Adaptive quadratically regularized Newton method for Riemannian optimization
From MaRDI portal
Publication:3176355
DOI10.1137/17M1142478zbMATH Open1415.65139arXiv1708.02016OpenAlexW2884819278WikidataQ115246937 ScholiaQ115246937MaRDI QIDQ3176355FDOQ3176355
Authors: Jiang Hu, Andre Milzarek, Zaiwen Wen, Yaxiang Yuan 
Publication date: 20 July 2018
Published in: SIAM Journal on Matrix Analysis and Applications (Search for Journal in Brave)
Abstract: Optimization on Riemannian manifolds widely arises in eigenvalue computation, density functional theory, Bose-Einstein condensates, low rank nearest correlation, image registration, and signal processing, etc. We propose an adaptive regularized Newton method which approximates the original objective function by the second-order Taylor expansion in Euclidean space but keeps the Riemannian manifold constraints. The regularization term in the objective function of the subproblem enables us to establish a Cauchy-point like condition as the standard trust-region method for proving global convergence. The subproblem can be solved inexactly either by first-order methods or a modified Riemannian Newton method. In the later case, it can further take advantage of negative curvature directions. Both global convergence and superlinear local convergence are guaranteed under mild conditions. Extensive computational experiments and comparisons with other state-of-the-art methods indicate that the proposed algorithm is very promising.
Full work available at URL: https://arxiv.org/abs/1708.02016
Recommendations
- A Broyden class of quasi-Newton methods for Riemannian optimization
- Riemannian stochastic variance-reduced cubic regularized Newton method for submanifold optimization
- Faster Riemannian Newton-type optimization by subsampling and cubic regularization
- Sequential quadratic optimization for nonlinear optimization problems on Riemannian manifolds
- Adaptive regularization with cubics on manifolds
Numerical mathematical programming methods (65K05) Methods of successive quadratic programming type (90C55) Numerical computation of eigenvalues and eigenvectors of matrices (65F15) Nonconvex programming, global optimization (90C26)
Cites Work
- KSSOLV -- a MATLAB toolbox for solving the Kohn-Sham equations
- A feasible method for optimization with orthogonality constraints
- Manopt, a Matlab toolbox for optimization on manifolds
- Adaptive subgradient methods for online learning and stochastic optimization
- Trust-region methods on Riemannian manifolds
- Title not available (Why is that?)
- The Geometry of Algorithms with Orthogonality Constraints
- Title not available (Why is that?)
- Title not available (Why is that?)
- Optimization Techniques on Riemannian Manifolds
- A Nonmonotone Line Search Technique and Its Application to Unconstrained Optimization
- Low-rank tensor completion by Riemannian optimization
- Low-rank matrix completion by Riemannian optimization
- Optimization methods on Riemannian manifolds and their application to shape space
- The Conjugate Gradient Method and Trust Regions in Large Scale Optimization
- A Riemannian symmetric rank-one trust-region method
- A Broyden class of quasi-Newton methods for Riemannian optimization
- An Extrinsic Look at the Riemannian Hessian
- An inexact successive quadratic approximation method for L-1 regularized optimization
- Adaptive cubic regularisation methods for unconstrained optimization. I: Motivation, convergence and numerical results
- A framework of constraint preserving update schemes for optimization on Stiefel manifold
- A regularized Newton method for computing ground states of Bose-Einstein condensates
- Projection-like retractions on matrix manifolds
- Optimality conditions for the nonlinear programming problems on Riemannian manifolds
- Exploiting negative curvature directions in linesearch methods for unconstrained optimization
- Minimizing a differentiable function over a differential manifold
- Globally convergent optimization algorithms on Riemannian manifolds: Uniform framework for unconstrained and constrained optimization
- Proximal Newton-type methods for minimizing composite functions
- A Riemannian BFGS method for nonconvex optimization problems
- Algebraic rules for quadratic regularization of Newton's method
- Accelerated methods for nonconvex optimization
- Finding approximate local minima faster than gradient descent
- The use of quadratic regularization with a cubic descent condition for unconstrained optimization
- The Riemannian Barzilai-Borwein method with nonmonotone line search and the matrix geometric mean computation
- Adaptive regularized self-consistent field iteration with exact Hessian for electronic structure calculation
- Gradient type optimization methods for electronic structure calculations
- A New First-Order Algorithmic Framework for Optimization Problems with Orthogonality Constraints
- A proximal gradient method for ensemble density functional theory
Cited In (36)
- A Riemannian dimension-reduced second-order method with application in sensor network localization
- Finding the global optimum of a class of quartic minimization problem
- A communication-efficient and privacy-aware distributed algorithm for sparse PCA
- New vector transport operators extending a Riemannian CG algorithm to generalized Stiefel manifold with low-rank applications
- Proximal quasi-Newton method for composite optimization over the Stiefel manifold
- An exact penalty approach for optimization with nonnegative orthogonality constraints
- Riemannian preconditioning
- Adaptive strategy for the damping parameters in an iteratively regularized Gauss-Newton method
- A brief introduction to manifold optimization
- Riemannian smoothing gradient type algorithms for nonsmooth optimization problem on compact Riemannian submanifold embedded in Euclidean space
- AN EFFICIENT METHOD FOR SOLVING A CLASS OF MATRIX TRACE FUNCTION MINIMIZATION PROBLEM IN MULTIVARIATE STATISTICAL
- A semismooth Newton based augmented Lagrangian method for nonsmooth optimization on matrix manifolds
- Effective algorithms for solving trace minimization problem in multivariate statistics
- Adaptive regularization with cubics on manifolds
- Structured Quasi-Newton Methods for Optimization with Orthogonality Constraints
- Riemannian Natural Gradient Methods
- An adaptive regularized proximal Newton-type methods for composite optimization over the Stiefel manifold
- Title not available (Why is that?)
- An extended delayed weighted gradient algorithm for solving strongly convex optimization problems
- Riemannian trust region methods for \(\mathrm{SC}^1\) minimization
- An entropy-regularized ADMM for binary quadratic programming
- A Decomposition Augmented Lagrangian Method for Low-Rank Semidefinite Programming
- Energy-adaptive Riemannian optimization on the Stiefel manifold
- Orthogonal canonical correlation analysis and applications
- Newton-based methods for finding the positive ground state of Gross-Pitaevskii equations
- A trust region method for solving multicriteria optimization problems on Riemannian manifolds
- Riemannian stochastic variance-reduced cubic regularized Newton method for submanifold optimization
- A class of smooth exact penalty function methods for optimization problems with orthogonality constraints
- Adaptive trust-region method on Riemannian manifold
- Exact penalty function for \(\ell_{2,1}\) norm minimization over the Stiefel manifold
- Global convergence of Riemannian line search methods with a Zhang-Hager-type condition
- On the geometric analysis of a quartic-quadratic optimization problem under a spherical constraint
- Sequential quadratic optimization for nonlinear optimization problems on Riemannian manifolds
- A penalty-free infeasible approach for a class of nonsmooth optimization problems over the Stiefel manifold
- Ground states of spin-\(F\) Bose-Einstein condensates
- Faster Riemannian Newton-type optimization by subsampling and cubic regularization
Uses Software
This page was built for publication: Adaptive quadratically regularized Newton method for Riemannian optimization
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3176355)
