L2O-with-guarantees

Code for running experiments on the Learning to Optimize with Guarantees framework.

Problem Settings

• least_squares.py
  Ill-conditioned least squares ((\kappa \approx 1000)).
  Baselines: GD, NAG

• bilinear_game.py
  Bilinear zero-sum games
  Baseline: Extragradient

• monotone_vi.py
  Strongly monotone games
  Baseline: Projected Gradient