v2007.09.13 - Convex Optimization

v2007.09.13 - Convex Optimization v2007.09.13 - Convex Optimization

convexoptimization.com
from convexoptimization.com More from this publisher
10.07.2015 Views

358 CHAPTER 5. EUCLIDEAN DISTANCE MATRIXThese preliminary findings lead one to speculate whether any concavenondecreasing composition of individual EDM entries d ij on R + will produceanother EDM; e.g., empirical evidence suggests that given EDM D , for eachfixed α ≥ 1 [sic] the composition [log 2 (1 + d 1/αij )] is also an EDM. Figure 89illustrates that composition’s concavity in d ij together with functions from(895) and (896).5.10.1 EDM by elliptopeFor some κ∈ R + and C ∈ S N + in the elliptope E N (5.9.1.0.1), Alfakih assertsany given EDM D is expressible [7] [77,31.5]D = κ(11 T − C) ∈ EDM N (899)This expression exhibits nonlinear combination of variables κ and C . Wetherefore propose a different expression requiring redefinition of the elliptope(880) by parametrization;E n t∆= S n + ∩ {Φ∈ S n | δ(Φ)=t1} (900)where, of course, E n = E n 1 . Then any given EDM D is expressiblewhich is linear in variables t∈ R + and E∈ E N t .D = t11 T − E ∈ EDM N (901)5.11 EDM indefinitenessBy the known result inA.7.2 regarding a 0-valued entry on the maindiagonal of a symmetric positive semidefinite matrix, there can be no positivenor negative semidefinite EDM except the 0 matrix because EDM N ⊆ S N h(704) andS N h ∩ S N + = 0 (902)the origin. So when D ∈ EDM N , there can be no factorization D =A T Anor −D =A T A . [247,6.3] Hence eigenvalues of an EDM are neither allnonnegative or all nonpositive; an EDM is indefinite and possibly invertible.

5.11. EDM INDEFINITENESS 3595.11.1 EDM eigenvalues, congruence transformationFor any symmetric −D , we can characterize its eigenvalues by congruencetransformation: [247,6.3][ ]VT−W T NDW = − D [ V N1 T1 ] = −[VTN DV N V T N D11 T DV N 1 T D1]∈ S N (903)BecauseW ∆ = [V N 1 ] ∈ R N×N (904)is full-rank, then (1312)inertia(−D) = inertia ( −W T DW ) (905)the congruence (903) has the same number of positive, zero, and negativeeigenvalues as −D . Further, if we denote by {γ i , i=1... N −1} theeigenvalues of −VN TDV N and denote eigenvalues of the congruence −W T DWby {ζ i , i=1... N} and if we arrange each respective set of eigenvalues innonincreasing order, then by theory of interlacing eigenvalues for borderedsymmetric matrices [149,4.3] [247,6.4] [244,IV.4.1]ζ N ≤ γ N−1 ≤ ζ N−1 ≤ γ N−2 ≤ · · · ≤ γ 2 ≤ ζ 2 ≤ γ 1 ≤ ζ 1 (906)When D ∈ EDM N , then γ i ≥ 0 ∀i (1249) because −V T N DV N ≽0 as weknow. That means the congruence must have N −1 nonnegative eigenvalues;ζ i ≥ 0, i=1... N −1. The remaining eigenvalue ζ N cannot be nonnegativebecause then −D would be positive semidefinite, an impossibility; so ζ N < 0.By congruence, nontrivial −D must therefore have exactly one negativeeigenvalue; 5.45 [77,2.4.5]5.45 All the entries of the corresponding eigenvector must have the same sign with respectto each other [61, p.116] because that eigenvector is the Perron vector corresponding tothe spectral radius; [149,8.2.6] the predominant characteristic of negative [sic] matrices.

358 CHAPTER 5. EUCLIDEAN DISTANCE MATRIXThese preliminary findings lead one to speculate whether any concavenondecreasing composition of individual EDM entries d ij on R + will produceanother EDM; e.g., empirical evidence suggests that given EDM D , for eachfixed α ≥ 1 [sic] the composition [log 2 (1 + d 1/αij )] is also an EDM. Figure 89illustrates that composition’s concavity in d ij together with functions from(895) and (896).5.10.1 EDM by elliptopeFor some κ∈ R + and C ∈ S N + in the elliptope E N (5.9.1.0.1), Alfakih assertsany given EDM D is expressible [7] [77,31.5]D = κ(11 T − C) ∈ EDM N (899)This expression exhibits nonlinear combination of variables κ and C . Wetherefore propose a different expression requiring redefinition of the elliptope(880) by parametrization;E n t∆= S n + ∩ {Φ∈ S n | δ(Φ)=t1} (900)where, of course, E n = E n 1 . Then any given EDM D is expressiblewhich is linear in variables t∈ R + and E∈ E N t .D = t11 T − E ∈ EDM N (901)5.11 EDM indefinitenessBy the known result inA.7.2 regarding a 0-valued entry on the maindiagonal of a symmetric positive semidefinite matrix, there can be no positivenor negative semidefinite EDM except the 0 matrix because EDM N ⊆ S N h(704) andS N h ∩ S N + = 0 (902)the origin. So when D ∈ EDM N , there can be no factorization D =A T Anor −D =A T A . [247,6.3] Hence eigenvalues of an EDM are neither allnonnegative or all nonpositive; an EDM is indefinite and possibly invertible.

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

Saved successfully!

Ooh no, something went wrong!