4-13【Question】 If X is a nonnegative continuous random variable, show that E ( X ) [1 F ( x )]dx 0 Apply this result to find the mean of the exponential distribution. 【Solution】 X is nonnegative continue r.v (i) E ( X ) xf ( x)dx 0 = lim b b 0 xf ( x)dx = lim ( x( F ( x) 1)) 0 ( F ( x) 1)dx b b 0 = lim b( F (b) 1) (1 F ( x))dx b 0 claim: lim b( F (b) 1) 0 b b b consider: 0 b(1 F (b)) b f ( x)dx xf ( x)dx 0 lim b(1 F (b)) 0 b E ( X ) (1 F ( x))dx 0 (ii) x ~ exp( ) F ( x) 1 e x 1 0 E ( X ) e x dx 4-18【Question】 If U 1 , , U n are independent uniform random variables, find E (U ( n ) U (1) ) . 【Solution】 U 1 , ,U n ~ uniform(0,1) f U ( n ) ,U (1) (u n , u1 ) n! 1 1 (u n u1 ) n 2 0!(n 2)!0! = n(n 1)(u n u1 ) n 2 R U ( n ) U (1) U (1) Y Y U (1) U ( n ) R Y J 0 1 1 1 1 f R,Y (r, y) n(n 1)r n2 1 , 0 y r y 1 1 r f R (r ) n(n 1)r n2 dy n(n 1)(1 r )r n2 , 0 r 1 0 1 E ( R) r n(n 1)(1 r )r n2 dr 0 1 1 0 0 = n(n 1)(1 r )r n1dr n(n 1) (1 r )r n1dr = n(n 1)(n)(2) n(n 1)( n 1)! n 1 (n 2) (n 1)! n 1 ( n, 2) 4-25【Question】 If X 1 and X 2 are independent random variables following a gamma distribution with parameters and , find E ( R 2 ) , where R 2 X12 X 22 【Solution】 X 1 , X 2 ~ Gamma( , ) , indep f X1 , X 2 ( x1 , x2 ) 1 x 1 x x1 e x2 e , x1 , x2 0 ( ) ( ) 1 2 E ( R 2 ) E ( X 12 X 22 ) = 0 0 ( x12 x22 ) f x1 , x2 ( x1 , x2 )dx1dx2 1 x 1 x x1 e x2 e dx1dx2 0 0 ( ) ( ) 1 x 1 x x e x2 e dx1dx2 0 0 ( ) 1 ( ) ( 2) 1 x = x e dx x2 1e x dx2 2 2 0 ( ) 2 0 ( ) ( ) ( 2) ( 2) 2 ( 1) = ( )2 ( )2 2 1 2 1 2 2 2 4-26【Question】 Referring to Example B in Section 4.1.2 what is the expected number of coupons needed to collect r different types, where r n ? 【Solution】 X 1 :the number of trails up to and including the trail on which the first coupon is collected. X 2 :the number of trails from that point up to and including the trail on which the next coupon different from the first is obtained,……and so on, up to X n . X k ~ geo( E( X k ) n r 1 ) n k 1, , n X k a r e i n d e. p n n r 1 Let X X 1 X 2 X r r E( X ) E( X 1 X r ) E( X k ) k 1 n n n ) =( n n 1 n r 1 nr 1 1 n k 1 k k 1 k = n(log n n ) n(log( n r ) nr ) n = n = n(log n n nr ) , r n nr 4-30【Question】 1 ) , where X is a Poisson random variable. X 1 【Solution】 X ~ p ( ) , 0 Find E ( p( X k ) e E( k k 0,1,2, k! 1 1 k k k 1 ) ( )e e e X 1 k 0 k 1 k! k 0 (k 1)! k 1 k! = e k k 0 k! ( 1) e (e 1) 1 e 4-34【Question】 Let X be uniform on [0,1], and let Y X . Find E (Y ) by (a) finding the density of Y and then finding the expectation and (b) using Theorem A of section 4.1.1. 