-plotted red lines are shown for reference.
Problem
I'm plotting right ascension ephemerides for planets, which have the property that they are cyclical: they hit a maximum value, 24, and then start again at 0. When I plot these using matplotlib, the "jump" from 24 to zero is joined so that I get horizontal lines running across my figure: How can I eliminate these lines? Is there an approach in matplotlib, or perhaps a way to split the lists at between the points where the jump occurs. Code to generate above figure: ``` from __future__ import division import ephem import matplotlib import matplotlib.pyplot import math fig, ax = matplotlib.pyplot.subplots() ax.set(xlim=[0, 24]) ax.set(ylim=[min(date_range), max(date_range)]) ax.plot([12*ep.ra/math.pi for ep in [ephem.Jupiter(base_date + d) for d in date_range]], date_range, ls='-', color='g', lw=2) ax.plot([12*ep.ra/math.pi for ep in [ephem.Venus(base_date + d) for d in date_range]], date_range, ls='-', color='r', lw=1) ax.plot([12*ep.ra/math.pi for ep in [ephem.Sun(base_date + d) for d in date_range]], date_range, ls='-', color='y', lw=3) ```