Nonlinear matrix recovery using optimization on the Grassmann manifold
2023 (English)In: Applied and Computational Harmonic Analysis, ISSN 1063-5203, E-ISSN 1096-603X, Vol. 62, p. 498-542Article in journal (Refereed) Published
Abstract [en]
We investigate the problem of recovering a partially observed high-rank matrix whose columns obey a nonlinear structure such as a union of subspaces, an algebraic variety or grouped in clusters. The recovery problem is formulated as the rank minimization of a nonlinear feature map applied to the original matrix, which is then further approximated by a constrained non-convex optimization problem involving the Grassmann manifold. We propose two sets of algorithms, one arising from Riemannian optimization and the other as an alternating minimization scheme, both of which include first- and second-order variants. Both sets of algorithms have theoretical guarantees. In particular, for the alternating minimization, we establish global convergence and worst-case complexity bounds. Additionally, using the Kurdyka-Lojasiewicz property, we show that the alternating minimization converges to a unique limit point. We provide extensive numerical results for the recovery of union of subspaces and clustering under entry sampling and dense Gaussian sampling. Our methods are competitive with existing approaches and, in particular, high accuracy is achieved in the recovery using Riemannian second-order methods.
Place, publisher, year, edition, pages
Elsevier, 2023. Vol. 62, p. 498-542
Keywords [en]
Nonlinear matrix recovery, Nonconvex optimization, Riemannian optimization, Second-order methods
National Category
Computational Mathematics Mathematical Analysis
Identifiers
URN: urn:nbn:se:umu:diva-202159DOI: 10.1016/j.acha.2022.11.001Scopus ID: 2-s2.0-85142157294OAI: oai:DiVA.org:umu-202159DiVA, id: diva2:1723635
Note
Errata: Florentin Goyens, Coralia Cartis, Armin Eftekhari, Corrigendum to "Nonlinear matrix recovery using optimization on the Grassmann manifold" [Appl. Comput. Harmon. Anal. 62 (2023) 498–542], Applied and Computational Harmonic Analysis, Volume 63, 2023, Page 93, ISSN 1063-5203, https://doi.org/10.1016/j.acha.2022.12.004
2023-01-032023-01-032023-01-03Bibliographically approved