Stan Workbench

Build

cd dkr
docker-compose build

Be patient; this takes a while to complete the build and the load. To stop, hit ctrl-c from the terminal.

The dkr/workbench host folder is mapped to the work folder in the notebook container.

Running

docker-compose up -d

Alternatively, combine the build & run steps

docker-compose up --build -d

Output

To inspect the logs

docker logs -f stanspace

Using the container's IPython

From the host's command line; useful if you wish to ssh in remotely

docker-compose exec lab ipython

Note that the stack needs to be up and running for this to work

Stopping

docker-compose down

Additional notes

PyStan in Jupyter

The PyStan package isn't compatible with the asyncio event loop used internally by Jupyter.
As a workaround there are two options:

1. Run the following at the top of any Python notebooks that will call PyStan:

import nest_asyncio
nest_asyncio.apply()
del nest_asyncio

2. Use the pystan-jupyter package which performs the above:

import stan_jupyter as stan