Lecture 2 Solution of Nonlinear Equations ( Root Finding Problems )

Similar documents
: Numerical Analysis Topic 2: Solution of Nonlinear Equations Lectures 5-11:

CISE301: Numerical Methods Topic 2: Solution of Nonlinear Equations

Numerical Methods Solution of Nonlinear Equations

Summary with Examples for Root finding Methods -Bisection -Newton Raphson -Secant

CHAPTER 4d. ROOTS OF EQUATIONS

Review of Taylor Series. Read Section 1.2

Single Variable Optimization

Chapter 3 Differentiation and Integration

Root Finding

Finite Difference Method

Chapter 4: Root Finding

OPTIMISATION. Introduction Single Variable Unconstrained Optimisation Multivariable Unconstrained Optimisation Linear Programming

36.1 Why is it important to be able to find roots to systems of equations? Up to this point, we have discussed how to find the solution to

Section 3.6 Complex Zeros

Chapter Newton s Method

EE 330 Lecture 24. Small Signal Analysis Small Signal Analysis of BJT Amplifier

Shuai Dong. Isaac Newton. Gottfried Leibniz

CIS526: Machine Learning Lecture 3 (Sept 16, 2003) Linear Regression. Preparation help: Xiaoying Huang. x 1 θ 1 output... θ M x M

Chapter 12. Ordinary Differential Equation Boundary Value (BV) Problems

2. PROBLEM STATEMENT AND SOLUTION STRATEGIES. L q. Suppose that we have a structure with known geometry (b, h, and L) and material properties (EA).

APPENDIX A Some Linear Algebra

A MODIFIED METHOD FOR SOLVING SYSTEM OF NONLINEAR EQUATIONS

SOLVING NON-LINEAR SYSTEMS BY NEWTON s METHOD USING SPREADSHEET EXCEL Tay Kim Gaik Universiti Tun Hussein Onn Malaysia

( ) [ ( k) ( k) ( x) ( ) ( ) ( ) [ ] ξ [ ] [ ] [ ] ( )( ) i ( ) ( )( ) 2! ( ) = ( ) 3 Interpolation. Polynomial Approximation.

Solution of Linear System of Equations and Matrix Inversion Gauss Seidel Iteration Method

Newton s Method for One - Dimensional Optimization - Theory

Polynomial Regression Models

CS 331 DESIGN AND ANALYSIS OF ALGORITHMS DYNAMIC PROGRAMMING. Dr. Daisy Tang

ME 501A Seminar in Engineering Analysis Page 1

Kernel Methods and SVMs Extension

On the Interval Zoro Symmetric Single-step Procedure for Simultaneous Finding of Polynomial Zeros

Lecture 26 Finite Differences and Boundary Value Problems

ORDINARY DIFFERENTIAL EQUATIONS EULER S METHOD

Topic 5: Non-Linear Regression

Lecture 13 APPROXIMATION OF SECOMD ORDER DERIVATIVES

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE

For now, let us focus on a specific model of neurons. These are simplified from reality but can achieve remarkable results.

Common loop optimizations. Example to improve locality. Why Dependence Analysis. Data Dependence in Loops. Goal is to find best schedule:

Chapter 14 Simple Linear Regression

EEE 241: Linear Systems

Curve Fitting with the Least Square Method

Chapter Newton-Raphson Method of Solving a Nonlinear Equation

Chapter 4 The Wave Equation

A new Approach for Solving Linear Ordinary Differential Equations

Practical Newton s Method

Anouncements. Multigrid Solvers. Multigrid Solvers. Multigrid Solvers. Multigrid Solvers. Multigrid Solvers

Least squares cubic splines without B-splines S.K. Lucas

A Hybrid Variational Iteration Method for Blasius Equation

NUMERICAL DIFFERENTIATION

Digital Signal Processing

Limited Dependent Variables

A Simple Research of Divisor Graphs

The Fundamental Theorem of Algebra. Objective To use the Fundamental Theorem of Algebra to solve polynomial equations with complex solutions

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems

Mathematical Economics MEMF e ME. Filomena Garcia. Topic 2 Calculus

Example: (13320, 22140) =? Solution #1: The divisors of are 1, 2, 3, 4, 5, 6, 9, 10, 12, 15, 18, 20, 27, 30, 36, 41,

Spring Force and Power

Linear Approximation with Regularization and Moving Least Squares

Population element: 1 2 N. 1.1 Sampling with Replacement: Hansen-Hurwitz Estimator(HH)

