Start by obtaining the docker image and source code. Pull the docker image:
docker pull morblockdock/numbat
The docker image can be run simply with:
docker run -it morblockdock/numbat zsh
from which you can navigate to the example files and run them as described elsewhere in the docs.
To have local access to the source files and simulation results the process is a bit more involved.
If on linux, you must either run the following commands in sudo, or first grant yourself permission (otherwise you'll get the
docker: Got permission denied while trying to connect to the Docker daemon socket error).
su - ${USER}
sudo usermod -aG docker ${USER}
Now, create a directory NumBAT-local
mkdir NumBAT-local
We now copy the source code into the NumBAT-local folder from the docker image. From the present working directory (ie without moving into NumBAT-local) execute the following:
docker run --rm --entrypoint tar morblockdock/numbat cC /home/NumBAT/ . | tar xvC ./NumBAT-local
This command runs the docker image, copies the source code and then removes the running docker container.
Now we run the image as before, but in addition we mount the local directory, allowing us to modify the files and obtain the results directly:
docker run -v $(pwd)/NumBAT-local/:/home/NumBAT/ -it morblockdock/numbat zsh
This runs the docker image, mounting the local directory to the path /home/NumBAT/ in the docker container, and runs the zsh shell).
Changes made in the NumBAT source code now occur for the container also, allowing for local development and easy access of results.
Scripts can be run directly in a container by passing the path, for example:
docker run -v $(pwd)/NumBAT-local/:/home/NumBAT/ -it morblockdock/numbat python3 /home/NumBAT/tutorials/sim-tut_01-first_calc.py
Runs the sim-tut_01-first_calc.py script directly from the local commandline
(without having to enter the docker container manually).
Finally, one can access a jupyterlab instance and run NumBAT in interactive jupyter notebooks from docker with the following:
docker run -v $(pwd)/NumBAT-local/:/home/NumBAT/ -itp 8888:8888 morblockdock/numbat
This will spawn an address to be used in a browser as per a typical jupyter instance. The terminal in jupyterlab can also be used to run the python scripts directly. Note that on recent versions of WSL the standard ip of 127.0.0.1 in the jupyter address may need to be replaced with [::1] due to ipv6 related issues. Also, don't forget to enable the dark mode jupyterlab theme found in the settings.
If you wish to build your own docker image, the local dockerfile is provided as an example. From the higher directory you can build a docker image using:
docker build -t "numbat" NumBAT
The dockerfile is a direct mapping of the setup script, in the syntax currently preferred in the docker image build process.