1. Excercise 1.2.8 Solution Ck , k = 1, 2, . . . is a sequence of increasing sets. (a) Ck = {x : 1/k ≤ x ≤ 3 − 1/k}, k = 1, 2, 3, . . . As k → ∞, 1/k → 0 and (3 − 1/k) → 3. ∞ [ lim Ck = n→∞ Ck = (0, 3) k=1 Note that the two points 0 and 3 are not in any of Ck for k = 1, 2, 3, . . .. Therefore, these two points are not included in limn→∞ Ck above. (b) Ck = {(x, y) : 1/k ≤ x2 + y 2 ≤ 4 − 1/k} As k → ∞, 1/k → 0 and (4 − 1/k) → 4. lim Ck = n→∞ ∞ [ Ck = {(x, y) : 0 < x2 + y 2 < 4} k=1 2. Excercise 1.2.9 Solution Ck , k = 1, 2, . . . is a sequence of decreasing sets. (a) Ck = {x : 2 − 1/k < x ≤ 2} As k → ∞, (2 − 1/k) → 2. The l.h.s of the inequality is approaching 2 from left then. By definition, ∞ \ lim = n→∞ Ck = {2} k=1 Since 2 is the only element that is in all Ck s. (b) Ck = {x : 2 < x ≤ 2 + 1/k} As k → ∞, (2 + 1/k) → 2. The r.h.s of the inequality is approaching 2 from right then. By definition, lim = n→∞ ∞ \ k=1 1 Ck = ∅ Since there is no single element could be found that is in all Ck s. To see this, we could assmue there exists a sufficiently small δ > 0 such that x = 2+δ. Certainly, x > 2 must hold then. However, we could always find another sufficiently large k such that 1/k < δ. Then it follows that x = 2+δ > 2+1/k, which contradicts our assumption. (c) Ck = {(x, y) : 0 ≤ x2 + y 2 ≤ 1/k} As k → ∞, 1/k → 0. Note the right part of the inequalities above includes equality. Consequently, 0 is the only element in the shrinking Ck sequence. Hence, lim = n→∞ ∞ \ Ck = {(0, 0)} k=1 3. Excercise 1.3.10 Solution (a) Follow the hint, denote Ci := {an exact match on the ith turn Also, by the general inclusion-exclusion formula, we could get pM = Pr {at least one exact match} = P (C1 ∪ C2 ∪ . . . ∪ C52 ) 52 52 52 52 = P (Ci ) − P (Ci ∩ Cj ) + P (Ci ∩ Cj ∩ Ck ) − . . . − P (∩52 i=1 Ci ) 1 2 3 52 52 Further, define p2 = 52 2 P (Ci ∩ Cj ), p3 = 3 P (Ci ∩ Cj ∩ Ck ), . . .. Then, pM = p1 − p2 + p3 − . . . − p52 1 1 1 = 1 − + − ... − 2! 3! 52! as desired. 2 (?) (b) Recall the Taylor expansion of ex , ∞ X xi 1 ex = 1 + x + x2 + . . . = 2 i! i=0 Evaluating the above equation at x = −1, we get e−1 = 1 − 1 + 1 1 − + ... 2 3! 1 − e−1 = 1 − 1 1 + − ... 2! 3! Then we could get Note that (?) is similar as the above equation, then we reach 1 − e−1 ≈ pM 4. Excercise 1.3.15 Solution The number of all possible outcomes: The number of possible outcoms of two red chips: The number of possible outcoms of two blue chips: The number of possible outcoms of two 1s: 22 · 60 The number of possible outcoms of two 2s: 22 · 60 The number of possible outcoms of two 3s: 22 · 60 8 2 5 2 · 3 2 · 6 0 6 0 Then, Pr {2 chips randomly drawn have either same color or same number} 6 6 6 5 3 2 2 · 0 + 2 · 0 +3· 2 · 0 = 8 2 5. Excercise 1.4.8 Solution Let D denote the event that a spring is defective; and I, II, III the events that 3 a spring is produced by Machcine I, II and III, respectively. We know the following probabilities P (I) = 1%, P (II) = 4%, P (III) = 2%, P (D | I) = 30% P (D | II) = 25% P (D | III) = 45% (a) Randomly draw a spring, find the probability that is defective. P (D) = P (D | I) · P (I) + P (D | II) · P (II) + P (D | III) · P (III) = 0.3 × 0.01 + 0.25 × 0.04 + 0.45 × 0.02 = 0.022 (b) Given a randomly drawn spring is defective, what the probability that is produced by Machcine II ? P (D | II) · P (II) P (D) 0.04 × 0.25 = 0.022 = 5/11 P (II|D) = 6. Excercise 1.4.19 Solution