STAT 410
Fall 2016
Functions of One Random Variable
Let X be a continuous random variable.
Let Y = g ( X ).
What is the probability distribution of Y ?
Cumulative Distribution Function approach:
FY( y ) = P( Y y ) = P( g( X ) y ) =
f X ( x ) dx
{ x: g (x ) y }
Moment-Generating Function approach:
M Y ( t ) = E ( e Y t ) = E ( e g( X ) t ) =
(x ) t f ( x ) dx
X
Change-of-Variable Technique:
X continuous r.v. with p.d.f. f X ( x ).
Y = g( X)
dx
/d y = d [ g
1( y ) ]
/d y
f Y ( y ) = f X ( g 1( y ) )
1.
g ( x ) one-to-one, differentiable
dx
dy
Consider a continuous random variable X with p.d.f.
5
f X ( x ) = 6 x
0 < x <1
o.w.
Find the probability distribution of Y = 1 X 2 .
Support of X = { 0 < x < 1 }
Y= 1
X2
Support of Y = { y > 1 }
g ( x ) = 1/x 2
dx
/d y = 12
g 1( y ) = 1
1
= y /2
y 3/2
dx
dy
f Y ( y ) = f X ( g 1( y ) )
1
5
3
= ( 6 y /2 ) ( y /2 ) = 3 y 4
2
y > 1.
OR
5
f X ( x ) = 6 x
0
FX( x ) = x6
1
0 < x <1
o.w.
FY( y ) = P( Y y ) = P( 1 X2 y ) = P( X 1
= 1 y 3,
y > 1.
Consider a continuous random variable X with p.d.f.
2x
0
fX( x ) =
a)
) = 1 FX( 1
y > 1.
f Y ( y ) = F 'Y ( y ) = 3 y 4,
2.
x<0
0 x <1
x 1
0 < x <1
o.w.
Find the probability distribution of Y =
2x
fX( x ) =
0
0 < x <1
o.w.
X.
FX( x ) = x 2
x<0
0 x <1
x 1
y<0
F Y ( y ) = P ( Y y ) = P ( X y ) = 0.
y0
F Y ( y ) = P ( Y y ) = P ( X y ) = P ( X y 2 ) = F X ( y 2 ).
0y<1
F Y ( y ) = F X ( y 2 ) = y 4.
y1
F Y ( y ) = F X ( y 2 ) = 1.
FY( y ) = y 4
1
y<0
0 y <1
y 1
f Y ( y ) = F 'Y ( y ) = 4 y
0 < y <1
o.w.
OR
g( x ) =
f Y ( y ) = f X ( g 1( y ) )
b)
dx
dy
FW( w) = P(W w ) = P(
1
1 )2 =
f W ( w ) = F 'W ( w ) =
0 < y < 1.
1
.
X +1
1
<w<1
2
0<x<1
/d y = 2 y
= ( 2 y 2 ) ( 2 y ) = 4 y 3,
Find the probability distribution of W =
= 1(
dx
g 1( y ) = y 2
1
1
1
w ) = P( X
1 ) = 1 FX(
1)
X +1
w
w
w w
2
1
< w < 1.
2
22w
1
< w < 1.
2
OR
g( x ) =
1
x +1
f Y ( y ) = f X ( g 1( y ) )
g 1( w ) =
dx
dy
[ 2 ( w1
dx
1)] (
w2
) =
22w
w3
/d w =
1
< w < 1.
2
w2
3.
Consider a continuous random variable X with the p.d.f. f X ( x ) =
a)
Let Y = 1
24
x4
, x > 2.
. Find the p.d.f. of Y, f Y ( y ).
Support of X = { x > 2 }
Y=1
Support of Y = { 0 < y < 1 2 }
g ( x ) = 1/x
dx
g 1 ( y ) = 1/y
dx
dy
f Y ( y ) = f X ( g 1( y ) )
= ( 24 y 4 ) ( y 2 ) = 24 y 2,
/d y = 1/y 2
0 < y < 1 2.
OR
FX( x ) = 1
x3
x > 2.
F Y ( y ) = P ( Y y ) = P ( 1 X y ) = P ( X 1 y ) = 1 F X ( 1 y ) = 8 y 3,
0 < y < 1 2.
f Y ( y ) = 24 y 2, 0 < y < 1/2 .
b)
Find the probability distribution of Y = 1 X 2 .
Support of X = { x > 2 }
Y= 1
g ( x ) = 1/x 2
Support of Y = { 0 < y < 1 4 }
g 1( y ) = 1
1
= y /2
dx
/d y = 12 y 3/2
f Y ( y ) = f X ( g 1( y ) )
dx
dy
= ( 24 y 2 ) (
1 3/2
y
) = 12 y 1/2 = 12
2
y,
0 < y < 1 4.
OR
FX( x ) = 1
x3
x > 2.
FY( y ) = P( Y y) = P( 1
= 1 FX( 1
X2
y) = P(X 1
) = 8 y 3/2,
0 < y < 1 4.
f Y ( y ) = 12 y 1/2 = 12 y , 0 < y < 1/4 .
4.
The p.d.f. of X is
f X ( x ) = x 1, 0 < x < 1, 0 < < . Let Y = 2 ln X.
How is Y distributed?
a)
Determine the probability distribution of Y by finding the c.d.f. of Y
F Y ( y ) = P ( Y y ) = P ( 2 ln X y ).
Hint: Find F X ( x ) first.
F X ( x ) = x ,
0 < x < 1.
0<x<1
y = 2 ln x
F Y ( y ) = P ( Y y ) = P ( 2 ln X y ) = P ( X e
f Y ( y ) = F Y' ( y ) =
1 y/2
e
,
2
y > 0.
y>0
y 2
/ )
= 1e
y 2
y > 0.
b)
Y has Exponential distribution with mean 2.
