v2010.10.26 - Convex Optimization
v2010.10.26 - Convex Optimization v2010.10.26 - Convex Optimization
314 CHAPTER 4. SEMIDEFINITE PROGRAMMING4.4.1.2.2 Exercise. Completely positive semidefinite ⎡ matrix. ⎤ [40]0.50 0.55 0.20Given rank-2 positive semidefinite matrix G = ⎣ 0.55 0.61 0.22 ⎦, find a0.20 0.22 0.08positive factorization G = X T X (900) by solvingfind X ≥ 0X∈R 2×3 [ I Xsubject to Z =X T GrankZ ≤ 2]≽ 0(740)via convex iteration.4.4.1.2.3 Exercise. Nonnegative matrix ⎡ factorization. ⎤17 28 42Given rank-2 nonnegative matrix X = ⎣ 16 47 51 ⎦, find a nonnegative17 82 72factorizationX = WH (741)by solvingfindA∈S 3 , B∈S 3 , W ∈R 3×2 , H∈R 2×3W , H⎡subject to Z = ⎣W ≥ 0H ≥ 0rankZ ≤ 2I W T HW A XH T X T Bwhich follows from the fact, at optimality,⎡ ⎤IZ ⋆ = ⎣ W ⎦ [I W T H ]H T⎤⎦≽ 0(742)(743)Use the known closed-form solution for a direction vector Y to regulate rankby convex iteration; set Z ⋆ = QΛQ T ∈ S 8 to an ordered diagonalization andU ⋆ = Q(:, 3:8)∈ R 8×6 , then Y = U ⋆ U ⋆T (4.4.1.1).
4.4. RANK-CONSTRAINED SEMIDEFINITE PROGRAM 315In summary, initialize Y then iterate numerical solution of (convex)semidefinite programminimize 〈Z , Y 〉A∈S 3 , B∈S 3 , W ∈R 3×2 , H∈R 2×3 ⎡ ⎤I W T Hsubject to Z = ⎣ W A X ⎦≽ 0H T X T BW ≥ 0H ≥ 0(744)with Y = U ⋆ U ⋆T until convergence (which is global and occurs in very fewiterations for this instance).Now, an application to optimal regulation of affine dimension:4.4.1.2.4 Example. Sensor-Network Localization or Wireless Location.Heuristic solution to a sensor-network localization problem, proposedby Carter, Jin, Saunders, & Ye in [72], 4.26 is limited to two Euclideandimensions and applies semidefinite programming (SDP) to littlesubproblems. There, a large network is partitioned into smaller subnetworks(as small as one sensor − a mobile point, whereabouts unknown) and thensemidefinite programming and heuristics called spaseloc are applied tolocalize each and every partition by two-dimensional distance geometry.Their partitioning procedure is one-pass, yet termed iterative; a termapplicable only in so far as adjoining partitions can share localized sensorsand anchors (absolute sensor positions known a priori). But there isno iteration on the entire network, hence the term “iterative” is perhapsinappropriate. As partitions are selected based on “rule sets” (heuristics, notgeographics), they also term the partitioning adaptive. But no adaptation ofa partition actually occurs once it has been determined.One can reasonably argue that semidefinite programming methods areunnecessary for localization of small partitions of large sensor networks.[278] [86] In the past, these nonlinear localization problems were solvedalgebraically and computed by least squares solution to hyperbolic equations;4.26 The paper constitutes Jin’s dissertation for University of Toronto [216] although hername appears as second author. Ye’s authorship is honorary.
- Page 263 and 264: 3.7. CONVEX MATRIX-VALUED FUNCTION
- Page 265 and 266: 3.7. CONVEX MATRIX-VALUED FUNCTION
- Page 267 and 268: 3.7. CONVEX MATRIX-VALUED FUNCTION
- Page 269 and 270: 3.8. QUASICONVEX 269exponential alw
- Page 271 and 272: 3.9. SALIENT PROPERTIES 2713.8.0.0.
