Numerical Solution of Non-Linear Ordinary Differential Equations via Collocation Method (Finite Elements) and Genetic Algorithms

Similar documents
x x x 2x x N ( ) p NUMERICAL METHODS UNIT-I-SOLUTION OF EQUATIONS AND EIGENVALUE PROBLEMS By Newton-Raphson formula

A Pseudo Spline Methods for Solving an Initial Value Problem of Ordinary Differential Equation

Comparison Study of Series Approximation. and Convergence between Chebyshev. and Legendre Series

Numerical Solutions of Second Order Boundary Value Problems by Galerkin Residual Method on Using Legendre Polynomials

A New Hybrid in the Nonlinear Part of Adomian Decomposition Method for Initial Value Problem of Ordinary Differential Equation

Chapter 2: Numerical Methods

Solving third order boundary value problem with fifth order block method

Research Article A New Second-Order Iteration Method for Solving Nonlinear Equations

Finite Difference Method for the Estimation of a Heat Source Dependent on Time Variable ABSTRACT

Recurrence Relations

Chapter 2 The Solution of Numerical Algebraic and Transcendental Equations

wavelet collocation method for solving integro-differential equation.

ALLOCATING SAMPLE TO STRATA PROPORTIONAL TO AGGREGATE MEASURE OF SIZE WITH BOTH UPPER AND LOWER BOUNDS ON THE NUMBER OF UNITS IN EACH STRATUM

The Method of Least Squares. To understand least squares fitting of data.

Chapter 4. Fourier Series

POWER SERIES SOLUTION OF FIRST ORDER MATRIX DIFFERENTIAL EQUATIONS

A NEW CLASS OF 2-STEP RATIONAL MULTISTEP METHODS

Application of a Two-Step Third-Derivative Block Method for Starting Numerov Method

On Exact Finite-Difference Scheme for Numerical Solution of Initial Value Problems in Ordinary Differential Equations.

Solution of Differential Equation from the Transform Technique

DECOMPOSITION METHOD FOR SOLVING A SYSTEM OF THIRD-ORDER BOUNDARY VALUE PROBLEMS. Park Road, Islamabad, Pakistan

METHOD OF FUNDAMENTAL SOLUTIONS FOR HELMHOLTZ EIGENVALUE PROBLEMS IN ELLIPTICAL DOMAINS

THE ENERGY BALANCE ERROR FOR CIRCUIT TRANSIENT ANALYSIS

On the convergence, consistence and stability of a standard finite difference scheme

ME 501A Seminar in Engineering Analysis Page 1

NUMERICAL DIFFERENTIAL 1

Computation of Hahn Moments for Large Size Images

PC5215 Numerical Recipes with Applications - Review Problems

Using Spreadsheets as a Computational Tool in Teaching Mechanical. Engineering

Numerical Solution of the Two Point Boundary Value Problems By Using Wavelet Bases of Hermite Cubic Spline Wavelets

Stability analysis of numerical methods for stochastic systems with additive noise

(c) Write, but do not evaluate, an integral expression for the volume of the solid generated when R is

8. Applications To Linear Differential Equations

62. Power series Definition 16. (Power series) Given a sequence {c n }, the series. c n x n = c 0 + c 1 x + c 2 x 2 + c 3 x 3 +

Information-based Feature Selection

1. Introduction. 2. Numerical Methods

M.Jayalakshmi and P. Pandian Department of Mathematics, School of Advanced Sciences, VIT University, Vellore-14, India.

LIMITS AND DERIVATIVES

Teaching Mathematics Concepts via Computer Algebra Systems

Uniform Strict Practical Stability Criteria for Impulsive Functional Differential Equations

Definitions and Theorems. where x are the decision variables. c, b, and a are constant coefficients.