Week3, Chapter 4. Position and Displacement. Motion in Two Dimensions. Instantaneous Velocity. Average Velocity

= z 20 z n. (k 20) + 4 z k = 4

Grid Generation around a Cylinder by Complex Potential Functions

Department of Quantitative Methods & Information Systems. Time Series and Their Components QMIS 320. Chapter 6

AS-Level Maths: Statistics 1 for Edexcel

The Multiple Classical Linear Regression Model (CLRM): Specification and Assumptions. 1. Introduction

Difference Equations

Additional Codes using Finite Difference Method. 1 HJB Equation for Consumption-Saving Problem Without Uncertainty

Propagation of error for multivariable function

: 5: ) A

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Lecture 21: Numerical methods for pricing American type derivatives

Lecture Notes on Linear Regression

Computational Biology Lecture 8: Substitution matrices Saad Mneimneh

Statistics for Economics & Business

Numerical Differentiation

Physics 2A Chapter 3 HW Solutions

PHYS 1441 Section 002 Lecture #15

International Mathematical Olympiad. Preliminary Selection Contest 2012 Hong Kong. Outline of Solutions

= 1.23 m/s 2 [W] Required: t. Solution:!t = = 17 m/s [W]! m/s [W] (two extra digits carried) = 2.1 m/s [W]

Parametric fractional imputation for missing data analysis. Jae Kwang Kim Survey Working Group Seminar March 29, 2010

Implicit Integration Henyey Method

Chapter 15 Student Lecture Notes 15-1

Statistics for Business and Economics

MATH 829: Introduction to Data Mining and Analysis The EM algorithm (part 2)

Introduction to Vapor/Liquid Equilibrium, part 2. Raoult s Law:

6.1 The function can be formulated as a fixed-point iteration as

Math1110 (Spring 2009) Prelim 3 - Solutions

EE 330 Fall 2016 Seating

Unit 5: Quadratic Equations & Functions

Bézier curves. Michael S. Floater. September 10, These notes provide an introduction to Bézier curves. i=0

ANALYSIS OF SOURCE LOCATION ALGORITHMS Part II: Iterative methods

Tracking with Kalman Filter

Journal of Universal Computer Science, vol. 1, no. 7 (1995), submitted: 15/12/94, accepted: 26/6/95, appeared: 28/7/95 Springer Pub. Co.

TEST 5 (phy 240) 2. Show that the volume coefficient of thermal expansion for an ideal gas at constant pressure is temperature dependent and given by

Integrals and Invariants of Euler-Lagrange Equations

Lecture 5 Decoding Binary BCH Codes

THE CHINESE REMAINDER THEOREM. We should thank the Chinese for their wonderful remainder theorem. Glenn Stevens

Section 8.3 Polar Form of Complex Numbers

Complex Variables. Chapter 18 Integration in the Complex Plane. March 12, 2013 Lecturer: Shih-Yuan Chen

Lecture 10 Support Vector Machines II

Transcription:

Lecture Soluton o Nonlnear Equatons Root Fndng Problems Dentons Classcaton o Methods Analytcal Solutons Graphcal Methods Numercal Methods Bracketng Methods Open Methods Convergence Notatons

Root Fndng Problems Many problems n Scence and Engneerng are epressed as: Gven a nd the value contnuous r uncton such that r, These problems are called root ndng problems.

Roots o Equatons A number r that satses an equaton s called a root o the equaton. The equaton : has our roots :, 3, 3, and..e., 4 3 3 7 4 3 5 7 5 8 8 3 The equaton has two smple roots and and a repeated root 3 wth multplcty. 3 3

Zeros o a Functon Let be a real-valued uncton o a real varable. Any number r or whch r s called a zero o the uncton. Eamples: and 3 are zeros o the uncton --3. 4

Graphcal Interpretaton o Zeros The real zeros o a uncton are the values o at whch the graph o the uncton crosses or touches the - as. Real zeros o 5

Smple Zeros has two smple zeros one at and one at 6

Multple Zeros has double zeros zero wth mulplcty at 7

Multple Zeros 3 3 has a zero wth mulplcty 3 at 8

Facts Any n th order polynomal has eactly n zeros countng real and comple zeros wth ther multplctes. Any polynomal wth an odd order has at least one real zero. I a uncton has a zero at r wth multplcty m then the uncton and ts rst m- dervatves are zero at r and the m th dervatve at r s not zero. 9