【Solution】 X ~ U [0,1] f X ( x) 1, x [0.1] Y X X Y2 dx 2y dy (a) f Y ( y) f X ( y 2 ) J 1 2 y 2 y E (Y ) 1 (b) E (Y ) 0 1 0 y 2 ydy x dx 1 0 2 y 2 dy 2 3 2 3 4-35【Question】 Find the mean of a negative binomial random variable.(Hint:Express the random variable as a sum.) 【Solution】 The negative binomial random variable Y is the number of the trials on which the r-th success occurs, in a sequence of independent trials with constant probability p of success on each trail. Let xi denote a random variable defined as the number of the trail on which the i-th success occurs, for i=1,2,….,r. Now define Wi X i X i 1 , i 1,2, , r r where X 0 is defined to be zero. Then we can write X r Y Wi . i 1 Notice that the random variable W1 ,W2 , ,Wr have identical geometric distribution and are mutually independent. r r r i 1 i 1 i 1 E (Y ) E ( Wi ) E (Wi ) 1 r p p 4-37【Questuon】 For what values of p is the group testing of Example C in section 4.1.2 inferior to testing every individual. 【Solution】 E ( N ) n(1 1 pk ) k E ( N ) n n(1 If p k n mk n, m, k N (from 課本 p.121) 1 1 1 pk ) n pk p k k k k 1 , then the group testing is inferior to testing every individual. k 4-38【Question】 Let X be an exponential random variable with standard deviation . Find p( X E ( X ) k ) for k 2,3,4, and compare the results to the bounds from Chebyshev’s inequality. 【Solution】 令 X ~ ( ) E( X ) 1 , Var ( X ) p( x k ) p( x 1 = p( x k k 1 1 2 ) = p( x ) 1 k 1 ) p( x 1 k ) ( k 2 1 k 0 p ( x 1 k ) 0) = k 1 e x dx e k 1 , k 2,3, Chebyshev’s inequality k p( x k ) e k 1 p ( x k ) 2 e 3 0.04979 1 0.25 4 3 e 4 0.01832 1 0.1111 9 1 0.0625 16 39 Show that Var(X - Y) Var(X) Var(Y) - 2Cov(X, Y). 4 e 5 0.006738 1 k2 Ans: Var ( X Y ) E ( X Y ) 2 [ E ( X Y )]2 E ( X 2 2 XY Y 2 ) [ E ( X ) E (Y )] 2 E ( X 2 ) 2 E ( XY ) E (Y 2 ) E 2 ( X ) 2 E ( X ) E (Y ) E 2 (Y ) E ( X 2 ) E 2 ( X ) E (Y 2 ) E 2 (Y ) 2[ E ( XY ) E ( X ) E (Y )] Var ( X ) Var (Y ) 2 cov( X , Y ) 41 Find the covariance and the correlation of N i and N j , where N1 , N 2 , …, N r are multinomial random variables.(Hint:Express them as sums.) Ans: N1 , N 2 , …, N r :multinomial r.v PN 1 ...N r ( n1 ...nr )= r n! n n P1 1 ...Pr r , n1!...nr ! ni =n, i 1 r P =1 i 1 i n n P N ( ni )= Pi i (1 Pi ) nni , N i ~ Bin (n, Pi ), i 1...r , E ( N i ) n Pi , i ni Var ( N i ) n Pi (1 Pi ). n E( Ni , N j ) nn j n n n j 0 ni 0 i j n! nn n n n Pi i Pj j (1 Pi Pj ) i j ni !n j !(n ni n j ) n n j n!Pj j Pi nn j (n n j 1)! nn n n 1 Pi i (1 Pi Pj ) i j = n j 0 n j !( n n j 1)! ni 0 (ni 1)!(n ni n j )! n nn j 1 n n j 1 ni 1 Pi (1 Pi Pj ) nni n j = n j 0 ( n j 1)!