• Jupyter Notebook 4.3
• Scientific Python Stack
• Python 3
• Python 2
• Nengo 2.3.1

After installing Docker, the following command downloads an image from DockerHub and starts a container using the image jjaguayo/nengo-scipy-jupyter. The image has a Jupyter notebook server, listening for HTTP connections on port 8888, capable of running Nengo models.

    docker run -it --restart=always -v <absolute-path-on-host>:/home/jovyan/work -p 8888:8888 jjaguayo/nengo-scipy-jupyter

Remember to replace <absolute-path-on-host> with the absolute path to the directory you want to share with the container.

This link gives more information on the Scientific Python Stack (provided by Jupyter) used as a base for this Docker image.

This link gives more information on Nengo, a Python library for building and simulating large-scale brain models.

The command above starts a restartable container that restarts if the application exits.

If you want to stop a running container, two commands can be used, docker ps and docker stop. The first command gives you a list of running containers and their ids. The second command stops the running container given the container id. For example, the following helps you find the id of the running container you want to stop (using docker ps) and the second command stops the running container.

    docker ps
    docker stop <container_id>

The option --restart=always indicates to Docker that the container should be restarted if the container is stopped or if the application exits.

If the user wants to change this behavior having the container simply exit and not restart, replace the --restart option with --rm. For example, the following will run a container and not restart it if the application ends or the container is stopped.

    docker run -it --rm -v <absolute-path-on-host>:/home/jovyan/work -p 8890:8888 jjaguayo/nengo-scipy-jupyter

Data stored in the jupyter notebook can be shared with a directory in your host using the -v option. For example, the following shares the data stored in the jupyter notebook with the directory <absolute-path-on-host> on my machine hosting the container.

    docker run -it --restart=always -v <absolute-path-on-host>:/home/jovyan/work -p 8888:8888 jjaguayo/nengo-scipy-jupyter

Remember to replace <absolute-path-on-host> with the absolute path to the directory you want to share with the container.

It is possible that port 8888 on your host machine is being used. If so, a different host port can be used by changing the first number in the -p option. For example, the following uses host port 8890 mapping it to port 8888 in the container.

    docker run -it --restart=always -v <absolute-path-on-host>:/home/jovyan/work -p <available-host-port-id>:8888 jjaguayo/nengo-scipy-jupyter

Remember to change <available-host-port-id> to an accessible port id on the host.

If you do not care to download the image from DockerHub, clone this repository and run the following from the directory containing the file Dockerfile.

    docker build -t <your_local_build_name> .
    docker run -it --restart=always -p 8888:8888 <your_local_build_name>

The first command builds an image called <your_local_build_name> and the second command runs it in a restartable container.

When starting the hub, a URL is provided by the container app. Copy and paste the URL into a browser to connect to the hub.

If you have not installed Docker, the following links provide instructions for installing on

• Ubuntu (14.04 or later)
• Mac OS X (Hardware releases 2010 or later)
• Windows 10

Note that there are minimum requirements for each platform so check the Docker documentation for more information on what is supported.