latexplotlib 0.4.0

Perfect matplotlib figures for latex

latexplotlib

Perfect matplotlib figures for latex.

Usage

This package has two basic functionalities. On the one hand, it sets sensible defaults for creating perfect figures for latex. This includes a color scheme optimized for color-blind people, correct font and font sizes, and sensible defaults to store the figure. On the other hand, it provides some functions to create perfectly sized figures. These figures fit your latex document without scaling and have the correct font size for your document.

latexplotlib styles

There are 6 different styles for matplotlib.

• latex10pt
• latex11pt
• latex12pt
• latex10pt-minimal
• latex11pt-minimal
• latex12pt-minimal

The *minimal versions change the font and the font sizes to ensure that the figures fonts match the latex font perfectly. This style is fully compatible with other styles.

The non-minimal versions set additional defaults to create perfect figures, that are accessible for color-blind people while still looking nice.

Both styles change the defaults of the plt.savefig command so that all saved figures look good. These new defaults are

plt.savefig(
    ...,
    bbox_inches=None,
    dpi=300,
    format="pdf",
    orientation="portrait",
    pad_inches=0.05
)

Example

Style latex10pt:

import matplotlib.pyplot as plt
import numpy as np
import latexplotlib as lpl


plt.style.use("latex10pt")

# lpl.size.set(200, 400)
with lpl.size.context(200, 400):
    fig, ax = lpl.subplots(1, 1)

x = np.linspace(1, 5, 100)

for t in range(4):
    label = f"$x^{t}$"
    ax.plot(x, x ** t, label=label)

ax.set_yscale("log")
ax.set_title("Perfect matplotlib figures for \\LaTeX")
ax.grid()

fig.legend()
fig.savefig("example_poly")
fig.savefig("example_poly.png")

Style latex10ptminimal:

import matplotlib.pyplot as plt
import numpy as np

import latexplotlib as lpl

plt.style.use("latex10pt-minimal")
# lpl.size.set(200, 400)
with lpl.size.context(200, 400):
    fig, ax = lpl.subplots(1, 1)

x = np.linspace(1, 5, 100)

for t in range(4):
    label = f"$x^{t}$"
    ax.plot(x, x ** t, label=label)

ax.set_yscale("log")
ax.set_title("Perfect matplotlib figures for \\LaTeX")
ax.grid()

fig.legend()
plt.savefig("example_poly_minimal")
plt.savefig("example_poly_minimal.png")

Get latex dimensions

You can find the dimensions of your document using the following command:

\the\textwidth

\the\textheight

Set and get latex page size

import latexplotlib as lpl

lpl.size.set(200, 400)

with lpl.size.context(100, 200):
    lpl.size()  # 100, 200

lpl.size()  # (200, 400)

Create figures for latex

import matplotlib.pyplot as plt

import latexplotlib as lpl


# A figure filling 75% of the latex page
_ = plt.figure(figsize=lpl.figsize(fraction=0.75))

# A subplot filling 80% of the latex page
fig, axes = lpl.subplots(3, 2, fraction=0.8)

# A subplot for 3 square plots next to each other
fig, axes = lpl.subplots(1, 3, fraction=0.8, ratio=1)

ratio keyword

The ratio keyword control the ratio of height to width. The default is the Golden ratio. ratio can also be max or any. In this case, the figure fills the available space.

import matplotlib.pyplot as plt

import latexplotlib as lpl

# A 3 by 2 figure where each subplots height to width ratio is the golden ratio
fig, axes = lpl.subplots(3, 2)

# A 3 by 2 figure where each subplot having a height to width ratio of 1:1
fig, axes = lpl.subplots(3, 2, ratio=1.0)

# A figure that is exactly 300pt height and 200pt wide
with lpl.size.context(200, 300):
    fig, axes = lpl.subplots(3, 2, ratio"max")

Include figures in Latex

\begin{figure}[tb]
    \centering
    \includegraphics{test.pdf}
    \caption{A test figure.}
\end{figure}

plt.tight_layout()

plt.tight_layout() changes the size of the produced figure. As such it is recommended to only use plt.tight_layout() with care! The same is true for savefig(..., bbox_inches=None)!

Instead one should use constrained_layout which produces nice figures. constrained_layout is used by default with all latexplotlib styles.

References

This package is inspired by the following sources:

• Code: https://pypi.org/project/SciencePlots/
• Figure sizes: https://jwalton.info/Embed-Publication-Matplotlib-Latex/
• Color palette (Okabe Ito): https://clauswilke.com/dataviz/color-basics.html