Union College Union | Digital Works Honors Theses Student Work 6-2012 The alcC ulus of Variations Erin Whitney Union College - Schenectady, NY Follow this and additional works at: https://digitalworks.union.edu/theses Part of the Logic and Foundations of Mathematics Commons, and the Mathematics Commons Recommended Citation Whitney, Erin, "The alcC ulus of Variations" (2012). Honors Theses. 920. https://digitalworks.union.edu/theses/920 This Open Access is brought to you for free and open access by the Student Work at Union | Digital Works. It has been accepted for inclusion in Honors Theses by an authorized administrator of Union | Digital Works. For more information, please contact [email protected]. The Calculus of Variations Advisor: Christina Tonnesen-Friedman Author: Erin Whitney Fall 2011 - Winter 2012 Chapter 1 An Introduction 1.1 Motivation The Calculus of Variations is a branch of mathematics that has commonly been used as a tool for other areas of study, including but not limited to physics, geometry, and economics. This area of mathematics involves finding the extrema of functionals: Definition 1.1.1. A functional is a mapping from a set of functions, S, to the real numbers, R. Calculus of Variations is used as a device for optimization problems that in- volve functionals, as opposed to elementary calculus involving functions. It is thus the case that we are working with extrema that are functions instead of vectors. These functionals often involve definite integrals. One application of this branch of mathematics is one in differential geome- try; finding the geodesic, or the shortest curve, i.e. the curve with minimum arclength, connecting any two given points. The most intuitive answer, and sometimes the correct answer, is a straight line between the two points. However, if the two points lie on a curved surface, a straight line might not be possible. It follows that we need another device to find the geodesics of this particular class of curves where a straight line is not the solution. The set up to find geodesics requires that we parametrize the surface with which we are working. We let the surface of interest be defined by Σ; and 3 let it be described by the position vector function r : σ ! R : Here, σ is a 2 nonempty, connected, open subset of R and (u; v) 2 σ: We use the follow- ing parametrization and assume that x; y; and z are smooth functions of 1 (u; v) 2 σ : r(u; v) = (x(u; v); y(u; v); z(u; v)); where r is a 1-1 and onto function of σ onto Σ: We refine our attention to smooth, simple curves, γ, on Σ: It follows that there is a curveγ ~ in σ that, under r, maps to γ: Assume thatγ ~ is parametrized by (u(t); v(t)) where t0 ≤ t ≤ t1: So, for the purpose of this motivating example, we will define the components of the first fundamental form and leave the details up to the reader, which can be found in reference [1]. We define the components as follows: 2 2 @r @r @r @r E = ;F = · ;G = : @u @u @v @v The arclength is thus defined by: Z t1 p L(γ) = Eu02 + 2F u0v0 + Gv02dt t0 where the functions u and v are functions of t: We will now demonstrate this abstract concept with the following concrete example: Example 1.1.1. We will now consider an example involving finding the geodesics on a unit sphere. First of all, we can parametrize an octant of a unit sphere, Σ: This parameterization is as follows: r(u; v) = (sin u cos v; sin u sin v; cos u) such that π π σ = f(u; v) j 0 < u < ; 0 < v < g: 2 2 We can then see that the components of the first fundamental form, E, F , and G, are as follows: E = j(cos u cos v; cos u sin v; − sin u)j2 = 1 F = (cos ucosv; cosusinv; − sin u) · (− sin u sin v; sin u cos v; 0) = 0 G = j(− sin u sin v; sin u cos v; 0)j2 = sin2 u: It then follows that the arclength is defined as Z t1 p L(γ) = u02 + sin2uv02dt: t0 We can use calculus of variations to determine the minimum of this ar- clength function as γ spans the set of all curves from γ(t0) = γ0 to γ(t1) = 2 γ1; that is, (u(t); v(t)) spans the set of all functions (u; v):[t0; t1] ! σ s.t. (u(t0); v(t0)) = (u0; v0) and (u(t1); v(t1)) = (u1; v1): The mathematics covered in this paper will make doing so possible [6]. In the remainder of the text, we will examine specific applications and prob- lems that utilize the calculus of variations. 