Alternating direction method of multipliers with variable step sizes
Provides a practical improvement to ADMM, a widely used optimization algorithm, for practitioners solving convex minimization problems.
The authors propose an adjustment rule for the step size in ADMM based on residual monotonicity, achieving significant improvements over established variants in numerical experiments.
The alternating direction method of multipliers (ADMM) is a flexible method to solve a large class of convex minimization problems. Particular features are its unconditional convergence with respect to the involved step size and its direct applicability. This article deals with the ADMM with variable step sizes and devises an adjustment rule for the step size relying on the monotonicity of the residual and discusses proper stopping criteria. The numerical experiments show significant improvements over established variants of the ADMM.