Implementing Global optimization of Lipschitz functions
The scripts that work (at least at time of writing) can be run with the following commands (from the root directory of the project).
This one runs the sequential optimizers (the --synthetic flag indicates that we
don't want to optimize the real world cros validation based functions):
python src/optimize.py results/syn-test --synthetic
This one produces parts of Figure 5 from the original paper:
python src/figure_five.py results/syn-test.pkl results/syn-test
These commands are generating results for the synthetic objective functions. The functionality to generate results corresponding to the 'real world' objective functions is currently being developed.
Currently the Docker container is designed simply to allow execution of the scripts on a host that doesn't have dependencies (eg, python or certain packages) available locally.
The container can be used interactively at this point by cloning this repo, building the image from the Dockerfile, and running it interactively:
git clone https://github.com/UBC-CS/lipo-python.git
cd lipo-python
docker build . -t lipo-python
docker run -v <path-to-cloned-repo>/:/home/ -it lipo-python