Roots o Equatons & Zeros o Functon Gven theequaton : Move all terms to one sde o the equaton : Dene 4 4 3 3 3 3 as : 7 7 4 3 5 5 8 3 7 8 5 8 The zeros o are the same as the roots o theequaton Whch are, 3, 3, and

Soluton Methods Several ways to solve nonlnear equatons are possble: Analytcal Solutons Possble or specal equatons only Graphcal Solutons Useul or provdng ntal guesses or other methods Numercal Solutons Open methods Bracketng methods

Analytcal Methods Analytcal Solutons are avalable or specal equatons only. Analytcal soluton o : a b c roots b ± b a 4ac No analytcal soluton s avalable or : e

Graphcal Methods Graphcal methods are useul to provde an ntal guess to be used by other methods. Solve e The root root [,].6 e Root 3

Numercal Methods Many methods are avalable to solve nonlnear equatons: Bsecton Method Newton s Method Secant Method False poston Method Muller s Method Barstow s Method Fed pont teratons. These wll be covered here 4

Bracketng Methods In bracketng methods, the method starts wth an nterval that contans the root and a procedure s used to obtan a smaller nterval contanng the root. Eamples o bracketng methods: Bsecton method False poston method 5

Open Methods In the open methods, the method starts wth one or more ntal guess ponts. In each teraton, a new guess o the root s obtaned. Open methods are usually more ecent than bracketng methods. They may not converge to a root. 6

Convergence Notaton A sequence,,..., n,... s sad to converge to to every ε > there ests N such that : n < ε n > N 7

8 Convergence Notaton C P C C p n n n n n n : order Convergence o Quadratc Convergence : Lnear Convergence :. to converge,...,, Let

Speed o Convergence We can compare derent methods n terms o ther convergence rate. Quadratc convergence s aster than lnear convergence. A method wth convergence order q converges aster than a method wth convergence order p q>p. Methods o convergence order p> are sad to have super lnear convergence. 9

Bsecton Method The Bsecton method s one o the smplest methods to nd a zero o a nonlnear uncton. It s also called nterval halvng method. To use the Bsecton method, one needs an ntal nterval that s known to contan a zero o the uncton. The method systematcally reduces the nterval. It does ths by dvdng the nterval nto two equal parts, perorms a smple test and based on the result o the test, hal o the nterval s thrown away. The procedure s repeated untl the desred nterval sze s obtaned.

Intermedate Value Theorem Let be dened on the nterval [a,b]. a Intermedate value theorem: a uncton s contnuous and a and b have derent sgns then the uncton has at least one zero n the nterval [a,b]. a b b

Eamples I a and b have the same sgn, the uncton may have an even number o real zeros or no real zeros n the nterval [a, b]. Bsecton method can not be used n these cases. a b The uncton has our real zeros a b The uncton has no real zeros

Two More Eamples I a and b have derent sgns, the uncton has at least one real zero. a b Bsecton method can be used to nd one o the zeros. The uncton has one real zero a b The uncton has three real zeros 3

Bsecton Method I the uncton s contnuous on [a,b] and a and b have derent sgns, Bsecton method obtans a new nterval that s hal o the current nterval and the sgn o the uncton at the end ponts o the nterval are derent. Ths allows us to repeat the Bsecton procedure to urther reduce the sze o the nterval. 4

Bsecton Method Assumptons: Gven an nterval [a,b] s contnuous on [a,b] a and b have opposte sgns. These assumptons ensure the estence o at least one zero n the nterval [a,b] and the bsecton method can be used to obtan a smaller nterval that contans the zero. 5

Bsecton Algorthm Assumptons: s contnuous on [a,b] a b < a Algorthm: Loop. Compute the md pont cab/. Evaluate c 3. I a c < then new nterval [a, c] I a c > then new nterval [c, b] End loop a c b b b a a a 6

Eample - - - - 7

Flow Chart o Bsecton Method Start: Gven a,b and ε u a ; v b c ab / ; w c no yes s u w < no s b-a /<ε yes Stop bc; v w ac; u w 8

Eample Can you use Bsecton method to nd a zero o : 3 3 n the nterval [,]? Answer: s contnuous on [,] and * 3 3> Assumptons are not satsed Bsecton method can not be used 9

3 Eample Answer: [,]? nterval n the 3 : o zero a nd to method Bsecton use you Can 3 used can be method Bsecton satsed are Assumptons - * and on [,] contnuous s <

