Introduction to the Calculus of Variations

Similar documents
Optimization using Calculus. Optimization of Functions of Multiple Variables subject to Equality Constraints

Lecture 9: Implicit function theorem, constrained extrema and Lagrange multipliers

Variational Principles

3 Applications of partial differentiation

Calculus of Variations Summer Term 2014

Math 201 Solutions to Assignment 1. 2ydy = x 2 dx. y = C 1 3 x3

Calculus of Variations Summer Term 2014

Lecture Notes for PHY 405 Classical Mechanics

Calculus of Variations

Calculus - II Multivariable Calculus. M.Thamban Nair. Department of Mathematics Indian Institute of Technology Madras

AP Calculus Testbank (Chapter 9) (Mr. Surowski)

Math 651 Introduction to Numerical Analysis I Fall SOLUTIONS: Homework Set 1

Advanced Calculus Math 127B, Winter 2005 Solutions: Final. nx2 1 + n 2 x, g n(x) = n2 x

ENGI Partial Differentiation Page y f x

Linearization and Extreme Values of Functions

A. MT-03, P Solve: (i) = 0. (ii) A. MT-03, P. 17, Solve : (i) + 4 = 0. (ii) A. MT-03, P. 16, Solve : (i)

(Most of the material presented in this chapter is taken from Thornton and Marion, Chap. 6)

Section Taylor and Maclaurin Series

Calculus of Variations and Computer Vision

Mathematics (Course B) Lent Term 2005 Examples Sheet 2

Math Refresher Course

MTH4101 CALCULUS II REVISION NOTES. 1. COMPLEX NUMBERS (Thomas Appendix 7 + lecture notes) ax 2 + bx + c = 0. x = b ± b 2 4ac 2a. i = 1.

Physics 129a Calculus of Variations Frank Porter Revision

Higher-order ordinary differential equations

Solutions: Problem Set 4 Math 201B, Winter 2007

McGill University April Calculus 3. Tuesday April 29, 2014 Solutions

Power Series. x n. Using the ratio test. n n + 1. x n+1 n 3. = lim x. lim n + 1. = 1 < x < 1. Then r = 1 and I = ( 1, 1) ( 1) n 1 x n.

Math 115 HW #5 Solutions

APPM 2350 FINAL EXAM FALL 2017

AP Calculus (BC) Chapter 9 Test No Calculator Section Name: Date: Period:

(3) Let Y be a totally bounded subset of a metric space X. Then the closure Y of Y

You can learn more about the services offered by the teaching center by visiting

1 Integration in many variables.

Calculus II Practice Test Problems for Chapter 7 Page 1 of 6

Paper 1 (Edexcel Version)

Optimization: Problem Set Solutions

Completion Date: Monday February 11, 2008

worked out from first principles by parameterizing the path, etc. If however C is a A path C is a simple closed path if and only if the starting point

a k 0, then k + 1 = 2 lim 1 + 1

A Short Essay on Variational Calculus

Mechanical Systems II. Method of Lagrange Multipliers

Ordinary Differential Equations (ODEs)

MATH529 Fundamentals of Optimization Constrained Optimization I

Math 152 Take Home Test 1

C2 Differential Equations : Computational Modeling and Simulation Instructor: Linwei Wang

Robotics. Islam S. M. Khalil. November 15, German University in Cairo

HOMEWORK 8 SOLUTIONS

MATH H53 : Final exam

1 Arithmetic calculations (calculator is not allowed)

Math 265H: Calculus III Practice Midterm II: Fall 2014

Implicit Functions, Curves and Surfaces

Practice problems for Exam 1. a b = (2) 2 + (4) 2 + ( 3) 2 = 29

MATH1231 CALCULUS. Session II Dr John Roberts (based on notes of A./Prof. Bruce Henry) Red Center Room 3065

49. Green s Theorem. The following table will help you plan your calculation accordingly. C is a simple closed loop 0 Use Green s Theorem

Example. Evaluate. 3x 2 4 x dx.

Taylor Series and stationary points

UNIVERSITY OF HOUSTON HIGH SCHOOL MATHEMATICS CONTEST Spring 2018 Calculus Test

Calculus I (Math 241) (In Progress)

HOMEWORK 7 SOLUTIONS

The University of British Columbia Midterm 1 Solutions - February 3, 2012 Mathematics 105, 2011W T2 Sections 204, 205, 206, 211.

Calculus of Variations Summer Term 2014

Find the indicated derivative. 1) Find y(4) if y = 3 sin x. A) y(4) = 3 cos x B) y(4) = 3 sin x C) y(4) = - 3 cos x D) y(4) = - 3 sin x

Book 4. June 2013 June 2014 June Name :

Multivariable Calculus and Matrix Algebra-Summer 2017

Calculus II Study Guide Fall 2015 Instructor: Barry McQuarrie Page 1 of 8

MATH Midterm 1 Sample 4

SECTION A. f(x) = ln(x). Sketch the graph of y = f(x), indicating the coordinates of any points where the graph crosses the axes.

