documentation by:
Jiachen Jin
National University of Defense Technology
2026
This package contains the code used in the paper "Adaptive Riemannian ADMM for Nonsmooth Optimization: Optimal Complexity without Smoothing".
This code has been tested to run in MATLAB R2023b.
If you use this code in an academic paper, please cite our paper:
Kangkang Deng, Jiachen Jin, Jiang Hu, and Hongxia Wang. Adaptive Riemannian ADMM for Nonsmooth Optimization: Optimal Complexity without Smoothing. In 39th Conference on Neural Information Processing Systems (NeurIPS 2025).
- spca_demo_ADMM : compares 4 Riemannian ADMM algorithms SOC, MADMM, RADMM and OADMM for solving spare PCA problem
- dpcp_demo_ADMM : compares 3 Riemannian ADMM algorithms SOC, MADMM and RADMM for solving Dual Principal Component Pursuit
If you have any questions or bugs, feel free to contact Jiachen Jin jinjiachen@126.com and Kangkang Deng freedeng1208@gmail.com.