如何在Docker容器的IPython环境中渲染并显示Matplotlib图表？

Got it, let's break down why your matplotlib plots aren't showing up and walk through the fixes that work for your setup:

Root Cause #

The core issue is that your Docker container doesn't have access to your host machine's graphical display system. The default agg backend is a non-interactive, headless renderer—it can generate plot data but can't pop up a window. Even %matplotlib inline won't work in a raw shell session because it relies on a frontend (like Jupyter's browser interface or an IDE's rendering panel) that isn't present in a basic bash shell attached to the container.

Solutions #

1. Allow Container to Access Host's X11 Display (Linux/macOS) #

This lets the container use your host's graphical system to pop up matplotlib windows directly.

For Linux:

• First, on your host machine, grant X11 access to Docker containers temporarily:

  xhost +
  

  (Run xhost - afterward to lock down X11 access again for security.)
• Modify your Docker run command to mount the X11 socket and set the DISPLAY environment variable:

  docker run --rm -it -p 8888:8888 \
  --mount type=bind,source=/project,target=/work \
  --mount type=bind,source=/tmp/.X11-unix,target=/tmp/.X11-unix \
  -e DISPLAY=$DISPLAY \
  python-3.9.1-jupyterlab
  

• Attach the shell again, then switch to an interactive backend in IPython:

  %matplotlib tk
  plt.plot([1.6, 2.7])
  plt.show()
  

  You should see a plot window pop up on your host desktop.

For macOS:

• First, install XQuartz (required for X11 support on macOS).
• Open XQuartz, then run this in a host terminal to grant local access:

  xhost +localhost
  

• Modify your Docker run command to point to the host's X11 display:

  docker run --rm -it -p 8888:8888 \
  --mount type=bind,source=/project,target=/work \
  -e DISPLAY=host.docker.internal:0 \
  python-3.9.1-jupyterlab
  

• Follow the same IPython steps above to switch to tk backend and show the plot.

2. Use VSCode's Jupyter Extension (Best for IDE Workflow) #

Since you're already using VSCode, skip the shell attach entirely and connect directly to the container's Jupyter server—this lets plots render natively in VSCode's interface.

• Start your container as usual, then copy the Jupyter access URL with token from the container's output (looks like http://127.0.0.1:8888/?token=xxxxxx).
• In VSCode, open the Jupyter extension via the sidebar icon.
• Click Connect to Jupyter Server → select Existing → paste the URL you copied.
• Create or open an .ipynb file in VSCode (you can access your /project files via the mounted /work directory).
• Run plot code—%matplotlib inline will work natively, and plots will display directly below the code cell in VSCode.

3. Save Plots to File (Quick Workaround) #

If you just need to view plots without graphical access, save them to the mounted directory and open them on your host:

In IPython:

plt.plot([1.6, 2.7])
plt.savefig('/work/plot.png')

The file will appear in your host's /project directory—open it with any image viewer.

内容的提问来源于stack exchange，提问作者Jeff