1.2 Background Information Let X be a set of functions, a function space, defined on the real numbers, R. Next, we will define the infinite vector space that we are working within by use of the following definitions, which all come out of reference [6]: Definition 1.2.1. A normed vector space is a vector space in which a norm is defined. The norm is a real-valued function on the vector space whose value at f; denoted kfk; satisfies the following properties: (1) kfk ≥ 0 (2) kfk = 0 () f = 0 (3) kαfk = jαjkfk (4) kf + gk ≤ kfk + kgk: Definition 1.2.2. An inner product on a vector space Y is a function on Y × Y , denoted h·; ·i, such that the following statements hold 8 f; g; h 2 Y , 8 α 2 R: (1) hf; fi ≥ 0 (2) hf; fi = 0 () f = 0 (3) hf + g; hi = hf; gi + hg; hi (4) hf; gi = hg; fi (5) hαf; gi = αhf; gi: A vector space with an inner product structure leads us to the definition of a norm on that vector space, and this norm is defined as: jjfjj = phf; fi: Definition 1.2.3. A Banach space is a complete normed vector space, i.e., a normed vector space in which every Cauchy sequence in the space converges. Definition 1.2.4. A Hilbert Space is a special type of Banach space in that it has an inner product structure. 3 For the purpose of this paper, we will be working within Hilbert Spaces. (n) Note that the function space denoted by C [x0, x1] is the set of functions th with at least an n order continuous derivative on the interval [x0, x1]. This is indeed an infinite dimensional vector space. To view it as a subset of a Hilbert Space, one could introduce the following inner product: Z x1 f(x)g(x)dx = 0: x0 The following theorems, from reference [6], offer a brief review of topics covered in calculus and are necessary to arriving at the remainder of our results. Theorem 1.2.1. (Mean Value Theorem) Let x0; x1 2 R be such that x0 <x1. Let f be a function continuous in [x0, x1] and differentiable in (x0, 0 x1). Then 9 ξ such that x0 < ξ < x1 and f(x1) = f(x0) + (x1 − x0)f (ξ). Theorem 1.2.2. (Taylor's Theorem) Let f be a function with the first n (n+1) derivatives continuous in [x0; x1] such that f (x) exists 8 x 2 (x0; x1): Then, 9 a number ξ 2 (x0; x1) such that: (x − x )2 f(x ) =f(x ) + (x − x )f 0(x ) 1 0 f 00(x ) + ::: 1 0 1 0 0 2 0 (x − x )n (x − x )n+1 ··· + 1 0 f (n)(x ) + 1 0 f (n+1)(ξ) n! 0 (n + 1)! We define the set S as follows: (2n) (k) k (k) k S = fy 2 C [x0; x1] j y (x0) = y0 and y (x1) = y1 for k = 0; : : : ; n − 1g k k such that y0 and y1 are fixed constants. For functionsy ^ 2 S in an - neighborhood of a function y 2 S, that is, jjy^ − yjj < , 9 η 2 X s.t. y^ = y + η: We define the set H by: (n) H = fη 2 C [x0; x1] j y + η 2 Sg: We will examine functionals of the form Z x1 J(y) = f(x; y; y0; : : : ; y(n))dx: x0 4 Definition 1.2.5. Let J: X ! R be a functional defined on the func- tion space (X, k · k). Let S ⊂ X. The functional J is said to have a local maximum in S at y 2 S if 9 > 0 s.t. J(^y) − J(y) ≤ 0 8 y 2 S s.t. ky^ −y k< . The functional is said to have a local minimum in S at y 2 S if y is a local maximum in S for −J. Definition 1.2.6. The functional J is said to have a local extremum, extrema if plural, in S at y 2 S if it is either a local maximum or a local minimum. Remark: In this text, we will first consider the case in which n = 1; and then we will consider the case in which n = 2: 1.3 First Case: n=1 2 Theorem 1.3.1. Let J: C [x0, x1] ! R be a functional of the form: Z x1 J(y) = f(x; y; y0)dx; x0 where f has continuous partial derivatives of second order with respect to x, 0 y, y and x0 <x1. If y 2 S is an extremum for J, then @f d @f − = 0 (1.1) @y dx @y0 8 x 2 [x0, x1].