Angles between subspaces and their tangents
arXiv:1209.0523 · doi:10.1515/jnum-2013-0013
Abstract
Principal angles between subspaces (PABS) (also called canonical angles) serve as a classical tool in mathematics, statistics, and applications, e.g., data mining. Traditionally, PABS are introduced via their cosines. The cosines and sines of PABS are commonly defined using the singular value decomposition. We utilize the same idea for the tangents, i.e., explicitly construct matrices, such that their singular values are equal to the tangents of PABS, using several approaches: orthonormal and non-orthonormal bases for subspaces, as well as projectors. Such a construction has applications, e.g., in analysis of convergence of subspace iterations for eigenvalue problems.
arXiv comment15 pages, 1 figure, 2 tables. Accepted to Journal of Numerical Mathematics
References in corpus (3)
Cited by in corpus (19)
- Unperturbed: spectral analysis beyond Davis-Kahan
- Multi-step Online Unsupervised Domain Adaptation
- A Theoretical Analysis of Catastrophic Forgetting through the NTK Overlap Matrix
- GraN-GAN: Piecewise Gradient Normalization for Generative Adversarial Networks
- Recovering sparse networks: Basis adaptation and stability under extensions
- Rayleigh-Ritz majorization error bounds of the mixed type
- Disentangling by Subspace Diffusion
- Guaranteed Simultaneous Asymmetric Tensor Decomposition via Orthogonalized Alternating Least Squares
- Bounds for fidelity of semiclassical Lagrangian states in K{ä}hler quantization
- Noisy Accelerated Power Method for Eigenproblems with Applications
- Angular values of nonautonomous and random linear dynamical systems: Part I -- Fundamentals
- Bias-Variance Tradeoffs in Joint Spectral Embeddings
- Optimal Convergence Rate of Self-Consistent Field Iteration for Solving Eigenvector-dependent Nonlinear Eigenvalue Problems
- Twice is enough for dangerous eigenvalues
- Random Differential Topology
- Recommendation on a Budget: Column Space Recovery from Partially Observed Entries with Random or Active Sampling
- Conceptually Diverse Base Model Selection for Meta-Learners in Concept Drifting Data Streams
- Strong Fast Invertibility and Lyapunov Exponents for Linear Systems
- Provable Data Clustering via Innovation Search