A Block Cipher Using Linear Congruences

Analysis of Analytical and Numerical Methods of Epidemic Models

Stability Analysis of the Euler Discretization for SIR Epidemic Model

Precalculus MATH Sections 3.1, 3.2, 3.3. Exponential, Logistic and Logarithmic Functions

University of Washington Department of Chemistry Chemistry 453 Winter Quarter 2015

Recursive Algorithm for Generating Partitions of an Integer. 1 Preliminary

Five Steps Block Predictor-Block Corrector Method for the Solution of ( )

Stability of fractional positive nonlinear systems

LIMITS AND DERIVATIVES NCERT

Partial Differential Equations

A NUMERICAL METHOD OF SOLVING CAUCHY PROBLEM FOR DIFFERENTIAL EQUATIONS BASED ON A LINEAR APPROXIMATION

Castiel, Supernatural, Season 6, Episode 18

Taylor polynomial solution of difference equation with constant coefficients via time scales calculus

A widely used display of protein shapes is based on the coordinates of the alpha carbons - - C α

Similarity Solutions to Unsteady Pseudoplastic. Flow Near a Moving Wall

Most text will write ordinary derivatives using either Leibniz notation 2 3. y + 5y= e and y y. xx tt t

Exact Solutions for a Class of Nonlinear Singular Two-Point Boundary Value Problems: The Decomposition Method

Finite Difference Approximation for First- Order Hyperbolic Partial Differential Equation Arising in Neuronal Variability with Shifts

Application of Homotopy Analysis Method for Solving various types of Problems of Ordinary Differential Equations

μ are complex parameters. Other

Determination of Energy Involved In a Stepwise Size Reduction of Maize, Using a Numerical Approach.

Finite Difference Approximation for Transport Equation with Shifts Arising in Neuronal Variability

Numerical Conformal Mapping via a Fredholm Integral Equation using Fourier Method ABSTRACT INTRODUCTION

TMA4205 Numerical Linear Algebra. The Poisson problem in R 2 : diagonalization methods

Subject: Differential Equations & Mathematical Modeling -III. Lesson: Power series solutions of Differential Equations. about ordinary points

A numerical Technique Finite Volume Method for Solving Diffusion 2D Problem

Scientific Research of the Institute of Mathematics and Computer Science

IP Reference guide for integer programming formulations.

The Numerical Solution of Singular Fredholm Integral Equations of the Second Kind

DETERMINATION OF MECHANICAL PROPERTIES OF A NON- UNIFORM BEAM USING THE MEASUREMENT OF THE EXCITED LONGITUDINAL ELASTIC VIBRATIONS.

CS537. Numerical Analysis and Computing

NUMERICAL METHOD FOR SINGULARLY PERTURBED DELAY PARABOLIC PARTIAL DIFFERENTIAL EQUATIONS

THE SOLUTION OF NONLINEAR EQUATIONS f( x ) = 0.

An Improved Self-Starting Implicit Hybrid Method

Subject: Differential Equations & Mathematical Modeling-III

AP Calculus BC Review Applications of Derivatives (Chapter 4) and f,

A New Solution Method for the Finite-Horizon Discrete-Time EOQ Problem

Find quadratic function which pass through the following points (0,1),(1,1),(2, 3)... 11

Recursive Algorithms. Recurrences. Recursive Algorithms Analysis

Research Article Some E-J Generalized Hausdorff Matrices Not of Type M

Newton Homotopy Solution for Nonlinear Equations Using Maple14. Abstract

Linear Differential Equations of Higher Order Basic Theory: Initial-Value Problems d y d y dy

Numerical Method for Blasius Equation on an infinite Interval

Introduction to Optimization Techniques. How to Solve Equations

(a) (b) All real numbers. (c) All real numbers. (d) None. to show the. (a) 3. (b) [ 7, 1) (c) ( 7, 1) (d) At x = 7. (a) (b)

Chapter 7 Isoperimetric problem

Algebra of Least Squares

LECTURE 2 LEAST SQUARES CROSS-VALIDATION FOR KERNEL DENSITY ESTIMATION

A) is empty. B) is a finite set. C) can be a countably infinite set. D) can be an uncountable set.

The Choquet Integral with Respect to Fuzzy-Valued Set Functions

Two-step Extrapolated Newton s Method with High Efficiency Index

Section 14. Simple linear regression.

Celestin Chameni Nembua University of YaoundéII, Cameroon. Abstract

DEGENERACY AND ALL THAT

Warped, Chirp Z-Transform: Radar Signal Processing

Variable selection in principal components analysis of qualitative data using the accelerated ALS algorithm

PAPER : IIT-JAM 2010

Period Function of a Lienard Equation

lim za n n = z lim a n n.

