You are on page 1of 260

Solutions Manual for

Probability and Random Processes for


Electrical and Computer Engineers

John A. Gubner
University of WisconsinMadison

File Generated July 13, 2007


CHAPTER 1
Problem Solutions

1. = {1, 2, 3, 4, 5, 6}.

2. = {0, 1, 2, . . . , 24, 25}.

3. = [0, ). RTT > 10 ms is given by the event (10, ).

4. (a) = {(x, y) IR2 : x2 + y2 100}.


(b) {(x, y) IR2 : 4 x2 + y2 25}.

5. (a) [2, 3] c = (, 2) (3, ).


(b) (1, 3) (2, 4) = (1, 4).
(c) (1, 3) [2, 4) = [2, 3).
(d) (3, 6] \ (5, 7) = (3, 5].

6. Sketches:

y y y

1
1 1
x x x

B0 B1 B1

y y y

x x x
3

C1 H3 J3

1
2 Chapter 1 Problem Solutions

y y
3
3
x x
3 3

U
H3 J3 = M 3 H3 U J3 = N 3

y 4
2 3

x x
2 3 4

U U
M2 N 3 = M2 M4 N3

     
7. (a) [1, 4] [0, 2] [3, 5] = [1, 4] [0, 2] [1, 4] [3, 5] = [1, 2] [3, 4].
(b)
 c
[0, 1] [2, 3] = [0, 1] c [2, 3] c
h i h i
= (, 0) (1, ) (, 2) (3, )
 h i
= (, 0) (, 2) (3, )
 h i
(1, ) (, 2) (3, )

= (, 0) (1, 2) (3, ).

\
(c) ( 1n , n1 ) = {0}.
n=1
\
1
(d) [0, 3 + 2n ) = [0, 3].
n=1
[
1
(e) [5, 7 3n ] = [5, 7).
n=1
[
(f) [0, n] = [0, ).
n=1
Chapter 1 Problem Solutions 3

8. We first let C A and show that for all B, (A B) C = A (B C). Write

A (B C) = (A B) (A C), by the distributive law,


= (A B) C, since C A A C = C.

For the second part of the problem, suppose (A B) C = A (B C). We must show
that C A. Let C. Then (A B) C. But then A (B C), which
implies A.
9. Let I := { : A B}. We must show that A I = A B.

: Let A I. Then A and I. Therefore, B, and then A B.


: Let A B. Then A and B. We must show that I too. In other
words, we must show that A B. But we already have B.

10. The function f : (, ) [0, ) with f (x) = x3 is not well defined because not all
values of f (x) lie in the claimed co-domain [0, ).
11. (a) The function will be invertible if Y = [1, 1].
(b) {x : f (x) 1/2} = [ /2, /6].
(c) {x : f (x) < 0} = [ /2, 0).
12. (a) Since f is not one-to-one, no choice of co-domain Y can make f : [0, ] Y
invertible.
(b) {x : f (x) 1/2} = [0, /6] [5 /6, ].
(c) {x : f (x) < 0} = .
13. For B IR,

X, 0 B and 1 B,

1 A, 1 B but 0
/ B,
f (B) = c,

A 0 B but 1
/ B,

, 0/ B and 1
/ B.

14. Let f : X Y be a function such that f takes only n distinct values, say y1 , . . . , yn .
Let B Y be such that f 1 (B) is nonempty. By definition, each x f 1 (B) has the
property that f (x) B. But f (x) must be one of the values y1 , . . . , yn , say yi . Now
f (x) = yi if and only if x Ai := f 1 ({yi }). Hence,
[
f 1 (B) = Ai .
i:yi B

15. (a) f (x) B c f (x) / f 1 (B) x f 1 (B) c .


/Bx

[
(b) f (x) Bn if and only if f (x) Bn for some n; i.e., if and only if x f 1 (Bn )
n=1

[
for some n. But this says that x f 1 (Bn ).
n=1
4 Chapter 1 Problem Solutions


\
(c) f (x) Bn if and only if f (x) Bn for all n; i.e., if and only if x f 1 (Bn )
n=1

\
for all n. But this says that x f 1 (Bn ).
n=1
S S S
16. If B = i {bi } and C = i {ci }, put a2i := bi and a2i1 := ci . Then A = i ai = B C
is countable.
S
17. Since each Ci is countable, we can write Ci = j ci j . It then follows that

[ [
[
B := Ci = {ci j }
i=1 i=1 j=1

is a doubly indexed sequence and is therefore countable as shown in the text.


S
18. Let A = m {am } be a countable set, and let B A. We must show that B is countable.
If B = , were done by definition. Otherwise, there is at least one element
S
of B in
A, say ak . Then put bn := an if an B, and put bn := ak if an
/ B. Then n {bn } = B
and we see that B is countable.
19. Let A B where A is uncountable. We must show that B is uncountable. We prove
this by contradiction. Suppose that B is countable. Then by the previous problem, A
is countable, contradicting the assumption that A is uncountable.
20. Suppose A is countable and B is uncountable. We must show that AB is uncountable.
We prove this by contradiction. Suppose that A B is countable. Then since B
A B, we would have B countable as well, contradicting the assumption that B is
uncountable.
21. MATLAB. OMITTED.
22. MATLAB. Intuitive explanation: Using only the numbers 1, 2, 3, 4, 5, 6, consider how
many ways there are to write the following numbers:
2 = 1+1 1 way, 1/36 = 0.0278
3 = 1+2 = 2+1 2 ways, 2/36 = 0.0556
4 = 1+3 = 2+2 = 3+1 3 ways, 3/36 = 0.0833
5 = 1+4 = 2+3 = 3+2 = 4+1 4 ways, 4/36 = 0.1111
6 = 1+5 = 2+4 = 3+3 = 4+2 = 5+1 5 ways, 5/36 = 0.1389
7 = 1+6 = 2+5 = 3+4 = 4+3 = 5+2 = 6+1 6 ways, 6/36 = 0.1667
8 = 2+6 = 3+5 = 4+4 = 5+3 = 6+2 5 ways, 5/36 = 0.1389
9 = 3+6 = 4+5 = 5+4 = 6+3 4 ways, 4/36 = 0.1111
10 = 4+6 = 5+5 = 6+4 3 ways, 3/36 = 0.0833
11 = 5+6 = 6+5 2 ways, 2/36 = 0.0556
12 = 6+6 1 way, 1/36 = 0.0278
36 ways, 36/36 = 1

23. Take := {1, . . . , 26} and put


|A| |A|
P(A) := = .
|| 26
Chapter 1 Problem Solutions 5

The event that a vowel is chosen is V = {1, 5, 9, 15, 21}, and P(V ) = |V |/26 = 5/26.
24. Let := {(i, j) : 1 i, j 26 and i 6= j}. For A , put P(A) := |A|/||. The event
that a vowel is chosen followed by a consonant is

Bvc = (i, j) : i = 1, 5, 9, 15, or 21 and j {1, . . . , 26} \ {1, 5, 9, 15, 21} .

Similarly, the event that a consonant is followed by a vowel is



Bcv = (i, j) : i {1, . . . , 26} \ {1, 5, 9, 15, 21} and j = 1, 5, 9, 15, or 21 .

We need to compute

|Bvc | + |Bcv | 5 (26 5) + (26 5) 5 21


P(Bvc Bcv ) = = = 0.323.
|| 650 65

The event that two vowels are chosen is



Bvv = (i, j) : i, j {1, 5, 9, 15, 21} with i 6= j ,

and P(Bvv ) = |Bvv |/|| = 20/650 = 2/65 .031.


25. MATLAB. The code for simulating the drawing of a face card is

% Simulation of Drawing a Face Card


%
n = 10000; % Number of draws.
X = ceil(52*rand(1,n));
faces = (41 <= X & X <= 52);
nfaces = sum(faces);
fprintf(There were %g face cards in %g draws.\n,nfaces,n)

26. Since 9 pm to 7 am is 10 hours, take := [0, 10]. The probability that the baby wakes
up during a time interval 0 t1 < t2 10 is
Z t2
1
P([t1 ,t2 ]) := d .
t1 10
R2
Hence, P([2, 10] c ) = P([0, 2]) = 0 1/10 d = 1/5.
27. Starting with the equations

SN = 1 + z + z2 + + zN2 + zN1
zSN = z + z2 + + zN2 + zN1 + zN ,

subtract the second line from the first. Canceling common terms leaves

SN zSN = 1 zN , or SN (1 z) = 1 zN .

If z 6= 1, we can divide both sides by 1 z to get SN = (1 zN )/(1 z).


6 Chapter 1 Problem Solutions

28. Let x = p(1). Then p(2) = 2p(1) = 2x, p(3) = 2p(2) = 22 x, p(4) = 2p(3) = 23 x,
p(5) = 24 x, and p(6) = 25 x. In general, p( ) = 2 1 x and we can write
6 6 5
1 26
1 = p( ) = 2 1 x = x 2 =
12
x = 63x.
=1 =1 =0

Hence, x = 1/63, and p( ) = 2 1 /63 for = 1, . . . , 6.


29. (a) By inclusionexclusion, P(A B) = P(A) + P(B) P(A B), which can be re-
arranged as P(A B) = P(A) + P(B) P(A B).
(b) Since P(A) = P(A B) + P(A B c ),

P(A B c ) = P(A) P(A B) = P(A B) P(B), by part (a).

(c) Since B and A B c are disjoint,

P(B (A B c )) = P(B) + P(A B c ) = P(A B), by part (b).

(d) By De Morgans law, P(A c B c ) = P([A B] c ) = 1 P(A B).


30. We must check the four axioms of a probability measure. First,

P() = P1 () + (1 )P2 () = 0 + (1 ) 0 = 0.

Second,

P(A) = P1 (A) + (1 )P2 (A) 0 + (1 ) 0 = 0.

Third,
     
[ [ [
P An = P1 An + (1 )P2 An
n=1 n=1 n=1

= P1 (An ) + (1 ) P2 (An )
n=1 n=1

= [ P1 (An ) + (1 )P2 (An )]
n=1

= P(An ).
n=1

Fourth, P() = P1 () + (1 )P2 () = + (1 ) = 1.


