Python implementation for the cutting plane algorithm in: Sparse high-dimensional regression: Exact scalable algorithms and phase transitions
Step1: Run data_generation.py with specific parameters:
- n: number of samples
- p: dimension
- k: number of nonzero components in the true regressor
$w$ -
$\rho$ : correlation coefficient
Step2: Run OA_process.py to get the results.
Note that: for large problem, turn off the warm-start attribute. i.e. s0_star, w0_star, s1 = OA_process(X, Y, k, gamma, False)