Lecture 2. Euler Equation and Legendre condition.pptVIP

Euler Equation andLegendre Condition Lecture Two * Three Methods ofDynamic Optimization The calculus of variations It can most easily be employed when all the functions describing the problem are smooth and the optimum is strictly inside the feasible region. Optimal control It is applicable in the instances above but can also accommodate boundary solutions. Dynamic programming It is especially useful in dealing with problems involving uncertainty and in differential games. * Simplest Problem Consider the problem The function F is assumed to be continuous in its three arguments t, x, x, and to have continuous partial derivatives with respect to x and x. The admissible class of functions x(t) consists of all continuously differentiable functions defined on the interval [t0, t1] satisfying the fixed endpoint conditions. * Euler Equation Suppose that the function x*(t), t0 ≤ t ≤ t1 provides the maximum of the problem. Let x(t) be some other admissible function, and define the functionh(t) = x(t) ? x*(t), then h(t0) = 0 and h(t1) = 0. The deviation h is admissible if the function x = x* + h is admissible. For any constant a, the function y(t) = x*(t) + ah(t) will also be admissible. * Euler Equation Given the function x* and the deviation h, the value of the integral in F for y(t) as a function of the parameter a, say g(a), is: Since x* is the maximization solution, the function g must assume its maximum at a = 0. This implies that g(0) = 0. * Euler Equation From the definition of g(a):where Fx and Fx, denote partial derivatives of F with respect to its second and third arguments, respectively. The expression is called the first variation. * Euler Equation Integrating the second term by parts Thus, the condition g(0) = 0 means * Euler Equation Recall that the function h was chosen arbitrarily, g(0) = 0 must be held for any choice of h. Thus This equation is called the Euler equation. Lemma: Suppose that g(t) is a given, continuous function def