Multipliers Correction Methods for Optimization Problems over the Stiefel Manifold

Multipliers Correction Methods for Optimization Problems over the Stiefel Manifold

Year:    2021

Author:    Lei Wang, Bin Gao, Xin Liu

CSIAM Transactions on Applied Mathematics, Vol. 2 (2021), Iss. 3 : pp. 508–531

Abstract

We propose a class of multipliers correction methods to minimize a differentiable function over the Stiefel manifold. The proposed methods combine a function value reduction step with a proximal correction step. The former one searches along an arbitrary descent direction in the Euclidean space instead of a vector in the tangent space of the Stiefel manifold. Meanwhile, the latter one minimizes a first-order proximal approximation of the objective function in the range space of the current iterate to make Lagrangian multipliers associated with orthogonality constraints symmetric at any accumulation point. The global convergence has been established for the proposed methods. Preliminary numerical experiments demonstrate that the new methods significantly outperform other state-of-the-art first-order approaches in solving various kinds of testing problems.

You do not have full access to this article.

Already a Subscriber? Sign in as an individual or via your institution

Journal Article Details

Publisher Name:    Global Science Press

Language:    English

DOI:    https://doi.org/10.4208/csiam-am.SO-2020-0008

CSIAM Transactions on Applied Mathematics, Vol. 2 (2021), Iss. 3 : pp. 508–531

Published online:    2021-01

AMS Subject Headings:    Global Science Press

Copyright:    COPYRIGHT: © Global Science Press

Pages:    24

Keywords:    Multipliers correction proximal approximation orthogonality constraint Stiefel manifold.

Author Details

Lei Wang

Bin Gao

Xin Liu

  1. Seeking Consensus on Subspaces in Federated Principal Component Analysis

    Wang, Lei | Liu, Xin | Zhang, Yin

    Journal of Optimization Theory and Applications, Vol. 203 (2024), Iss. 1 P.529

    https://doi.org/10.1007/s10957-024-02523-1 [Citations: 0]
  2. A communication-efficient and privacy-aware distributed algorithm for sparse PCA

    Wang, Lei | Liu, Xin | Zhang, Yin

    Computational Optimization and Applications, Vol. 85 (2023), Iss. 3 P.1033

    https://doi.org/10.1007/s10589-023-00481-4 [Citations: 3]
  3. Decentralized Optimization Over the Stiefel Manifold by an Approximate Augmented Lagrangian Function

    Wang, Lei | Liu, Xin

    IEEE Transactions on Signal Processing, Vol. 70 (2022), Iss. P.3029

    https://doi.org/10.1109/TSP.2022.3182883 [Citations: 7]
  4. Smoothing algorithms for nonsmooth optimization over the Stiefel manifold with applications to the graph Fourier basis problem

    Zhu, Jinlai | Huang, Jianfeng | Yang, Lihua | Li, Qia

    Analysis and Applications, Vol. 22 (2024), Iss. 05 P.937

    https://doi.org/10.1142/S0219530524500088 [Citations: 1]
  5. Stochastic Gauss–Newton Algorithms for Online PCA

    Zhou, Siyun | Liu, Xin | Xu, Liwei

    Journal of Scientific Computing, Vol. 96 (2023), Iss. 3

    https://doi.org/10.1007/s10915-023-02289-0 [Citations: 0]
  6. Smoothing Gradient Tracking for Decentralized Optimization over the Stiefel Manifold with Non-Smooth Regularizers

    Wang, Lei | Liu, Xin

    2023 62nd IEEE Conference on Decision and Control (CDC), (2023), P.126

    https://doi.org/10.1109/CDC49753.2023.10383237 [Citations: 1]
  7. A Theory of the NEPv Approach for Optimization on the Stiefel Manifold

    Li, Ren-Cang

    Foundations of Computational Mathematics, Vol. (2024), Iss.

    https://doi.org/10.1007/s10208-024-09687-2 [Citations: 0]
  8. Maximizing sum of coupled traces with applications

    Wang, Li | Zhang, Lei-Hong | Li, Ren-Cang

    Numerische Mathematik, Vol. 152 (2022), Iss. 3 P.587

    https://doi.org/10.1007/s00211-022-01322-y [Citations: 3]
  9. Dissolving Constraints for Riemannian Optimization

    Xiao, Nachuan | Liu, Xin | Toh, Kim-Chuan

    Mathematics of Operations Research, Vol. 49 (2024), Iss. 1 P.366

    https://doi.org/10.1287/moor.2023.1360 [Citations: 2]