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.

54 CHAPTER 2. CONVEX GEOMETRY<br />

Any matrix A∈ R M×M can be written as the sum of its symmetric hollow<br />

and antisymmetric antihollow parts: respectively,<br />

( ) ( 1 1<br />

A =<br />

2 (A +AT ) − δ 2 (A) +<br />

2 (A −AT ) + δ (A))<br />

2 (62)<br />

The symmetric hollow part is orthogonal in R M2 to the antisymmetric<br />

antihollow part; videlicet,<br />

))<br />

1 1<br />

tr((<br />

2 (A +AT ) − δ (A))( 2 2 (A −AT ) + δ 2 (A) = 0 (63)<br />

In the ambient space of real matrices, the antisymmetric antihollow subspace<br />

is described<br />

{ }<br />

∆ 1<br />

=<br />

2 (A −AT ) + δ 2 (A) | A∈ R M×M ⊆ R M×M (64)<br />

S M⊥<br />

h<br />

because any matrix in S M h is orthogonal to any matrix in S M⊥<br />

h . Yet in<br />

the ambient space of symmetric matrices S M , the antihollow subspace is<br />

nontrivial;<br />

S M⊥<br />

h<br />

∆<br />

= { δ 2 (A) | A∈ S M} = { δ(u) | u∈ R M} ⊆ S M (65)<br />

In anticipation of their utility with Euclidean distance matrices (EDMs)<br />

in5, for symmetric hollow matrices we introduce the linear bijective<br />

vectorization dvec that is the natural analogue to symmetric matrix<br />

vectorization svec (49): for Y = [Y ij ]∈ S M h<br />

⎡<br />

dvecY = ∆ √ 2<br />

⎢<br />

⎣<br />

⎤<br />

Y 12<br />

Y 13<br />

Y 23<br />

Y 14<br />

Y 24<br />

Y 34 ⎥<br />

.<br />

⎦<br />

Y M−1,M<br />

∈ R M(M−1)/2 (66)<br />

Like svec (49), dvec is an isometric isomorphism on the symmetric hollow<br />

subspace.

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

Saved successfully!

Ooh no, something went wrong!