The Proximal Augmented Lagrangian Method for Nonsmooth Composite Optimization

• Published in: IEEE transactions on automatic control 2019-07, Vol.64 (7), p.2861-2868
• Main Authors: Dhingra, Neil K., Khong, Sei Zhen, Jovanovic, Mihailo R.
• Format: Article
• Language: English
• Subjects: Algorithms; Ascent; Asymptotic methods; Asymptotic stability; Augmented Lagrangian; Continuity (mathematics); control for optimization; Convergence; global exponential stability; Gradient flow; Gradient methods; Linear programming; method of multipliers; Minimization; Multipliers; non-smooth optimization; Nonlinear programming; Optimal control; Optimization; primal-dual dynamics; proximal algorithms; proximal augmented Lagrangian; Regularization; regularization for design; structured optimal control
• ISSN: 0018-9286; 1558-2523
• DOI: 10.1109/TAC.2018.2867589

We study a class of optimization problems in which the objective function is given by the sum of a differentiable but possibly nonconvex component and a nondifferentiable convex regularization term. We introduce an auxiliary variable to separate the objective function components and utilize the Moreau envelope of the regularization term to derive the proximal augmented Lagrangian-a continuously differentiable function obtained by constraining the augmented Lagrangian to the manifold that corresponds to the explicit minimization over the variable in the nonsmooth term. The continuous differentiability of this function with respect to both primal and dual variables allows us to leverage the method of multipliers (MM) to compute optimal primal-dual pairs by solving a sequence of differentiable problems. The MM algorithm is applicable to a broader class of problems than proximal gradient methods and it has stronger convergence guarantees and a more refined step-size update rules than the alternating direction method of multipliers (ADMM). These features make it an attractive option for solving structured optimal control problems. We also develop an algorithm based on the primal-descent dual-ascent gradient method and prove global (exponential) asymptotic stability when the differentiable component of the objective function is (strongly) convex and the regularization term is convex. Finally, we identify classes of problems for which the primal-dual gradient flow dynamics are convenient for distributed implementation and compare/contrast our framework to the existing approaches.