Utilities for generating PGF figures from Matplotlib

pip install matplotlib-pgfutils==1.5.0



The Portable Graphics Format (PGF) is a language for producing vector graphics within TeX documents. There is also a higher-level language TikZ (TikZ ist kein Zeichenprogramm -- TikZ is not a drawing program) which uses PGF.

Since version 1.2, the Python plotting library Matplotlib has included a PGF backend to generate figures ready for inclusion in a TeX document. In order to get consistent figures that fit with the style of the document, this requires some configuration. The aim of pgfutils is to simplify this configuration and allow figures to be easily generated from a Python script.

The module provides two functions which set up and then save the figure. The actual plotting is performed by standard Matplotlib functions. For example, to generate a plot showing the 1st, 3rd, 5th, 7th, 9th and 11th harmonics of an ideal square wave and the resulting sum:

# Set up the figure environment.
from pgfutils import setup_figure, save
setup_figure(width=0.9, height=0.4)

import numpy as np
from matplotlib import pyplot as plt

# Generate a square wave from a few terms of its Fourier series.
f = 3
t = np.linspace(0, 1, 501)
square = np.zeros(t.shape)
for n in range(1, 12, 2):
    # Create this harmonic and plot it as a dashed
    # partially-transparent line.
    component = np.sin(2 * n * np.pi * f * t) / n
    plt.plot(t, component, '--', alpha=0.4)

    # Add it to the overall signal.
    square += component

# Scale the final sum.
square *= 4 / np.pi

# Plot and label the figure.
plt.plot(t, square, 'C0')
plt.xlim(0, 1)
plt.ylim(-1.2, 1.2)
plt.xlabel("Time (s)")
plt.ylabel("Amplitude (V)")

# Save as a PGF image.


Using pgfutils requires Python 3. Each commit is currently tested with release versions of Python 3.6, 3.7, 3.8 and 3.9.

The only external dependency is matplotlib. All the other dependencies are part of the standard Python library.


pgfutils comes with some examples which demonstrate its usage and integration into a build system. Depending on your installation method, these may be present somewhere in your filesystem (e.g., on a Linux system, they might be at /usr/share/matplotlib-pgfutils/examples). They can also be found in the extras/examples directory.


Documentation for pgfutils can be found online at

Alternatively, you can find the source of this documentation in Markdown format in the doc/ directory of the source:

An example configuration file showing the default settings is given in extras/pgfutils.cfg.

Unit testing

Build Status codecov

Extensive unit tests are included in the tests/ directory of the source. Each commit to the source repository is automatically tested thanks to Travis CI. The test coverage (that is, how many of the lines of code in the source were executed during the tests) is monitored by Codecov. The badges above show the status of the last commit made to the source.

You can also run the tests on a local copy of the code. They are designed to be run with the pytest framework and employ the package via the pytest-cov plugin to measure the coverage. If you have these packages installed, run pytest from the top-level directory to execute the tests. A basic test coverage report will be printed on the terminal, and the full report can be viewed by opening the htmlcov/index.html file in your web browser.


pgfutils is released under the three-clause BSD license. The terms of this license can be found in the LICENSE file in the source, or online at

The Cotham Sans font used in some unit tests is copyright (c) 2015 Sebastien Sanfilippo and is licensed under the SIL Open Font License, Version 1.1. The license can be found in the source at tests/sources/fonts/Cotham/OFL.txt or online at and the font itself can be found at