Inverse Problems and Optimal Design in Electricity and Magnetism

Similar documents
Contents. Preface. 1 Introduction Optimization view on mathematical models NLP models, black-box versus explicit expression 3

Numerical Analysis of Electromagnetic Fields

GEOPHYSICAL INVERSE THEORY AND REGULARIZATION PROBLEMS

fiziks Institute for NET/JRF, GATE, IIT-JAM, JEST, TIFR and GRE in PHYSICAL SCIENCES

M E M O R A N D U M. Faculty Senate approved November 1, 2018

LINEAR AND NONLINEAR PROGRAMMING

Optimization and Root Finding. Kurt Hornik

Iterative Methods for Smooth Objective Functions

CHAPTER 2. COULOMB S LAW AND ELECTRONIC FIELD INTENSITY. 2.3 Field Due to a Continuous Volume Charge Distribution

Data Fitting and Uncertainty

Optimization: Nonlinear Optimization without Constraints. Nonlinear Optimization without Constraints 1 / 23

Numerical Optimization Professor Horst Cerjak, Horst Bischof, Thomas Pock Mat Vis-Gra SS09

Nonlinear Optimization: What s important?

Optimization Concepts and Applications in Engineering

Numerical optimization

NUMERICAL MATHEMATICS AND COMPUTING

Numerical optimization. Numerical optimization. Longest Shortest where Maximal Minimal. Fastest. Largest. Optimization problems

Chapter 4. Unconstrained optimization

Numerisches Rechnen. (für Informatiker) M. Grepl P. Esser & G. Welper & L. Zhang. Institut für Geometrie und Praktische Mathematik RWTH Aachen

Numerical Methods in Matrix Computations

IPAM Summer School Optimization methods for machine learning. Jorge Nocedal

Constrained optimization. Unconstrained optimization. One-dimensional. Multi-dimensional. Newton with equality constraints. Active-set method.

Scientific Computing: An Introductory Survey

Scientific Computing: An Introductory Survey

Numerical computation II. Reprojection error Bundle adjustment Family of Newtonʼs methods Statistical background Maximum likelihood estimation

MATHEMATICS FOR COMPUTER VISION WEEK 8 OPTIMISATION PART 2. Dr Fabio Cuzzolin MSc in Computer Vision Oxford Brookes University Year

Scientific Computing: Optimization

NUMERICAL COMPUTATION IN SCIENCE AND ENGINEERING

Unit-1 Electrostatics-1

Programming, numerics and optimization

Optimization. Totally not complete this is...don't use it yet...

ELECTROMAGNETISM. Second Edition. I. S. Grant W. R. Phillips. John Wiley & Sons. Department of Physics University of Manchester

Condensed Table of Contents for Introduction to Stochastic Search and Optimization: Estimation, Simulation, and Control by J. C.

OPTIMAL ESTIMATION of DYNAMIC SYSTEMS

Numerical Optimization

Coordinate Update Algorithm Short Course Proximal Operators and Algorithms

Preface to the Second Edition. Preface to the First Edition

A THEORETICAL INTRODUCTION TO NUMERICAL ANALYSIS

REGULARIZATION PARAMETER SELECTION IN DISCRETE ILL POSED PROBLEMS THE USE OF THE U CURVE

Conjugate Directions for Stochastic Gradient Descent

Methods that avoid calculating the Hessian. Nonlinear Optimization; Steepest Descent, Quasi-Newton. Steepest Descent

UNIT-I Static Electric fields

Appendix A Taylor Approximations and Definite Matrices

Contents. Preface to the Third Edition (2007) Preface to the Second Edition (1992) Preface to the First Edition (1985) License and Legal Information

Chapter 3 Numerical Methods

Contents. Preface... xi. Introduction...

UNIT I ELECTROSTATIC FIELDS

IDL Advanced Math & Stats Module

Numerical Optimization Techniques

Optimization Tutorial 1. Basic Gradient Descent

Iterative regularization of nonlinear ill-posed problems in Banach space

Preface to Second Edition... vii. Preface to First Edition...

