Logga in

Priserna visas inklusive moms och du betalar med Klarna


Priserna visas exklusive moms, du kan betala med Klarna eller faktura

Priserna visas inklusive moms och du betalar med Klarna


Priserna visas exklusive moms, du kan betala med Klarna eller faktura

Varukorg

Varukorgen är tom!

Varukorgen inkl. moms 0 kr


Elektronisk distribution

Frakt inkl. moms 0 kr


Varav moms (6 %) 0 kr

Varav moms (25 %) 0 kr

Öresutjämning 0 kr


Att betala inkl. moms 0 kr


Till kassan

Nonlinear Optimization

A Basic Course
Skickas följande arbetsdag

This book is intended to be used in a first course treating nonlinear optimization problems. It starts with a review of the necessary concepts of calculus and linear algebra, including positive definite matrices. The classical numerical optimization methods with line search algorithms are motivated and derived, and some of their properties investigated. The theories of constrained optimization, convexity and duality are developed rigorously and presented with focus on solving concrete optim...

This book is intended to be used in a first course treating nonlinear optimization problems. It starts with a review of the necessary concepts of calculus and linear algebra, including positive definite matrices. The classical numerical optimization methods with line search algorithms are motivated and derived, and some of their properties investigated. The theories of constrained optimization, convexity and duality are developed rigorously and presented with focus on solving concrete optimization problems. There are plenty of solved examples as well as exercises to which answers and in many cases complete solutions are provided.

Preface 7

 

CHAPTER 1 Introduction and mathematical background 9

1.1 Overview 9

1.2 Notation and some calculus 11

1.3 Conditions for unconstrained local minima 18

1.4 Characterizations of positive semidefinite matrices 21

 

CHAPTER 2 A prototype algorithm for unconstrained

optimization 39

2.1 A prototype algorithm 39

2.2 Termination criteria 40

2.3 Descent direction 40

2.4 Speed of convergence 41

 

CHAPTER 3 Line search 43

3.1 Exact, approximate and inexact line searches 43

3.2 Bracketing—finding an initial interval 44

3.3 Uniform search 45

3.4 Dichotomous search 47

3.5 Golden Section search 47

3.6 Bisection search 49

3.7 Quadratic fit 49

3.8 Newton’s method 50

3.9 Armijo’s rule 51

3.10 Wolfe’s conditions 53

 

CHAPTER 4 Unconstrained numerical optimization methods 57

4.1 The Steepest Descent method 58

4.2 Newton’s method and modifications 59

4.3 Methods for non-differentiable functions 62

4.4 Conjugate directions 67

4.5 The Conjugate Gradient method 70

4.6 Quasi-Newton methods 74

4.7 Least-squares problems and the Gauß–Newton method 82

 

CHAPTER 5 Penalty and barrier functions 87

5.1 Penalty functions 87

5.2 Barrier functions 96

 

CHAPTER 6 Convexity 101

6.1 Convex functions of one variable 101

6.2 Convex sets 111

6.3 Convex optimization problems 116

6.4 Characterizations of convex functions and sets 120

6.5 Separating hyperplane and Farkas’ lemma 126

 

CHAPTER 7 Theory of constrained optimization 135

7.1 First-order necessary conditions: inequality constraints 135

7.2 First-order necessary conditions: mixed constraints 150

7.3 First-order necessary conditions: affine constraints 154

7.4 First-order sufficient conditions 155

7.5 Strategy to find a global solution 159

7.6 Second-order sufficient conditions 161

7.7 Sensitivity 169

 

CHAPTER 8 Duality 173

8.1 The dual problem 173

8.2 Necessary and sufficient conditions for strong duality 176

8.3 Saddle points 182

8.4 Geometrical interpretation of duality gap 183

8.5 Linear problems and duality 188

 

CHAPTER 9 Quadratic problems 201

9.1 General properties 201

9.2 Elimination of equality constraints 202

9.3 KKT solution with equality constraints 205

9.4 Solution by duality 207

 

CHAPTER 10 Exercises 209

CHAPTER 11 Answers or solutions to exercises 241

 

Index 295

 

Information

Författare:
Stefan Diehl
Språk:
Engelska
ISBN:
9789144183053
Utgivningsår:
2024
Artikelnummer:
47343-01
Upplaga:
Första
Sidantal:
300

Författare

Stefan Diehl

Stefan Diehl är civilingenjör och docent i tillämpad matematik vid Lunds Tekniska Högskola. Han har ­erfarenhet av undervisning, kurs- och programu...

 ;