The Intrinsic Riemannian Proximal Gradient Method for Nonconvex Optimization
Preprint
Abstract
We consider the proximal gradient method on Riemannian manifolds for functions that are possibly not geodesically convex. Starting from the forward-backward-splitting, we define an intrinsic variant of the proximal gradient method that uses proximal maps defined on the manifold and therefore does not require or work in the embedding. We investigate its convergence properties and illustrate its numerical performance, particularly for nonconvex or nonembedded problems that are hence out of reach for other methods.Keywords
All Related Versions
This publication has 28 references indexed in Scilit:
- The Difference of Convex Algorithm on Hadamard ManifoldsJournal of Optimization Theory and Applications, 2024
- A Grassmann manifold handbook: basic geometry and computational aspectsAdvances in Computational Mathematics, 2024
- An inexact Riemannian proximal gradient methodComputational Optimization and Applications, 2023
- Riemannian proximal gradient methodsMathematical Programming, 2021
- Fenchel Duality Theory and a Primal-Dual Algorithm on Riemannian ManifoldsFoundations of Computational Mathematics, 2021
- A modified proximal point method for DC functions on Hadamard manifoldsComputational Optimization and Applications, 2020
- MADMM: A Generic Algorithm for Non-smooth Optimization on Manifolds2016
- On Some Basic Results Related to Affine Functions on Riemannian ManifoldsJournal of Optimization Theory and Applications, 2016
- A proximal point algorithm for DC fuctions on Hadamard manifoldsJournal of Global Optimization, 2015
- Low-rank retractions: a survey and new resultsComputational Optimization and Applications, 2014