A FIRST COURSE IN OPTIMIZATION THEORY RANGARAJAN K. SUNDARAM New York University CAMBRIDGE (NIVERSITY PRESSPreface Acknowledgements Contents 1 Mathematical Preliminaries 1.1 Notation and Preliminary Definitions 1 b&b LLL 1.1.2 Integers, Rationals, Reals, R” Inner Product, Norm, Metric Sets and Sequences in R" 1.2.1 Sequences and Limits Subsequences and Limit Points Cauchy Sequences and Completeness Suprema, Infima, Maxima, Minima Monotone Sequences in R The Lim Sup and Lim Inf 1.2.7 Open Balls, Open Sets, Closed Sets 1.2.8 Bounded Sets and Compact Sets 1,2.9 Convex Combinations and Convex Sets 1.2.10 Unions, Intersections, and Other Binary Operations Matrices 1.3.1 Sum, Product, Transpose 1.3.2. -Some Important Classes of Matrices 1.3.3 Rank of a Matrix 13.4 The Determinant 1.3.5 The Inverse 1.3.6 Calculating the Determinant Functions 14.1 Continuous Functions 1.4.2 _ Differentiable and Continuously Differentiable Functions vii page xiii xvii yanne i 14 7 18 22 23 23 24 30 30 32 33 35 38 39 41 41 43viii Contents w 1.4.3. Partial Derivatives and Differentiability 1.4.4. Directional Derivatives and Differentiability 1.4.5. Higher Order Derivatives 1.5. Quadratic Forms: Definite and Semidefinite Matrices 1.5.1 Quadratic Forms and Definiteness 1.5.2. Identifying Definiteness and Semidefiniteness 1.6 Some Important Results 1.6.1 Separation Theorems 1.6.2 The Intermediate and Mean Value Theorems 1.6.3 The Inverse and Implicit Function Theorems 1.7 Exercises Optimization in R" 2.1. Optimization Problems in R" 2.2. Optimization Problems in Parametric Form 2.3 Optimization Problems: Some Examples 2.3.1 Utility Maximization 2.3.2. Expenditure Minimization 2.3.3. Profit Maximization 2.3.4. Cost Minimization 2.3.5 Consumption-Leisure Choice 2.3.6 Portfolio Choice 2.3.7 Identifying Pareto Optima 2,3.8 Optimal Provision of Public Goods 23.9 Optimal Commodity Taxation 2.4- Objectives of Optimization Theory 2.5 A Roadmap 2.6. Exercises Existence of Solutions: The Weierstrass Theorem 3.1 The Weierstrass Theorem 3.2 The Weierstrass Theorem in Applications 3.3 A Proof of the Weierstrass Theorem 3.4. Exercises Unconstrained Optima 4.1 “Unconstrained” Optima 4.2 First-Order Conditions 4.3 Second-Order Conditions 44 Using the First- and Second-Order Conditionsw a 45 46 4.7 Contents A Proof of the First-Order Conditions A Proof of the Second-Order Conditions Exercises Equality Constraints and the Theorem of Lagrange 5.1 5.2 53 54 55 5.6 57 5.8 Constrained Optimization Problems Equality Constraints and the Theorem of Lagrange 5.2.1 Statement of the Theorem 5.2.2. The Constraint Qualification 5.2.3 The Lagrangean Multipliers Second-Order Conditions Using the Theorem of Lagrange 5.4.1 A “Cookbook” Procedure 5.4.2. Why the Procedure Usually Works 5.4.3 When It Could Fail 5.4.4 A Numerical Example Two Examples from Economics 5.5.1 An Illustration from Consumer Theory 5.5.2 An Illustration from Producer Theory 5.5.3 Remarks A Proof of the Theorem of Lagrange A Proof of the Second-Order Conditions Exercises Inequality Constraints and the Theorem of Kuhn and Tucker 6.1 6.2 63 ‘The Theorem pf Kuhn and Tucker 6.1.1 Statement of the Theorem 6.1.2. The Constraint Qualification 6.1.3 The Kuhn-Tucker Multipliers Using the Theorem of Kuhn and Tucker 6.2.1 A “Cookbook” Procedure 6.2.2. Why the Procedure Usually Works 6.2.3. When It Could Fail 6.24 A Numerical Example Illustrations from Economics 6.3.1 An Illustration from Consumer Theory 6.3.2 An Illustration from Producer Theory The General Case: Mixed Constraints A Proof of the Theorem of Kuhn and Tucker Exercises 106 108 110 112 112 113 114 1s 116 117 121 121 122 123 127 128 128 130 132 135 137 142 145 145 145 147 148 150 150 151 152 155 157 158 161 165 168x KR 0 Contents Convex Structures in Optimization Theory 7.1 Convexity Defined 7.1.1 Concave and Convex Functions 7.1.2. Strictly Concave and Strictly Convex Functions 7.2. Implications of Convexity 7.2.1 Convexity and Continuity 7.2.2. Convexity and Differentiability 7.2.3 Convexity and the Properties of the