A Sobolev trust-region method for numerical solution of the Ginz

Stable Adaptive Momentum for Rapid Online Learning in Nonlinear Systems

INVERSE DETERMINATION OF SPATIAL VARIATION OF DIFFUSION COEFFICIENTS IN ARBITRARY OBJECTS CREATING DESIRED NON- ISOTROPY OF FIELD VARIABLES

INTRODUCTION TO ELECTRODYNAMICS

CS 542G: Robustifying Newton, Constraints, Nonlinear Least Squares

Introduction. New Nonsmooth Trust Region Method for Unconstraint Locally Lipschitz Optimization Problems

UNIT-I INTRODUCTION TO COORDINATE SYSTEMS AND VECTOR ALGEBRA

Preconditioned conjugate gradient algorithms with column scaling

Optimization Methods

Outline. Scientific Computing: An Introductory Survey. Optimization. Optimization Problems. Examples: Optimization Problems

Mathematical optimization

Optimization Methods for Circuit Design

Lecture 11: CMSC 878R/AMSC698R. Iterative Methods An introduction. Outline. Inverse, LU decomposition, Cholesky, SVD, etc.

NONLINEAR. (Hillier & Lieberman Introduction to Operations Research, 8 th edition)

4M020 Design tools. Algorithms for numerical optimization. L.F.P. Etman. Department of Mechanical Engineering Eindhoven University of Technology

Scientific Computing: An Introductory Survey

ADAM PIŁAT Department of Automatics, AGH University of Science and Technology Al. Mickiewicza 30, Cracow, Poland

j=1 r 1 x 1 x n. r m r j (x) r j r j (x) r j (x). r j x k

ELECTRO MAGNETIC FIELDS

Optimization Methods

Trust-region methods for rectangular systems of nonlinear equations

An Iterative Descent Method

CHETTINAD COLLEGE OF ENGINEERING & TECHNOLOGY NH-67, TRICHY MAIN ROAD, PULIYUR, C.F , KARUR DT.

Numerical Mathematics

Minimization of Static! Cost Functions!

8 Numerical methods for unconstrained problems

Parameter Identification in Partial Differential Equations

Trajectory-based optimization

Part 3: Trust-region methods for unconstrained optimization. Nick Gould (RAL)

Numerical Methods for Engineers

Numerical Analysis for Statisticians

1 Newton s Method. Suppose we want to solve: x R. At x = x, f (x) can be approximated by:

Combining Conjugate Direction Methods with Stochastic Approximation of Gradients

EAD 115. Numerical Solution of Engineering and Scientific Problems. David M. Rocke Department of Applied Science

KINGS COLLEGE OF ENGINEERING DEPARTMENT OF ELECTRONICS AND COMMUNICATION ENGINEERING QUESTION BANK

Medical Physics & Science Applications

Trust Region Methods. Lecturer: Pradeep Ravikumar Co-instructor: Aarti Singh. Convex Optimization /36-725

Ranking from Crowdsourced Pairwise Comparisons via Matrix Manifold Optimization

OPER 627: Nonlinear Optimization Lecture 14: Mid-term Review

Engineering Electromagnetic Fields and Waves

Multidisciplinary System Design Optimization (MSDO)

Machine Learning Support Vector Machines. Prof. Matteo Matteucci

The speed of Shor s R-algorithm

ADAPTIVE FILTER THEORY

Primal-Dual Interior-Point Methods for Linear Programming based on Newton s Method

GLOBALLY CONVERGENT GAUSS-NEWTON METHODS

Chapter 5. Magnetostatics

Recovery of anisotropic metrics from travel times

Transcription:

Inverse Problems and Optimal Design in Electricity and Magnetism P. Neittaanmäki Department of Mathematics, University of Jyväskylä M. Rudnicki Institute of Electrical Engineering, Warsaw and A. Savini Department of Electrical Engineering, University ofpavia CLARENDON PRESS OXFORD

Contents I Mathematical methodology 1. Mathematical preliminaries 3 1.1. Basic notation 3 1.2. Vectors, matrices, norms 4 1.2.1. Matrices 4 1.2.2. Independence, orthogonality, subspaces 6 1.2.3. Special matrices 7 1.2.4. Block matrices 8 1.2.5. Vector norms 9 1.2.6. Matrix norms 11 1.3. Functions 12 1.4. Domains 13 1.5. Function spaces 14 1.6. Classical integral formulae 21 1.7. References 23 2. Boundary-value problems 24 2.1. Variational methods for elliptic problems 24 2.1.1. Classical formulation 24 2.1.2. Variational formulation 25 2.2. Integral-equation formulation for partial differential 33 equations 2.3. References 36 3. Numerical methods for boundary-value problems 38 3.1. Introduction 38 3.2. Finite-element method for elliptic problems 42 3.2.1. Triangulation of the domain 42 3.2.2. Finite-element equations 45 3.3. Boundary-element method 51 3.4. References 54 4. Regularization 56 4.1. Ill-posed problems 56 4.2. Regularization schemes 57 4.3. Discrepancy principle 59 4.4. Linear integral equations of the first kind 61 4.5. Tikhonov regularization 62 4.6. Truncated singular-value decomposition 66 4.7. Regularization parameter 67

x Contents 4.8. Solving Fredholm integral equations of the first 69 kind using regularization 4.9. References 72 Numerical methods for systems of equations 76 5.1. Solving linear systems 76 5.1.1. Elementary iterative methods 77 5.1.2. The conjugate gradient method 78 5.1.3. The preconditioned biconjugate gradient method 79 5.2. Solving nonlinear systems 80 5.2.1. Slope methods 80 5.2.2. Newton-Raphson method 82 5.2.3. Powell's hybrid method 83 5.2.4. Brown-Brent methods 84 5.3. References 87 Unconstrained optimization 89 6.1. Optimality conditions 94 6.2. Search methods 100 6.3. Steepest-descent method 102 6.4. Conjugate gradient method 103 6.5. Newton method with a linear search 105 6.6. Quasi-Newton or variable-metric method 107 6.7. Polytope method 111 6.8. Stochastic optimization 115 6.8.1. Random-search methods 115 6.8.2. Genetic algorithms 116 6.8.3. Simulated annealing 116 6.8.4. Simulated annealing optimization code 119 6.9. Neural computing 123 6.9.1. Introduction 123 6.9.2. Preliminaries 123 6.9.3. Multilayer feedforward networks 124 6.9.4. Optimization by neural networks 128 6.10. References 131 Constrained optimization 137 7.1. Linear programming 137 7.2. Optimality (Karush-Kuhn-Tucker) conditions 139 7.2.1. Smooth case 139 7.2.2. Nonsmooth case 141 7.3. Sequential linear programming 142 7.4. Sequential quadratic programming 147 7.5. Nonsmooth and multicriteria optimization 151 7.5.1. Smooth multicriteria optimization 152 7.5.2. Unconstrained convex optimization 152 7.5.3. Unconstrained nonconvex optimization 153 7.5.4. Constrained optimization 154

Contents xi 7.5.5. Multicriteria optimization 154 7.5.6. Proximal bundle nonsmooth optimization code 155 7.6. Concluding remarks 158 7.7. References 159 Linear least-squares 161 8.1. Overdetermined systems of equations 161 8.2. Normal equations and ill-conditioning 163 8.3. Singular-value decomposition 164 8.4. Constrained linear least-squares 167 8.5. References 170 Nonlinear least-squares 172 9.1. Gauss-Newton method 173 9.2. Levenberg-Marquardt method 175 9.3. Powell's hybrid method 178 9.4. Derivative-free methods 179 9.5. Large-residual problems 180 9.6. Trust-region method 181 9.7. Constrained nonlinear least-squares 183 9.8. References 185 II Fundamentals of electromagnetism 10. Introduction 191 11. Maxwell's equations 192 11.1. Basic equations 192 11.2. Interface conditions 195 11.3. References 197 12. Potential equations in electricity and magnetism 198 12.1. Laplace's equation 198 12.2. Poisson's equation 200 12.3. Nonlinear magnetostatic fields in R 2 201 12.4. Quasistatic linear electromagnetic fields 202 12.5. Electromagnetic fields in linear isotropic media 205 12.6. Further elements of electromagnetic theory 207 12.6.1. Energy, power and forces 207 12.6.2. Coulomb's law and the Biot-Savart law 208 12.7. References 209 13. Numerical methods in electromagnetism 210 13.1. Introduction 210 13.2. Approximation of the static case 210 13.2.1. Least-squares variational methods in 210 electromagnetism 13.2.2. Least-squares finite-element methods 212 for electric fields 13.2.3. The nonlinear case 213

