MATH 5010–001 SUMMER 2003 SOLUTIONS TO ASSIGNMENT 4 Problems from pp. 171–180 1. The possible values of X are −2, −1, 0, 1, 2, 4, and the corresponding probabilities are: 8 P {X = −2} = P {X = −1} = P {X = 0} = P {X = 1} = P {X = 2} = P {X = 4} = 17. • • • 2 14 , 2 8 2 1 1 14 , 2 2 2 14 , 2 4 8 1 1 14 , 2 4 2 1 1 14 , 2 4 2 14 . 2 (a) You may recall from your reading that P {X = i} = F (i) − F (i−). This gives: 1 1 P {X = 1} = F (1) − F (1−) = 12 + 1−1 4 − 4 = 4 ; 1 2−1 = 16 . P {X = 2} = F (2) − F (2−) = 11 12 − 2 + 4 1 P {X = 3} = F (3) − F (3−) = 1 − 11 12 = 12 . 17. (b) P { 12 < X < 32 } = F ( 32 −) − F ( 12 ) = 1 2 + (3/2)−1 4 − (1/2) 4 = 12 . 18. Note that X is a binomial with parameters n = 4 and p = 12 . Therefore, P {X − 2 = 4 for k = −2, −1, 0, 1, 2. k} = P {X = 2 + k} = k4 (1/2)2+k (1/2)n−(2+k) = 2−4 2+k 24. (a) If A chooses 1, and if the winnings of B are written as X, then P {X = 1} = p and P {X = −3/4} = 1 − p. Therefore, in this case, E(X) = p − (3/4)(1 − p) = 74 p − 34 . 24. (b) If A chooses 2, then P {X = 2} = 1 − p and P {X = −3/4} = p. Therefore, p. E(X) = 2(1 − p) − (3/4)p = 2 − 11 4 Since B does not know what A is going to choose, B chooses the p that makes the two expected-value lines of (a) and (b) intersect. (See the diagram given during the lectures.) That is, the p for which 74 p − 34 = 2 − 11 p; i.e., p = 11 . 4 18 As for player A, she expects to win q − 34 q if B chooses 1, and − 34 q + 2(1 − q) if B chooses 2. Therefore, the optimal choice for A is also where these two lines meet and that is q = 11 18 . 1 Theoretical Problems from pp. 180–184 13. Recall that n k k = 0, . . . , n. P {X = k} = p (1 − p)n−k , k As far as the variable p is concerned, nk is a constant. So we are asked to maximize f (p) = pk (1 − p)n−k . Of course, f 0 (p) = kpk−1 (1 − p)n−k − (n − k)pk (1 − p)n−k−1 . (1) So we need to analyse f and f 0 . We do this with some care: Case 1: k = 0. In this case, f 0 (p) = −n(1 − p)n−1 is always negative. This means that f is decreasing, so the maximzer is p = 0. Case 2: k = n. In this case, f 0 (p) = npn−1 is always positive. This means that f is increasing, so the maximizer is p = 1. Case 3: 1 ≤ k ≤ n − 1. In this case, f 0 (p) = 0 yields the equation, kpk−1 (1 − p)n−k = (n − k)pk (1 − p)n−k−1 . Because the maximizer is some p strictly inside (0, 1), we can divide by pk−1 (1 − p)n−k−1 and get p = (k/n). To check that this is indeed a maximizer, need to check that f 00 (k/n) < 0. But f 00 (p) = k(k − 1)pk−2 (1 − p)n−k − 2k(n − k)pk−1 (1 − p)n−k−1 + (n − k)(n − k − 1)pk (1 − p)n−k−2 . To check that f 00 (k/n) < 0, we can divide f 00 (p) by pk−2 (1 − p)n−k−2 , and then let p := (k/n) to check for the following inequality: 2 2 k k k k − 2k(n − k) < 0? k(k − 1) 1 − 1− + (n − k)(n − k − 1) n n n n A few lines of calculations show that the left-hand side is equal to −k + (k 2 /n). So f 00 (kn) < 0 if and only if (k 2 /n) < k; i.e., if k 2 < nk. Because 1 ≤ k ≤ n − 1, this is the case and we are done. 2