Professional Documents
Culture Documents
Chapter 2, problems 61, 67, 69, Chapter 3, problems 8, 12, 18, 19, 20, 34, 37. Due 9/23/99. . Find the density of cX when X follows a gamma distribution. Show that only is a ected by such a transformation, which justi es calling a scale parameter. Let Y = cX . Then 1 fY (y ) = fX (y=c): jcj
2.61
(y=c)
y
1
y=c =c)y
( )
x=
; x
0;
> 0;
> 0:
Note: there is a typo in my copy of the book! In my copy, at any rate, the exponent is missing a minus sign. Most of you didn't seem to have this problem, so I hope that got corrected at a later printing. a) Find the density function. Ok, di erentiating with respect to x
fx (x) = @ @x = ( = )x 1e e
(
x=
(x= )
(
x=
b) Show that if W follows a Weibull distribution, then X = (W= ) follows an exponential distribution. Let g (x) = (x= ) . Then g 1(x) = x1= . We know that
fX (x) = fW (g 1(x)) @ 1 g (x) @x = ( = ) x1= ] 1 e x ( = )x(1 = 1 1 + 1+1 x( 1)= +(1 )= e = e x; 0 < x < 1
= x
)
so X follows an exponential distribution. c) How could Weibull random variables be generated from a uniform random number generator? Let U be a uniform(0,1) random variable. We showed in class that log(U ) follows an exponential distribution. Combining this with the result from part b), Y = log(U )]1= will follow a Weibull( ; ) distribution. 1
2.69. If the radius of a sphere is an exponential random variable, nd the density function of the volume. Let 3 x 1=3 : 4 g (x) = x3 ; g 1 (x) = 3 4 Letting X denote the radius of the sphere and Y the volume, Y = g (X ) so
fY (y ) = fX (g 1(y ))
"
@ 1 g (y ) @y #
=
3.8
. Let X and Y have the joint density 6 fXY (x; y ) = (x + y )2 ; 0 x 1; 0 y 1: 7 1), (iii)
P (X
a) By integrating over the appropriate regions, nd (i) P (X > Y ), (ii) P (X + Y 1=2). (i) We can let the limits of integration be 0 to 1 for x, and 0 to x for y , so
P (X > Y ) =
Z1Zx
1
7 3 (x + y ) 0 dx Z12 (8x3 x3 )dx = 7 Z01 2x3dx = 0 1 x4 1 = 1 : = 2 0 2 For this part, we could have also found this result by symmetry, noting that P (X > Y ) + P (X < Y ) = 1 and realizing that the joint density is symmetric in x and y so that each of the two terms should contribute equally. (ii) Here, we can again let the range of integration be 0 to 1 for x, but the range for y is now from 0 to 1 x. Thus, Z 1Z 1 x 6 2 P (X + Y 1) = 7 (x + y ) dydx 0 0 =
3 0
fXY dydx
= =
Z
0 0
= 2 x 1 x4 7 4
Z 2 7 (1 x )dx
1 3 1
6 1 (x + y )3 1 7 3 0 = 3: 14 0
dx
(iii) In this case the limits are easy; from 0 to 1=2 for x, and from 0 to 1 for y . Z 1=2 Z 1 6 2 P (X 1=2) = 7 (x + y ) dydx Z01=2 20 3 3 = 7 (x + 1) x ]dx = 2 7 1 = 14
0
b) Find the marginal densities of X and Y . This is largely carryover from part a);
fX = fX (x) =
Z1
1 0
1 2 = 7 (x + y )3 0 2 (x + 1)3 x3 ] = 7 2 = 7 3x2 + 3x + 1]; 0 x 1: As the joint density is symmetric in x and y , an identical exercise yields 2 fY (y ) = 3y 2 + 3y + 1]; 0 y 1: 7
Z 6 (x + y ) dy 7
2
fXY dy
(x + y )2 2 3y 2 + 3y + 1] 7 + 2 = 3 3y(x+ 3y )+ 1] ; 0 x 1: 2 y
6 7
fXY fY
(x + y )2 ; 0 y 1: 3x2 + 3x + 1] 3
3.12
. Let
x < 1;
x:
a) Find c. To do this, we need to integrate the joint density over its entire allowable region, set the result equal to 1, and solve for c. 1 = = = =
Z1Z1 Z
0 0
1 1 1Z x
x
fXY dydx y3
Z1 Z1
0
c(x2 y 2 )e x dydx
2
cx y c
3 e
x3
x x
4 c Z 1 x3 e x dx: = 3 0 Now at this point, we can proceed in two equivalent ways. We can do the integration by parts, getting (after a few splitting into parts)
xe
3
2cx
2c 3 e x dx
dx
x j1
0
3x e
2
x j1
0
6xe
x j1 +
0
Z1
0
6e x dx = 6;
or we can recognize that the above integral is the Gamma function with parameter 3+1 = 4, and (4) = (4 1)! = 6. This integral occurs su ciently often that it is worth learning to look for it. At any rate, 1 = 4 c 6 = 8c ! c = 1 : 3 8 b) Find the marginal densities. Well, we went a long way towards nding the marginal density for x in solving part a);
fX (x) =
Zx
= 4 cx3e x 3 = 1 x3 e x ; 0 x < 1 6 which we recognize as a Gamma( = 4; = 1) density. In solving for fY , we need to be a bit careful with the limits of integration; for a speci c value of y , x ranges from jy j to 1. It is also easier to split up the x2 and y 2 parts of the joint density in doing the integration.
