Math 564, Theory of Probability, HW08

Zihan Zhu
Student ID: 23555445

1. X1 be the larger value of two fair die and X2 the smaller.

(a) Denote X and Y to be value of fair die, which follow discrete uniform
distribution and are independent. Then X1 = max(X, Y ) and X2 =
min(X, Y ).
P (X1 = x1 , X2 = x2 ) = P (max(X, Y ) = x1 , min(X, Y ) = x2 )
= P (X = x1 , Y = x2 ) + P (X = x2 , Y = x1 ) − P (X = Y )
= P (X = x1 )P (Y = x2 ) + P (X = x2 )P (Y = x1 ) − P (X − Y )
11 11 11
= + −
66 66 66
1
=
36
(1)
here x1 ∈ {1, 2, 3, 4, 5, 6}, x2 ∈ {1, 2, 3, 4, 5, 6}.
(b)
#{x1 = x2 } 1
P (X1 = X2 ) = = (2)
36 6
(c)
P (X1 = x1 ) = P (max(X, Y ) = x1 )
= P (X = x1 , Y ≤ x1 ) + P (Y = x1 , X ≤ x1 ) − P (X = x1 , Y = x1 )
= P (X = x1 )P (Y ≤ x1 ) + P (Y = x1 )P (X ≤ x1 ) − P (X = x1 , Y = x1 )
1 x1 1 x1 1
= + −
6 6 6 6 36
2x1 − 1
=
36
(3)

1
 here x1 ∈ {1, 2, 3, 4, 5, 6}.
P (X2 = x2 ) = P (min(X, Y ) = x2 )
= P (X = x2 , Y ≥ x2 ) + P (Y = x2 , X ≥ x2 ) − P (X = x2 , Y = x2 )
= P (X = x2 )P (Y ≥ x2 ) + P (Y = x2 )P (X ≥ x2 ) − P (X = x2 , Y = x2 )
1 7 − x2 1 7 − x2 1
= + −
6 6 6 6 36
13 − 2x2
=
36
(4)
here x2 ∈ {1, 2, 3, 4, 5, 6}.
(d)
P (X1 , X2 )
P (X2 = x2 |X1 ) =
P (X1 )
(5)
1
=
2X1 − 1
6
X 1 6
E(X2 |X1 ) = = (6)
x2 =1
2X1 − 1 2X1 − 1

2
2. (a)
Z Z Z Z
f (x1 , x2 )dx1 dx2 = x2 exp(−x1 x2 )dx1 dx2
x1 ≥0 0≤x2 ≤1
Z Z
=− exp(−x1 x2 )d(−x1 x2 )dx2
0≤x2 ≤1 x1 ≥0
Z
=− exp(−x1 x2 )|∞
0 dx2
0≤x2 ≤1
Z
= dx2 = 1
0≤x2 ≤1
(7)
and f (x1 , x2 ) ≥ 0, so this is bona fide density.
(b)
Z
fX1 (x1 ) = f (x1 , x2 )dx2
Z
= x2 exp(−x1 x2 )dx2
0≤x2 ≤1
Z
1
=− x2 exp(−x1 x2 )d(−x1 x2 )
x1 0≤x2 ≤1
Z (8)
x2 1 1
= − exp(−x1 x2 )|0 + exp(−x1 x2 )dx2
x1 x1 0≤x2 ≤1
1 1
= − exp(−x1 ) − 2 exp(−x1 x2 )|10
x1 x1
1 1 1
= 2 − ( + 2 )exp(−x1 )
x1 x1 x1
Z
fX2 (x2 ) = f (x1 , x2 )dx1
Z
= x2 exp(−x1 x2 )dx1
x1 ≥0
Z (9)
=− exp(−x1 x2 )d(−x1 x2 )
x1 ≥0
= −exp(−x1 x2 )|∞
0 = 1

(c)
f (x1 , x2 )
f (x1 |x2 ) =
f (x2 ) (10)
= x2 exp(−x1 x2 ), x1 ≥ 0

3
 f (x1 , x2 )
f (x2 |x1 ) =
f (x1 )
x2 exp(−x1 x2 )
= 1
x2
− ( x11 + x12 )exp(−x1 )
1 1
(11)
x21 x2 exp(−x1 x2 )
=
1 − (x1 + 1)exp(−x1 )
x21
= [x2 exp(−x2 )]
exp(x1 ) − (1 + x1 )
(d)
R ∞ R 1/2
1 0
f (x1 , x2 )dx1 x2
P (X1 > 1|X2 < 1/2) =
P (X2 < 1/2)
Z ∞ Z 1/2
=2 x2 exp(−x1 x2 )dx1 x2
1Z 0

