Cartesian coordinates

In mpltern, the x- and y-limits are initially [−3^−1/2, +3^−1/2] and [0, 1], respectively. The triangle vertices are, by default, (0, 1), (−3^−1/2, 0), (+3^−1/2, 0).

In most methods, we can still plot in Cartesian coordinates rather than in ternary coordinates by using Matplotlib transforms like ax.transData, ax.transAxes, fig.transFigure.

import numpy as np
import matplotlib.pyplot as plt
import mpltern


def plot_frame(ax):
    """Plot frame of the original Axes."""
    ax.fill(
        [0, 0, 1, 1],
        [0, 1, 1, 0],
        ec="k",
        fc="none",
        ls=":",
        clip_on=False,
        transform=ax.transAxes,
    )


ax = plt.subplot(projection='ternary')
plot_frame(ax)

This may be useful to plot e.g. the subfigure label as well as to understand the legend position.

fig = plt.figure(figsize=(10.8, 4.8))
fig.subplots_adjust(wspace=0.3)

ax = fig.add_subplot(1, 2, 1, projection='ternary')
plot_frame(ax)
x = [-1.0 / np.sqrt(3.0), +0.5 / np.sqrt(3.0)]
y = [0.0, 0.5]
ax.plot(x, y, label="a", transform=ax.transData)
ax.legend(loc=1)
ax.text(0.02, 0.94, "(a)", ha="left", transform=ax.transAxes)

ax = fig.add_subplot(1, 2, 2, projection='ternary')
plot_frame(ax)
x = [+1.0 / np.sqrt(3.0), -0.5 / np.sqrt(3.0)]
y = [0.0, 0.5]
ax.plot(x, y, label="b", transform=ax.transData)
ax.legend(loc=2)
ax.text(0.98, 0.94, "(b)", ha="right", transform=ax.transAxes)

plt.show()