Derivative 7.3 Convexity and Optimization 7.3.1 Some General Observations 7.3.2 Convexity and Unconstrained Optimization 7.3.3, Convexity and the Theorem of Kuhn and Tucker 7.4. Using Convexity in Optimization 7.5 A Proof of the First-Derivative Characterization of Convexity 7.6 A Proof of the Second-Derivative Characterization of Convexity 7.7 A Proof of the Theorem of Kuhn and Tucker under Convexity 78 Exercises Quasi-Convexity and Optimization 8.1 Quasi-Coneave and Quasi-Convex Functions 8.2 Quasi-Convexity as a Generalization of Convexity 8.3. Implications of Quasi-Convexity 8.4 Quasi-Convexity and Optimization 8.5 Using Quasi-Convexity in Optimization Problems 8.6 A Proof of the First-Derivative Characterization of Quasi-Convexity 8.7. A Proof of the Second-Derivative Characterization of Quasi-Convexity 8.8 A Proof of the Theorem of Kuhn and Tucker under Quasi-Convexity 8.9 Exercises Parametric Continuity: The Maximum Theorem 9.1. Correspondences 9.1.1 Upper- and Lower-Semicontinuous Correspondences 9.1.2 Additional Definitions 9.1.3 A Characterization of Semicontinuous Correspondences 9.1.4 Semicontinuous Functions and Semicontinuous Cortespondences 9.2, Parametric Continuity: The Maximum Theorem 9.2.1 The Maximum Theorem 9.2.2 The Maximum Theorem under Convexity 172 173 174 176 177 177 179 183 185 185 187 187 189 190 191 194 198 203 204 205 209 213 215 216 217 220 221 224 225 225 228 229 233 235 235 237: : S is} 93 94 95 Contents An Application to Consumer Theory 93.1 Continuity of the Budget Correspondence 9.3.2 The Indirect Utility Function and Demand Correspondence ‘An Application to Nash Equilibrium 9.4.1 Normal-Form Games 9.4.2 The Brouwer/Kakutani Fixed Point Theorem 9.4.3. Existence of Nash Equilibrium Exercises Supermodularity and Parametric Monotonicity 10.1 10.2 10.3 10.4 10.5 Lattices and Supermodularity 10.1.1 Lattices 10.1.2 Supermodularity and Increasing Differences Parametric Monotonicity An Application to Supermodular Games 10.3.1 Supermodular Games 10.3.2 The Tarski Fixed Point Theorem 10.3.3 Existence of Nash Equilibrium A Proof of the Second-Derivative Characterization of Supermodularity Exercises Finite-Horizon Dynamic Programming 1d 11.2 113 114 ILS 11.6 1L7 Dynamic Programming Problems Finite-Horizon/Dynamic Programming Histories, Strategies, and the Value Function Markovian Strategies Existence of an Optimal Strategy An Example: The Consumption-Savings Problem Exercises Stationary: Discounted Dynamic Programming 12.1 12.2 12.3 12.4 Description of the Framework Histories, Strategies, and the Value Function The Bellman Equation A Technical Digression 12.4.1 Complete Metric Spaces and Cauchy Sequences 12.4.2 Contraction Mappings 12.4.3 Uniform Convergence xi 240 242 243 243 244 246 247 253 254 254 255 258 262 262 263 263 264 266 268 268 268 269 271 272 276 278 281 281 282 283 286 286 287 289xii Contents 12.5. Existence of an Optimal Strategy 12.5.1 A Preliminary Result 12.5.2. Stationary Strategies 12.5.3 Existence of an Optimal Strategy 12.6 An Example: The Optimal Growth Model 12.6.1 The Model 12.6.2 Existence of Optimal Strategies 12.6.3. Characterization of Optimal Strategies 12.7 Exercises Appendix A Set Theory and Logic: An Introduction A.L Sets, Unions, Intersections A.2 Propositions: Contrapositives and Converses A.3- Quantifiers and Negation A4_ Necessary and Sufficient Conditions Appendix B_ The Real Line B.1 Construction of the Real Line B.2_ Properties of the Real Line Appendix C Structures on Vector Spaces C.1_ Vector Spaces C2. Inner Product Spaces C3 Normed Spaces C.4 / Metric Spaces C4.1 Definitions C.4.2 Sets and Sequences in Metric Spaces C.4.3 Continuous Functions on Metric Spaces 4.4 Separable Metric Spaces C45 Subspaces C5 Topological Spaces C.5.1 Definitions C.5.2 Sets and Sequences in Topological Spaces C.5.3 Continuous Functions on Topological Spaces C54 Bases ~. C6 Exercises . Bibliography Index 291 292 294 295 298 299 300 301 309 315 315 316 318 320 323 323 326 330 330 332 333 336 336 337 339 340 341 342 342 343 343 343 345 349 351