A simple, Ubuntu-based docker image, containing a bunch of utilities to do mathematical optimization inside Jupyter notebooks based on AMPL, a powerful mathematical optimization modelling language.
It includes python modules for data processing and analysis (numpy, pandas, matplotlib), and amplpy, a python interface for the AMPL language, so that you can seamlessly switch between AMPL and Python code to process and visualize the data from your optimization problems.
You can get the image either from docker hub:
docker pull leotac/ampl-notebook
or build it locally from the Dockerfile:
git clone git@github.com:sbebo/ampl-docker.git
cd ampl-docker
docker build -t ampl-notebook .
Note: AMPL is a commercial product, and docker build will automatically download a (limited)
demo version from http://ampl.com/, containing several commercial optimization solvers,
such as cplex, gurobi, xpress, baron, snopt.
If you want to use your own fully licensed copy of AMPL and solvers, you must provide the location of an archive containing the AMPL binary for linux and solvers, as follows:
docker build --build-args AMPL=ampl.tgz -t ampl-notebook .
The archive must be within the Docker context, that is, in the subtree rooted in the directory where you run the docker build command.
The default entrypoint is the script start_notebook.sh, that starts a Jupyter notebook. The container can be run, for example, as follows:
docker run --rm -it -v $(pwd):/root/mycode -p 8888:8888 ampl-notebook
where the current directory is mounted in the container in /root/mycode, and we published the port 8888 so that it can be reached from the host. The server will be available from the host at 0.0.0.0:8888.