/ S, () = 0. Second, by definition, (A) 0. Third, for disjoint
31. First, since 0
An , suppose 0 n An . Then 0 Am for some S m,and 0 / An for n 6= m. Then
(Am ) = 1 and (An ) = 0 for n 6= m. Hence, n An = 1 and n (An ) = (Am ) =
S S 
1. A similar analysis shows that if 0 / n An then n An and n (An ) are both
zero. Finally, since 0 , () = 1.
Chapter 1 Problem Solutions 7

32. Starting with the assumption that for any two disjoint events A and B, P(A B) =
P(A) + P(B), we have that for N = 2,
 N  N
[
P An = P(An ). ()
n=1 n=1

Now we must show that if () holds for any N 2, then () holds for N + 1. Write
 N+1   N  
[ [
P An = P An AN+1
n=1 n=1
 N 
[
= P An + P(AN+1 ), additivity for two events,
n=1
N
= P(An ) + P(AN+1 ), by (),
n=1
N+1
= P(An ).
n=1

c F c F , it is easy to see that


33. Since An := Fn Fn1 1 n

N
[ N
[
An Fn .
n=1 n=1
S
The hard part is to show the reverse inclusion . Suppose Nn=1 Fn . Then Fn
for some n in the range 1, . . . , N. However, may belong to Fn for several values of
n since the Fn may not be disjoint. Let
k := min{n : Fn and 1 n N}.
In other words, 1 k N and Fk , but
/ Fn for n < k; in symbols,
c
Fk Fk1 F1c =: Ak .
S S S
Hence, Ak Nn=1 An . The proof that n=1 An n=1 Fn is similar except that
k := min{n : Fn and n 1}.
34. For arbitrary events Fn , let An be as in the preceding problem. We can then write
   
[ [
P Fn = P An = P(An ), since the An are disjoint,
n=1 n=1 n=1
N
= lim
N
P(An ), by def. of infinite sum,
n=1
 N 
[
= lim P An
N
n=1
 N 
[
= lim P Fn .
N
n=1
8 Chapter 1 Problem Solutions

35. For arbitrary events Gn , put Fn := Gnc . Then


   
\ [
P Gn = 1 P Fn , by De Morgans law,
n=1 n=1
 N 
[
= 1 lim P Fn , by the preceding problem,
N
n=1
  N 
\
= 1 lim 1 P Gn , by De Morgans law,
N
n=1
 N 
\
= lim P Gn .
N
n=1

36. By the inclusionexclusion formula,

P(A B) = P(A) + P(B) P(A B) P(A) + P(B).

This establishes the union bound for N = 2. Now suppose the union bound holds for
some N 2. We must show it holds for N + 1. Write
 N+1   N  
[ [
P Fn = P Fn FN+1
n=1 n=1
 N 
[
P Fn + P(FN+1 ), by the union bound for two events,
n=1
N
P(Fn ) + P(FN+1 ), by the union bound for N events,
n=1
N+1
= P(Fn ).
n=1

37. To establish the union bound for a countable sequence of events, we proceed as fol-
c F c F be disjoint with S A = S F . Then
lows. Let An := Fn Fn1 1 n n=1 n n=1 n
   
[ [
P Fn = P An
n=1 n=1

= P(An ), since the An are disjoint,
n=1

P(Fn ), since An Fn .
n=1
S
38. Following the hint, we put Gn := k=n Bk so that we can write
     N 
\ [ \ \
P Bk = P Gn = lim P Gn , limit property of P,
N
n=1 k=n n=1 n=1
Chapter 1 Problem Solutions 9

= lim P(GN ), since Gn Gn+1 ,


N
 
[
= lim P Bk , definition of GN ,
N
k=N

lim
N
P(Bk ), union bound.
k=N

This last limit must be zero since


k=1 P(Bk ) < .

39. In this problem, the probability of an interval is its length.


(a) P(A0 ) = 1, P(A1 ) = 2/3, P(A2 ) = 4/9 = (2/3)2 , and P(A3 ) = 8/27 = (2/3)3 .
(b) P(An ) = (2/3)n .
(c) Write
   N 
\ \
P(A) = P An = lim P An , limit property of P,
N
n=1 n=1
= lim P(AN ), since An An+1 ,
N
= lim (2/3)N = 0.
N

40. Consider the collection consisting of the empty set along with all unions of the form
S
i Aki for some finite subsequence of distinct elements from {1, . . . , n}. We first show
that this collection is a -field. First, it contains by definition. Second, since
A1 , . . . , An is a partition,  
[ c [
Aki = Ami ,
i i

where mi is the subsequence {1, . . . , n} \ {ki }. Hence, the collection is closed under
complementation. Third,
[ [ 
[
Akn,i = Am j ,
n=1 i j

where an integer l {1, . . . , n} is in {m j } if and only if kn,i = l for some n and some i.
This shows that the collection is a -field. Finally, since every element in our collec-
tion must be contained in every -field that contains A1 , . . . , An , our collection must
be (A1 , . . . , An ).
41. We claim that A is not a -field. Our proof is by contradiction: We assume A is a
-field and derive a contradiction. Consider the set

\
[0, 1/2n ) = {0}.
n=1

Since [0, 1/2n ) Cn An A , the intersection must be in A since we are assuming


A is a -field. Hence, {0} A . Now, any set in A must belong to some An . By the
10 Chapter 1 Problem Solutions

preceding problem, every set in An must be a finite union of sets from Cn . However,
the singleton set {0} cannot be expressed as a finite union of sets from any Cn . Hence,
{0} /A.
42. Let Ai := X 1 ({xi }) for i = 1, . . . , n. By the problems mentioned in the hint, for any
subset B, if X 1 (B) 6= , then
[
X 1 (B) = Ai (A1 , . . . , An ).
i:xi B

It follows that the smallest -field containing all the X 1 (B) is (A1 , . . . , An ).

43. (a) F = , A, B, {3}, {1, 2}, {4, 5}, {1, 2, 4, 5},
(b) The corresponding probabilities are 0, 5/8, 7/8, 1/2, 1/8, 3/8, 1/2, 1.
(c) Since {1}
/ F , P({1}) is not defined.
44. Suppose that a -field A contains an infinite sequence Fn of sets. If the sequence is
not disjoint, we can construct a new sequence An that is disjoint with each An A .
Let a = a1 , a2 , . . . be an infinite sequence of zeros and ones. Then A contains each
union of the form [
Ai .
i:ai =1

Furthermore, since the Ai are disjoint, each sequence a gives a different union, and
we know from the text that the number of infinite sequences a is uncountably infinite.
T T
45. (a) First, since is in each A , A . Second, if TA A , then A A
for each , and so A c A for each . Hence, A c A . Third, if An A
S
for all n,
S
then for
T
each n and each , An A . Then n An A for each ,
and so n An A .
(b) We first note that

A1 = {, {1}, {2}, {3, 4}, {2, 3, 4}, {1, 3, 4}, {1, 2}, }

and
A2 = {, {1}, {3}, {2, 4}, {2, 3, 4}, {1, 2, 4}, {1, 3}, }.
It is then easy to see that

A1 A2 = {, {1}, {2, 3, 4}, }.


T
(c) First note that by part (a), A :C A A is a -field, and since C A for each
T
A , the -field A :C A A contains C . Finally, if D is any -field that contains
C , then D is one of the A s in the intersection. Hence,
\
C A D.
A :C A
T
Thus A :C A A is the smallest -field that contains C .
Chapter 1 Problem Solutions 11

46. The union of two -fields is not always a -field. Here is an example. Let :=
{1, 2, 3, 4}, and put
 
F := , {1, 2}, {3, 4}, and G := , {1, 3}, {2, 4}, .

Then 
F G = , {1, 2}, {3, 4}, {1, 3}, {2, 4},
is not a -field since it does not contain {1, 2} {1, 3} = {1}.
47. Let denote the positive integers, and let A denote the collection of subsets A such
that either A or A c is finite.

(a) Let E denote the subset of even integers. Then E does not belong to A since
neither E nor E c (the odd integers) is a finite set.
(b) To show that A is closed under finite unions, we consider two cases. First
suppose that A1 , . . . , An are all finite. Then
n n
[
Ai |Ai | < ,

i=1 i=1
Sn
and so i=1 Ai A . In the second case, suppose that some A cj is finite. Then
 n c n
[ \
Ai = Aic A cj .
i=1 i=1
Sn
Hence, the complement of i=1 Ai is finite, and so the union belongs to A .
S
(c) A is not a -field. To see this, put Ai := {2i} for i = 1, 2, . . . . Then i=1 Ai =
E
/ A by part (a).

48. Let be an uncountable set. Let A denote the collection of all subsets A such that
either A is countable or A c is countable. We show that A is a -field. First, the empty
set is countable. Second, if A A , we must show that A c A . There are two cases.
If A is countable, then the complement of A c is A, and so A c A . If A c is countable,
then A c A . Third, let A1S, A2 , . . . belong to A . There are two cases to consider. If
all An are countable, then n An is also countable by an earlier problem. Otherwise,
if some Amc is countable, then write
 c
[ \
An = Anc Amc .
n=1 n=1

Since
S
the subset of a countable set is countable, we see that the complement of
n=1 n is countable, and thus the union belongs to A .
A
T 1 1
49. (a) Since (a, b] = n=1 (a, b + n ), and since each (a, b + n ) B, (a, b] B.
T
(b) Since {a} = 1 1 1 1
n=1 (a n , a+ n ), and since each (a n , a+ n ) B, the singleton
{a} B.
12 Chapter 1 Problem Solutions

(c) Since by part (b), singleton sets are Borel sets, and since A is a countable union
of Borel sets, A B; i.e., A is a Borel set.
(d) Using part (a), write
 
 \
1
(a, b] = (a, b + n)
n=1
 N 
\
= lim (a, b + 1n ) , limit property of probability,
N
n=1

= lim (a, b + N1 ) , decreasing sets,
N
= lim (b + N1 ) a, characterization of ,
N
= b a.

Similarly, using part (b), we can write


 
 \
1 1
{a} = (a n,a+ n)
n=1
 N 
\
1 1
= lim (a n,a+ n) , limit property of probability,
N
n=1

= lim (a N1 , a + N1 ) , decreasing sets,
N
= lim 2/N, characterization of ,
N
= 0.

