Cholesky QR-based retraction on the generalized Stiefel manifold
When optimizing on a Riemannian manifold, it is important to use an efficient retraction, which maps a point on a tangent space to a point on the manifold. In this paper, we prove a map based on the QR factorization to be a retraction on the generalized Stiefel manifold. In addition, we propose an efficient implementation of the retraction based on the Cholesky QR factorization. Numerical experiments show that the proposed retraction is more efficient than the existing one based on the polar factorization.
KeywordsRiemannian optimization Generalized Stiefel manifold Retraction Cholesky QR factorization
Mathematics Subject Classification90C30 65K05 65F30
The authors would like to thank the editor and the reviewer for their careful reading and constructive comments, especially on Theorem 3.2. The authors would also like to thank Dr. Akira Imakura (University of Tsukuba) and Dr. Yusaku Yamamoto (The University of Electro-Communications) for their helpful advice. This study was supported in part by Grant Numbers JP16K17647 and JP18K18064 from the Grants-in-Aid for Scientific Research Program (KAKENHI) of the Japan Society for the Promotion of Science (JSPS).
- 6.Boumal, N., Absil, P.A., Cartis, C.: Global rates of convergence for nonconvex optimization on manifolds. IMA J. Numer. Anal. https://doi.org/10.1093/imanum/drx080 (2018)
- 9.Imakura, A., Yamamoto, Y.: Efficient implementations of the modified Gram–Schmidt orthogonalization with a non-standard inner product. arXiv preprint arXiv:1703.10440 (2017)
- 10.Kasai, H., Sato, H., Mishra, B.: Riemannian stochastic recursive gradient algorithm with retraction and vector transport and its convergence analysis. In: International Conference on Machine Learning, pp. 2521–2529 (2018)Google Scholar
- 11.Lowery, B.R., Langou, J.: Stability analysis of QR factorization in an oblique inner product. arXiv preprint arXiv:1401.5171 (2014)
- 19.Yger, F., Berar, M., Gasso, G., Rakotomamonjy, A.: Adaptive canonical correlation analysis based on matrix manifolds. In: Proceedings of the 29th International Conference on Machine Learning, pp. 299–306 (2012)Google Scholar
- 20.Yger, F., Berar, M., Gasso, G., Rakotomamonjy, A.: Oblique principal subspace tracking on manifold. In: Proceedings of the 2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 2429–2432. IEEE (2012)Google Scholar