Define A := {at least 4 draws to obtain a spade}. Then we could have the following, Ac = {at most 3 draws to obtain a spade} = {obtain a spade at the 1st draw}+ {obtain a spade at the 2st draw}+ {obtain a spade at the 3st draw} 4 (a) Draw cards with replacement. P (A) = 1 − P (Ac ) 13 52 − 13 13 52 − 13 2 13 =1− + · +( ) · 52 52 52 52 52 27 = 64 (b) Draw cards without replacement. P (A) = 1 − P (Ac ) 13 52 − 13 13 52 − 13 38 13 + · + · · =1− 52 52 51 52 51 50 13 3 13 3 38 13 =1− + · + · · 52 4 51 4 51 50 7. Excercise 1.4.32 Solution Suppose that we could select p1 and p2 satisfies the following equation P (zero hits) = P (one hit) = P (two hits) Note that by assuming independence, and plugging p1 , p2 to the chain of equation above (1 − p1 )(1 − p2 ) = (1 − p1 )p2 + (1 − p2 )p1 (1 − p )p + (1 − p )p = p p 1 2 2 1 1 2 Solving for the above system, we could get p1 = 1/3p2 , 3p22 − 3p2 + 1 = 0 It then turns out that we couldn’t get a real root for p2 . Thus, it is not possible to select p1 , p2 satisfying the condition. 8. Find pmf of Y = X 3 , where pX (x) = 1 2 2 0 x = 1, 2, . . . otherwise 5 Excercise 1.6.8 Solution First, identify the support of Y , SY = {1, 8, 27, . . .}. pY (y) = P (Y = y) X = P (X = x) x∈g −1 (y) X = pX (x) x∈g −1 (y) √ = pX ( 3 y) √ 3y 1 = 2 Hence, we could have the pmf of Y as pY (y) = 1 2 √ 3y For y ∈ SY 0 otherwise 9. Excercise 1.7.9 Solution Median of distribution. (a) p(x) = 4! x!(4−x)! 1 x 4 3 4−x 4 0 x = 0, 1, 2, 3, 4 otherwise Computing P (X ≤ x) and P (X < x) for x = 1, 2, 3, 4, it easy to see that 3 3 4 3 4! 1 3 3 189 1 + = = > 4 3! 4 4 4 256 2 81 1 P (X < 1) = p(0) = < 256 2 P (X ≤ 1) = p(0) + p(1) = Then the median for this distribution is 0. 6 (b) 3x2 f (x) = 0 0<x<1 otherwise Since point probability of continuous r.v. is 0, then P (X ≤ x) = P (X < x) =⇒ P (X ≤ x) = 1/2 Calculate the cdf, F (x) = P (X ≤ x) Z x 3t2 dt = 0 x = t3 0 = x3 Then, by solving x3 = 1/2 yields the median for this distribution is p 3 1/2. (c) f (x) = 1 , −∞ < x < ∞ π(1 + x2 ) Calculate the cdf as F (X) = P (X ≤ x) Z x 1 = dt 2 −∞ π(1 + t ) Z 1 x 1 = dt π −∞ 1 + t2 1 = [arctan t]x−∞ π 1 1 = arctan x + π 2 Then, by solving 1 π arctan x + 1 2 = 1 2 yields the median for this distribution is x = 0. 7 10. Excercise 1.7.21 Solution 2xe−x2 f (x) = 0 0<x<∞ otherwise Let X , Y denote the support of the r.v. X and Y, respectively. Note the transformation Y = g(X) = X 2 is monotone on X ; the support transformation is then X = {x : 0 < x < ∞}, Y = {y : 0 < y < ∞} 1 Identify the inverse as g −1 (y) = y 2 , for y ∈ Y. Compute the Jacobian of transformation d −1 1 − 1 J = g (y) = y 2 dy 2 Now we could calculate the pdf of Y within the support Y fY (y) = fX (g −1 (y)) · J = e−y Then, we could get the pdf of Y as e−y fY (y) = 0 0<y<∞ otherwise 11. Excercise 1.8.10 Solution 2x f (x) = 0 0<x<1 otherwise 8 (a) E[1/x] Compute the expectation directly as 1 E[ ] = X Z 1 0 Z 1 = 0 1 · f (x) dx x 1 · 2x dx x =2 (b) Compute the cdf and pdf of Y = 1/X Identify the inverse as g −1 (y) = 1/y and Y = (1, ∞), also note that g(·) is monotonically decreasing. Also get the cdf of X as FX (x) = Rx 0 2t dt = x2 , then FY (y) = 1 − FX (g −1 (y)) = 1 − y −2 d 1 −2 Compute the Jacobian of transformation as J = dy y = y , then the cdf is fY (y) = 2 y3 0 1<y<∞ otherwise (c) Calculate the expectation by definition ∞ Z y · fY (y) dx EY = Z−∞ ∞ = y · 2y −3 dy 1 ∞ = −2 · y −1 1 =2 9