Math 5010-1, Spring 2005 Assignment 5 Problems #31, p. 175. Let R denote the event that it will rain tomorrow. The metereologist truly feels that P (R) = p∗ . Now, suppose we are told that P (R) = p (when, in fact, it is p∗ ). Then, we give weight 1 − (1 − p)2 if it rains, and 1 − p2 otherwise. So, if W denotes the weight we assigned to the metereologist, then E[W ] = 1 − (1 − p)2 P (R) + 1 − p2 P (Rc ) = 1 − (1 − p)2 p∗ + 1 − p2 (1 − p∗ ). Call this h(p). The metereologist’s goal is to report a p that maximizes h(p). We can use calculus to do this: Remembering that p is the variable and p∗ ∈ [0, 1] is a fixed constant, we have h0 (p) = 2(1 − p)p∗ − 2p(1 − p∗ ) = 2p∗ − 2p, h00 (p) = −2. So the answer is p = p∗ . Please note the correction to the previous solution. #33, p. 175. Let M denote the number of papers that the paperboy buys each day. Also let X denote the paperboy’s demand, so that X is Bin(n = 10, p = 1/3). The paperboy’s profit is the following random variable: Z = 15 min(X, M ) − 10M. [The min has to be there, because the paperboy can sell at most M copies now that he has bought M copies.] Therefore, f (M ) = E[Z] = 15E [min (X, M )] − 10M. (1) This is too hard to compute. So it pays to be tricky: Note that min(X, M ) ≤ X. Therefore, E[min(X, M )] ≤ E[X] = np = 10/3. Therefore, f (M ) ≤ 15(10/3) − 10M = 50 − 10M . Also, min(X, M ) ≤ M , so that f (M ) ≤ 15M − 10M = 5M . Therefore, f (M ) ≤ min (50 − 10M, 5M ) = g(M ). Therefore, max f ≤ max g, which happens where 50 − 10M = 5M (why?). Equivalently, where M = 50/15. Because M is an integer the closest is M = 3. That is, we have proved that maxM f (M ) ≤ g(3) = 15. To finish we must show that maxM f (M ) is not too far below 15. If so, then M = 3 is (nearly) optimal. But i 10−i 10 i 10−i X 2 1 2 1 E [min (X, M )] = min(i, M ) ≥M 3 3 3 3 i=0 10 X i=M = M P {X ≥ M } . 1 Therefore, f (M ) ≥ 15M P {X ≥ M } − 10M. In particular, maxM f (M ) ≥ 45P {X ≥ 3} − 30. Note that P {X ≥ 3} = 1 − P {X ≤ 2} 0 10 1 9 2 8 2 10 1 2 10 1 2 10 1 − − =1− 3 3 1 3 3 2 3 3 0 10 9 2 8 2 1 2 1 2 =1− − 10 − 45 3 3 3 3 3 ≥ 0.7. Therefore, maxM f (M ) ≥ 45 × 0.7 − 30 = 1.5 > 1. #1, p. 228. R1 (a) Set c −1 (1 − x2 ) dx = 1 to find that c = 3/4. (b) If a ≤ −1 then F (a) = 0. If a > 1, then F (a) = 1. If −1 < a < 1 then Z 3 1 3 a (1 − x2 ) dx = (1 + a) − (a3 + 1). F (a) = 4 −1 4 4 #4, p. 228. R∞ (a) P {X > 20} = 20 (10/x2 ) dx = 1/2. (b) If a ≤ 10 then F (a) = 0. If a > 10 then Z a 1 1 10 10 dx = 10 − =1− . F (a) = 2 10 a a 10 x (c) Let N denote the number of types of devices that will funciton for at least 15 R∞ hours. Then N is Bin(n = 6, p), where p = P {X > 15} = 15 (10/x2 ) dx = (2/3). Therefore, we are asked to find P {N ≥ 3} = 6 i 6−i X 1 6 2 i=3 #6, p. 228. (a) Recall that E[X] = 3 i 3 . R∞ xf (x) dx. Therefore, Z Z ∞ 1 ∞ 2 −x/2 E[X] = x e dx = 2 y 2 e−y dy. 4 0 0 R 0 R 0 Recall integration by parts: uv = (uv) − vu . Therefore, apply this with u(y) = y 2 and v(y) = −e−y ; u0 (y) = 2y and v 0 (y) = e−y . Whence we obtain Z ∞ Z ∞ Z ∞ ∞ 2 −y 2 −y −y y e dy = −y e + 2 ye dy = 2 ye−y dy. 0 −∞ 0 2 0 0 Apply integration by parts (again!) to find that the last integral is one. I.e., E[X] = 4. (b) We saw in #1 that c = 3/4. Therefore, 3 E[X] = 4 Z 1 x(1 − x2 ) dx = 0, −1 by symmetry (for instance). (c) This one is a bit tricky: Z ∞ E[X] = 5 5 1 x 2 dx = 5 x ∞ Z 5 dx = ∞. x #16, p. 229. Let X denote the number of years we have to wait until the first year with rainfall over 40 inches. Then X has a geometric distribution with parameter p = P {Z > 15}, where Z is normally distributed with parameters µ = 40 and σ 2 = 16. We are asked to find ∞ ∞ X X i−1 P {X ≥ 10} = p(1 − p) =p (1 − p)i−1 = (1 − p)10 . i=10 i=10 It remains to find p. But recalling that (Z − 40)/4 is standard normal, we have 15 − 40 p = P {Z > 15} = P N (0, 1) > = P {N (0, 1) > −6.25} ≈ 1. 4 Theoretical Exercises #19, p. 186. (a) Let X denote the number of games played this way. The possible values for X are 5, 6, 7, · · ·. Now P {X = 5} = P {game #5 is a loss} = 1 − p. More generally, if k ≥ 5, then P {X = k} = P {games #5 through k are wins; game #(k + 1) is a loss} = pk−5 (1 − p). Therefore, E[X] = ∞ X kp k−5 (1 − p) = (1 − p) k=5 = (1 − p) = (1 − p) ∞ X kpk−5 k=5 ∞ X k=5 ∞ X (k − 5)p k−5 + (1 − p)5 ∞ X k=5 ipi + 5(1 − p) i=1 ∞ X i=0 3 pi . pk−5 The first sum can be written as p ∞ X i−1 ip i=1 The second sum is d =p dp P∞ i=0 ∞ X ! p i i=0 R∞ −∞ 1 1−p = p . (1 − p)2 pi = 1/(1 − p). Therefore, E[X] = #1, p. 232. We know that d =p dp p + 5. 1−p f (x) dx = 1. Thus, Z √ a ∞ 3 −y2 1= ax e y e dx = 2 dy (y = x b) b 0 0 Z ∞ a 1 −1/2 3/2 −z = 2 z e z dz (z = y 2 ) b 0 2 Z ∞ a a = 2 ze−z dz = 2 . 2b 0 2b Z ∞ 3 −bx2 Therefore, a = 2b2 . 4