( n n j 1)! ni 0 n j 1 n n!Pj j Pi n n n2 Pi Pjn j (1 Pj ) nn j 1 = n(n 1) n j 0 n j 1 n2 n 2 n j 1 n 2 n1 1 Pj (1 Pj ) nn j 1 P ( 1 P ) = n(n 1) Pi Pj j j n 1 n 1 n j 1 j = n(n 1) Pi PJ (1) n2 0 n(n 1) Pi PJ ∴Cov(n i ,n j )=E(n i ,n j )-E(n i )E(n j )=n(n-1)P i P j -nP i nP j =- nP i P j P= Cov(ni , n j ) Var (ni )Var (n j ) nPi Pj n Pi (1 Pi ) n Pj (1 Pj ) nPi Pj n Pi Pj (1 Pi )(1 Pj ) Pi Pj (1 Pi )(1 Pj ) 45 Two independent measurements, X and Y, are taken of a quantity μ. E(X)=E(Y)= μ, but σ X and σ Y are unequal. The two measurements are combined by means of a weighted average to give Z = αX + (1-α)Y Where α is a scalar and 0 α 1. a Show that E(Z) = μ. b Find α in terms of σ X and σ Y to minimize Var(Z). c Under what circumstances is it better to use the average ( X Y ) 2 than either X or Y alone? Ans:(a) E ( Z ) E (X (1 )Y ) E ( X ) (1 ) E (Y ) (b) Var (Z ) Var (X (1 )Y ) 2 X2 (1 ) 2 Y2 f ( ) f ( ) 2 X2 (1 ) Y2 f ( ) 2 X2 2 Y2 ﹥0 f ( ) 0 時有 min 2 X2 2(1 ) Y2 2 X2 2 Y2 2 Y2 (c) V a (rZ ) V a (r X2 Y2 Y2 X2 Y2 2 Y2 X Y ) X , Var ( X ) X2 , Var (Y ) Y2 2 4 X2 Y2 3 X2 4 Y2 Y2 X2 3 Y2 4 2 X 49 X Y 1 2 is better, when X2 3 2 3 Y Let T = k 1 X k , where the X k are independent random variables with means μ n and variances σ 2 . Find E(T) and Var(T). n n n k 1 k 1 Ans: E (T ) E ( k X k ) E (k X k ) kE( X k ) k 1 n(n 1) 2 n V a (rT ) V a (r k X k ) ( X 1 , X 2 ,, X n are independent) k 1 n n k 1 k 1 Var (k X k ) k 2Var ( X k ) 51 n(n 1)( 2n 1) 2 6 If X and Y are independent random variables, find Var(XY) in terms of the means and variances of X and Y Ans: X , Y independent Var ( XY ) E ( XY ) 2 E 2 ( XY ) E ( X 2Y 2 ) E 2 X E 2Y ( X , Y indep , E ( XY ) EX EY ) EX 2 EY 2 E 2 X E 2Y ( X , Y indep, X 2 ,Y 2 indep, E ( X 2 Y 2 ) EX 2 EY 2 ) (VarX E 2 X ) (VarY E 2Y ) E 2 X E 2Y VarX VarY VarX E 2Y E 2 X VarY E 2 X E 2Y E 2 X E 2Y VarX VarY VarX E 2Y E 2 X VarY X2 Y2 Y2 X2 X2 Y2 53 Let (X,Y) be a random point uniformly distributed on a unit disk. Show that Cov(X,Y)=0, but that X and Y are not independent. Ans: f X ,Y ( x, y ) 1 1 x 2 f X ( x) 1 x 2 同理 f y ( y ) , 1 2 x2 y2 1 dy 2 1 x2 , 1 y2 , x 1,1 y 1,1 f x, y ( x, y) f x ( x) f y ( y) x 2 y 2 1 X , Y a r en o ti n d e p e ntd e n 1 EX 1 2 x 1 x 2 dx 0 ( 2 x 1 x 2 為奇 函 數) 同理, EY 0 1 x 2 E( X ,Y ) 54 2 x y 1 2 1 1 1 1 2 1 x x y x 0dx 0 2 dx 1 x 1 1 x 2 1 2 1 Cov( X , Y ) E ( XY ) EX EY 0 0 0 Cov( X , Y ) 0, but that X and Y are not i n d e p e ntd. e n 1 dydx Let Y have a density that is symmetric about zero, and let X=SY, where S is an independent random variable taking on the values +1 and –1 with probability each. Show that Cov(X,Y)=0, but that X and Y are not independent. Ans: E ( S ) 1 P( S 1) (1) P( S 1) 1 1 0 2 2 Cov( X , Y ) Cov( SY , Y ) E ( SY Y ) E ( SY ) E (Y ) E ( S ) E (Y 2 ) E ( S ) E (Y ) E (Y ) ( S和Y indep.) 