Transcription:

Proceedigs of te 6t WSEAS It. Cof. o EVOLUTIONARY COPUTING Lisbo Portugal Jue 6-8 5 pp6- Numerical Solutio of No-Liear Ordiary Differetial Equatios via Collocatio etod Fiite Elemets ad Geetic Algoritms Nikos E. astorakis ilitary Istitutes of Uiversity Educatio ASEI Helleic Naval Academy Terma Hatzikyriakou 859 Piraeus GREECE ttp://www.wseas.org/mastorakis Abstract: - I tis paper a ew metod for solvig o-liear ordiary differetial equatios is proposed. Te metod is based o fiite elemets collocatio metod as well as o geetic algoritms. Te metod seems to ave some advatages i compariso wit te typical sequetial oe step ad multi step metods. Key-Words: - Ordiary Differetial Equatios Fiite Elemets Geetic Algoritms Evolutioary Computig Collocatio Itroductio Researc i umerical solutio of Ordiary Differetial Equatios ODEs is a ope field durig te last ceturies ad may umerical metods ave bee adopted to solve iitial value problems. Te importace of ODE is great because matematical models tat occur i sciece pysics cemistry biology ecoomy geology space sciece etc ad egieerig electrical mecaical civil etc are usually ordiary differetial equatios. Euler s metod Heu s metod Polygo metod Ruge Kutta metods are te most usual oe step metods wile ile Simpso Adams Basfort oulto metods are te most usual multi step metod. Suppose tat a first order dy ODE is epressed i te form: f y d Commo feature i oe step metods is te discovery ad te usage of a discrete sequece y i yi F i yi were F is a fuctio or procedure or algoritm i wic i ad y i are ivolved. Commo feature i te so called multistep metods is te discovery ad te usage of a discrete sequece yi yi F i yi i yi i yi L ik yik Te previous formula describes a k step metod. ore details ca be foud i may tetbooks ad surveys like [] []. A disadvatage of all tese metods is tat te umerical errors are accumulated ad so te solutio of te iitial value problems is affected by tese errors. Especially we te iterval of solutio is big we ave to use small step of icrease wic affects te compleity of te metod. Terefore to icrease accuracy we must decrease te step of te metod wic uavoidably leads to big computatioal load ad compleity. I tis paper a attempt is made to solve tis problem by usig fiite elemets collocatio metod ad geetic algoritms. Fiite elemets metod yields a set of o liear algebraic equatios. Tese o liear equatios ca be solved via Geetic Algoritm GA []. Te solutio of te o liear algebraic equatios via GAs is eamied i [] wit more details ad specific eamples. I sectio we preset te mai fiite elemets. I sectio a formulatio of te geeral iitial value problem is give. I sectio a formulatio of te geeral boudary value problem is preseted. Fially some cocludig remarks are provided i 5. Overview of te ai Fiite Elemets Fiite Elemets are bases of fuctios suc tat te solutio of a iitial or boudary value problem to be able to be epressed as a liear combiatio of tese fuctios. Below we remid te most commo fiite elemets. Fiite elemets are basis fuctios tat are ozero oly i small regios te elemets.

Suppose tat we ave to solve te followig iitial value problem over te iterval ] [ y f d dy wit te iitial coditio y y.a. Fiite Elemets: Piecewise Liear Fuctios at fuctios I te iterval ] [ we cosider te poits: L as well as te fuctios... It is proved tat te fuctios compose a basis of te space of fuctios tat are cotiuous i te iterval ] [. I applicatios we cosider te solutio of our No-Liear Ordiary Differetial Equatios as a liear combiatio of te fiite subset L of tis basis. Usually te poits L are equispaced. So 5. 5 5.. Fiite Elemets: Piecewise Square Fuctios I te iterval ] [ we cosider te poits: L We defie We also cosider Proceedigs of te 6t WSEAS It. Cof. o EVOLUTIONARY COPUTING Lisbo Portugal Jue 6-8 5 pp6-

6 ad 7 Especially ad. It is proved tat te fuctios... L compose a basis of te space of fuctios tat are cotiuous i te iterval ] [. I applicatios we cosider te solutio of our No-Liear Ordiary Differetial Equatios as a liear combiatio of te fiite subset L of tis basis.. Fiite Elemets: Hermite Fuctios Te basic Hermite fuctios are defied as follows 8 9 For L we defie as well as... Proceedigs of te 6t WSEAS It. Cof. o EVOLUTIONARY COPUTING Lisbo Portugal Jue 6-8 5 pp6-

