v2009.01.01 - Convex Optimization
v2009.01.01 - Convex Optimization v2009.01.01 - Convex Optimization
410 CHAPTER 5. EUCLIDEAN DISTANCE MATRIX In terms of V N , the unit simplex (265) in R N−1 has an equivalent representation: S = {s ∈ R N−1 | √ 2V N s ≽ −e 1 } (987) where e 1 is as in (814). Incidental to the EDM assertion, shifting the unit-simplex domain in R N−1 translates the polyhedron P in R n . Indeed, there is a map from vertices of the unit simplex to members of the list generating P ; p : R N−1 ⎛⎧ ⎪⎨ p ⎜ ⎝ ⎪⎩ −β e 1 − β e 2 − β . e N−1 − β ⎫⎞ ⎪⎬ ⎟ ⎠ ⎪⎭ → = ⎧ ⎪⎨ ⎪⎩ R n x 1 − α x 2 − α x 3 − α . x N − α ⎫ ⎪⎬ ⎪⎭ (988) 5.9.1.0.5 Proof. EDM assertion. (⇒) We demonstrate that if D is an EDM, then each distance-square ‖p(y)‖ 2 described by (985) corresponds to a point p in some embedded polyhedron P − α . Assume D is indeed an EDM; id est, D can be made from some list X of N unknown points in Euclidean space R n ; D = D(X) for X ∈ R n×N as in (798). Since D is translation invariant (5.5.1), we may shift the affine hull A of those unknown points to the origin as in (928). Then take any point p in their convex hull (78); P − α = {p = (X − Xb1 T )a | a T 1 = 1, a ≽ 0} (989) where α = Xb ∈ A ⇔ b T 1 = 1. Solutions to a T 1 = 1 are: 5.43 a ∈ {e 1 + √ } 2V N s | s ∈ R N−1 (990) where e 1 is as in (814). Similarly, b = e 1 + √ 2V N β . P − α = {p = X(I − (e 1 + √ 2V N β)1 T )(e 1 + √ 2V N s) | √ 2V N s ≽ −e 1 } = {p = X √ 2V N (s − β) | √ 2V N s ≽ −e 1 } (991) 5.43 Since R(V N )= N(1 T ) and N(1 T )⊥ R(1) , then over all s∈ R N−1 , V N s is a hyperplane through the origin orthogonal to 1. Thus the solutions {a} constitute a hyperplane orthogonal to the vector 1, and offset from the origin in R N by any particular solution; in this case, a = e 1 .
5.9. BRIDGE: CONVEX POLYHEDRA TO EDMS 411 that describes the domain of p(s) as the unit simplex Making the substitution s − β ← y S = {s | √ 2V N s ≽ −e 1 } ⊂ R N−1 + (987) P − α = {p = X √ 2V N y | y ∈ S − β} (992) Point p belongs to a convex polyhedron P−α embedded in an r-dimensional subspace of R n because the convex hull of any list forms a polyhedron, and because the translated affine hull A − α contains the translated polyhedron P − α (931) and the origin (when α∈A), and because A has dimension r by definition (933). Now, any distance-square from the origin to the polyhedron P − α can be formulated {p T p = ‖p‖ 2 = 2y T V T NX T XV N y | y ∈ S − β} (993) Applying (940) to (993) we get (985). (⇐) To validate the EDM assertion in the reverse direction, we prove: If each distance-square ‖p(y)‖ 2 (985) on the shifted unit-simplex S −β ⊂ R N−1 corresponds to a point p(y) in some embedded polyhedron P − α , then D is an EDM. The r-dimensional subspace A − α ⊆ R n is spanned by p(S − β) = P − α (994) because A − α = aff(P − α) ⊇ P − α (931). So, outside the domain