= −2 exp(−x1 x2 )|∞1 dx2

(12)
0≤x2 ≤1/2
Z
=2 exp(−x2 )dx2
0≤x2 ≤1/2
1/2 2
= −2exp(−x2 )|0 =2− √
e
(e) Z 1
E(X2 |X1 ) = f (x2 |X1 )dx2
0
1
X12
Z
= [x2 exp(−x2 )]x2
0 exp(X1 ) − (1 + X1 )
Z 1 (13)
X12
= [x2 exp(−x2 )]x2
exp(X1 ) − (1 + X1 ) 0
2 X12
= (1 − )
e exp(X1 ) − (1 + X1 )

4
(f) Z ∞
E(X1 |X2 ) = x1 f (x1 |X2 )dx1
0
Z ∞
= x1 X2 exp(−x1 X2 )dx1
0
Z ∞
=− x1 exp(−x1 X2 )d(−x1 X2 )
0
Z ∞ (14)
= exp(−x1 X2 )dx1
0
Z ∞
1
= exp(−x1 X2 )dx1
X2 0
1
=
X2
Z ∞
E(X12 |X2 ) = x21 f (x1 |X2 )dx1
Z0 ∞
= x21 X2 exp(−x1 X2 )dx1
0
Z ∞
=− x21 exp(−x1 X2 )d(−x1 X2 )
Z 0∞ (15)
=2 x1 exp(−x1 X2 )dx1
0
Z ∞
2
= exp(−x1 X2 )dx1
X2 0
2
= 2
X2
V ar(X1 |X2 ) = E(X12 |X2 ) − E 2 (X1 |X2 )
2 1
= 2 − ( )2
X2 X2 (16)
1
= 2
X2

5
3. (a)
m x n m s n
P (X1 = x, S = s) = ( ) ( )1−x Ck−1
s
( )( )k−1−s
m+n m+n m+n m+n
s m s+x n k−s−x
= Ck−1 ( ) ( )
m+n m+n
(17)
(b)
P (S = s) = P (X1 = 1, S = s) + P (X1 = 0, S = s)
s m s+1 n
= Ck−1 ( ) ( )k−s−1
m+n m+n
s m s n (18)
+ Ck−1 ( )( )k−s
m+n m+n
s m s n
= Ck−1 ( )( )k−s−1
m+n m+n
P (X1 = 1, S = s)
P (X1 = 1|S = s) =
P (S = s)
s m s+1 n
Ck−1 ( m+n ) ( m+n )k−s−1
= (19)
s m s n
Ck−1 ( m+n ) ( m+n )k−s−1
m
=
m+n
(c)
x 1−x C s k−1−s
Cm Cn m−x Cn−1+x
P (X1 = x, S = s) = 1 s
Cm+n Cm+n−1
(20)
C s C k−1−s
= m sn
Cm+n
P (S = s) = P (X1 = 1, S = s) + P (X1 = 0, S = s)
s m s+1 n
= Ck−1 ( ) ( )k−s−1
m+n m+n
s m s n (21)
+ Ck−1 ( )( )k−s
m+n m+n
s m s n
= Ck−1 ( )( )k−s−1
m+n m+n
P (X1 = 1, S = s)
P (X1 = 1|S = s) =
P (S = s)
s m s+1 n
Ck−1 ( m+n ) ( m+n )k−s−1
= (22)
s m s n
Ck−1 ( m+n ) ( m+n )k−s−1
m
=
m+n

6
4. (a)
n
X
MX (t) = E(exp(tX)) = exp(it)Cni pi (1 − p)n−i
i=0
n
X (23)
= Cni [exp(t) + p]i (1 − p)n−i
i=0
= [1 − p + pexp(t)]n
KX (t) = logMX (t) = nlog[1 − p + pexp(t)] (24)
(b)
0 npexp(t) EXexp(t)
KX (t) = = =x (25)
1 − p + pexp(t) 1 − p + pexp(t)
x(1 − p) x(1 − p)
exp(t) = −→ t∗ (x) = log (26)
(n − x)p (n − x)p
(c)
∗
KX (x) = inf {KX (t) − tx} (27)
According to results in part (b),

∗ n(1 − p) n (1 − p)x
KX (x) = −KX (t∗ ) + t∗ x = −[ ] + xlog (28)
n−x (n − x)p

(d)
∗ n(1 − p) n (1 − p)np
−KX (np) = −[ ] + nplog =1 (29)
n − np (n − np)p
1
So P (X > np) < e