matplotlib arrowheads and aspect ratio

Question

If I run this script:

import matplotlib.pyplot as plt
import pylab as plab

plt.figure()
plt.plot([0,2], [2,0], color='c', lw=0.5)
plt.plot([1,2], [2,1], color='k', lw=0.5)
plt.arrow(1,1,0.5,0.5, head_width=0.1, width=0.01, head_length=0.1, color='r')
plt.arrow(1.25,0.75,0.5,0.5, head_width=0.1, width=0.01, head_length=0.1, color='g')

plab.axes().set_aspect(0.5)

plt.show()

I get this: Notice that the backs of the arrowheads are flush with that black line that stretches from (1,2) to (2,1). That makes sense, but I want the backs of the arrowheads to be VISUALLY perpendicular to the tails WITHOUT changing the aspect ratio. How do I do this?

Answer 1

I've been annoyed with that problem for (almost) ever, and mostly end up using annotate() to draw arrows. For example (lacking a lot of tweaking to end up with the identical result as your plot...):

import matplotlib.pyplot as plt
import pylab as plab

plt.figure()
ax=plt.subplot(211)
plt.plot([0,2], [2,0], color='c', lw=0.5)
plt.plot([1,2], [2,1], color='k', lw=0.5)
plt.arrow(1,1,0.5,0.5, head_width=0.1, width=0.01, head_length=0.1, color='r')
ax.set_aspect(0.5)

ax=plt.subplot(212)
plt.plot([0,2], [2,0], color='c', lw=0.5)
plt.plot([1,2], [2,1], color='k', lw=0.5)
ax.annotate("",
            xy=(1.5, 1.5), xycoords='data',
            xytext=(1, 1), textcoords='data',
            arrowprops=dict(arrowstyle="-|>",
                            connectionstyle="arc3"),
            )
ax.set_aspect(0.5)

plt.show()