0 0 1 f x│y ( x│y ) 2 1 2 f x│y f x x y x y f X ( x) 1 1 f Y ( x) f Y ( x) 2 2 x y X , Y are noti n d e p e ntd. e n 55 In Section 3.7, the joint density of the minimum and maximum of n 1 2 independent uniform random variables was found. In the case n =2, this amounts to X and Y, the minimum and maximum, respectively, of two independent random variables uniform on [0,1], having the joint density f ( x, y ) 2, 0 x y a Find the covariance and the correlation of X and Y. Does the sigh of the correlation make sense intuitively? b Find E(X│Y=y) and E(Y│X=x). Do these results male sense intuitively? c Find the probability density functions of the random variables E(X│Y) and E(Y│X). d What is the linear predictor of Y in terms of X (denoted by Yˆ =a + bX) that has minimal mean squared error? What is the mean square prediction error? e What is the predictor of Y in terms of X[ Yˆ =h(x)] that has minimal mean squared error? What is the mean square prediction error? Ans: f X ,Y ( x, y ) 2, 0 x y 1 a. y y f Y ( y ) f X ,Y ( x, y )dx 2dx 2 y, 0 0 1 E ( XY ) 0 y 0 1 f X ,Y ( x, y )dxdy 0 1 1 0 0 EX xf X ( x)dx 2 x(1 x)dx 1 1 0 0 y 0 2 xydxdy 1 3 EX 2 x 2 f X ( x)dx 2 x 2 (1 x)dx 1 1 0 0 EY yf Y ( y )dy 2 y 2 dx 1 1 0 0 y (0,1) 1 6 2 3 EY 2 y 2 f Y ( y )dx 2 y 3 dy 1 2 1 1 2 1 ( ) 6 3 18 1 2 1 VarY EY 2 E 2Y ( ) 2 2 3 18 1 1 2 1 Cov( X , Y ) E ( XY ) EX EY 4 3 3 36 1 Cov( X , Y ) 1 36 2 VarX VarY 1 1 18 18 VarX EX 2 E 2 X 1 4 b. f X│Y ( x│y) f X ,Y ( x, y) fY ( y) 1 , y (0,1), x (0, y) X│Y y ~ U (0, y) y y , y (0,1) 2 f ( x, y ) 1 ( y│x) X ,Y , x (0,1), y ( x,1) Y│X x ~ U ( x,1) f X ( x) 1 x E ( X│Y y ) f ,Y│X E (Y│X x) c. x 1 , x (0,1) 2 令W1 ( X│Y ) W2 (Y│X ) y 1 , fW1 (W1 ) 2 f Y (2W1 ) 8W1 , W1 (0, ) 2 2 X 1 1 , fW2 (W2 ) 2 f X (2W2 1) 8(1 W2 ), W2 ( ,1) 2 2 d. Y之最佳線性預測值:Yˆ a bX a Y b X 1 b ( Y ) X 2 2 1 1 1 3 2 3 2 Yˆ 1 1 X, 2 2 x (0,1) 1 1 1 1 1 此時之M.S.E.:E (Y X ) 2 Y2 (1 2 ) (1 ) 2 2 18 4 24 X 1 , e. Y之最佳預測值:Yˆ E (Y│X ) 2 x (0,1) 此時M.S.E: . E (Y E (Y│X )) 2 EY 2 E ( E 2 (Y│X )) 1 ( x 1) 2 E( ) 2 4 1 1 ( x 1) 2 f X ( x)dx 2 0 4 1 1 1 ( x 1) 2 (1 x) dx 2 2 0 1 11 2 24 1 24 58 Let X and Y be jointly distributed random variables with correlation XY ; define ~ ~ ~ the s tan dardized random variables X and Y as X = ( X E( X )) ~ and Y = (Y E(Y )) Var( X ) ~ ~ Var (Y ) . Show that Cov( X , Y )= XY . ~ ( X EX ) ~ (Y EY ) Ans: X , Y , correlatio n: X ,Y VarX VarY ( X EX ) (Y EY ) X EX Y EY ~ ~ Cov( X , Y ) E ( ) E( ) E( ) VarX VarY VarX VarY E ( X EX ) (Y EY ) V a r X V a r Y X ,Y 67 A fair coin is tossed n times, and the number of heads, N, is counted. The coin is then tossed N more times. Find the expected total number of heads generated by this process. 