5.9 Representations of Functions as a Power Series

Review for the Final Exam

AP Calculus 2004 AB FRQ Solutions

1MA6 Partial Differentiation and Multiple Integrals: I

d(x n, x) d(x n, x nk ) + d(x nk, x) where we chose any fixed k > N

minimize x subject to (x 2)(x 4) u,

ECM Calculus and Geometry. Revision Notes

Strauss PDEs 2e: Section Exercise 1 Page 1 of 6

MATH 162. Midterm 2 ANSWERS November 18, 2005

DEPARTMENT OF MATHEMATICS AND STATISTICS UNIVERSITY OF MASSACHUSETTS. MATH 233 SOME SOLUTIONS TO EXAM 2 Fall 2018

Math 210, Final Exam, Spring 2012 Problem 1 Solution. (a) Find an equation of the plane passing through the tips of u, v, and w.

Advanced Dynamics. - Lecture 4 Lagrange Equations. Paolo Tiso Spring Semester 2017 ETH Zürich

Calculus III: Practice Final

Duality, Dual Variational Principles

2.2 Separable Equations

Cálculo de Variaciones I Tarea # 3

Chapter 2: Differentiation

TAYLOR AND MACLAURIN SERIES

18.02 Multivariable Calculus Fall 2007

Calculus of Variations

AP Calculus BC Spring Final Part IA. Calculator NOT Allowed. Name:

Partial Derivatives. w = f(x, y, z).

Maxima/minima with constraints

This is example 3 on page 44 of BGH and example (b) on page 66 of Troutman.

MATHEMATICS FOR ENGINEERING DIFFERENTIATION TUTORIAL 2 ADVANCED DIFFERENTIATION

Parametric Equations and Polar Coordinates

Lecture notes for Math Multivariable Calculus

Lagrange Multipliers

Numerical Optimization

Multivariable Calculus Notes. Faraad Armwood. Fall: Chapter 1: Vectors, Dot Product, Cross Product, Planes, Cylindrical & Spherical Coordinates

(x 3)(x + 5) = (x 3)(x 1) = x + 5. sin 2 x e ax bx 1 = 1 2. lim

CP1 REVISION LECTURE 3 INTRODUCTION TO CLASSICAL MECHANICS. Prof. N. Harnew University of Oxford TT 2017

Physics 5153 Classical Mechanics. Canonical Transformations-1

Transcription:

236861 Numerical Geometry of Images Tutorial 1 Introduction to the Calculus of Variations Alex Bronstein c 2005 1

Calculus Calculus of variations 1. Function Functional f : R n R Example: f(x, y) =x 2 + y 2 f : F R, in particular f(u) = Ω F (x, u(x),u (x),...) dx Example: f (u) = Ω u(x, y) 2 dxdy 2. Derivative Variation df (x) dx = lim f(x + x) f(x) δf(u) x 0 x δu = lim f(u + ɛδu) f(u) ɛ 0 ɛ df = ɛ f(x + ɛ x) ɛ=0 δf = ɛ f(u + ɛδu) ɛ=0 2

Calculus Calculus of variations 3. Local (relative) minimum f(x ) f(x) f(u ) f(u) x : x x α u : max x Ω u(x) u (x) α 4. Necessary condition for local extremum df (x) dx =0 δf(u) δu =0 5. Sufficient condition for local extremum d 2 f(x) dx 2 0 More complex theory 3

Calculus Calculus of variations 3. Constrained local minimum min f(x) min F (x, u(x),u (x))dx x u(x) Ω s.t. g(x) = 0 s.t. G(x, u(x),u (x)) = 0 4. Lagrangian l(x) =f(x)+λg(x) l(u) = Ω (F + λ(x)g) dx 5. Method of Lagrange multipliers dl(x ) dx =0 δl(u ) δu =0 g(x) =0 G(x, u (x),u (x)) = 0 4

Examples of functionals 1. Curve length L(y) = x1 x 0 1+y 2 dx The length of a non-parametric curve y(x). 1 2. Curve length L(x, y) = x 2 + y 2 dt The length of a parametric curve (x(t),y(t)). 3. Surface area A(z) = 1+zx 2 + zydxdy 2 S The area of a non-parametric surface S =(x, y, z(x, y)). 4. Total variation TV(y) = 0 x1 x 0 y dx The oscillation strength of a non-parametric curve y(x). 5