S − β of linear surjection p(y) , the simplex complement \S − β ⊂ R N−1 must contain the domain of the distance-square ‖p(y)‖ 2 = p(y) T p(y) to remaining points in the subspace A − α ; id est, to the polyhedron’s relative exterior \P − α . For ‖p(y)‖ 2 to be nonnegative on the entire subspace A − α , −V T N DV N must be positive semidefinite and is assumed symmetric; 5.44 −V T NDV N ∆ = Θ T pΘ p (995) where 5.45 Θ p ∈ R m×N−1 for some m ≥ r . Because p(S − β) is a convex polyhedron, it is necessarily a set of linear combinations of points from some 5.44 The antisymmetric part ( −V T N DV N − (−V T N DV N) T) /2 is annihilated by ‖p(y)‖ 2 . By the same reasoning, any positive (semi)definite matrix A is generally assumed symmetric because only the symmetric part (A +A T )/2 survives the test y T Ay ≥ 0. [176,7.1] 5.45 A T = A ≽ 0 ⇔ A = R T R for some real matrix R . [287,6.3]
- Page 359 and 360: 5.4. EDM DEFINITION 359 5.4.2.2.1 E
- Page 361 and 362: 5.4. EDM DEFINITION 361 ten affine
- Page 363 and 364: 5.4. EDM DEFINITION 363 spheres: Th
- Page 365 and 366: 5.4. EDM DEFINITION 365 By eliminat
- Page 367 and 368: 5.4. EDM DEFINITION 367 where Φ ij
- Page 369 and 370: 5.4. EDM DEFINITION 369 5.4.2.2.6 D
- Page 371 and 372: 5.4. EDM DEFINITION 371 10 5 ˇx 4
- Page 373 and 374: 5.4. EDM DEFINITION 373 corrected b
- Page 375 and 376: 5.4. EDM DEFINITION 375 by translat
- Page 377 and 378: 5.4. EDM DEFINITION 377 Crippen & H
- Page 379 and 380: 5.4. EDM DEFINITION 379 where ([√
- Page 381 and 382: 5.4. EDM DEFINITION 381 because (A.
- Page 383 and 384: 5.5. INVARIANCE 383 5.5.1.0.1 Examp
- Page 385 and 386: 5.5. INVARIANCE 385 x 2 x 2 x 3 x 1
- Page 387 and 388: 5.6. INJECTIVITY OF D & UNIQUE RECO
- Page 389 and 390: 5.6. INJECTIVITY OF D & UNIQUE RECO
- Page 391 and 392: 5.6. INJECTIVITY OF D & UNIQUE RECO
- Page 393 and 394: 5.7. EMBEDDING IN AFFINE HULL 393 5
- Page 395 and 396: 5.7. EMBEDDING IN AFFINE HULL 395 F
- Page 397 and 398: 5.7. EMBEDDING IN AFFINE HULL 397 5
- Page 399 and 400: 5.8. EUCLIDEAN METRIC VERSUS MATRIX
- Page 401 and 402: 5.8. EUCLIDEAN METRIC VERSUS MATRIX
- Page 403 and 404: 5.8. EUCLIDEAN METRIC VERSUS MATRIX
- Page 405 and 406: 5.8. EUCLIDEAN METRIC VERSUS MATRIX
- Page 407 and 408: 5.9. BRIDGE: CONVEX POLYHEDRA TO ED
- Page 409: 5.9. BRIDGE: CONVEX POLYHEDRA TO ED
- Page 413 and 414: 5.10. EDM-ENTRY COMPOSITION 413 of
- Page 415 and 416: 5.10. EDM-ENTRY COMPOSITION 415 The
- Page 417 and 418: 5.11. EDM INDEFINITENESS 417 5.11.1
- Page 419 and 420: 5.11. EDM INDEFINITENESS 419 we hav
- Page 421 and 422: 5.11. EDM INDEFINITENESS 421 So bec
- Page 423 and 424: 5.11. EDM INDEFINITENESS 423 where
- Page 425 and 426: 5.12. LIST RECONSTRUCTION 425 where
- Page 427 and 428: 5.12. LIST RECONSTRUCTION 427 (a) (
- Page 429 and 430: 5.13. RECONSTRUCTION EXAMPLES 429 D
