PANTR: A proximal algorithm with trust-region updates for nonconvex constrained optimization

This work presents PANTR, an efficient solver for nonconvex constrained optimization problems, that is well-suited as inner augmented Lagrangian method. The proposed scheme combines forward-backward iterations with solutions to trust-region subproblems: the former ensures global convergence, whereas latter enables fast update directions. We discuss how algorithm able exploit exact Hessian information of smooth objective term through a linear Newton approximation, while benefiting from structure box-constraints or ℓ1-regularization. An open-source C++ implementation PANTR made available part NLP library ALPAQA. Finally, effectiveness method demonstrated in nonlinear model predictive control applications.