12.07.2015 Views

v2007.11.26 - Convex Optimization

v2007.11.26 - Convex Optimization

v2007.11.26 - Convex Optimization

SHOW MORE
SHOW LESS
  • No tags were found...

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

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

6.6. CORRESPONDENCE TO PSD CONE S N−1+ 4256.6 Correspondence to PSD cone S N−1+Hayden & Wells et alii [134,2] assert one-to-one correspondence of EDMswith positive semidefinite matrices in the symmetric subspace. Becauserank(V DV )≤N−1 (5.7.1.1), that positive semidefinite cone correspondingto the EDM cone can only be S N−1+ . [6,18.2.1] To clearly demonstrate thiscorrespondence, we invoke inner-product form EDM definition[ ]D(Φ) =∆ 01δ(Φ)T + 1 [ 0 δ(Φ) ] [ ] 0 0 T T− 2∈ EDM N0 Φ⇔(851)Φ ≽ 0Then the EDM cone may be expressedEDM N = { D(Φ) | Φ ∈ S N−1+}(1049)Hayden & Wells’ assertion can therefore be equivalently stated in terms ofan inner-product form EDM operatorD(S N−1+ ) = EDM N (853)V N (EDM N ) = S N−1+ (854)identity (854) holding because R(V N )= N(1 T ) (741), linear functions D(Φ)and V N (D)= −VN TDV N (5.6.2.1) being mutually inverse.In terms of affine dimension r , Hayden & Wells claim particularcorrespondence between PSD and EDM cones:r = N −1: Symmetric hollow matrices −D positive definite on N(1 T ) correspondto points relatively interior to the EDM cone.r < N −1: Symmetric hollow matrices −D positive semidefinite on N(1 T ) , where−VN TDV N has at least one 0 eigenvalue, correspond to points on therelative boundary of the EDM cone.r = 1: Symmetric hollow nonnegative matrices rank-one on N(1 T ) correspondto extreme directions (1041) of the EDM cone; id est, for some nonzerovector u (A.3.1.0.7)rankV T N DV N =1D ∈ S N h ∩ R N×N+}⇔{D ∈ EDM N−VTD is an extreme direction ⇔ N DV N ≡ uu TD ∈ S N h(1050)

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

Saved successfully!

Ooh no, something went wrong!