50. Let I denote the collection of open intervals, and let O denote the collection of open
sets. We need to show that (I ) = (O). Since I O, every -field containing
O also contains I . Hence, the smallest -field containing O contains I ; i.e., I
(O). By the definition of the smallest -field containing I , it follows that (I )
(O). Now, if we can show that O (I ), then it will similarly follow that (O)
(I ). Recall that in the problem statement, it was shown that every open set U can
be written as a countable union of open intervals. This means U (I ). This proves
that O (I ) as required.
51. MATLAB. Chips from S1 are 80% reliable; chips from S2 are 70% reliable.
52. Observe that
 N(Od,S1 ) 
N(Ow,S1 ) = N(OS1 ) N(Od,S1 ) = N(OS1 ) 1
N(OS1 )

and
 N(Od,S2 ) 
N(Ow,S2 ) = N(OS2 ) N(Od,S2 ) = N(OS2 ) 1 .
N(OS2 )
Chapter 1 Problem Solutions 13

53. First write


P(A [B C]) P(B C) P([A B] C)
P(A|B C) P(B|C) = = = P(A B|C).
P(B C) P(C) P(C)
From this formula, we can isolate the equation
P([A B] C)
P(A|B C) P(B|C) = .
P(C)
Multiplying through by P(C) yields P(A|B C) P(B|C) P(C) = P(A B C).
54. (a) P(MM) = 140/(140 + 60) = 140/200 = 14/20 = 7/10 = 0.7. Then P(HT) =
1 P(MM) = 0.3.
(b) Let D denote the event that a workstation is defective. Then

P(D) = P(D|MM)P(MM) + P(D|HT)P(HT)


= (.1)(.7) + (.2)(.3)
= .07 + .06 = 0.13.

(c) Write
P(D|MM)P(MM) .07 7
P(MM|D) = = = .
P(D) .13 13
55. Let O denote the event that a cell is overloaded, and let B denote the event that a call
is blocked. The problem statement tells us that

P(O) = 1/3, P(B|O) = 3/10, and P(B|O c ) = 1/10.

To find P(O|B), first write


P(B|O)P(O) 3/10 1/3
P(O|B) = = .
P(B) P(B)
Next compute

P(B) = P(B|O)P(O) + P(B|O c )P(O c )


= (3/10)(1/3) + (1/10)(2/3) = 5/30 = 1/6.

We conclude that
1/10 6 3
P(O|B) = = = = 0.6.
1/6 10 5
56. The problem statement tells us that P(R1 |T0 ) = and P(R0 |T1 ) = . We also know
that
1 = P() = P(T0 T1 ) = P(T0 ) + P(T1 ).
The problem statement tells us that these last two probabilities are the same; hence
they are both equal to 1/2. To find P(T1 |R1 ), we begin by writing
P(R1 |T1 )P(T1 )
P(T1 |R1 ) = .
P(R1 )
14 Chapter 1 Problem Solutions

Next, we note that P(R1 |T1 ) = 1 P(R0 |T1 ) = 1 . By the law of total probability,

P(R1 ) = P(R1 |T1 )P(T1 ) + P(R1 |T0 )P(T0 )


= (1 )(1/2) + (1/2) = (1 + )/2.

So,
(1 )(1/2) 1
P(T1 |R1 ) = = .
(1 + )/2 1 +

57. Let H denote the event that a student does the homework, and let E denote the event
that a student passes the exam. Then the problem statement tells us that

P(E|H) = .8, P(E|H c ) = .1, and P(H) = .6.

We need to compute P(E) and P(H|E). To begin, write

P(E) = P(E|H)P(H) + P(E|H c )P(H c )


= (.8)(.6) + (.1)(1 .6) = .48 + .04 = .52.

Next,
P(E|H)P(H) .48 12
P(H|E) = = = .
P(E) .52 13

58. The problem statement tells us that

P(AF |CF ) = 1/3, P(AF |CFc ) = 1/10, and P(CF ) = 1/4.

We must compute

P(AF |CF )P(CF ) (1/3)(1/4) 1/12


P(CF |AF ) = = = .
P(AF ) P(AF ) P(AF )

To compute the denominator, write

P(AF ) = P(AF |CF )P(CF ) + P(AF |CFc )P(CFc )


= (1/3)(1/4) + (1/10)(1 1/4) = 1/12 + 3/40 = 19/120.

It then follows that


1 120 10
P(CF |AF ) = = .
12 19 19
59. Let F denote the event that a patient receives a flu shot. Let S, M, and R denote the
events that Sue, Minnie, or Robin sees the patient. The problem tells us that

P(S) = .2, P(M) = .4, P(R) = .4, P(F|S) = .6, P(F|M) = .3, and P(F|R) = .1.

We must compute

P(F|S)P(S) (.6)(.2) .12


P(S|F) = = = .
P(F) P(F) P(F)
Chapter 1 Problem Solutions 15

Next,

P(F) = P(F|S)P(S) + P(F|M)P(M) + P(F|R)P(R)


= (.6)(.2) + (.3)(.4) + (.1)(.4) = .12 + .12 + .04 = 0.28.

Thus,
12 100 3
P(S|F) = = .
100 28 7
60. (a) Let = {1, 2, 3, 4, 5} with P(A) := |A|/||. Without loss of generality, let 1
and 2 correspond to the two defective chips. Then D := {1, 2} is the event that
a defective chip is tested. Hence, P(D) = |D|/5 = 2/5.
(b) Your friends information tells you that of the three chips you may test, one is
defective and two are not. Hence, the conditional probability that the chip you
test is defective is 1/3.
(c) Yes, your intuition is correct. To prove this, we construct a sample space and
probability measure and compute the desired conditional probability. Let

:= {(i, j, k) : i < j and k 6= i, k 6= j},

where i, j, k {1, 2, 3, 4, 5}. Here i and j are the chips taken by the friend, and
k is the chip that you test. We again take 1 and 2 to be the defective chips. The
10 possibilities for i and j are

12 13 14 15
23 24 25
34 35
45

For each pair in the above table, there are three possible values of k:

345 245 235 235


145 135 134
125 124
123

Hence, there are 30 triples in . For the probability measure we take P(A) :=
|A|/||. Now let Fi j denote the event that the friend takes chips i and j with
i < j. For example, if the friend takes chips 1 and 2, then from the second table,
k has to be 3 or 4 or 5; i.e.,

F12 = {(1, 2, 3), (1, 2, 4), (1, 2, 5)}.

The event that the friend takes two chips is then

T := F12 F13 F14 F15 F23 F24 F25 F34 F35 F45 .

Now the event that you test a defective chip is

D := {(i, j, k) : k = 1 or 2 and i < j with i, j 6= k}.


16 Chapter 1 Problem Solutions

We can now compute


P(D T )
P(D|T ) = .
P(T )
Since the Fi j that make up T are disjoint, |T | = 103 = 30 and P(T ) = |T |/|| =
1. We next observe that

D T = [D F13 ] [D F14 ] [D F15 ]


[D F23 ] [D F24 ] [D F25 ]
[D F34 ] [D F35 ] [D F45 ].

Of the above intersections, the first six intersections are singleton sets, and the
last three are pairs. Hence, |DT | = 61+32 and so P(DT ) = 12/30 = 2/5.
We conclude that P(D|T ) = P(D T )/P(T ) = (2/5)/1 = 2/5, which is the
answer in part (a).

Remark. The model in part (c) can be used to solve part (b) by observing
that the probability in part (b) is

P(D|F12 F13 F14 F15 F23 F24 F25 ),

which can be similarly evaluated.


61. (a) If two sets A and B are disjoint, then by definition, A B = .
(b) If two events A and B are independent, then by definition, P(AB) = P(A)P(B).
(c) If two events A and B are disjoint, then P(A B) = P() = 0. In order for
them to be independent, we must have P(A)P(B) = 0; i.e., at least one of the
two events must have zero probability. If two disjoint events both have positive
probability, then they cannot be independent.
62. Let W denote the event that the decoder outputs the wrong message. Of course, W c
is the event that the decoder outputs the correct message. We must find P(W ) =
1 P(W c ). Now, W c occurs if only the first bit is flipped, or only the second bit is
flipped, or only the third bit is flipped, or if no bits are flipped. Denote these disjoint
events by F100 , F010 , F001 , and F000 , respectively. Then

P(W c ) = P(F100 F010 F001 F000 )


= P(F100 ) + P(F010 ) + P(F001 ) + P(F000 )
= p(1 p)2 + (1 p)p(1 p) + (1 p)2 p + (1 p)3
= 3p(1 p)2 + (1 p)3 .

Hence,
P(W ) = 1 3p(1 p)2 (1 p)3 = 3p2 2p3 .
If p = 0.1, then P(W ) = 0.03 0.002 = 0.028.
63. Let Ai denote the event that your phone selects channel i, i = 1, . . . , 10. Let B j denote
the event that your neighbors phone selects channel j, j = 1, . . . , 10. Let P(Ai ) =
Chapter 1 Problem Solutions 17

P(B j ) = 1/10, and assume Ai and B j are independent. Then


 10  10 10 10
[ 1 1
P [Ai B j ] = P(Ai B j ) = P(Ai )P(B j ) = 10 10 = 0.1.
i=1 i=1 i=1 i=1

64. Let L denote the event that the left airbag works properly, and let R denote the event
that the right airbag works properly. Assume L and R are independent with P(L c ) =
P(R c ) = p. The probability that at least one airbag works properly is

P(L R) = 1 P(L c R c ) = 1 P(L c )P(R c ) = 1 p2 .

