SOLUTIONS ECONOMICS 172B Midterm exam, Wednesday, January 30, 2008 1. (18 minutes, 20 points) Recall but do not write the definitions of convex set, concave function, and upper contour set. Prove that every upper contour set of a concave function is a convex set, noting where the various definitions are used in the proof. Let f be any concave function. Let t be any real value and consider the upper contour set defined by t. Let x and y be any two points in the upper contour set, and let λ be any value in [0, 1]. In order to show that the upper contour set is convex, it must be shown that λx + (1 − λ)y is also a point of the upper contour set. By definition of the upper contour set, f (x) ≥ t and f (y) ≥ t ((*)) Because the function is concave f (λx + (1 − λ)y) ≥ λf (x) + (1 − λ)f (y) From relation (*) the right hand side of this relation is ≥ λt + (1 − λ)t = t. Therefore f (λx + (1 − λ)y) ≥ t and therefore λx + (1 − λ)y is in the upper contour set. Since these facts are true for all choices of concave f, real t, x, y, and λ in [0, 1], every upper contour set of a concave function is a convex set. The proof is much shorter if more mathematical notation is used and you repeat the obvious less frequently than I did. 2. (18 minutes, 20 points) Consider the bisection method of maximizing a concave function of one variable. In 100 words or fewer, explain how the method works. You may use as many diagrams or equations as you wish, or you may describe the spreadsheet that applies the method. Initialize by choosing xlow and xhigh so that f 0 (xlow ) > 0 and f 0 (xhigh ) < 0. By concavity of f , the maximum lies in the interval (xlow , xhigh ). Choose a small > 0 as the convergence criterion. Step: Let x0 be the midpoint of the interval (xlow , xhigh ). Evaluate f 0 (x0 ). If f 0 (x0 ) > 0 replace xlow by x0 .If f 0 (x0 ) < 0 replace xhigh by x0 . Now if xhigh − xlow < the iterations are complete. and you take the midpoint of the interval as the approximate solution. If xhigh −xlow ≥ repeat the step using, of course, the new interval (xlow , xhigh ). The maximum of f is always in the interval (xlow , xhigh ), which is progressively narrower. Eventually it must converge. (82 words) A diagram would be nice. f(x) xlow x’ x* xhigh 3. (9 minutes, 10 points) Consider the function f (x1 , x2 ) = −x21 − 3x22 . 8 (a) Write the gradient. µ ¶ ¶ µ f1 (x1 , x2 ) −2x1 ∇f (x1 , x2 ) = = f2 (x1 , x2 ) −6x2 (b) In 50 words or fewer, explain the significance of the gradient of a concave function. The gradient vector lies in the direction of steepest ascent. In seeking to maximize a concave function f by choice of x1 and x2 , it is useful to explore in that direction, that is to look at g(t) = f (x̄1 − 2x̄1 t, x̄2 − 6x̄2 t) One follows g (t) to its maximum and then starts over from that point. (46 words) 9