Math 5010-1, Spring 2005 Assignment 10 Problems #1, p. 379. Let H = {the coin lands heads}. Let X denote her winning, and D the number of dots that are rolled. Then, 1 1 E[X | H] + E[X | Hc ] 2 2 1 1 = E[2D | H] + E[D/2 | Hc ] 2 2 1 = E[D | H] + E[D | Hc ]. 4 E[X] = But the outcome of the die is independent of the outcome of the coin. So, E[D | H] = E[D] = 3.5, and E[D | Hc ] = 3.5, for the same reasons. Therefore, E[X] = 3.5 + (3.5/4) ≈ 4.38. #3, p. 379. By independence, the joint pdf of (X, Y ) is f (x, y) = 1 0 ≤ x, y ≤ 1. Therefore, Z α 1 Z 1 |x − y|α dy dx E [|X − Y | ] = 0 Z 0 1 Z Z α 1 Z 1 (x − y) dy dx + = 0 = x 0 2 1+α 0 1 Z x1+α dx = 0 (y − x)α dy dx x 2 . (1 + α)(2 + α) #17, p. 381. Let Ii = 1 if your ith guess is correct; else let Ii = 0. You ought to check that N = I1 + · · · + In . (a) E[N ] = E[I1 ] + · · · + E[In ]. Now E[Ii ] is equal to the probability that the ith guess is correct; this probability is 1/n for all i = 1, . . . , n. Therefore, E[N ] = (1/n) + · · · + (1/n) = 1, as asserted. This is the gist of the hint. (b) Suppose we are told the first i − 1 cards, and we wish to make the ith guess. Our choice is then, “any one of the remaining n − (i − 1) cards with equal probability.” So the probability that we get the ith one right is then E[Ii ] = 1 1 . n−i+1 Therefore, E[N ] = n X n X E[Ii ] = i=1 i=1 n X1 1 = ≈ n − i + 1 j=1 j Z n 1 dx = ln n. x #50, p. 385. First we compute the marginal pdf of Y : ∞ Z fY (y) = 0 e−x/y e−y dx = e−y y y > 0. That is, Y is exponential with parameter λ = 1. Next we find the conditional pdf of X given that Y = y: fX|Y (x|y) = = f (x, y) fY (y) e−x/y y x > 0. So, conditionally on {Y = y}, X is exponentially distributed with parameter λ = 1/y. Therefore, Z ∞ 2 E[X | Y = y] = x2 fX|Y (x|y) dx −∞ Z ∞ dx = x2 e−x/y y 0 Z ∞ = y2 z 2 e−z dz (z = x/y) 0 2 = y Γ(3) = 2y 2 . #75, p. 389. Recall that M (t) = exp(λ(et −1)) is the moment generating function of a Poisson with parameter λ. Therefore, X is Poisson with parameter 2. However, MY (t) = (1/4)10 is not a moment generating function. For instance, MY (0) must be one, but it is not. So the problem is ill-posed. #77(b), p. 385. We need the individual marginal pdf’s. Let us start with Y : Z ∞ 2 1 fY (y) = √ e−y e−(x−y) /2 dx = e−y y > 0. 2π −∞ Therefrom we have Z MY (t) = = ∞ ty −y e e Z ∞ dy = e−y(1−t) dy 0 1 1−t 0 if −∞ < t < 1, ∞ otherwise. 2 To find the mgf of X, we will R R not find fX first; it is too complicated. Instead we use the formula, E[g(X, Y )] = g(x, y)f (x, y) dx dy. In this way, we obtain Z ∞Z ∞ tX 2 1 etx e−y e−(x−y) /2 dx dy MX (t) = E e =√ 2π 0 −∞ Z ∞ Z ∞ 1 −y tx −(x−y)2 /2 √ e = e e dx dy. 2π −∞ 0 Now the term inside (· · ·) is the moment generating function of a N (y, 1). Therefore, Z ∞ 2 MX (t) = e−y ety+t /2 dy 0 Z ∞ t2 /2 =e e−y+ty dy 0 2 t /2 e if −∞ < t < 1, = 1−t ∞ otherwise. #3, p. 427. Let X1 , . . . , Xn denote the grades of n students. We know that the Xi ’s are independent and idedntically distributed with E[Xi ] = 75. If n is large, then by the CLT, the class average (X1 + · · · + Xn )/n is approximately normally distributed with mean µ = 75 and variance σ 2 /n, where σ 2 is the variance of her test. We wish to find (a large) n such that X1 + · · · + Xn P 70 ≤ ≤ 80 ≥ 0.9. n By the CLT, the left-hand side is approximately √ √ σ2 5 n 5 n ≤ 80 = Φ −Φ − P 70 ≤ N 75, n σ σ √ 5 n − 1. = 2Φ σ Set this equal to 0.9 to find that n must satisfy √ 5 n Φ ≈ 0.95. σ Now use the table on page 203 to find that the preceding implies that √ 5 n ≈ 1.65. σ Therefore, the minimum n is n ≈ 0.1089σ 2 , provided that n (equivalently, σ) is large enough that the appeal to the CLT is valid. For example, suppose σ = 25. Then, 0.1089σ 2 ≈ 68, which is large enough. This would then imply that if the standard deviation of the grades is 25, then 69 students would suffice. 3