- Page 431 and 432: 5.13. RECONSTRUCTION EXAMPLES 431 T
- Page 433 and 434: 5.13. RECONSTRUCTION EXAMPLES 433 w
- Page 435 and 436: 5.14. FIFTH PROPERTY OF EUCLIDEAN M
- Page 437 and 438: 5.14. FIFTH PROPERTY OF EUCLIDEAN M
- Page 439 and 440: 5.14. FIFTH PROPERTY OF EUCLIDEAN M
- Page 441 and 442: 5.14. FIFTH PROPERTY OF EUCLIDEAN M
- Page 443 and 444: 5.14. FIFTH PROPERTY OF EUCLIDEAN M
- Page 445 and 446: Chapter 6 Cone of distance matrices
- Page 447 and 448: 6.1. DEFINING EDM CONE 447 6.1 Defi
- Page 449 and 450: 6.2. POLYHEDRAL BOUNDS 449 This con
- Page 451 and 452: 6.3. √ EDM CONE IS NOT CONVEX 451
- Page 453 and 454: 6.4. A GEOMETRY OF COMPLETION 453 (
- Page 455 and 456: 6.4. A GEOMETRY OF COMPLETION 455 (
- Page 457 and 458: 6.4. A GEOMETRY OF COMPLETION 457 F
- Page 459 and 460: 6.5. EDM DEFINITION IN 11 T 459 by
410 CHAPTER 5. EUCLIDEAN DISTANCE MATRIX<br />
In terms of V N , the unit simplex (265) in R N−1 has an equivalent<br />
representation:<br />
S = {s ∈ R N−1 | √ 2V N s ≽ −e 1 } (987)<br />
where e 1 is as in (814). Incidental to the EDM assertion, shifting the<br />
unit-simplex domain in R N−1 translates the polyhedron P in R n . Indeed,<br />
there is a map from vertices of the unit simplex to members of the list<br />
generating P ;<br />
p : R N−1<br />
⎛⎧<br />
⎪⎨<br />
p<br />
⎜<br />
⎝<br />
⎪⎩<br />
−β<br />
e 1 − β<br />
e 2 − β<br />
.<br />
e N−1 − β<br />
⎫⎞<br />
⎪⎬<br />
⎟<br />
⎠<br />
⎪⎭<br />
→<br />
=<br />
⎧<br />
⎪⎨<br />
⎪⎩<br />
R n<br />
x 1 − α<br />
x 2 − α<br />
x 3 − α<br />
.<br />
x N − α<br />
⎫<br />
⎪⎬<br />
⎪⎭<br />
(988)<br />
5.9.1.0.5 Proof. EDM assertion.<br />
(⇒) We demonstrate that if D is an EDM, then each distance-square ‖p(y)‖ 2<br />
described by (985) corresponds to a point p in some embedded polyhedron<br />
P − α . Assume D is indeed an EDM; id est, D can be made from some list<br />
X of N unknown points in Euclidean space R n ; D = D(X) for X ∈ R n×N<br />
as in (798). Since D is translation invariant (5.5.1), we may shift the affine<br />
hull A of those unknown points to the origin as in (928). Then take any<br />
point p in their convex hull (78);<br />
P − α = {p = (X − Xb1 T )a | a T 1 = 1, a ≽ 0} (989)<br />
where α = Xb ∈ A ⇔ b T 1 = 1. Solutions to a T 1 = 1 are: 5.43<br />
a ∈<br />
{e 1 + √ }<br />
2V N s | s ∈ R N−1<br />
(990)<br />
where e 1 is as in (814). Similarly, b = e 1 + √ 2V N β .<br />
P − α = {p = X(I − (e 1 + √ 2V N β)1 T )(e 1 + √ 2V N s) | √ 2V N s ≽ −e 1 }<br />
= {p = X √ 2V N (s − β) | √ 2V N s ≽ −e 1 } (991)<br />
5.43 Since R(V N )= N(1 T ) and N(1 T )⊥ R(1) , then over all s∈ R N−1 , V N s is a<br />
hyperplane through the origin orthogonal to 1. Thus the solutions {a} constitute a<br />
hyperplane orthogonal to the vector 1, and offset from the origin in R N by any particular<br />
solution; in this case, a = e 1 .