Install#

OptFlow is currently formed as a Python package. To install OptFlow, please download the .whl file from the link provided below:

and use the pip install command within your Python environment:

pip install optflow-0.1.3-py3-none-any.whl

Note

OptFlow is in Alpha test. The functionality is evolving rapidly and the APIs may subject to change in further versions without notice in advance.

Dependencies#

conda install numpy 
conda install cvxpy                             # for programming-based optimizer, https://www.cvxpy.org
pip install tensorflow                          # for learning-to-optimize model training
pip install tensorflow_probability
pip install llvmlite                            # for JIT compilation
pip install matplotlib
pip install graphviz                            # python API of Graphviz for graph visualization, https://github.com/xflr6/graphviz

Optional Dependencies#

conda install --channel conda-forge pyscipopt   # if you want to use SCIP as programming-based optimizer, https://github.com/scipopt/PySCIPOpt#installation
conda install graphviz                          # https://www.graphviz.org/
pip install gym[classic_control]    # for the gym simulation environment
pip install moviepy                 # for gym result visualization in jupyter notebook

Legacy Version#