A simple problem with two simple solutions, the choice just depends on if we think ax.scatter(oqu.Quaternion) should be allowed.
If you run this code:
import matplotlib.pyplot as plt
import numpy as np
import orix.plot as opl
import orix.quaternion as oqu
opl.register_projections()
# NOTE, this is a Quaternion, not a Rotation/Misorientation
q = oqu.Quaternion.random(1000)
fig = plt.figure()
ax = fig.add_subplot(111,projection='homochoric')
# plot the quaternion, which gets misinterpreted as a Vector3D by RotationPlot
ax.scatter(q)
# Plot C1 fundamental zone and clean up plot to help with visualization.
fz =oqu.OrientationRegion.from_symmetry(oqu.symmetry.C1)
ax.plot_wireframe(fz)
ax.set_aspect('equal')
ax.axis('off')
You would expect a cloud of points filling an orb with the same radius as the C1 fundamental zone. Instead though, you get this:
This is because the logic in orix.plot.rotation_plot.RotationPlot.transform() checks if the input is a Rotation, and if it isn't but it is an Object3d, it assume it's a Vector3D. The end result being that [q.a,q.b,q.c] from a quaternion are recast as the xyz of a Vector3D and the qd is discarded. since qa>0, the resulting plot is also a half-circle.
Solution 1: Quaternions should be plottable
If we think a quaternion should be plottable, we just change line 80 of orix.plot.rotatoin_plot.RotationPlot.transform() from
if isinstance(xs, Rotation):
to
if isinstance(xs, Quaternion):
which adds capability but might cause a problem if a later function assumes all plot elements have an improper attribute,
Solution 1: Quaternions should NOT be plottable
If we think quaternions should not be plotted directly, we add an extra elif isinstance(xs, Quaternion) catch that raises an error.
Either way, I can add this to a PR along with unit tests, I just want a second person's input.
After typing this out, I'm leading vaguely to Solution 2 just to avoid a future edge case, but I could be convinced to either way
Thoughts?
A simple problem with two simple solutions, the choice just depends on if we think
ax.scatter(oqu.Quaternion)should be allowed.If you run this code:
You would expect a cloud of points filling an orb with the same radius as the C1 fundamental zone. Instead though, you get this:
This is because the logic in
orix.plot.rotation_plot.RotationPlot.transform()checks if the input is aRotation, and if it isn't but it is anObject3d, it assume it's aVector3D. The end result being that[q.a,q.b,q.c]from a quaternion are recast as the xyz of a Vector3D and theqdis discarded. sinceqa>0, the resulting plot is also a half-circle.Solution 1: Quaternions should be plottable
If we think a quaternion should be plottable, we just change line 80 of
orix.plot.rotatoin_plot.RotationPlot.transform()fromto
which adds capability but might cause a problem if a later function assumes all plot elements have an
improperattribute,Solution 1: Quaternions should NOT be plottable
If we think quaternions should not be plotted directly, we add an extra
elif isinstance(xs, Quaternion)catch that raises an error.Either way, I can add this to a PR along with unit tests, I just want a second person's input.
After typing this out, I'm leading vaguely to Solution 2 just to avoid a future edge case, but I could be convinced to either way
Thoughts?