1 Ans: N ~ B (n, ) 2 令 X 標丟 N=m 次,出現頭之次數,m=0,1,2,…,n 1 X│N m ~ B(m, ), 2 x 0,1,2,, m EX E ( E ( X│N )) E ( EN N n ) 2 4 ( Note:若X ~ B(n, p), 則EX np) n 2 ∴The expected total number of heads generated by this process is 3 n 4 70 Let the point (X,Y) be uniformly distributed over the half disk x 2 y 2 1 , where y 0. If you observe X, what is the best prediction for Y ? If you observe Y , what is the best prediction for X ? For both questions , “best” means having the minimum mean squared error . sol f X ,Y ( x, y ) = 2 . x 2 y 2 1 , y>0 f X (x) = 0 f Y 1 x 2 fx, y( x. y)dy 2 1 x 2 0 1 y 2 ( y) f x , y ( x, y)dx 1 y 2 2 1 y 2 1 y 2 dy dx 2 1 x 2 ,x (1,1) 4 1 y 2 ,y (0,1) 2 f X ,Y ( x, y ) f y x ( y x) f X ( x) 2 f Y ( y) 1 x 1 x2 2 f X ,Y ( x, y) f X Y ( x y) 1 4 1 y2 2 1 2 1 y 2 , x (1,1), y (0, 1 x 2 ) , x (0,1), y ( 1 y 2 , 1 y 2 ) 在給定 X x 之下, Y 之最佳預測值 E (Y X x) 1 x2 , x (1,1) , (Note: Y X x ~ U (0, 1 x 2 ) ) 2 在給定 Y y 之下, X 之最佳預測值 E( X Y y) 0, y (0,1) , (Note: X Y y ~ U ( 1 y 2 , 1 y 2 ) ) P159 #71 Let X and Y have the joint density a Find Cov(X,Y) and the correlation of X and Y. b Find E ( X Y y) and E (Y X x). c Find the density functions of the random variables E ( X Y ) and E (Y X ). sol a. f X ( x) x f X ,Y ( x, y)dy x e y dy e y , x (0, ) 1 EX 1 ( X ~ (1) 且若 Y ~ ( ) , EY ,VarY f Y ( y) y 0 f X ,Y ( x, y)dx y 0 1 ) VarX 1 2 e y dy ye y , y (0, ) EY f Y ( y)dy y 2 e y dy 2 , EY 2 y 2 f Y ( y)dy y 3 e y dy 6 0 0 0 0 E ( XY ) 0 y 0 xyf X ,Y ( x, y)dxdy 0 y 0 xye y dxdy 0 1 3 y y e dy 3 2 Cov( X , Y ) E ( XY ) EX EY 3 1 2 1 , VarY EY 2 E 2Y 6 4 2 Cov( x, y ) 1 1 correlatio n XY VarX VarY 1 2 2 b. f X ,Y ( x, y ) e y 1 f X Y (x y , x (0, y ), y (0, ) y f Y ( y ) ye y y , y (0, ), ( X Y y ~ U (0, y )) 2 f X ,Y ( x, y ) e y f Y X ( y x) x e ( y x ) , x (0, ), y ( x, ) f X ( x) e E ( X Y y) x x E (Y X x) y f Y X ( y x)dy y e ( y x ) dy x 1, x (0, ) 令W1 E ( X Y ) Y , f W1 ( w1 ) 2 f Y (2w1 ) 4W1e 2 w1 , w1 (0, ) 2 令W2 E(Y X ) x 1, fW2 (w2 ) f X (w2 1) e ( w2 1) , w2 (1, ) P160 #75 Find the moment-generating function of a Bernoulli random variable , and use it to find the mean, variance, and third moment. sol P( X 0) 1 p, P( X 1) p M X (t ) E (e tx ) (1 p ) e 0 pe pe t , t R EX M X (0) p, EX 2 M X (0) p, EX 3 M X (0) p VarX EX 2 E 2 X p p 2 p (1 p ) P160 #84 Assuming that X ~ N (0, 2 ) ,use the m.g.f. to show that the odd moment are zero and the even moments are 2n (2n)! 