“…The problem structure (1) arises in various contexts and applications, e.g., nonlinear least-squares formulations for estimation and tracking [5,24,34], nonlinear matrix inequalities for reduced order controller design [12,38], nonlinear model predictive control with ellipsoidal feasible sets [7,18,41,44], robustified inequalities in nonlinear optimization [9,29], tube-following optimal control problems [39], non-smooth composite minimization [2,23], as well as training of deep neural networks [25-27, 33, 42].…”