The Euler-Lagrange equation Let us be given the functional f(u) = x1 x 0 F (x, u(x),u (x)) dx with F C 3 and all admissible u(x) having fixed boundary values u(x 0 )=u 0 and u(x 1 )=u 1. An extremum of f(u) satisfies the differential equation F u d dx F u = 0 with the boundary conditions u(x 0 )=u 0 and u(x 1 )=u 1. 6

The second Euler-Lagrange equation A second, less known Euler-Lagrange equation is satisfied along u (x). d dx (F u F u ) F x = 0 7

The E-L equation (independent on x) If the integrand does not depend on the independent variable x, f(u) = x1 x 0 F (u(x),u (x)) dx, the second E-L equation becomes a first-order differential equation or d dx (F u F u ) = 0 F u F u = const. 8

The E-L equation (independent on u(x)) If the integrand does not depend on u(x), f(u) = x1 x 0 F (u(x),u (x)) dx, the E-L equation becomes a first-order differential equation or d dx F u = 0 F u = const. 9

The E-L equation (independent on u (x)) If the integrand does not depend on u (x), f(u) = x1 x 0 F (u(x),u (x)) dx, the E-L equation becomes an algebraic equation F u (x, u(x)) = 0. 10

The E-L equation (high-order functionals) Given the functional f(u) = x1 x 0 F ( x, u(x),u (x),..., u (n) (x) ) dx with F C n+2 and fixed boundary values u (i) (x 0 )=u (i) 0 and u (i) (x 1 )=u (i) 1 for i =0,..., n 1. The Euler-Lagrange equation (also known as the Euler-Poisson equation) is F u d dx F u + d2 dx F dn 2 u +( 1)n dx F n u (n) = 0. 11

The E-L equation (multiple independent variables) Given the functional f(u) = with x R n and u( Ω) = u Ω. Ω F (x, u(x),u x1 (x),..., u xn (x)) dx An extremum of f(u) satisfies the differential equation F u x 1 F u x1... x n with the boundary condition u( Ω) = u Ω. F u xn = 0 12

The E-L equation (multiple functions) Given the functional f(u) = x1 x 0 F (x, u 1 (x),..., u n (x),u 1(x),..., u n(x)) dx with F C 3 and all admissible u(x) having fixed boundary values u i (x 0 )=u 0 i and u i (x 1 )=u 1 i. An extremum of f(u) satisfies the system of differential equations F ui d dx F u = 0 i with the boundary conditions u(x 0 )=u 0 and u(x 1 )=u 1. 13

Problem 1: Minimum distance on a plane Prove that the family of curves minimizing the distance in the plane are straight lines. L = 1 1 1+y 2 dx, Solution The Euler-Lagrange equation Integration w.r.t. x yields 0 = d dx F y F y = d dx ( ) y. 1+y 2 y 1+y 2 = γ = const. 14

Problem 1: Minimum distance on a plane (cont.) Substitute y = tan θ: y 1+y 2 = sin θ cos θ 1+ sin2 θ cos 2 θ = sin θ cos θ sin 2 +cos 2 θ cos 2 θ = sin θ cos 2 cos θ 1 = sinθ, from where sin θ = γ. Substituting again yields y = tan θ = sin θ cos θ = ± sin θ 1 sin 2 θ = ±γ 1 γ = α = const, from where y = αx + β, β = const. The solution describes a line in the plane; the exact values of α, β are determined from the endpoint values. 15

Problem 2: Constrained maximum Find a curve y(x) with fixed endpoints y(±1) = 0 and perimeter which maximizes the area S = 1 1 A(y) = 1+y 2 dx, 1 1 ydx. Solution Construct the Lagrangian L(y) = and denote 1 1 ( y + λ ) 1+y 2 dx + const, F (x, y, y ) = y + λ 1+y 2. 16

Problem 2: Constrained maximum (cont.) The Euler-Lagrange equation Integration w.r.t. x yields 0 = d dx F y F y = d dx ( ) λy 1+y 2 1. λy 1+y 2 = x α, where α = const. Substitute y = tan θ: y λ 1+y 2 = λ sin θ cos θ 1+ sin2 θ cos 2 θ = λ sin θ cos θ sin 2 +cos 2 θ cos 2 θ = λ sin θ cos 2 cos θ 1 = λ sin θ, from where sin θ = x α λ. 17

Problem 2: Constrained maximum (cont.) Substituting again y = tan θ = sin θ cos θ = ± sin θ 1 sin 2 θ = ±(x α) λ 1 (x α)2 λ 2 = ±(x α) λ2 (x α) 2. Integration w.r.t. x yields (table of integrals or Mathematica) y = λ 2 (x α) 2 + β 2, or (x α) 2 +(y β) 2 = λ 2 where β = const. The latter equation describes a part of a circle with radius λ centered at (α, β). The exact values of the constants are determined by demanding the endpoint conditions and the perimeter constraint. 18