65. The probability that the dart never lands within 2 cm of the center is
   N 
\ \
c c
P An = lim P An , limit property of P,
N
n=1 n=1
N
= lim
N
P(Anc ), independence,
n=1
N
= lim
N
(1 p)
n=1
= lim (1 p)N = 0.
N

66. Let Wi denote the event that you win on your ith play of the lottery. The probability
that you win at least once in n plays is
 n   n  n
[ \
c
P Wi = 1 P Wi = 1 P(Wic ), by independence,
i=1 i=1 i=1
n
= 1 (1 p) = 1 (1 p)n .
i=1

We need to choose n so that 1 (1 p)n > 1/2, which happens if and only if
ln 2
1/2 > (1 p)n or ln 2 > n ln(1 p) or < n,
ln(1 p)

where the last step uses the fact that ln(1 p) is negative. For p = 106 , we need
n > 693147.
67. Let A denote the event that Anne catches no fish, and let B denote the event that Betty
catches no fish. Assume A and B are independent with P(A) = P(B) = p. We must
compute
P(A [A B]) P(A)
P(A|A B) = = ,
P(A B) P(A B)
where the last step uses the fact that A A B. To compute the denominator, write

P(A B) = 1 P(A c B c ) = 1 P(A c )P(B c ) = 1 (1 p)2 = 2p p2 = p(2 p).


18 Chapter 1 Problem Solutions

Then
p 1
P(A|A B) = = .
p(2 p) 2 p

68. We show that A and B \C are independent as follows. First, since C B,

P(B) = P(C) + P(B \C).

Next, since A and B are independent and since A and C are independent,

P(A B) = P(A)P(B) = P(A)[P(C) + P(B \C)] = P(A C) + P(A)P(B \C).

Again using the fact that C B, we now write

P(A B) = P(A [C B \C]) = P(A C) + P(A B \C).

It follows that P(A B \C) = P(A)P(B \C), which establishes the claimed indepen-
dence.
69. We show that A, B, and C are mutually independent. To begin, note that P(A) =
P(B) = P(C) = 1/2. Next, we need to identify the events

A B = [0, 1/4)
A C = [0, 1/8) [1/4, 3/8)
B C = [0, 1/8) [1/2, 5/8)
A B C = [0, 1/8)

so that we can compute

P(A B) = P(A C) = P(B C) = 1/4 and P(A B C) = 1/8.

We find that

P(A B) = P(A)P(B), P(A C) = P(A)P(C), P(B C) = P(B)P(C),

and
P(A B C) = P(A)P(B)P(C).

70. From a previous problem we have that P(A C|B) = P(A|B C)P(C|B). Hence,
P(A C|B) = P(A|B)P(C|B) if and only if P(A|B C) = P(A|B).
71. We show that the probability of the complementary event is zero. By the union bound,
   
[ \ \
c c
P Bk P Bk .
n=1 k=n n=1 k=n

We show that every term on the right is zero. Write


   N 
\ \
c c
P Bk = lim P Bk , limit property of P,
N
k=n k=n
Chapter 1 Problem Solutions 19

N
= lim
N
P(Bkc ), independence,
k=n
N
= lim
N
[1 P(Bk )]
k=n
N
lim
N
exp[P(Bk )], the hint,
k=n
 N 
= lim exp P(Bk )
N
k=n
 N 
= exp lim
N
k ,
P(B ) since exp is continuous,
k=n
 
= exp P(Bk )
k=n

= e = 0,

where the second-to-last step uses the fact that


k=1 P(Bk ) = k=n P(Bk ) = .

72. There are 3 5 7 = 105 possible systems.


73. There are 2n n-bit numbers.
74. There are 100! different orderings of the 100 message packets. In order that the first
header packet to be received is the 10th packet to arrive, the first 9 packets to be
received must come from the 96 data packets, the 10th packet must come from the 4
header packets, and the remaining 90 packets can be in any order. More specifically,
there are 96 possibilities for the first packet, 95 for the second, . . . , 88 for the ninth, 4
for the tenth, and 90! for the remaining 90 packets. Hence, the desired probability is
96 88 4 90! 96 88 4 90 89 88 4
= = = 0.02996.
100! 100 91 100 99 98 97
5
75. 2 = 10 pictures are needed.
76. Suppose the player chooses distinct digits wxyz. The player wins if any of the 4! = 24
permutations of wxyz occurs. Since each permutation has probability 1/10 000 of
occurring, the probability of winning is 24/10 000 = 0.0024.

77. There are 83 = 56 8-bit words with 3 ones (and 5 zeros).

78. The probability that a random byte has 4 ones and 4 zeros is 84 /28 = 70/256 =
0.2734.
79. In the first case, since the prizes are different, order is important. Hence, there are
41 40 39 = 63 960 outcomes. In the second case, since the prizes are the same, order
is not important. Hence, there are 41 3 = 10 660 outcomes.

80. There are 52 14 possible hands. Since the deck contains 13 spades, 13 hearts, 13
 13 13 13
diamonds, and 13 clubs, there are 13 2 3 4 5 hands with 2 spades, 3 hearts, 4
20 Chapter 1 Problem Solutions

diamonds, and 5 clubs. The probability of such a hand is


13 13 13 13
2 3 4 5
52
= 0.0116.
14

81. All five cards are of the same suit if and only if they are all spades or all hearts or all
diamonds or all clubs. These are four disjoint events. Hence, the answer is four times
the probability of getting all spades:
13
5 1287
4 52  = 4 = 0.00198.
5
2 598 960

n 
82. There are k1 ,...,km such partitions.
83. The general result is  
n
mn .
k0 , . . . , km1
4 
When n = 4 and m = 10 and a player chooses xxyz, we compute 2,1,1 /10 000 =
4
0.0012. For xxyy, we compute 2,2 /10 000 = 0.0006. For xxxy, we compute
4
3,1 /10 000 = 0.0004.

84. Two apples and three carrots corresponds to (0, 0, 1, 1, 0, 0, 0). Five apples corre-
sponds to (0, 0, 0, 0, 0, 1, 1).
CHAPTER 2
Problem Solutions

1. (a) { : X( ) 3} = {1, 2, 3}.


(b) { : X( ) > 4} = {5, 6}.
(c) P(X 3) = P(X = 1) + P(X = 2) + P(X = 3) = 3(2/15) = 2/5, and
P(X > 4) = P(X = 5) + P(X = 6) = 2/15 + 1/3 = 7/15.
2. (a) { : X( ) = 2} = {1, 2, 3, 4}.
(b) { : X( ) = 1} = {41, 42, . . . , 52}.
(c) P(X = 1 or X = 2) = P({1, 2, 3, 4} {41, 42, . . . , 52}). Since these are disjoint
events, the probability of their union is 4/52 + 12/52 = 16/52 = 4/13.
3. (a) { [0, ) : X( ) 1} = [0, 1].
(b) { [0, ) : X( ) 3} = [0, 3].
R
(c) P(X 1) = 01 e d = 1 e1 . P(X 3) = 1 e3 , P(1 < X 3) = P(X
3) P(X 1) = e1 e3 .
4. First, since X 1 () = , () = P(X 1 ()) = P() = 0. Second, (B) =
P(X 1 (B)) 0. Third, for disjoint Bn ,
      
[ [ [
1
Bn = P X Bn =P X (Bn ) = P(X 1 (Bn )) = (Bn ).
1

n=1 n=1 n=1 n=1 n=1

Fourth, (IR) = P(X 1 (IR)) = P() = 1.


5. Since

P(Y > n 1) = P(Y = k) = P(Y = n) + P(Y = k) = P(Y = n) + P(Y > n),
k=n k=n+1

it follows that P(Y = n) = P(Y > n 1) P(Y > n).


6. P(Y = 0) = P({TTT,THH,HTH,HHT}) = 4/8 = 1/2, and
P(Y = 1) = P({TTH,THT,HTT,HHH}) = 4/8 = 1/2.
7. P(X = 1) = P(X = 2) = P(X = 3) = P(X = 4) = P(X = 5) = 2/15 and P(X = 6) =
1/3.
8. P(X = 2) = P({1, 2, 3, 4}) = 4/52 = 1/13. P(X = 1) = P({41, 42, . . . , 52}) = 12/52 =
3/13. P(X = 0) = 1 P(X = 2) P(X = 1) = 9/13.
9. The possible values of X are 0, 1, 4, 9, 16. We have P(X = 0) = P({0}) = 1/7, P(X =
1) = P({1, 1}) = 2/7, P(X = 4) = P({2, 2}) = 2/7, P(X = 9) = P({3}) = 1/7,
and P(X = 16) = P({4}) = 1/7.

21
22 Chapter 2 Problem Solutions

10. We have

P(X > 1) = 1 P(X 1) = 1 [P(X = 0) + P(X = 1)]


= 1 [e + e ] = 1 e (1 + ).

When = 1, P(X > 1) = 1 e2 (2) = 1 2/e = 0.264.


11. The probability that the sensor fails to activate is

P(X < 4) = P(X 3) = P(X = 0) + + P(X = 3) = e (1 + + 2 /2! + 3 /3!).

If = 2, P(X < 4) = e2 (1 + 2 + 2 + 4/3) = e2 (19/3) = 0.857. The probability


that the sensor activates is 1 P(X < 4) = 0.143.
12. Let {Xk = 1} correspond to the event that the kth student gets an A. This event has
probability P(Xk = 1) = p. Now, the event that only the kth student gets an A is

{Xk = 1 and Xl = 0 for l 6= k}.

Hence, the probability that exactly one student gets an A is


 15  15
[
P {Xk = 1 and Xl = 0 for l 6= k} = P({Xk = 1 and Xl = 0 for l 6= k})
k=1 k=1
15
= p(1 p)14
k=1
= 15(.1)(.9)14 = 0.3432.