fY (y ) =
c(x2 y 2 )e x dydx
Z1
jyj
cx2e x dx
2
Z1
jy j
cy 2e x dx
jyj
cx e
2
x j1 jyj
2cxe
x j1 + jy j
Z1
2ce
jy j x j1 + cy 2e x j1 jy j jyj
2ce
x dx
Z1
cy 2e x dx
and
fX jY (xjy ) =
jyj x 1:
3.18
and 0 elsewhere. a) Sketch the region of integration over which the density is positive and use it in determining limits of integration to answer the following questions. This is a bit di cult to do in a typesetting environment - the region is triangular with vertices at (0; 0), (1; 0), and (1; 1). b) Find k. Again, we need to integrate the density over its entire positive region, set the value to 1, and solve for k. 1 = = =
Z Zx
1 0 1 0 0 1
Z Z
k(x y )dydx y2
k xy x2 k dx
3 1
!x
0
dx
= kx = k 6 0 6 so k = 6. c) Find the marginal densities of X and Y . Starting with stu from part b),
fX (x) =
Zx
0
k(x y )dy
fY (y ) =
Z
y
k(x y )dx x2
= k 2 xy y ! 1 y k y2 y2 = k 2 2 ! 2 = k y2 y + 1 2 2 = 3(1 y ) ; 0 < y < 1: d) Find the conditional densities of Y given X and X given Y . f (x; y ) 6(x y ) fY jX (y jx) = XY = = 2x 2 y ; 0 < y < x 2 fX (x) 3x x fXY (x; y ) 6(x y ) = = 2 x y 2; y < x < 1 fX jY (xjy ) = 2 fY (y ) 3(1 y ) (1 y )
T1 and T2, with parameters and respectively. Find a) P (T1 > T2). OK. The limits of integration here are 0 to t1 for t2 , and 0 to 1 for t1 . P (T1 > T2 ) =
3.19
Z 1 Z t1
0 0 0 0
= = = =
Z 1 Z t1 Z1 Z1
0
fT1 (t1 )
Z t1
0
Z1
0 0
i dt cdf of an exponential Z1
0
( + ) 1
+ 0 = 1 + = + : Note that this does something sensible - if we let = , so that the two random variables are identically distributed, then the chances are equal that one is greater than the other. b) P (T1 > 2T2). This is much like part a), but the limits of integration for t2 are now from 0 to t1 =2.
P (T1 > 2T2) =
= 1
( + ) 1
dt1
Z 1 Z t1=
0 0
Z1
0
= 1 = 1
Z1
0
fT 1 1 e e
( +
t1 =2
=2)t1 dt
= +=2=2 : + =2 Note that this is equivalent to what we got for part a) if we change the value of the parameter of T2 from to =2. This ts with the result we found earlier (problem 2.61) where we found that if X has a Gamma( ; ) distribution, then cX has a Gamma( ; =c) distribution. The exponential distribution is a gamma distribution with = 1, so letting c = 2 the result follows. . If X1 is uniform on 0; 1], and, conditional on X1, X2 is uniform on 0; X1], nd the joint and marginal distributions of X1 and X2. Well, from the statement of the problem, the densities that we are given are fX1 and fX2 jX1 . These can easily be used to construct the joint density function, as
3.20
1 1 = 1 x = x ; 0 < x1 < 1; 0 < x2 < x1: 1 1 Now, starting with the joint we can integrate to get the marginals:
fX1 (x1 ) = 1; 0 < x1 < 1 (given) fX2 (x2 ) =
Z 1 dx x2 x = log(x )jx2
1 1
3.34
. Let N1 and N2 be independent random variables following Poisson distributions with parameters 1 and 2 . Show that the distribution of N = N1 + N2 is Poisson with parameter 1 + 2 .
pN (n) =
1 X
n1 = 1
pN1 N2 (n1 ; n n1 )
= = =
pN1 (n1 )pN2 (n n1 ) independence n1 = 1 n X pN1 (n1 )pN2 (n n1 ) pmfs both positive n1 =0 n X n1 e 1 n n1 e 2
1
1 X
n1 =0
n1 !
(n n1 )! 7
( 1+ = e n!
n! n1 n1 !(n n1 )! 1 n1 =0 ! n ( 1 + 2) X e n n1 n n1 = n! n1 =0 n1 1 2 ( 1 + 2) = e n! ( 1 + 2)n
2)
n X
n n1
+ 2.
. Let X and Y by independent standard normal random variables. Find the density of Z = X + Y , and show that Z is normally distributed as well. (Hint: Use the technique of completing the square to help in evaluating the integral.) As X and Y are independent,
fZ (z ) =
Z1
and as both x and y range from 1 to 1 we leave the integration limits alone. Z1 1 1 1 p exp 2 x2 p1 exp 2 (z x)2 dx fZ (z ) = 2 1 2 1 Z 1 exp 1 x2 + (z x)2] dx = 2 Z 1 2 1 1 exp 1 2x2 2xz + z 2 ] dx = 2 Z 1 2 1 1 exp 1 2(x 1 z )2 + 1 z 2 ] dx = 2 2 2 2 1 exp 1 z 2 Z 1 exp @ 1 x p1 z 2 A dx 2 = 2 4 2 1= 2 1 ! z 1 exp 1 p 2 q2 (1=p2)2 = 2 2 2 ! 1 p 2 z 1 = q p exp 2 2 2 ( 2)2
1
fX (x)fY (z x)dx
"
#1