Best Estmate and Error Level Bsecton method obtans an nterval that s guaranteed to contan a zero o the uncton. The best estmate o the zero o the uncton ater the rst teraton o the Bsecton method s the md pont o the ntal nterval: b a Estmate o the zero : r b a Error 3

Stoppng Crtera Two common stoppng crtera. Stop ater a ed number o teratons. Stop when the absolute error s less than a speced value How are these crtera related? 3

Stoppng Crtera c n : s the mdpont o the nterval at the n th teraton c n s usually used as the estmate o the root. r : s the zero o the uncton. Ater n teratons : error r -c n E n a b a n n 33

Convergence Analyss Gven, a, b, and ε How many teratons are needed such that : - r ε where r s the zero o and s the bsecton estmate.e., c k? log b a log ε n log 34

Convergence Analyss Alternatve Form log b a log ε n log wdth o ntal nterval n log log desred error b a ε 35

Eample a 6, b 7, ε.5 How many teratons are needed such that : - r ε? n log b a log ε log log log.5 log.9658 n 36

Eample Use Bsecton method to nd a root o the equaton cos wth absolute error <. assume the ntal nterval [.5,.9] Queston : What s? Queston : Are the assumptons satsed? Queston 3: How many teratons are needed? Queston 4: How to compute the new estmate? 37

CISE3_Topc 38

Bsecton Method Intal Interval a-.3776 b.784 a.5 c.7 b.9 Error <. -.3776 -.648.784.5.7.9 Error <. -.648.33.784.7.8.9 Error <.5 39

Bsecton Method -.648.83.33.7.75.8 -.648 -.35.83.7.75.75 Error <.5 Error <.5 Intal nterval contanng the root: [.5,.9] Ater 5 teratons: Interval contanng the root: [.75,.75] Best estmate o the root s.7375 Error <.5 4

A Matlab Program o Bsecton Method a.5; b.9; ua-cosa; vb-cosb; or k:5 cab/ cc-cosc u*c< bc ; vc; else ac; uc; end end c.7 c -.648 c.8 c.33 c.75 c.83 c.75 c -.35 4

4 Eample Fnd the root o: root the nd to used can be method Bsecton, * contnuous s * nterval:[,] n the 3 3 < b a

Eample Iteraton a b c ab c b-a.5 -.375.5.5.5.66.5 3.5.5.375-7.3E-3.5 4.5.375.35 9.3E-.65 5.35.375.34375 9.37E-3.35 43

Bsecton Method Advantages Smple and easy to mplement One uncton evaluaton per teraton The sze o the nterval contanng the zero s reduced by 5% ater each teraton The number o teratons can be determned a pror No knowledge o the dervatve s needed The uncton does not have to be derentable Dsadvantage Slow to converge Good ntermedate appromatons may be dscarded 44

Newton-Raphson Method Also known as Newton s Method Gven an ntal guess o the root, Newton-Raphson method uses normaton about the uncton and ts dervatve at that pont to nd a better guess o the root. Assumptons: s contnuous and the rst dervatve s known An ntal guess such that s gven 45

Newton Raphson Method - Graphcal Depcton - I the ntal guess at the root s, then a tangent to the uncton o that s s etrapolated down to the -as to provde an estmate o the root at. 46

Dervaton o Newton s Method Gven: Queston Taylor Therorem : Fnd h h : such that ' A new guess an ntal How do we obtan a o guess the root : o h h the root o. better estmate ' h ' Newton Raphson Formula? 47

48 Newton s Method end n or Assumputon Gven ' : ', ', end X FP X F X X k or X PROGRAM MATLAB / :5 4 % X X FP X FP FP uncton X X F X F F uncton 6* ^ 3* ] [ ^ 3* ^3 ] [ F.m FP.m

49 Eample.3 9.74.369.4375 ' Iteraton 3:.4375 6 9 3 ' Iteraton : 3 33 33 4 ' Iteraton : 4 3 ' 4, 3 the uncton zero o Fnd a 3 3

Eample k Iteraton k k k k k k 4 33 33 3 3 9 6.4375.565.4375.369 9.74.3.45 3.3.564 6.844.756.384 4.756.65 6.4969.746. 5

5 Convergence Analyss ' mn ' ' ma such that ests then there '. where r at be contnuous ' ' ', Let Theorem : C C -r -r -r r I r and -r -r k k δ δ δ δ >

Proo 5 ' '' '' ' ' '' '! : Raphson - Newton ;! ], [ : about o epanson Taylor seres The r r r r r r r r r r r r ξ ξ ξ ξ