13. Let X1 , X2 , X3 be the random digits of the drawing. Then P(Xi = k) = 1/10 for k =
0, . . . , 9 since each digit has probability 1/10 of being chosen. Then if the player
chooses d1 d2 d3 , the probability of winning is

P {X1 = d1 , X2 = d2 , X3 = d3 } {X1 = d1 , X2 = d2 , X3 6= d3 }

{X1 = d1 , X2 6= d2 , X3 = d3 } {X1 6= d1 , X2 = d2 , X3 = d3 } ,

which is equal to .13 +3[.12 (.9)] = 0.028 since the union is disjoint and since X1 , X2 , X3
are independent.
14.
 m   m 
[ \
P {Xk < 2} = 1 P {Xk 2}
k=1 k=1
m m
= 1 P(Xk 2) = 1 [1 P(Xk 1)]
k=1 k=1

= 1 [1 {e +e m
}] = 1 [1 e (1 + )]m .
Chapter 2 Problem Solutions 23

15. (a)
 n   n  n
[ \
P {Xi 2} = 1 P {Xi 1} = 1 P(Xi 1)
i=1 i=1 i=1
n
= 1 [P(Xi = 0) + P(Xi = 1)]
i=1

= 1 [e + e ]n = 1 en (1 + )n .
 n  n n
\
(b) P {Xi 1} = P(Xi 1) = [1 P(Xi = 0)] = (1 e )n .
i=1 i=1 i=1
 n  n
\
(c) P {Xi = 1} = P(Xi = 1) = ( e )n = n en .
i=1 i=1

16. For the geometric0 pmf, write



1
(1 p)pk = (1 p) pk = (1 p) 1 p
= 1.
k=0 k=0

For the geometric1 pmf, write



1
(1 p)pk1 = (1 p) pk1 = (1 p) pn = (1 p)
1 p
= 1.
k=1 k=1 n=0

17. Let Xi be the price of stock i, which is a geometric0 (p) random variable. Then
 29   29  29
[ \
P {Xi > 10} = 1 P {Xi 10} = 1 [(1 p)(1 + p + + p10 )]
i=1 i=1 i=1
 1 p11 29
= 1 (1 p) = 1 (1 p11 )29 .
1 p

Substituting p = .7, we have 1 (1 .711 )29 = 1 (.98)29 = 1 .560 = 0.44.


18. For the first problem, we have
 n  n n
\
P(min(X1 , . . . , Xn ) > `) = P {Xk > `} = P(Xk > `) = p` = pn` .
k=1 k=1 k=1

Similarly,
 n  n n
\
P(max(X1 , . . . , Xn ) `) = P {Xk `} = P(Xk `) = (1 p` ) = (1 p` )n .
k=1 k=1 k=1

19. Let Xk denote the number of coins in the pocket of the kth student. Then the Xk are
independent and is uniformly distributed from 0 to 20; i.e., P(Xk = i) = 1/21.
24 Chapter 2 Problem Solutions

 25  25
\
(a) P {Xk 5} = 16/21 = (16/21)25 = 1.12 103 .
k=1 k=1
 25   25 
[ \
(b) P {Xk 19} = 1 P {Xk 18} = 1 (1 2/21)25 = 0.918.
k=1 k=1
(c) The probability that only student k has 19 coins in his or her pocket is
 
\
P {Xk = 10} {Xl 6= 19} = (1/21)(20/21)24 = 0.01477.
l6=k

Hence, the probability that exactly one student has 19 coins is


 25 
[ \
P {Xk = 10} {Xl 6= 19} = 25(0.01477) = 0.369.
k=1 l6=k

20. Let Xi = 1 if block i is good. Then P(Xi = 1) = p and



P(Y = k) = P {X1 = 1} {Xk1 = 1}{Xk = 0} = pk1 (1 p), k = 1, 2, . . . .

Hence, Y geometric1 (p).


21. (a) Write

P(X > n) = (1 p)pk1 = (1 p)p`+n
k=n+1 `=0

1
= (1 p)pn p` = (1 p)pn = pn .
`=0 1 p

(b) Write

P(X > n + k, X > n) P(X > n + k) pn+k


P(X > n + k|X > n) = = = n = pk .
P(X > n) P(X > n) p

22. Since P(Y > k) = P(Y > n + k|Y > n), we can write

P(Y > n + k,Y > n) P(Y > n + k)


P(Y > k) = P(Y > n + k|Y > n) = = .
P(Y > n) P(Y > n)

Let p := P(Y > 1). Taking k = 1 above yields P(Y > n + 1) = P(Y > n)p. Then
with n = 1 we have P(Y > 2) = P(Y > 1)p = p2 . With n = 2 we have P(Y > 3) =
P(Y > 2)P(Y > 1) = p3 . In general then P(Y > n) = pn . Finally, P(Y = n) = P(Y >
n 1) P(Y > n) = pn1 pn = pn1 (1 p), which is the geometric1 (p) pmf.
23. (a) To compute pX (i), we sum row i of the matrix. This yields pX (1) = pX (3) = 1/4
and pX (2) = 1/2. To compute pY ( j), we sum column j to get pY (1) = pY (3) =
1/4 and pY (2) = 1/2.
Chapter 2 Problem Solutions 25

(b) To compute P(X < Y ), we sum pXY (i, j) over i and j such that i < j. We have

P(X < Y ) = pXY (1, 2) + pXY (1, 3) + pXY (2, 3) = 0 + 1/8 + 0 = 1/8.

(c) We claim that X and Y are not independent. For example, pXY (1, 2) = 0 is not
equal to pX (1)pY (2) = 1/8.
24. (a) To compute pX (i), we sum row i of the matrix. This yields pX (1) = pX (3) = 1/4
and pX (2) = 1/2. To compute pY ( j), we sum column j to get pY (1) = pY (3) =
1/6 and pY (2) = 2/3.
(b) To compute P(X < Y ), we sum pXY (i, j) over i and j such that i < j. We have

P(X < Y ) = pXY (1, 2) + pXY (1, 3) + pXY (2, 3) = 1/6 + 1/24 + 1/12 = 7/24.

(c) Using the results of part (a), it is easy to verify that pX (i)pY ( j) = pXY (i, j) for
i, j = 1, 2, 3. Hence, X and Y are independent.
25. To compute the marginal of X, write

e3 3j e3 3
pX (1) =
3 =
3
e = 1/3.
j=0 j!

Similarly,

4e6 6j 4e6 6
pX (2) =
6 =
6
e = 2/3.
j=0 j!

Alternatively, pX (2) = 1 pX (1) = 1 1/3 = 2/3. Of course pX (i) = 0 for i 6= 1, 2.


We clearly have pY ( j) = 0 for j < 0 and

3 j1 e3 6 j1 e6
pY ( j) = +4 , j 0.
j! j!
Since pX (1)pY ( j) 6= pXY (1, j), X and Y are not independent.
26. (a) For k 1,

(1 p)pk1 kn ek
pX (k) = n!
n=0

kn
= (1 p)pk1 ek n! = (1 p)pk1 ek ek = (1 p)pk1 ,
n=0

which we recognize as the geometric1 (p) pmf.


(b) Next,

1 p
pY (0) = (1 p)pk1 ek = (p/e)k1
e k=1
k=1
1 p 1 p 1 1 p
= (p/e)m = e 1 p/e = e p .
e m=0
26 Chapter 2 Problem Solutions

(c) Since pX (1)pY (0) = (1 p)2 /(e p) is not equal to pXY (1, 0) = (1 p)/e, X
and Y are not independent.
27. MATLAB. Here is a script:

p = ones(1,51)/51;
k=[0:50];
i = find(g(k) >= -16);
fprintf(The answer is %g\n,sum(p(i)))

where

function y = g(x)
y = 5*x.*(x-10).*(x-20).*(x-30).*(x-40).*(x-50)/1e6;

28. MATLAB. If you modified your program for the preceding problem only by the way
you compute P(X = k), then you may get only 0.5001 = P(g(X) 16 and X 50).
Note that g(x) > 0 for x > 50. Hence, you also have to add P(X 51) = p51 = 0.0731
to 0.5001 to get 0.5732.
29. MATLAB. OMITTED.
30. MATLAB. OMITTED.
31. MATLAB. OMITTED.
32. E[X] = 2(1/3) + 5(2/3) = 12/3 = 4.
33. E[I(2,6) (X)] = 5k=3 P(X = k) = (1 p)[p3 + p4 + p5 ]. For p = 1/2, we get E[I(2,6) (X)] =
7/64 = 0.109375.
34. Write

n e
1 e n+1 e n
E[1/(X + 1)] = n+1 n!
=
n=0 (n + 1)!
= n!
n=1
n=0
 
e n e 1 e
=
n=0 n!
1 =

[e 1] =

.

35. Since var(X) = E[X 2 ] (E[X])2 , E[X 2 ] = var(X) + (E[X])2 . Hence, E[X 2 ] = 7 + 22 =
7 + 4 = 11.
36. Since Y = cX, E[Y ] = E[cX] = cm. Hence,

var(Y ) = E[(Y cm)2 ] = E[(cX cm)2 ] = E[c2 (X m)2 ]


= c2 E[(X m)2 ] = c2 var(X) = c2 2 .

37. We begin with

E[(X +Y )3 ] = E[X 3 + 3X 2Y + 3XY 2 +Y 3 ]


= E[X 3 ] + 3E[X 2Y ] + 3E[XY 2 ] + E[Y 3 ]
= E[X 3 ] + 3E[X 2 ]E[Y ] + 3E[X]E[Y 2 ] + E[Y 3 ], by independence.
Chapter 2 Problem Solutions 27

Now, as noted in the text, for a Bernoulli(p) random variable, X n = X, and so E[X n ] =
E[X] = p. Similarly E[Y n ] = q. Thus,

E[(X +Y )3 ] = p + 3pq + 3pq + q = p + 6pq + q.

38. The straightforward approach is to put

f (c) := E[(X c)2 ] = E[X 2 ] 2mc + c2

and differentiate with respect to c to get f 0 (c) = 2m + 2c. Solving f 0 (c) = 0 results
in c = m. An alternative approach is to write

E[(X c)2 ] = E[{(X m) + (m c)}2 ] = 2 + (m c)2 .

From this expression, it is obvious that c = m minimizes the expectation.


39. The two sketches are:

x /a
1/2
( x /a )
1 1 I[ a , )( x )
I[ a , )( x )

8
8

a a

40. The right-hand side is easy: E[X]/2 = (3/4)/2 = 3/8 = 0.375. The left-hand side is
more work:

P(X 2) = 1 P(X 1) = 1 [P(X = 0) + P(X = 1)] = 1 e (1 + ).

