10.03.2015 Views

v2009.01.01 - Convex Optimization

v2009.01.01 - Convex Optimization

v2009.01.01 - Convex Optimization

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

508 CHAPTER 7. PROXIMITY PROBLEMS<br />

projection of an ordered vector of eigenvalues (in diagonal matrix Λ) on a<br />

subset of the monotone nonnegative cone (2.13.9.4.2)<br />

K M+ = {υ | υ 1 ≥ υ 2 ≥ · · · ≥ υ N−1 ≥ 0} ⊆ R N−1<br />

+ (383)<br />

Of interest, momentarily, is only the smallest convex subset of the<br />

monotone nonnegative cone K M+ containing every nonincreasingly ordered<br />

eigenspectrum corresponding to a rank ρ subset of the positive semidefinite<br />

cone S N−1<br />

+ ; id est,<br />

K ρ M+<br />

∆<br />

= {υ ∈ R ρ | υ 1 ≥ υ 2 ≥ · · · ≥ υ ρ ≥ 0} ⊆ R ρ + (1231)<br />

a pointed polyhedral cone, a ρ-dimensional convex subset of the<br />

monotone nonnegative cone K M+ ⊆ R N−1<br />

+ having property, for λ denoting<br />

eigenspectra,<br />

[<br />

ρ<br />

] K<br />

M+<br />

= π(λ(rank ρ subset)) ⊆ K N−1 ∆<br />

0<br />

M+<br />

= K M+ (1232)<br />

For each and every elemental eigenspectrum<br />

γ ∈ λ(rank ρ subset)⊆ R N−1<br />

+ (1233)<br />

of the rank ρ subset (ordered or unordered in λ), there is a nonlinear<br />

surjection π(γ) onto K ρ M+ .<br />

7.1.3.0.2 Exercise. Smallest spectral cone.<br />

Prove that there is no convex subset of K M+ smaller than K ρ M+ containing<br />

every ordered eigenspectrum corresponding to the rank ρ subset of a positive<br />

semidefinite cone (2.9.2.1).<br />

<br />

7.1.3.0.3 Proposition. (Hardy-Littlewood-Pólya) Inequalities. [153,X]<br />

[48,1.2] Any vectors σ and γ in R N−1 satisfy a tight inequality<br />

π(σ) T π(γ) ≥ σ T γ ≥ π(σ) T Ξπ(γ) (1234)<br />

where Ξ is the order-reversing permutation matrix defined in (1608),<br />

and permutator π(γ) is a nonlinear function that sorts vector γ into<br />

nonincreasing order thereby providing the greatest upper bound and least<br />

lower bound with respect to every possible sorting.<br />

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!