Walking in the Shadow: A New Perspective on Descent Directions for Constrained Minimization Code

This is the code for our paper titled Walking in the Shadow: A New Perspective on Descent Directions for Constrained Minimization (https://arxiv.org/abs/2006.08426)

---

Requirements:

• This code requires the aeroplane_data_small.mat data set, which is included in this repository and could also be downloaded at https://github.com/Simon-Lacoste-Julien/linearFW
• This code also requires Gurobi (with a valid license) to be installed to the python environment used - this could be done at https://www.gurobi.com/documentation/9.0/quickstart_mac/the_grb_python_interface_f.html

---

The complete code with explanantions is given in the Jupyter Notebook file:

Walking_in_the_shadow_code.ipynb

We also provide separate python files for each experiment:

• To run the video - co-localization experiment:
  python Video Co-localization.py
  
  

• To run the Lasso Regression (small instance) Experiment:
  python lasso_regression_small.py
  
  

• To run the Lasso Regression (large instance) Experiment:
  python lasso_regression_large.py