I frequently find myself trying to create plots that use a secondary axis
to indicate data in a second set of units. For example, I might plot a data
set with an x-axis in data number (e.g. the output of an analog to digital
converter), and then wish to display the calibrated units on a secondary
x-axis (e.g. volts).
There are quite a few examples that do this by creating a secondary axis
using twiny(), and then setting the limits of the secondary x-axis to the
scaled limits of the primary, and possibly setting the ticks to line up as
import matplotlib.pyplot as plt
import numpy as np
from scipy.stats import norm
f, ax = plt.subplots(1)
x_dn = np.arange(4096)
y = norm.pdf(x_dn, 2048, 64)
v = lambda x: x*5.0/4096
ax_top = ax.twiny()
ax_top.set_xticks([v(x) for x in ax.get_xticks()]) # optional, aligns the
ax_top.set_xlim([v(x) for x in ax.get_xlim()])
This isn't too painful if you're only doing it once. However, if you
subsequently want to change the limits (from the command line) you have to
explicitly set the limits of both axes or they will become out of sync
(aside: they also can get out of sync if you set the secondary limits
before the ticks, because setting the ticks can change the limits!). If you
do set the ticks to line up the grids, then you also have to recompute
those for the secondary axis.
ax.set_xlim([1500, 2500]) # now they're out of sync
ax_top.set_xticks([v(x) for x in ax.get_xticks()]) # ticks correct, but in
ax_top.set_xlim([v(x) for x in ax.get_xlim()]) # all's well again.
It just seems like there ought to be a better way. I apologize if it's out
there and I missed it.
-------------- next part --------------
An HTML attachment was scrubbed...