. We cosider agai te solutio of te differetial equatio as a liear combiatio of te fiite subset L of tis basis.. Fiite Elemets: Splies Fuctios Te basic procedure splie fuctio is defied as follows ] [ ] [ So te Splie basis is L were L. Obviously a sligt modificatio is eeded for Hece... We ca cosider agai te solutio of te differetial equatio as a liear combiatio of te fiite subset... of tis basis. Formulatio ad Solutio of te Geeral Iitial Value Problem via Collocatio ad Geetic Algoritm Cosider te geeral first order iitial problem. y f d dy y y.a Suppose tat { } is a subset of te basis of fuctios like. or. or. or. o wic our solutio is epaded. Te our solutio ca be epressed as a liear combiatio c y Proceedigs of te 6t WSEAS It. Cof. o EVOLUTIONARY COPUTING Lisbo Portugal Jue 6-8 5 pp6-

Proceedigs of te 6t WSEAS It. Cof. o EVOLUTIONARY COPUTING Lisbo Portugal Jue 6-8 5 pp6- I te metod of collocatio [] we select poits i te iterval ] L. [ Tese poits ca be equispaced or o equispaced. Te we ave te equatios y c 5. c 5. f c 5. f We ca solve te Equatio 5. wit respect to oe c say c l were l {... }. Substitutig tis c l ito 5. 5. we obtai a system of o liear equatios i ukows: c c... cl cl... c. Tese equatios are umbered as 5a. 5a.. Please ote tat tese equatios do ot cotai c l c 5a. f c 5a. f Te solutio of 5a. 5a. ca be obtaied via a Geetic Algoritm followig te metod of [6]. Te Geetic Algoritm is used to elimiate a error i 5a. 5a.. Te reaso tat we solve 5. wit respect to c l as well as te reaso for eecutig te Geetic Algoritm o 5a. 5a. istead of 5.5. 5. is tat we must satisfy te iitial coditio.a wit zero error. A brief overview of te GAs metodology could be te followig: Suppose tat we ave to maimize miimize te fuctio Q wic is ot ecessary cotiuous or differetiable. GAs are searc algoritms wic iitially were isiped by te process of atural geetics reproductio of a origial populatio performace of crossover ad mutatio selectio of te best. Te mai idea for a optimizatio problem is to start our searc o wit oe iitial poit but wit a populatio of iitial poits. Te umbers poits of tis iitial set called populatio quite aalogously to biological systems are coverted to te biary system. I te sequel tey are cosidered as cromosomes actually sequeces of ad. Te et step is to form pairs of tese poits wo will be cosidered as parets for a "reproductio" see te followig figure...... parets... }... cildre "Parets" come to "reproductio" were tey itercage parts of teir "geetic material". Tis is acieved by te so-called crossover see te previous figure wereas always a very small probability for a utatio eists. utatio is te peomeo were quite radomly - wit a very small probability toug - a becomes or a becomes. Assume tat every pair of "parets" gives k cildre. By te reproductio te populatio of te "parets" are eaced by te "cildre" ad we ave a icreasemet of te origial populatio because ew members were added parets always belog to te cosidered populatio. Te ew populatio as ow k members. Te te process of atural selectio is applied. Accordig te cocept of atural selectio from te k members oly survive. Tese members are selected as te members wit te iger values of QQ if we attempt to acieve maimizatio of Q or wit te lower values of Q if we attempt to acieve miimizatio of QQ. By repeated iteratios of reproductio uder crossover ad mutatio ad atural selectio we ca fid te miimum or maimum of Q as te poit to wic te best values of our populatio coverge. Te termiatio criterio is fulfilled if te mea value of Q i te -members populatio is o loger improved maimized or miimized. ore detailed overviews of GAs ca be foud i [] [] [] ad [].

