Abstract
We introduce and analyze a natural geometric version of Renegar's condition number R for the homogeneous convex feasibility problem associated with a regular cone C ⊆ ℝ n. Let Gr n,m denote the Grassmann manifold of m-dimensional linear subspaces of ℝ n, and consider the projection distance d p(W 1,W 2) := ∥ΠW 1 - ΠW 2 ∥ (spectral norm) between W 1,W 2 ∈ Gr n,m, where ΠW i denotes the orthogonal projection onto W i. We call ℒ (W) := max{d p(W,W′) -1 | W′ ∈ Σ m} the Grassmann condition number of W ∈ Gr n,m, where the set of ill-posed instances Σ m ⊂ Gr n,m is defined as the set of linear subspaces touching C. We show that if W = im(A T ) for a matrix A ∈ ℝ m×n, then ℒ (W) ≤ R(A) ≤ ℒ (W) κ(A), where κ(A) = ∥A∥∥A† ∥ denotes the matrix condition number. This extends work by Belloni and Freund in [Math. Program. Ser. A, 119 (2009), pp. 95-107]. Furthermore, we show that ℒ (W) can also be characterized in terms of the Riemannian distance metric on Gr n,m. This differential geometric characterization of ℒ (W) is the starting point of the sequel [D. Amelunxen and P. Bürgisser, Probabilistic analysis of the Grassman condition number, preprint, http:arxiv.org/abs/1112.2603, 2011] to this paper, where the first probabilistic analysis of Renegar's condition number for an arbitrary regular cone C is achieved. © 2012 Society for Industrial and Applied Mathematics.
| Original language | English |
|---|---|
| Pages (from-to) | 1029-1041 |
| Journal | SIAM Journal on Optimization |
| Volume | 22 |
| Issue number | 3 |
| DOIs | |
| Publication status | Published - 2012 |
| Externally published | Yes |
Research Keywords
- Condition number
- Convex programming
- Perturbation
Fingerprint
Dive into the research topics of 'A coordinate-free condition number for convex programming'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver