Introduction to unconstrained optimization - direct search methods

Similar documents
Constrained optimization: direct methods (cont.)

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

8 Numerical methods for unconstrained problems

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

Optimization Methods

ECE580 Exam 1 October 4, Please do not write on the back of the exam pages. Extra paper is available from the instructor.

NonlinearOptimization

Nonlinear Optimization: What s important?

Optimization Concepts and Applications in Engineering

MATH 4211/6211 Optimization Basics of Optimization Problems

Statistics 580 Optimization Methods

ECE580 Fall 2015 Solution to Midterm Exam 1 October 23, Please leave fractions as fractions, but simplify them, etc.

Unconstrained optimization

Numerical Optimization

Computational Optimization. Augmented Lagrangian NW 17.3

Lecture 3. Optimization Problems and Iterative Algorithms

Introduction to Unconstrained Optimization: Part 2

MATH 5720: Unconstrained Optimization Hung Phan, UMass Lowell September 13, 2018

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

, b = 0. (2) 1 2 The eigenvectors of A corresponding to the eigenvalues λ 1 = 1, λ 2 = 3 are

14. Nonlinear equations

Scientific Computing: An Introductory Survey

Scientific Computing: An Introductory Survey

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

Math (P)refresher Lecture 8: Unconstrained Optimization

On prediction. Jussi Hakanen Post-doctoral researcher. TIES445 Data mining (guest lecture)

Constrained Optimization

Higher-Order Methods

1 Numerical optimization

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

1 Overview. 2 A Characterization of Convex Functions. 2.1 First-order Taylor approximation. AM 221: Advanced Optimization Spring 2016

CE 191: Civil and Environmental Engineering Systems Analysis. LEC 05 : Optimality Conditions

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

nonrobust estimation The n measurement vectors taken together give the vector X R N. The unknown parameter vector is P R M.

Optimization Methods

Scientific Computing: Optimization

Comparative study of Optimization methods for Unconstrained Multivariable Nonlinear Programming Problems

IE 5531: Engineering Optimization I

Lecture V. Numerical Optimization

Optimization. Escuela de Ingeniería Informática de Oviedo. (Dpto. de Matemáticas-UniOvi) Numerical Computation Optimization 1 / 30

Lecture Notes: Geometric Considerations in Unconstrained Optimization

Optimization. Next: Curve Fitting Up: Numerical Analysis for Chemical Previous: Linear Algebraic and Equations. Subsections

5 Quasi-Newton Methods

HW3 - Due 02/06. Each answer must be mathematically justified. Don t forget your name. 1 2, A = 2 2

AM 205: lecture 18. Last time: optimization methods Today: conditions for optimality

ISM206 Lecture Optimization of Nonlinear Objective with Linear Constraints

Written Examination

Math 273a: Optimization Basic concepts

Convex Optimization CMU-10725

Quasi-Newton Methods

1 Numerical optimization

Computational Finance

Advanced Mathematical Programming IE417. Lecture 24. Dr. Ted Ralphs

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

Review of Classical Optimization

2. Quasi-Newton methods

Motivation: We have already seen an example of a system of nonlinear equations when we studied Gaussian integration (p.8 of integration notes)

Line Search Methods for Unconstrained Optimisation

Gradient Descent. Dr. Xiaowei Huang

Mathematical optimization

Numerical Optimization

LECTURE 22: SWARM INTELLIGENCE 3 / CLASSICAL OPTIMIZATION

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

Static unconstrained optimization

Functions of Several Variables

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

Gradient Descent. Ryan Tibshirani Convex Optimization /36-725

Computational Optimization. Convexity and Unconstrained Optimization 1/29/08 and 2/1(revised)

ECS550NFB Introduction to Numerical Methods using Matlab Day 2

Lagrange Multipliers

Introduction to Scientific Computing

Unconstrained minimization of smooth functions

Introduction to Optimization

Examination paper for TMA4180 Optimization I

Lecture 3: Basics of set-constrained and unconstrained optimization

Scientific Data Computing: Lecture 3

Numerical Optimization of Partial Differential Equations

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

Convex Optimization. Problem set 2. Due Monday April 26th

Nonlinear Programming

Lecture 5: September 15

Optimization. Yuh-Jye Lee. March 21, Data Science and Machine Intelligence Lab National Chiao Tung University 1 / 29

Optimization Tutorial 1. Basic Gradient Descent

Optimization II: Unconstrained Multivariable

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

MA/OR/ST 706: Nonlinear Programming Midterm Exam Instructor: Dr. Kartik Sivaramakrishnan INSTRUCTIONS

CHAPTER 2: QUADRATIC PROGRAMMING

Methods for Unconstrained Optimization Numerical Optimization Lectures 1-2

Optimization Methods. Lecture 19: Line Searches and Newton s Method

Numerical Optimization. Review: Unconstrained Optimization

Computational Optimization. Mathematical Programming Fundamentals 1/25 (revised)

Multidisciplinary System Design Optimization (MSDO)

4TE3/6TE3. Algorithms for. Continuous Optimization

Unconstrained Multivariate Optimization

Performance Surfaces and Optimum Points

Numerical Optimization: Basic Concepts and Algorithms

Shiqian Ma, MAT-258A: Numerical Optimization 1. Chapter 3. Gradient Method

CoE 3SK3 Computer Aided Engineering Tutorial: Unconstrained Optimization

An Iterative Descent Method

On the Local Quadratic Convergence of the Primal-Dual Augmented Lagrangian Method

Transcription:

Introduction to unconstrained optimization - direct search methods Jussi Hakanen Post-doctoral researcher jussi.hakanen@jyu.fi

Structure of optimization methods Typically Constraint handling converts the problem to (a series of) unconstrained problems In unconstrained optimization a search direction is determined at each iteration The best solution in the search direction is found with line search Constraint handling method Unconstrained optimization Line search

Group discussion 1. What kind of optimality conditions there exist for unconstrained optimization (x R n )? 2. List methods for unconstrained optimization? what are their general ideas? Discuss in small groups (3-4) for 15-20 minutes Each group has a secretary who writes down the answers of the group At the end, we summarize what each group found

Reminder: gradient and hessian Definition: If function f: R n R is differentiable, then the gradient f(x) consists of the partial derivatives f(x) i.e. f x = f(x),, f(x) x 1 x n Definition: If f is twice differentiable, then the matrix 2 f(x) 2 f(x) x 1 x 1 x 1 x n H x = 2 f(x) 2 f(x) x n x 1 x n x n is called the Hessian (matrix) of f at x Result: If f is twice continuously differentiable, then 2 f(x) x i x j = 2 f(x) x j x i T x i

Reminder: Definite Matrices Definition: A symmetric n n matrix H is positive semidefinite if x R n x T Hx 0. Definition: A symmetric n n matrix H is positive definite if x T Hx > 0 0 x R n Note: If (> <), then H is negative semidefinite (definite). If H is neither positive nor negative semidefinite, then it is indefinite. Result: Let S R n be open convex set and f: S R twice differentiable in S. Function f is convex if and only if H(x ) is positive semidefinite for all x S.

Unconstraint problem min f x, s. t. x R n Necessary conditions: Let f be twice differentiable in x. If x is a local minimizer, then f x = 0 (that is, x is a critical point of f) and H x is positive semidefinite. Sufficient conditions: Let f be twice differentiable in x. If f x = 0 and H(x ) is positive definite, then x is a strict local minimizer. Result: Let f: R n R is twice differentiable in x. If f x = 0 and H(x ) is indefinite, then x is a saddle point.

Unconstraint problem Adopted from Prof. L.T. Biegler (Carnegie Mellon University)

Descent direction Definition: Let f: R n R. A vector d R n is a descent direction for f in x R n if δ > 0 s.t. f x + λd < f(x ) λ (0, δ]. Result: Let f: R n R be differentiable in x. If d R n s.t. f x T d < 0 then d is a descent direction for f in x.

Model algorithm for unconstrained minimization Let x h be the current estimate for x 1) [Test for convergence.] If conditions are satisfied, stop. The solution is x h. 2) [Compute a search direction.] Compute a nonzero vector d h R n which is the search direction. 3) [Compute a step length.] Compute α h > 0, the step length, for which it holds that f x h + α h d h < f(x h ). 4) [Update the estimate for minimum.] Set x h+1 = x h + α h d h, h = h + 1 and go to step 1. From Gill et al., Practical Optimization, 1981, Academic Press

On convergence Iterative method: a sequence {x h } s.t. x h x when h Definition: A method converges linearly if α [0,1) and M 0 s.t. h M x h+1 x α x h x, superlinearly if M 0 and for some sequence α h 0 it holds that h M x h+1 x α h x h x, with degree p if α 0, p > 0 and M 0 s.t. h M x h+1 x α x h x p. If p = 2 (p = 3), the convergence is quadratic (cubic).

Summary of group discussion for methods 1. Newton s method 1. Utilizes tangent 2. Golden section method 1. For line search 3. Downhill Simplex 4. Cyclic coordinate method 1. One coordinate at a time 5. Polytopy search (Nelder-Mead) 1. Idea based on geometry 6. Gradient descent (steepest descent) 1. Based on gradient information

Direct search methods Univariate search, coordinate descent, cyclic coordinate search Hooke and Jeeves Powell s method

From Miettinen: Nonlinear optimization, 2007 (in Finnish) Coordinate descent f x = 2x 1 2 + 2x 1 x 2 + x 2 2 + x 1 x 2

From Miettinen: Nonlinear optimization, 2007 (in Finnish) Idea of pattern search

From Miettinen: Nonlinear optimization, 2007 (in Finnish) Hooke and Jeeves f x = x 1 2 4 + x 1 2x 2 2

From Miettinen: Nonlinear optimization, 2007 (in Finnish) Hooke and Jeeves with fixed step length f x = x 1 2 4 + x 1 2x 2 2

Powell s method Most efficient pattern search method Differs from Hooke and Jeeves so that for each pattern search step one of the coordinate directions is replaced with previous pattern search direction.