Proceedigs of te 6t WSEAS It. Cof. o EVOLUTIONARY COPUTING Lisbo Portugal Jue 6-8 5 pp6- So we we ave to solve a system of equatios i ukow variables. f... f f...... Te square fuctio Q f f L f or te absolute value fuctio Q f f L f are defied or i r geeral ay suitable orm of f f f... f ad our problem is mi Q If te global miimum of Q is at te poit * * * * * * L te L is a solutio of te aforemetioed systems of o-liear equatios. Te metod as also bee used i []. Formulatio ad Solutio of te Geeral Boudary Value Problem via Collocatio ad Geetic Algoritm Cosider te geeral secod order iitial value problem. d y f y y 6 d dy d dy d ya 6.a y b 6.b Suppose tat { } is a subset of te basis of fuctios like. or. or. or. o wic our solutio is epaded as follows: y c. We ca use te metod of collocatio agai i te iterval ]. [ Te followig system of o liear equatios is obtaied y c 7. c f c 7. f c c 7. y 7. We ca solve te systems of te Equatio 7. 7. wit respect to two variables say c l c l were l l {... }. Substitutig tese epressios of c l c l ito 7. 5. we obtai a system of o liear equatios i ukows: c c... cl...... cl cl cl c. Tese equatios ew equatios after te elimiatio of c l c l are umbered as 7a. 7a.. Please ote tat tese equatios do ot cotai c c l l c f c 7a. f c 7a. Te solutio of 7a. 7a. ca be obtaied via a Geetic Algoritm.. Te Geetic Algoritm is used to elimiate a error i 7a. 7a.. Te reaso tat we solve te 7. ad 7. wit respect to c l c l as well as te reaso for eecutig te Geetic Algoritm i 7a. 7a. istead of 7.7. 7.7. is tat we must

Proceedigs of te 6t WSEAS It. Cof. o EVOLUTIONARY COPUTING Lisbo Portugal Jue 6-8 5 pp6- satisfy te boudary value coditios 6.a ad 6.b wit zero error. 5 Cocludig Remarks ad Future Researc As GAs is a powerful tool for te solutio of systems of o liear equatios tey ca fid applicatios i te solutio of o liear Ordiary Differetial Equatios. Te collocatio metod is used ad a system of o liear equatios is obtaied. Tis system is solved by GAs. Numerical eamples ca outlie te validity ad efficiecy of our proposed metod. usig Geetic Algoritms IEEE Trasactios o Circuits ad Systems I: Fudametal Teory ad Applicatios Part I Vol. 5 No. 5 pp. 6-69 ay. [] astorakis N.E. Goos I.F. Swamy.N.S.: Stability of ultidimesioal Systems usig Geetic Algoritms IEEE Trasactios o Circuits ad Systems Part I Vol. 5 No. 7 pp. 96-965 July. [] Neil Gersefeld 999 Te Nature of atematical odelig Cambridge Uiversity Press. Refereces: [] Goldberg D.E. 989 Geetic Algoritms i Searc Optimizatio ad acie Learig Addiso-Wesley Secod Editio 989 [] Grefestette J.J. Optimizatio of cotrol parameters for Geetic Algoritms IEEE Tras. Systems a ad Cyberetics SC 6 Ja/Feb 986 pp. 8 [] Eberart R. Simpso P. ad Dobbis R. 996 Computatioal Itelligece PC Tools AP Professioals. [] Kosters W.A. Kok J.N. ad Floree P. Fourier Aalysis of Geetic Algoritms Teoretical Computer Sciece Elsevier 9 99 pp. - 75. [5] E. Balagusuramy Numerical etods Tata cgraw Hill New Deli 999 [6] Nikos E. astorakis Solvig No-liear Equatios via Geetic Algoritms Proceedigs of te 6t WSEAS Iteratioal Coferece o Evolutioary Computig Lisbo Portugal Jue 6-8 5. [7] Ioais F. Goos Lefteris I. Virirakis Nikos E. astorakis.n.s. Swamy "Evolutioary Desig of -Dimesioal Recursive Filters via te Computer Laguage GENETICA" to appear i IEEE Trasactios o Circuits ad Systems I: Fudametal Teory ad Applicatios. 5 [8] Goos I.F. astorakis N.E. Swamy.N.S.: A Geetic Algoritm Approac to te Problem of Factorizatio of Geeral ultidimesioal Polyomials IEEE Trasactios o Circuits ad Systems I: Fudametal Teory ad Applicatios Part I Vol. 5 No. pp. 6- Jauary. [9] astorakis N.E. Goos I.F. Swamy.N.S.: Desig of -Dimesioal Recursive Filters