Determine the probability distribution of Y by finding the m.g.f. of Y
M Y ( t ) = E ( e Y t ) = E ( e 2 ln X t ).
M Y ( t ) = E ( e Y t ) = E ( e 2 ln X t ) = E ( X 2 t ) =
(x
2 t
x 1 dx
0
1
2 t 1
dx =
c)
2t
1
,
1 2t
t<
1
.
2
Y has Exponential distribution with mean 2.
Determine the probability distribution of Y by finding the p.d.f. of Y, f Y ( y ),
using the change-of-variable technique.
y = g ( x ) = 2 ln x
dx
fY( y ) = fX( g
/d y =
y
x = g 1 ( y ) = = e /2
1 y/2
e
2
dx
(y))
dy
y 2
= e
1 y 2
e
2
1 y/2
e
,
2
y > 0.
Y has Exponential distribution with mean 2.
5.
Let Z be a N ( 0, 1 ) standard normal random variable.
Show that X = Z 2 has a chi-square distribution with 1 degree of freedom.
2
MX( t ) = E( et Z ) =
t z2
z 2 2 (1 2 t )
since
( 1 2 t )1 2 e (z 2
X has a
) (1 2 t )
z2 2
dz =
dz
(1 2 t )
12
t < 1/2,
is the p.d.f. of a N ( 0,
2(1)
1
) random variable.
1 2t
distribution.
OR
FX( x ) = P( X x ) = P( Z2 x ) = P(
z2 2
x Z
x ) F Z ( x ).
d z = FZ(
x)
f X ( x ) = F 'X ( x ) =
f Z ( x )
f Z ( x )
2 x
2 x
x 2
x 2
1 1
1 1
e
e
=
2
2
2
2
x
x
X has a
2(1)
21
1 2
1
(1 2 ) 2 1 2
distribution.
x 2
(1 2 ) 1
x 2
x > 0.
6.
Consider a continuous random variable X with p.d.f.
3 < x < 1
0< x<2
otherwise
0.2
f X ( x ) = 0.3
0
Find the probability distribution of Y = X 2.
Case 1:
y<0
P( X2 y ) = 0
y0
FY( y ) = P( Y y ) = P( X2 y ) = P(
0y<1
F Y ( y ) = 0.3
Case 2:
1y<4
F Y ( y ) = 0.
y <1
y <2
y.
F Y ( y ) = 0.2 ( 1 +
y ) + 0.3 y .
y X
y)
Case 3:
4y<9
F Y ( y ) = 0.2 ( 1 +
Case 4:
y9
y <3
y ) + 0.6.
F Y ( y ) = 1.
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
Consider a continuous random variable X, with p.d.f. f and c.d.f. F, where F is strictly
increasing on some interval I, F = 0 to the left of I, and F = 1 to the right of I. I may be a
bounded interval or an unbounded interval such as the whole real line. F 1 ( u ) is then well
defined for 0 < u < 1.
Fact 1:
Let U ~ Uniform ( 0, 1 ), and let X = F 1 ( U ). Then the c.d.f. of X is F.
Proof:
P ( X x ) = P ( F 1 ( U ) x ) = P ( U F ( x ) ) = F ( x ).
Fact 2:
Let U = F ( X ); then U has a Uniform ( 0, 1 ) distribution.
Proof:
P ( U u ) = P ( F ( X ) u ) = P ( X F 1 ( u ) ) = F ( F 1 ( u ) ) = u.
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
7.
Let X have a logistic distribution with p.d.f.
f(x) =
ex
(1 + e )
x 2
< x < .
Show that
Y =
1
1 + eX
has a U ( 0, 1 ) distribution.
FX( x ) =
1
1 + e x
< x < .
Y = F X ( X ) has a Uniform ( 0, 1 ) distribution by Fact 2.
OR
Weibull distribution:
b
f X ( x ) = a b x b 1 e a x ,
x > 0,
where a > 0, b > 0.
8.
Let X have an exponential distribution with = 1; that is, the p.d.f. of X is
f ( x ) = e x, 0 < x < . Let T be defined by T = ln X.
a)
Show that the p.d.f. of T is
t
g ( t ) = e t e e , < t < ,
which is the p.d.f. of an extreme value distribution.
x>0
dx
= et
dt
x = et
t = ln ( x )
<t<
g (t ) = fT(t ) = fX(et )
dx
dt
t
= et e e ,
< t < .
OR
FX( x ) = 1 e
x > 0.
t
t
t
FT( t ) = P( T t ) = P( X e ) = FX( e ) = 1 e e ,
< t < .
t
g ( t ) = f T ( t ) = e t e e , < t < .
b)
Let W be defined by T = + ln W, where < < and > 0.
Show that W has a Weibull distribution.
dt
=
w
dw
t = + ln w
f W ( t ) = f T ( + ln w )
+ ln w
dt
= e + ln w e e
dw
w
= e w 1 e e w ,
w > 0.
9.
Let X be a continuous random variable with probability density function
f X ( x ) = x 1 e x ,
where > 0, > 0.
x > 0,
( X has a Weibull distribution. ) Consider Y = X . What is the probability
distribution of Y?
Y = X
x>0
X = g 1( y ) = Y 1
dx
1 1 1
=
.
y
dy
y>0
f Y ( y ) = f X ( g 1( y ) )
dx
dy
= y ( 1 ) /
e y y
1
1 1
= e
OR
x
FX( x ) =
x
1 u du =
,
1 e
u e
x > 0.
0
y
FY( y ) = P( Y y ) = P( X y1 ) = FX( y1 ) = 1 e
,
fY( y ) = e
y > 0.
Y has Exponential distribution with mean 1 .
y > 0.
y > 0.
