rosen_style

Consistent, readable Matplotlib defaults for Rosen Research Group papers and presentations.

Install

pip install git+https://github.com/Quantum-Accelerators/rosen_style.git

Use

import matplotlib.pyplot as plt
import rosen_style

rosen_style.use("paper")  # or "presentation"
fig, ax = plt.subplots()
ax.plot([0, 1, 2], [0, 1, 0])
ax.set(xlabel="Time (s)", ylabel="Response (a.u.)")

Use a context manager to apply a style temporarily:

import matplotlib.pyplot as plt
import rosen_style

with rosen_style.context("paper"):
    fig, ax = plt.subplots()
    ax.scatter([1, 2, 3], [2.1, 3.8, 6.2])
    ax.set(xlabel="Concentration (mol/L)", ylabel="Response (a.u.)")
    fig.savefig("response.png")  # saved at the style default of 600 DPI

Paper figures default to a 3.25-inch single-column width, with height chosen using the golden ratio. Use columns=2 for a 7-inch double-column figure:

with rosen_style.context("paper", columns=2):
    fig, ax = plt.subplots()
    ax.plot([0, 1, 2], [0, 1, 0])
    fig.savefig("double-column.png")

Pass square=True for equal figure width and height, which is useful for parity plots and heatmaps. See the main Matplotlib settings for the complete defaults.

Outside the with block, Matplotlib's previous settings are restored.

The defaults use 600 DPI for display and saved output, a color-vision-friendly categorical cycle, the perceptually uniform plasma image colormap, readable labels, white saved backgrounds, no grid lines, and minor ticks in paper mode. Figure titles are intentionally left to captions or surrounding presentation content. Pair color with markers, line styles, or direct labels when it carries meaning.

Examples

Line plot:

Scatter plot (with square=True):

Heatmap with a perceptually uniform color scale:

CI regenerates and commits these images when their source or the styles change.

Design references

• Claus O. Wilke, Fundamentals of Data Visualization

There are also many excellent Python examples on The Python Graph Gallery and Python Charts websites. For what not to do, check out the "Friends Don't Let Friends Make Bad Graphs" repository.