- Page 273 and 274: Chapter 4Semidefinite programmingPr
- Page 275 and 276: 4.1. CONIC PROBLEM 275(confer p.162
- Page 277 and 278: 4.1. CONIC PROBLEM 277PCsemidefinit
- Page 279 and 280: 4.1. CONIC PROBLEM 279is the affine
- Page 281 and 282: 4.1. CONIC PROBLEM 281faces of S 3
- Page 283 and 284: 4.1. CONIC PROBLEM 2834.1.2.3 Previ
- Page 285 and 286: 4.2. FRAMEWORK 285Semidefinite Fark
- Page 287 and 288: 4.2. FRAMEWORK 287On the other hand
- Page 289 and 290: 4.2. FRAMEWORK 2894.2.2.1 Dual prob
- Page 291 and 292: 4.2. FRAMEWORK 291For symmetric pos
- Page 293 and 294: 4.2. FRAMEWORK 293has norm ‖x ⋆
- Page 295 and 296: 4.2. FRAMEWORK 295minimize 1 TˆxX
- Page 297 and 298: 4.2. FRAMEWORK 297asminimize ‖ỹ
- Page 299 and 300: 4.3. RANK REDUCTION 2994.3 Rank red
- Page 301 and 302: 4.3. RANK REDUCTION 301A rank-reduc
- Page 303 and 304: 4.3. RANK REDUCTION 303(t ⋆ i)
- Page 305 and 306: 4.3. RANK REDUCTION 3054.3.3.0.1 Ex
- Page 307 and 308: 4.3. RANK REDUCTION 3074.3.3.0.2 Ex
- Page 309 and 310: 4.4. RANK-CONSTRAINED SEMIDEFINITE
- Page 311 and 312: 4.4. RANK-CONSTRAINED SEMIDEFINITE
- Page 313: 4.4. RANK-CONSTRAINED SEMIDEFINITE
- Page 317 and 318: 4.4. RANK-CONSTRAINED SEMIDEFINITE
- Page 319 and 320: 4.4. RANK-CONSTRAINED SEMIDEFINITE
- Page 321 and 322: 4.4. RANK-CONSTRAINED SEMIDEFINITE
- Page 323 and 324: 4.4. RANK-CONSTRAINED SEMIDEFINITE
- Page 325 and 326: 4.4. RANK-CONSTRAINED SEMIDEFINITE
- Page 327 and 328: 4.4. RANK-CONSTRAINED SEMIDEFINITE
- Page 329 and 330: 4.4. RANK-CONSTRAINED SEMIDEFINITE
- Page 331 and 332: 4.4. RANK-CONSTRAINED SEMIDEFINITE
- Page 333 and 334: 4.5. CONSTRAINING CARDINALITY 333mi
- Page 335 and 336: 4.5. CONSTRAINING CARDINALITY 3350R
- Page 337 and 338: 4.5. CONSTRAINING CARDINALITY 337it
- Page 339 and 340: 4.5. CONSTRAINING CARDINALITY 339m/
- Page 341 and 342: 4.5. CONSTRAINING CARDINALITY 341we
- Page 343 and 344: 4.5. CONSTRAINING CARDINALITY 343fl
- Page 345 and 346: 4.5. CONSTRAINING CARDINALITY 345We
- Page 347 and 348: 4.5. CONSTRAINING CARDINALITY 3474.
- Page 349 and 350: 4.5. CONSTRAINING CARDINALITY 349R
- Page 351 and 352: 4.5. CONSTRAINING CARDINALITY 351pe
- Page 353 and 354: 4.6. CARDINALITY AND RANK CONSTRAIN
- Page 355 and 356: 4.6. CARDINALITY AND RANK CONSTRAIN
- Page 357 and 358: 4.6. CARDINALITY AND RANK CONSTRAIN
- Page 359 and 360: 4.6. CARDINALITY AND RANK CONSTRAIN
- Page 361 and 362: 4.6. CARDINALITY AND RANK CONSTRAIN
- Page 363 and 364: 4.6. CARDINALITY AND RANK CONSTRAIN
314 CHAPTER 4. SEMIDEFINITE PROGRAMMING4.4.1.2.2 Exercise. Completely positive semidefinite ⎡ matrix. ⎤ [40]0.50 0.55 0.20Given rank-2 positive semidefinite matrix G = ⎣ 0.55 0.61 0.22 ⎦, find a0.20 0.22 0.08positive factorization G = X T X (900) by solvingfind X ≥ 0X∈R 2×3 [ I Xsubject to Z =X T GrankZ ≤ 2]≽ 0(740)via convex iteration.4.4.1.2.3 Exercise. Nonnegative matrix ⎡ factorization. ⎤17 28 42Given rank-2 nonnegative matrix X = ⎣ 16 47 51 ⎦, find a nonnegative17 82 72factorizationX = WH (741)by solvingfindA∈S 3 , B∈S 3 , W ∈R 3×2 , H∈R 2×3W , H⎡subject to Z = ⎣W ≥ 0H ≥ 0rankZ ≤ 2I W T HW A XH T X T Bwhich follows from the fact, at optimality,⎡ ⎤IZ ⋆ = ⎣ W ⎦ [I W T H ]H T⎤⎦≽ 0(742)(743)Use the known closed-form solution for a direction vector Y to regulate rankby convex iteration; set Z ⋆ = QΛQ T ∈ S 8 to an ordered diagonalization andU ⋆ = Q(:, 3:8)∈ R 8×6 , then Y = U ⋆ U ⋆T (4.4.1.1).