Convergence Analyss Remarks When the guess s close enough to a smple root o the uncton then Newton s method s guaranteed to converge quadratcally. Quadratc convergence means that the number o correct dgts s nearly doubled at each teraton. 53

Problems wth Newton s Method I the ntal guess o the root s ar rom the root the method may not converge. Newton s method converges lnearly near multple zeros { r r }. In such a case, moded algorthms can be used to regan the quadratc convergence. 54

Multple Roots 3 has three zeros at zeros has two at - 55

Problems wth Newton s Method - Runaway - The estmates o the root s gong away rom the root. 56

Problems wth Newton s Method - Flat Spot - The value o s zero, the algorthm als. I s very small then wll be very ar rom. 57

Problems wth Newton s Method - Cycle - 3 5 4 The algorthm cycles between two values and 58

59 Newton s Method or Systems o Non Lnear Equatons [ ] M M M ',,...,,..., ' ' root o the o guess an ntal : X F X F X F X F X X Iteraton s Newton F X Gven k k k k

6 Eample Solve the ollowng system o equatons:, guess Intal 5 5 y y y. y, 5 5 ', 5 5 X y F y y. y F

6 Soluton Usng Newton s Method.6.33 -.5.65 7.5.5.5.5.5 7.5.5.5 ', -.5.65 : Iteraton.5.5 5 6 6 5 5 ', 5 5 5 : Iteraton X F F. X y F. y y. y F

6 Eample Try ths Solve the ollowng system o equatons:, Intal guess y y y y, 4 ', X y F y y y F

63 Eample Soluton.98.557.98.557.969.587..6 5 4 3 X k Iteraton

Newton s Method Revew Assumptons :, ', Newton' s Method new estmate: Problem : ' ' s not avalable, or dcult to obtan analytcally. ' are avalable, 64

65 Secant Method ' ponts : are two ntal ' and h h

66 Secant Method New estmate Secant Method : ponts ntal Two Assumptons : that such and

67 Secant Method.5

68 Secant Method - Flowchart ;,, <ε Stop NO Yes

69 Moded Secant Method the method may dverge. properly, not selected I? How to select Problem : ' needed : guess s only one ntal moded Secant method, ths In δ δ δ δ δ δ δ

Eample 5 Fnd the roots o : Intal ponts 5 and 3 3. 4 3 - wth error <. - -3-4 - -.5 - -.5.5.5 7

Eample - -.. -.. -..585 -.6. 6 -.6. -.5.9 -.5. -.5. 7

Convergence Analyss The rate o convergence o the Secant method s super lnear: r r α C, α.6 r : root : estmate o the root at the th teraton. It s better than Bsecton method but not as good as Newton s method. 7

Summary Method Pros Cons Bsecton Newton Secant - Easy, Relable, Convergent - One uncton evaluaton per teraton - No knowledge o dervatve s needed - Fast near the root - Two uncton evaluatons per teraton - Fast slower than Newton - One uncton evaluaton per teraton - No knowledge o dervatve s needed - Slow - Needs an nterval [a,b] contanng the root,.e., ab< - May dverge - Needs dervatve and an ntal guess such that s nonzero - May dverge - Needs two ntal ponts guess, such that - s nonzero 73

74 Eample.5 ponts ntal Two : root o the nd to Secant method Use 6 and

Soluton k k k. -..5 8.896.56 -.76 3.836 -.4645 4.47.3 5.33 -.65 6.347 -.5 75

76 Eample.. or., or teratons, three ater Stop. : pont ntal the Use : root o a nd to Method Newton's Use 3 < < k k k

Fve Iteratons o the Soluton k k k k ERROR. -...5.875 5.75.5.3478.7 4.4499.6 3.35. 4.685.5 4.347. 4.646. 5.347. 4.646. 77

78 Eample.. or., or teratons, three ater Stop. : pont ntal Use the : root o a nd to Method Newton's Use < < k k k e

79 Eample -. -.567..567 -. -.567..567 -.9 -.584.46.5379.46 -.3679 -.63. ' ' ', : root o a nd to Method Newton's Use k k k k k e e

Eample Estmates o the root o: -cos..6 Intal guess.74473944598 correct dgt.73994768864 4 correct dgts.739853347 correct dgts.739853356 4 correct dgts 8

Eample In estmatng the root o: -cos, to get more than 3 correct dgts: 4 teratons o Newton.8 43 teratons o Bsecton method ntal nterval [.6,.8] 5 teratons o Secant method.6,.8 8