武大讲义dynamicoptimizationiii1.docVIP

III-1 Control Theory: Basic Techniques We are usually convinced more easily by reasons we have found ourselves than by those which have occurred to others. – Pascal 1670 Optimal control theory which is a modern generalization of the classical calculus of variations theory that offers a unified method for treating very general dynamic optimization problems in continuous time. This new theory is now used by a large number of economists, even when they face a problem that can be solved with the calculus of variations. The calculus of variations actually has a rather long history. Some of its main results were established by Euler and Lagrange as early as in the 18th century. In economics, some of its first applications were by Ramsey 1928 to an optimal savings problem, and by Hotelling 1931 to a problem of how to extract a natural resource. 1 The Basic Problem Consider a system whose state at time is characterized by a number , the state variable. The process that causes to change can be controlled, at least partially, by a control function . We assume that the rate of change of depends on , and . The state at some initial point is typically known, . Hence the revolution of is described by a controlled differential equation 1 Suppose we choose some control function defined for . Inserting this function into 1 gives a first-order differential equation for alone. Because the initial point is fixed, a unique solution of 1 is usually obtained. By choosing different control functions , the system can be steered along many different paths, not all of which are equally desirable. As usual in economic analysis, assume that it is possible to measure the benefits associated with each path. More specifically, assume that the benefits can be measured by means of the integral 2 where is a given function. Here, is called the objective function or the criterion function. Certain restrictions are often placed on the final state . Moreover, the time at which the pr