Non-standard axis directions

[AndrewFalkowski]

Looking to replicate a non-standard ternary plot where the left and bottom axis increase from the bottom left corner (example). This structure is fairly standard in my field, but I haven't found a single ternary plotting library capable of replicating it. Wondering if there is some way to define custom axis directions in your library?

Thanks!

[yuzie007]

Thank you @AndrewFalkowski for taking a look on mpltern. Please find the following snippet, which may mostly reproduce the tick style in your example.

import matplotlib.pyplot as plt
import mpltern


ax = plt.subplot(projection="ternary")

ax.plot([0.1, 0.2], [0.2, 0.4], [0.7, 0.4])

ax.set_tlabel("Sn")
ax.set_llabel("Bi")
ax.set_rlabel("Cu")

ax.taxis.set_tick_params(
    which="both",
    tick1On=True,
    label1On=False,
    tick2On=True,
    label2On=True,
)
ax.laxis.set_tick_params(
    which="both",
    tick1On=True,
    label1On=False,
    tick2On=True,
    label2On=False,
)
ax.raxis.set_tick_params(
    which="both",
    tick1On=True,
    label1On=True,
    tick2On=True,
    label2On=False,
)

plt.show()

[carlosgmartin]

@yuzie007 How can I add labels along the edges of the triangle, rather than the vertices, like the "Mole fraction Sn" and "Mole fraction Cu" labels shown in the example image?

[yuzie007]

@carlosgmartin You can do, e.g.,

ax.taxis.set_label_position('tick1')

For details, please have a look on https://mpltern.readthedocs.io/en/latest/gallery/axis_and_tick/00.axis_label_position.html#sphx-glr-gallery-axis-and-tick-00-axis-label-position-py

[carlosgmartin]

@yuzie007 Thanks for your prompt response. What I meant was, is it possible to have both vertex and edge labels, separately?

[yuzie007]

@carlosgmartin Thank you for your interesting question.

There are no simple APIs to add axis labels both at the corner and at the side, similarly that Matplotlib cannot put x-axis labels both at the bottom and at the top, unless we use ax.secondary_xaxis() or ax.twiny().
(By philosophy, I would not add something that cannot be done using Matplotlib.)

One possible workaround is making another TernaryAxes on top of the first one. The following is an example.
I hope this helps you even a bit.

import matplotlib.pyplot as plt
from mpltern.datasets import get_spiral


t, l, r = get_spiral()

fig = plt.figure()

ax = fig.add_subplot(1, 1, 1, projection='ternary')

ax.plot(t, l, r)

ax.set_tlabel('Top')
ax.set_llabel('Left')
ax.set_rlabel('Right')

# another axes exactly at the same position
tax = fig.add_axes(ax.get_position(), projection='ternary')

# remove spines
for spine in tax.spines.values():
    spine.set_visible(False)

# make the plotting region transparent
tax.set_facecolor((0, 0, 0, 0))

tax.set_tlabel('Top Side')
tax.set_llabel('Left Side')
tax.set_rlabel('Right Side')

# put axis labels on the sides
tax.taxis.set_label_position('tick1')
tax.laxis.set_label_position('tick1')
tax.raxis.set_label_position('tick1')

# (optional) plot something rather on the second axes
tax.plot(l, r, t, 'C1')

plt.savefig('test.svg')

plt.show()