Dual Descent ALM and ADMM

Classical primal-dual algorithms attempt to solve $\max_{\mu}\min_{x} $\mathcal{L}$(x,\mu)$byalternativelyminimizingovertheprimalvariable$x$throughprimaldescentandmaximizingthedualvariable$\mu$ through dual ascent. However, when $\mathcal{L}(x,\mu )$ is highly nonconvex with complex constraints in $x$, the minimization over $x$ may not achieve global optimality, and hence the dual ascent step loses its valid intuition. This observation motivates us to propose a new class of primal-dual algorithms for nonconvex constrained optimization with the key feature to reverse dual ascent to a conceptually new dual descent, in a sense, elevating the dual variable to the same status as the primal variable. Surprisingly, this new dual scheme achieves some best iteration complexities for solving nonconvex optimization problems. In particular, when the dual descent step is scaled by a fractional constant, we name it scaled dual descent (SDD), otherwise, unscaled dual descent (UDD). For nonconvex multiblock optimization with nonlinear equality constraints, we propose SDD-ADMM and show that it finds an $ϵ$-stationary solution in $\mathcal{O}({ϵ}^{-4})$ iterations. The complexity is further improved to $\mathcal{O}({ϵ}^{-3})$ and $\mathcal{O}({ϵ}^{-2})$ under proper conditions. We also propose UDD-ALM, combining UDD with ALM, for weakly convex minimization over affine constraints. We show that UDD-ALM finds an $ϵ$-stationary solution in $\mathcal{O}({ϵ}^{-2})$ iterations. These complexity bounds for both algorithms either achieve or improve the best-known results in the ADMM and ALM literature. Moreover, SDD-ADMM addresses a long-standing limitation of existing ADMM frameworks.