For = 3/4, P(X 2) = 0.1734. So the bound is a little more than twice the value
of the probability.
41. The Chebyshev bound is ( + 2 )/4. For = 3/4, the bound is 0.3281, which is
a little better than the Markov inequality bound in the preceding problem. The true
probability is 0.1734.
42. Comparing the definitions of XY and cov(X,Y ), we find XY = cov(X,Y )/(X Y ).
Hence, cov(X,Y ) = X Y XY . Since cov(X,Y ) := E[(X mX )(Y mY ), if Y = X,
we see that cov(X, X) = E[(X mX )2 ] =: var(X).
43. Put

f (a) := E[(X aY )2 ] = E[X 2 ] 2aE[XY ] + a2 E[Y 2 ] = X2 2aX Y + a2 Y2 .

Then
f 0 (a) = 2X Y + 2aY2 .
Setting this equal to zero and solving for a yields a = (X /Y ).
28 Chapter 2 Problem Solutions

44. Since P(X = 1) = P(X = 2) = 1/4, E[X] = 0. Similarly, since P(XY = 1) = 1/4
and P(XY = 4) = 1/4, E[XY ] = 0. Thus, E[XY ] = 0 = E[X]E[Y ] and we see that X
and Y are uncorrelated. Next, since X = 1 implies Y = 1, P(X = 1,Y = 1) = P(X =
1) = 1/4 while P(Y = 1) = P(X = 1 or X = 1) = 1/2. Thus,

P(X = 1)P(Y = 1) = (1/4)(1/2) = 1/8, but P(X = 1,Y = 1) = 1/4.

45. As discussed in the text, for uncorrelated random variables, the variance of the sum
is the sum of the variances. Since independent random variables are uncorrelated, the
same results holds for them too. Hence, for Y = X1 + + XM ,
M
var(Y ) = var(Xk ).
k=1

We also have E[Y ] = Mk=1 E[Xk ]. Next, since the Xk are i.i.d. geometric1 (p), E[Xk ] =
1/(1 p) and var(Xk ) = p/(1 p)2 . It follows that var(Y ) = M p/(1 p)2 and E[Y ] =
M/(1 p). We conclude by writing

Mp M2 M(p + M)
E[Y 2 ] = var(Y ) + (E[Y ])2 = 2
+ = .
(1 p) (1 p)2 (1 p)2

46. From E[Y ] = E[dX s(1 X)] = d p s(1 p) = 0, we find that d/s = (1 p)/p.
47. (a) p = 1/1000.
(b) Since (1 p)/p = (999/1000)/(1/1000) = 999, the fair odds against are 999 :1.
(c) Since the fair odds of 999 :1 are not equal to the offered odds of 500 :1, the game
is not fair. To make the game fair, the lottery should pay $900 instead of $500.
48. First note that
1 1
Z p1 , p 6= 1,
1 1 p t 1
dt =
1 tp

lnt , p = 1.
1
For p > 1, the integral is equal to 1/(p 1). For p 1, the integral is infinite.
For 0 < p 1, write
Z k+1 Z
1 1 1
kp t p
dt =
1 tp
dt = .
k=1 k=1 k

For p > 1, it suffices to show that p


k=2 1/k < . To this end, write
Z k+1 Z
1 1 1 1
kp = (k + 1) p k t p
dt =
1 tp
dt < .
k=2 k=1 k=1

49. First write


C1
p 1

1
E[X n ] = kn p
= C p pn
.
k=1 k k=1 k
Chapter 2 Problem Solutions 29

By the preceding problem this last sum is finite for p n > 1, or equivalently, n <
p 1. Otherwise the sum is infinite; the case 1 p n > 0 being handled by the
preceding problem, and the case 0 p n being obvious.
50. If all outcomes are equally likely,
n
1 1 n
H(X) = pi log pi = log n = log n.
n i=1
i=1

If X is a constant random variable with pi = 0 for i 6= j, then


n
1 1
H(X) = pi log pi = p j log
pj
= 1 log 1 = 0.
i=1

51. Let P(X = xi ) = pi for i = 1, . . . , n. Then


n  n 
E[g(X)] = g(xi )pi and g(E[X]) = g xi pi .
i=1 i=1

For n = 2, Jensens inequality says that

p1 g(x1 ) + p2 g(x2 ) g(p1 x1 + p2 x2 ).

If we put = p1 , then 1 = p2 and the above inequality becomes

g(x1 ) + (1 )g(x2 ) g( x1 + (1 )x2 ),

which is just the definition of a convex function. Hence, if g is convex, Jensens


inequality holds for n = 2. Now suppose Jensens inequality holds for some n 2.
We must show it holds for n + 1. The case of n is
n  n 
g(xi )pi g xi pi , if p1 + + pn = 1.
i=1 i=1

Now suppose that p1 + + pn+1 = 1, and write


n+1  n 
pi
g(xi )pi = (1 pn+1 ) g(xi ) 1 pn+1
+ pn+1 g(xn+1 ).
i=1 i=1

Let us focus on the quantity in brackets. Since


n
pi p1 + + pn 1 pn+1
1 pn+1 =
pn+1
=
1 pn+1
= 1,
i=1

Jensens inequality for n terms yields


n  n 
pi pi
g(xi ) 1 pn+1 g xi 1 pn+1 .
i=1 i=1
30 Chapter 2 Problem Solutions

Hence,
n+1  n 
pi
g(xi )pi (1 pn+1 )g xi 1 pn+1
+ pn+1 g(xn+1 ).
i=1 i=1

Now apply the two-term Jensen inequality to get


n+1   n  
pi
g(xi )pi g (1 pn+1 ) xi 1 pn+1
+ pn+1 xn+1
i=1 i=1
 n+1 
= g pi xi .
i=1

52. With X = |Z| and g(x) = x / , we have

E[g(X)] = E[X / ] = E[(|Z| ) / ] = E[|Z| ]

and
E[X] = E[|Z| ].
Then Jensens inequality tells us that

E[|Z| ] (E[|Z| ]) / .

Raising both sides the 1/ power yields Lyapunovs inequality.


53. (a) For all discrete random variables, we have i P(X = xi ) = 1. For a nonnegative
random variable, if xk < 0, we have

1 = P(X = xi ) I[0,) (xi )P(X = xi ) + P(X = xk ) = 1 + P(X = xk ).


i i

From this it follows that 0 P(X = xk ) 0, and so P(X = xk ) = 0.


(b) Write

E[X] = xi P(X = xi ) = xi P(X = xi ) + xi P(X = xk ).


i i:xi 0 k:xk <0

By part (a), the last sum is zero. The remaining sum is obviously nonnegative.
(c) By part (b), 0 E[X Y ] = E[X] E[Y ]. Hence, E[Y ] E[X].
CHAPTER 3
Problem Solutions
 
1
1. First, E[X] = G0X (1) = 6 + 43 z = 1
6 + 4
3 = 9
6 = 32 . Second, E[X(X 1)] =
z=1
G00X (1) = 4/3. Third, from E[X(X 1)] = E[X 2 ]E[X], we have E[X 2 ] = 4/3+3/2 =
17/6. Finally, var(X) = E[X 2 ] (E[X])2 = 17/6 9/4 = (34 27)/12 = 7/12.
   

2. pX (0) = GX (0) = 16 + 61 z + 23 z2 = 1/6, pX (1) = G0X (0) = 16 + 43 z = 1/6,
z=0 z=0
and pX (2) = G00X (0)/2 = (4/3)/2 = 2/3.
3. To begin, note that G0X (z) = 5((2 + z)/3)4 /3, and G00X (z) = 20((2 + z)/3)3 /9. Hence,
E[X] = G0X (1) = 5/3 and E[X(X 1)] = 20/9. Since E[X(X 1)] = E[X 2 ] E[X],
E[X 2 ] = 20/9 + 5/3 = 35/9. Finally, var(X) = E[X 2 ] (E[X])2 = 35/9 25/9 =
10/9.
4. For X geometric0 (p),

1
GX (z) = zn P(X = n) = zn (1 p)pn = (1 p) (zp)n = (1 p) 1 pz .
n=0 n=0 n=0

Then
1 p 1 p p
E[X] = G0X (1) = p = p= .
(1 pz)2 z=1 (1 p)2 1 p
Next,

(1 p)p
2
E[X ] E[X] = E[X(X 1)] = G00X (1) = 2p(1 pz)
(1 pz)4 z=1
(1 p)p 2p 2p2
= 3
= .
(1 p) (1 p)2
This implies that
2p2 p 2p2 + p(1 p) p + p2
E[X 2 ] = + = = .
(1 p)2 1 p (1 p)2 (1 p)2
Finally,
p + p2 p2 p
var(X) = E[X 2 ] (E[X])2 = 2
2
= .
(1 p) (1 p) (1 p)2

For X geometric1 (p),



1 p
GX (z) = zn P(X = n) = zn (1 p)pn1 = (zp)n
p n=1
n=1 n=1
1 p pz (1 p)z
= = .
p 1 pz 1 pz

31
32 Chapter 3 Problem Solutions

Now
(1 pz)(1 p) + (1 p)pz 1 p
G0X (z) = = .
(1 pz)2 (1 pz)2
Hence,
1
E[X] = G0X (1) = .
1 p
Next,
(1 p) 2p(1 p)
G00X (z) = 2(1 pz)p = .
(1 pz)4 (1 pz)3
We then have
2p(1 p) 2p
E[X 2 ] E[X] = E[X(X 1)] = G00X (1) = = ,
(1 p)3 (1 p)2

and
2p 1 2p + 1 p 1+ p
E[X 2 ] = 2
+ = 2
= .
(1 p) 1 p (1 p) (1 p)2
Finally,
1+ p 1 p
var(X) = E[X 2 ] (E[X])2 = = .
(1 p)2 (1 p)2 (1 p)2

5. Since the Xi are independent Poisson(i ), we use probability generating functions to


find GY (z), which turns out to be Poisson( ) with := 1 + + n . It then follows
that P(Y = 2) = 2 e /2. It remains to write
n n
GY (z) = E[zY ] = E[zX1 ++Xn ] = E[zX1 zXn ] = E[zXi ] = ei (z1)
i=1 i=1
 n 
= exp i (z 1) = e (z1) ,
i=1

which is the Poisson( ) pgf. Hence, Y Poisson( ) as claimed.