2 n 2 n (n!) sol M (t )在t 0之n次導數皆存在,n 1 由propertyB知X之任意次動差皆存 在 1 又利用e 2 2t 2 1 M (t ) e 2 2t 2 之Tayior展開式得 : 2 n1 0, 2 n P161 (2n)! 2 n t 2 n nt n 1 1 2 2 n 2nt 2n ( t ) n n n! n 0 n! 2 n 0 2 ( n!) n 0 2 ( n!) ( 2n)! n 0 (2n)! 2 n ,n 0 2 n (n!) #88 If X is a nonnegative integer-valued random variable, the probability-generating function of X is defined to be G ( s ) s k Pk k 0 where pk P( X k ) . a Show that pk b 1 dk G ( s ) s 0 k! ds k Show that dG s 1 E ( X ) ds d 2G s 1 E X ( X 1) ds 2 c Express the probability-generating function in terms of moment-generating function. d Find the probability-generating function of the Poisson distribution. sol b. G(s) uniformly converges for s 1 可將此power series 逐eri 而得G ( s )之導數, 即 g ( s ) kpk s k 1 ......(1) k 1 g ( s) k (k 1) p k s k 2 ......(2) k 2 k 一般有G ( n ) ( s) k (k 1)( k 2).....( k n 1) p k s k n n! p k s k n k n k n n 此級數在 s 1 仍收斂 在上一式中若令s 0, 則除了常數項其餘均為0, 故得 p n g (n) (0) n! 若EX存在, 則由(1)可得 EX g (1) 而若EX(X 1)亦存在,由(2)可得 EX(X 1) g (1) c 假定s 0 令X之 mgf 為 M X (t) G X ( s ) E ( s X ) E (e X (ln S ) ) M X (ln S ) for some s 0 d 由課本 P144 example A 知 , 若X ~ P( ), 則 X 之mgf : M X (t ) exp (e t 1) 由 (c)之結果得G X ( s) M X (ln S ) exp ( s 1) for all s >0 P161 #93 Find expressions for the approximate mean and variance of Y = g(X) for (a) g(x)= x , (b) g(x)=logx, and (c) g(x)= sin 1 x . sol a g ( x) x , E( X ) , Var ( X ) 2 E ( g ( x)) g ( E ( X )) Var ( g ( x)) ( ) b 2 Var ( X ) 2 4 E( X ) g ( x) log X Var ( X ) 2 E ( g ( x)) g ( E ( X )) log Var ( g ( x)) (log ) Var ( X ) 2 2 2 c E( X ) g ( x) sin 1 Var ( X ) 2 E ( g ( x)) g ( E ( X )) sin 1 Var ( g ( x)) (sin 1 ) 2 Var ( X ) 2 1 2 P161 #96 Two sides, x 0 and y 0 , of a right triangle are independently measured as X and Y, where E(X)= x 0 and E(Y)= y 0 and Var( X ) Var(Y ) 2 . The angle between the two sides is then determined as Y X Find the approximate mean and variance of . t a n1 sol 由課本 P152 知 Z g(x, y), 則 1 g(μ( 2 1 2 g(μ( 2 2 g(μ( EZ g(μ( σ 2X ( ) σY ( ) σ X, Y 2 x 2 y xy g(μ( 2 g(μ( 2 g(μ( g(μ( VarZ σ 2X ( ) σ 2Y ( ) 2σ X, Y ( )( ), x y x y where μ denote the point (μ X , μ Y ) X, Y are independen t random variables . EX x 0 , EY y 0 VarX VarY σ 2 Y θ tan 1 ( ) X y 令θ(x,y) tan 1 ( ) x θ y θ x 2θ 2xy 2θ 2xy 2θ y2 x 2 2 , x x y 2 y x 2 y 2 x 2 (x 2 y 2 ) 2 y 2 (x 2 y 2 ) 2 xy (x 2 y 2 ) 2 EZ tan 1 ( y0 x y σ2 x y σ2 y ) 0 0 2 20 0 2 2 tan 1 ( 0 ) ( X, Y independen t σ xy 0) 2 2 x0 x0 (x 0 y 0 ) x 0 y0 y σ x σ σ2 VarZ 2 (Note : X, Y independen t ) (x 0 y 02 ) 2 (x 02 y 02 ) 2 x 02 y 02 2 0 2 2 0 2