Proximal gradient algorithm with trust region scheme on Riemannian manifold
Abstract We consider the problem of minimizing the sum of a smooth function and nonsmooth function over a Riemannian manifold. We develop a Riemannian proximal gradient algorithm with a trust region scheme for the problem. Global convergence is presented under natural assumptions. We establish an \(O(1/{\varepsilon ^2})\) iteration complexity bound that matches the best-known complexity bound of Riemannian trust region methods for smooth optimization. Moreover, we give a nonmonotone version and an accelerated version of the Riemannian proximal gradient algorithm with a trust region scheme. Their global convergence is established. Numerical results demonstrate the effectiveness of the proposed methods.
Abstract
We consider the problem of minimizing the sum of a smooth function and nonsmooth function over a Riemannian manifold. We develop a Riemannian proximal gradient algorithm with a trust region scheme for the problem. Global convergence is presented under natural assumptions. We establish an \(O(1/{\varepsilon ^2})\) iteration complexity bound that matches the best-known complexity bound of Riemannian trust region methods for smooth optimization. Moreover, we give a nonmonotone version and an accelerated version of the Riemannian proximal gradient algorithm with a trust region scheme. Their global convergence is established. Numerical results demonstrate the effectiveness of the proposed methods.
What's Your Reaction?