6. If GX (z) = k
k=0 z P(X = k), then GX (1) = k=0 P(X = k) = 1. For the particular
formula in the problem, we must have 1 = GX (1) = (a0 + a1 + a2 + + an )m /D, or
D = (a0 + a1 + a2 + + an )m .
7. From the table on the inside of the front cover of the text, E[Xi ] = 1/(1 p). Thus,
n
n
E[Y ] = E[Xi ] =
1 p
.
i=1

Second, since the variance of the sum of uncorrelated random variables is the sum of
the variances, and since var(Xi ) = p/(1 p)2 , we have
n
np
var(Y ) = var(Xi ) =
(1 p)2
.
i=1
Chapter 3 Problem Solutions 33

It then easily follows that


np n2 n(p + n)
E[Y 2 ] = var(Y ) + (E[Y ])2 = 2
+ = .
(1 p) (1 p)2 (1 p)2
Since Y is the sum of i.i.d. geometric1 (p) random variables, the pgf of Y is the product
of the individual pgfs. Thus,
 
n
(1 p)z (1 p)z n
GY (z) = = .
i=1 1 pz 1 pz

8. Starting with GY (z) = [(1 p) + pz]n , we have




E[Y ] = GY0 (1) = n[(1 p) + pz]n1 p = np.
z=1

Next,


E[Y (Y 1)] = GY00 (1) = n(n 1)[(1 p) + pz]n2 p2 = n(n 1)p2 .
z=1

It then follows that

E[Y 2 ] = E[Y (Y 1)] + E[Y ] = n(n 1)p2 + np = n2 p2 np2 + np.

Finally,

var(Y ) = E[Y 2 ] (E[Y ])2 = n2 p2 np2 + np (np)2 = np(1 p).

9. For the binomial(n, p) random variable,


n n  
n
GY (z) = P(Y = k)zk = k pk (1 p)nk zk .
k=0 k=0

Hence,  
n
n
1 = GY (1) = k pk (1 p)nk .
k=0

10. Starting from  


n
n
k pk (1 p)nk = 1,
k=0
we follow the hint and replace p with a/(a + b). Note that 1 p = b/(a + b). With
these substitutions, the above equation becomes
n   k  nk n  
n a b n ak bnk
a+b a+b
= 1 or k nk
= 1.
k=0 k k=0 k (a + b) (a + b)

Multiplying through by (a + b)n we have


n  
n
k ak bnk = (a + b)n .
k=0
34 Chapter 3 Problem Solutions

11. Let Xi = 1 if bit i is in error, Xi = 0 otherwise. Then Yn := X1 + + Xn is the number


of errors in n bits. Assume the Xi are independent. Then
GYn (z) = E[zYn ] = E[zX1 ++Xn ] = E[zX1 zXn ]
= E[zX1 ] E[zXn ], by independence,
= [(1 p) + pz]n , since the Xi are i.i.d. Bernoulli(p).
We recognize this last expression
 as the binomial(n, p) probability generating func-
tion. Thus, P(Yn = k) = nk pk (1 p)nk .
12. Let Xi binomial(ni , p) denote the number of students in the ith room. Then Y =
X1 + + XM is the total number of students in the school. Next,
GY (z) = E[zY ] = E[zX1 ++XM ]
= E[zX1 ] E[zXM ], by independence,
M
= [(1 p) + pz]ni , since Xi binomial(ni , p),
i=1
= [(1 p) + pz]n1 ++nM .
Setting n := n1 + + nM , we see that Y binomial(n, p). Hence, P(Y = k) =
n k
k p (1 p)nk .
13. Let Y = X1 + +Xn , where the Xi are i.i.d. with P(Xi = 1) = 1 p and P(Xi = 2) = p.
Observe that Xi 1 Bernoulli(p). Hence,
n
Y = n + (Xi 1) .
i=1
| {z }
=: Z

Since Z is the sum of i.i.d. Bernoulli(p) random variables, Z binomial(n, p). Hence,
P(Y = k) = P(n + Z = k) = P(Z = k n)
 
n
= pkn (1 p)2nk , k = n, . . . , 2n.
kn

14. Let Xi be i.i.d. Bernoulli(p), where Xi = 1 means bit i is flipped. Then Y := X1 + +


Xn (n = 10) is the number of bits flipped. A codeword cannot be decoded if Y > 2.
We need to find P(Y > 2). Observe that
n
GY (z) = E[zY ] = E[zX1 ++Xn ] = E[zX1 zXn ] = E[zXi ] = [(1 p) + pz]n .
i=1

This is the pgf of a binomial(n, p) random variable. Hence,


P(Y > 2) = 1 P(Y 2) = 1 [P(Y = 0) + P(Y = 1) + P(Y = 2)]
      
10 10 10 2
= 1 (1 p)10 + p(1 p)9 + p (1 p)8
0 1 2
= 1 (1 p)8 [(1 p)2 + 10p(1 p) + 45p2 ].
Chapter 3 Problem Solutions 35

15. For n = 150 and p = 1/100, we have


   
k P Binomial(n, p) = k P Poisson(np) = k
0 0.2215 0.2231
1 0.3355 0.3347
2 0.2525 0.2510
3 0.1258 0.1255
4 0.0467 0.0471
5 0.0138 0.0141
16. If the Xi are i.i.d. with mean m, then
 n 
1 1 n 1 n nm
E[Mn ] = E
n i=1
Xi =
n i=1
E[Xi ] =
n i=1
m =
n
= m.

If X is any random variable with mean m, then E[cX] = cE[X] = cm, and

var(cX) = E[(cX cm)2 ] = E[c2 (X m)2 ] = c2 E[(X m)2 ] = c2 var(X).

17. In general, we have

P(|Mn m| < ) 0.9 P(|Mn m| ) < 0.1.

By Chebyshevs inequality,

2
P(|Mn m| ) < 0.1
n 2

if n > 2 /(.1) 2 . For 2 = 1 and = 0.25, we require n > 1/(.1)(.25)2 = 1/.00625 =


160 students. If instead = 1, we require n > 0.1 = 10 students.
18. (a) E[Xi ] = E[IB (Zi )] = P(Zi B). Setting p := P(Zi B), we see that Xi = IB (Zi )
Bernoulli(p). Hence, var(Xi ) = p(1 p).
(b) In fact, the Xi are independent. Hence, they are uncorrelated.
19. Mn = 0 if and only if all the Xi are zero. Hence,
 n  n
\
P(Mn = 0) = P {Xi = 0} = P(Xi = 0) = (1 p)n .
i=1 i=1

In particular, if p = 1/1000, then P(M100 = 0) = (1 p)100 = 0.999100 = 0.905.


Hence, the chances are more than 90% that when we run a simulation, M100 = 0 and
we learn nothing!
20. If Xi = Z Bernoulli(1/2) for all i, then

1 n 1 n
Mn =
n i=1
Xi = Z = Z,
n i=1
36 Chapter 3 Problem Solutions

and m = E[Xi ] = E[Z] = 1/2. So,

P(|Mn m| 1/4) = P(|Z 1/2| 1/4).

Now, Z 1/2 = 1/2, and |Z 1/2| = 1/2 with probability one. Thus,

P(|Z 1/2| 1/4) = P(1/2 1/4) = 1 6 0.

21. From the discussion of the weak law in the text, we have

2
P(|Mn m| n ) .
nn2

If nn2 as n , then probability on the left will go to zero as n .



22. We have from the example that with p := /( + ), pX|Z (i| j) = ij pi (1 p) ji
for i = 0, . . . , j. In other words, as a function of i, pX|Z (i| j) is a binomial( j, p) pmf.
Hence,
j
E[X|Z = j] = ipX|Z (i| j)
i=0

is just the mean of a binomial( j, p) pmf. The mean of such a pmf is j p. Hence,
E[X|Z = j] = j p = j /( + ).

23. The problem is telling us that P(Y = k|X = i) = nk pki (1 pi )nk . Hence

P(Y < 2|X = i) = P(Y = 0|X = i) + P(Y = 1|X = i) = (1 pi )n + npi (1 pi )n1 .

24. The problem is telling us that P(X = k|Y = j) = jk e j /k!. Hence,

P(X > 2|Y = j) = 1 P(X 2|Y = j)


= 1 [P(X = 0|Y = j) + P(X = 1|Y = j) + P(X = 2|Y = j)]
= 1 [e j + j e j + j2 e j /2]
= 1 e j [1 + j + j2 /2].

25. For the first formula, write

P(X = xi ,Y = y j ) P(X = xi )P(Y = y j )


pX|Y (xi |y j ) := = = P(X = xi ) = pX (xi ).
P(Y = y j ) P(Y = y j )

Similarly, for the other formula,

P(Y = y j , X = xi ) P(Y = y j )P(X = xi )


pY |X (y j |xi ) := = = P(Y = y j ) = pY (y j ).
P(X = xi ) P(X = xi )
Chapter 3 Problem Solutions 37

26. We use the law of total probability to write



P(T = n) = P(X Y = n) = P(X Y = n|Y = k)P(Y = k)
k=0

= P(X k = n|Y = k)P(Y = k), by the substitution law,
k=0

= P(X = n + k|Y = k)P(Y = k)
k=0

= P(X = n + k)P(Y = k), by independence,
k=0

= (1 p)pn+k (1 q)qk
k=0

(1 p)(1 q)pn
= (1 p)(1 q)pn (pq)k = .
k=0 1 pq

27. The problem is telling us that


n e n e
P(Y = n|X = 1) = and P(Y = n|X = 2) = .
n! n!
The problem also tells us that P(X = 1) = P(X = 2) = 1/2. We can now write

P(X = 1,Y = 2) P(Y = 2|X = 1)P(X = 1) ( 2 e /2)(1/2)


P(X = 1|Y = 2) = = = .
P(Y = 2) P(Y = 2) P(Y = 2)
It remains to use the law of total probability to compute
2
P(Y = 2) = P(Y = 2|X = i)P(X = i)
i=1
= [P(Y = 2|X = 1) + P(Y = 2|X = 2)]/2
= [ 2 e /2 + 2 e /2]/2 = [ 2 e + 2 e ]/4.

