10.03.2015 Views

v2009.01.01 - Convex Optimization

v2009.01.01 - Convex Optimization

v2009.01.01 - Convex Optimization

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

446 CHAPTER 6. CONE OF DISTANCE MATRICES<br />

a resemblance to EDM definition (798) where<br />

S N h<br />

∆<br />

= { A ∈ S N | δ(A) = 0 } (59)<br />

is the symmetric hollow subspace (2.2.3) and where<br />

S N⊥<br />

c = {u1 T + 1u T | u∈ R N } (1876)<br />

is the orthogonal complement of the geometric center subspace (E.7.2.0.2)<br />

S N c<br />

∆<br />

= {Y ∈ S N | Y 1 = 0} (1874)<br />

6.0.1 gravity<br />

Equality (1173) is equally important as the known isomorphisms (912) (913)<br />

(924) (925) relating the EDM cone EDM N to positive semidefinite cone S N−1<br />

+<br />

(5.6.2.1) or to an N(N −1)/2-dimensional face of S N + (5.6.1.1). 6.1 But<br />

those isomorphisms have never led to this equality relating whole cones<br />

EDM N and S N + .<br />

Equality (1173) is not obvious from the various EDM definitions such as<br />

(798) or (1099) because inclusion must be proved algebraically in order to<br />

establish equality; EDM N ⊇ S N h ∩ (S N⊥<br />

c − S N +). We will instead prove (1173)<br />

using purely geometric methods.<br />

6.0.2 highlight<br />

In6.8.1.7 we show: the Schoenberg criterion for discriminating Euclidean<br />

distance matrices<br />

D ∈ EDM N<br />

⇔<br />

{<br />

−V<br />

T<br />

N DV N ∈ S N−1<br />

+<br />

D ∈ S N h<br />

(817)<br />

is a discretized membership relation (2.13.4, dual generalized inequalities)<br />

between the EDM cone and its ordinary dual.<br />

6.1 Because both positive semidefinite cones are frequently in play, dimension is explicit.

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

Saved successfully!

Ooh no, something went wrong!