xii Contents 13.3. Approximation of quasistatic electric and 216 magnetic fields 13.4. Approximation of the wave equation for electric 221 scalar potentials 13.5. Concluding remarks 223 13.6. References 224 III Inverse problems and optimal design in electromagnetic applications 14. Inverse problems and optimal design 231 14.1. Introduction 231 14.2. Inverse electromagnetic problems: methodology 232 14.3. Optimal design techniques for solving inverse 233 electromagnetic problems 14.3.1. Deterministic optimization methods (DOMs) 235 14.3.2. Stochastic optimization methods (SOMs) 235 14.4. References 236 15. Synthesis of sources 237 15.1. Synthesis of the magnetic field in a solenoid 237 15.2. Synthesis of the magnetic field on a solenoid axis 239 15.3. Synthesis of an electric field due to a point charge 243 15.4. Synthesis of an electric field due to a surface charge 245 15.4.1. Parallel-plate capacitors 245 15.4.2. Thin conducting plate 247 15.5. Synthesis of a magnetic field in a wire system 249 15.6. Synthesis of electromagnets 256 15.7. References 258 16. Synthesis of boundary conditions 259 16.1. Synthesis of the electric field in a finite domain 259 16.2. Synthesis of an electric field due to a boundary 262 potential 16.3. Synthesis of the magnetic field due to a boundary 266 current 16.4. References 268 17. Synthesis of material properties 269 17.1. Synthesis of permittivity 270 17.2. References 275 18. Optimal shape synthesis 276 18.1. Optimal shape design of a solenoid 276 18.2. Optimal shape design of the shim coil of a solenoid 279 18.3. Optimal shape design of an electromagnet 282 18.4. Optimal shape design of an air-filled capacitor 285 18.5. Optimal shape design of a pole profile in a linear 288 H-shaped magnet

Contents xiii 18.6. Optimal shape design of a pole profile in a nonlinear 291 H-shaped magnet 18.7. References 297 19. Remarks on inverse and design problems 298 19.1. Survey of solved problems 298 19.2. References 304 20. Artificial neural networks (ANNs) for inverse electromagnetic 309 modelling 20.1. Remarks on artificial neural networks 309 20.2. References 312 IV Implementation of the FEM, design-sensitivity and shape design procedures 21. Introduction 315 22. Implementation of the finite-element method 316 22.1. Linear elliptic boundary-value problems 316 22.1.1. Discretization of the problem 316 22.1.2. Isoparametric elements 317 22.1.3. Data structures 320 22.1.4. General program structure 322 22.2. A nonlinear FEM solver using a quasi-newton method 322 22.3. References 324 23. Finite-element design-sensitivity analysis 325 23.1. Setting of the optimal shape design problem 325 23.2. Design-sensitivity analysis for linear problems 327 23.3. Sensitivity for the nonlinear potential equation 330 23.4. Implementation of optimal shape design procedures 333 23.5. Automatic differentiation of computer programs 335 23.6. References 338 24. Subroutine libraries 339 24.1. General-purpose software libraries 339 24.2. Partial differential equations and electromagnetic 341 software 24.3. Software libraries for nonsmooth and multicriterion 348 optimization 24.4. Artificial intelligence tools and software for optimal 349 shape design 24.5. References 349 Author index 355 Subject index 359