We conclude by writing

2 e /4 1
P(X = 1|Y = 2) = = .
[ 2 e + 2 e ]/4 1 + ( / )2 e

28. Let X = 0 or X = 1 according to whether message zero or message one is sent. The
problem tells us that P(X = 0) = P(X = 1) = 1/2 and that

P(Y = k|X = 0) = (1 p)pk and P(Y = k|X = 1) = (1 q)qk ,

where q 6= p. We need to compute

P(Y = k|X = 1)P(X = 1) (1 q)qk (1/2)


P(X = 1|Y = k) = = .
P(Y = k) P(Y = k)
38 Chapter 3 Problem Solutions

We next use the law of total probability to compute

P(Y = k) = [P(Y = k|X = 0) + P(Y = k|X = 1)]/2 = [(1 p)pk + (1 q)qk ]/2.

We can now compute

(1 q)qk (1/2) 1
P(X = 1|Y = k) = = .
[(1 p)pk + (1 q)qk ]/2 (1p)pk
1 + (1q)qk

29. Let R denote the number of red apples in a crate, and let G denote the number of green
apples in a crate. The problem is telling us that R Poisson( ) and G Poisson( )
are independent. If T = R + G is the total number of apples in the crate, we must
compute
P(T = k|G = 0)P(G = 0)
P(G = 0|T = k) = .
P(T = k)
We first use the law of total probability, substitution, and independence to write

P(T = k|G = 0) = P(R+G = k|G = 0) = P(R = k|G = 0) = P(R = k) = k e /k!.

We also note from the text that the sum of two independent Poisson random variables
is a Poisson random variable whose parameter is the sum of the individual parameters.
Hence, P(T = k) = ( + )k e( + ) /k!. We can now write
 k
k e /k! e
P(G = 0|T = k) = = .
( + )k e( + ) /k! +

30. We begin with


n
P(Y = 1|X = n)P(X = n)
1
n!
e
P(X = n|Y = 1) = = n+1 .
P(Y = 1) P(Y = 1)

Next, we compute

1 n e
P(Y = 1) = P(Y = 1|X = n)P(X = n) = n+1 n!
n=0 n=0
 
e n+1 e k e k
=
(n + 1)! =
k!
=
1
n=0 k=1 k=0 k!
e 1 e
= [e 1] = .

We conclude with
n n
n!
1
e
n+1
1
e n+1
P(X = n|Y = 1) = = n+1 n! = .
P(Y = 1) (1 e )/ (e 1)(n + 1)!
Chapter 3 Problem Solutions 39

31. We begin with


n k nk n e /n!
P(Y = k|X = n)P(X = n) k p (1 p)
P(X = n|Y = k) = = .
P(Y = k) P(Y = k)

Next,
 
n k
P(Y = k) = P(Y = k|X = n)P(X = n) = p (1 p)nk n e /n!
n=0 n=k k

pk k e [(1 p) ]nk pk k e [(1 p) ]m
=
k! (n k)!
=
k! m!
n=k m=0
pk k e (1p) (p )k ep
= e = .
k! k!

Note that Y Poisson(p ). We continue with


n k nk n e /n! n k nk n e /n!
k p (1 p) k p (1 p)
P(X = n|Y = k) = =
P(Y = k) (p )k ep /k!
[(1 p) ]nk e(1p)
= .
(n k)!

32. First write



 P {X > k} {max(X,Y ) > k} P(X > k)

P X > k max(X,Y ) > k =  = ,
P max(X,Y ) > k P max(X,Y ) > k

since {X > k} {max(X,Y ) > k}. We next compute


 
P max(X,Y ) > k = 1 P max(X,Y ) k = 1 P(X k)P(Y k).

If we put k := P(X k) and use the fact that X and Y have the same pmf, then

 1 k 1 k 1
P X > k max(X,Y ) > k = = = .
1 k2 (1 k )(1 + k ) 1 + k

With n = 100 and p = .01, we compute

1 = P(X 1) = P(X = 0) + P(X = 1) = .99100 + .9999 = .366 + .370 = .736.

It follows that the desired probability is 1/(1 + 1 ) = 1/1.736 = 0.576.


33. (a) Observe that

P(XY = 4) = P(X = 1,Y = 4) + P(X = 2,Y = 2) + P(X = 4,Y = 1)


= (1 p)(1 q)[pq4 + p2 q2 + p4 q].
40 Chapter 3 Problem Solutions

(b) Write

pZ ( j) = pY ( j i)pX (i)
i

= pY ( j i)pX (i), since pX (i) = 0 for i < 0,
i=0
j
= pY ( j i)pX (i), since pY (k) = 0 for k < 0,
i=0
j j
= (1 p)(1 q) pi q ji = (1 p)(1 q)q j (p/q)i .
i=0 i=0

Now, if p = q,
pZ ( j) = (1 p)2 p j ( j + 1).
If p 6= q,

1 (p/q) j+1 q j+1 p j+1


pZ ( j) = (1 p)(1 q)q j = (1 p)(1 q) .
1 p/q q p

34. For j = 0, 1, 2, 3,
j
pZ ( j) = (1/16) = ( j + 1)/16.
i=0
For j = 4, 5, 6,
3
pZ ( j) = (1/16) = (7 j)/16.
i= j3

For other values of j, pZ ( j) = 0.


35. We first write
P(Y = j|X = 1) P(X = 0)

P(Y = j|X = 0) P(X = 1)
as
1j e1 / j! 1 p
.
0j e0 / j! p
We can further simplify this to
 j 1 p 1 0
1
e .
0 p
Taking logarithms and rearranging, we obtain
h 1 p i.
j 1 0 + ln ln(1 /0 ).
p
Observe that the right-hand side is just a number (threshold) that is computable from
the problem data. If we observe Y = j, we compare j to the threshold. If j is greater
than or equal to this number, we decide X = 1; otherwise, we decide X = 0.
Chapter 3 Problem Solutions 41

36. We first write


P(Y = j|X = 1) P(X = 0)

P(Y = j|X = 0) P(X = 1)
as
(1 q1 )q1j 1 p
.
(1 q0 )q0j p
We can further simplify this to
q j (1 p)(1 q0 )
1
.
q0 p(1 q1 )
Taking logarithms and rearranging, we obtain
h (1 p)(1 q ) i.
0
j ln ln(q1 /q0 ),
p(1 q1 )
since q1 < q0 implies ln(q1 /q0 ) < 0.
37. Starting with P(X = xi |Y = y j ) = h(xi ), we have

P(X = xi ,Y = y j ) = P(X = xi |Y = y j )P(Y = y j ) = h(xi )pY (y j ).

If we can show that h(xi ) = pX (xi ), then it will follow that X and Y are independent.
Now observe that the sum over j of the left-hand side reduces to P(X = xi ) = pX (xi ).
The sum over j of the right-hand side reduces to h(xi ). Hence, pX (xi ) = h(xi ) as
desired.
38. First write

pXY (1, j) = (1/3)3 j e3 / j! and pXY (2, j) = (4/6)6 j e6 / j!

Notice that 3 j e3 / j! is a Poisson(3) pmf and 6 j e6 / j! is a Poisson(6) pmf. Hence,


pX (1) = j=0 pXY (1, j) = 1/3 and pX (2) = j=0 pXY (2, j) = 2/3. It then follows
that pY |X ( j|1) is Poisson(3) and pY |X ( j|2) is Poisson(6). With these observations, it
is clear that
E[Y |X = 1] = 3 and E[Y |X = 2] = 6,
and

E[Y ] = E[Y |X = 1](1/3) + E[Y |X = 2](2/3) = 3(1/3) + 6(2/3) = 1 + 4 = 5.

To obtain E[X|Y = j], we first compute

pY ( j) = pXY (1, j) + pXY (2, j) = (1/3)3 j e3 / j! + (2/3)6 j e6 / j!

and
(1/3)3 j e3 / j! 1
pX|Y (1| j) = =
(1/3)3 j e3 / j! + (2/3)6 j e6 / j! 1 + 2 j+1 e3
and
(2/3)6 j e6 / j! 1
pX|Y (2| j) = = .
(1/3)3 j e3 / j! + (2/3)6 j e6 / j! 1 + 2( j+1) e3
42 Chapter 3 Problem Solutions

We now have
1 1
E[X|Y = j] = 1 +2
1 + 2 j+1 e3 1 + 2( j+1) e3
1 2 j+1 e3 1 + 2 j+2 e3
= +2 = .
1 + 2 j+1 e3 1 + 2 j+1 e3 1 + 2 j+1 e3

39. Since Y is conditionally Poisson(k) given X = k, E[Y |X = k] = k. Hence,



1
E[Y ] = E[Y |X = k]P(X = k) = kP(X = k) = E[X] =
1 p
,
k=1 k=1

since X geometric1 (p). Next



E[XY ] = E[XY |X = k]P(X = k) = E[kY |X = k]P(X = k), by substitution,
k=1 k=1

= kE[Y |X = k]P(X = k) = k2 P(X = k) = E[X 2 ]
k=1 k=1
p 1 1+ p
= var(X) + (E[X])2 = 2
+ 2
= .
(1 p) (1 p) (1 p)2

Since E[Y 2 |X = k] = k + k2 ,

E[Y 2 ] = E[Y 2 |X = k]P(X = k) = (k + k2 )P(X = k) = E[X] + E[X 2 ]
k=1 k=1
1 1+ p 2
= + = .
1 p (1 p)2 (1 p)2

Finally, we can compute

2 1 1
var(Y ) = E[Y 2 ] (E[Y ])2 = = .
(1 p)2 (1 p)2 (1 p)2

40. From the solution of the example, it is immediate that E[Y |X = 1] = and E[Y |X =
0] = /2. Next,

E[Y ] = E[Y |X = 0](1 p) + E[Y |X = 1]p = (1 p) /2 + p .

Similarly,

E[Y 2 ] = E[Y 2 |X = 0](1 p) + E[Y 2 |X = 1]p


= ( /2 + 2 /4)(1 p) + ( + 2 )p.

To conclude, we have

var(Y ) = E[Y 2 ] (E[Y ])2 = ( /2 + 2 /4)(1 p) + ( + 2 )p [(1 p) /2 + p ]2 .


Chapter 3 Problem Solutions 43

41. Write
1 1
E[(X + 1)Y 2 ] = E[(X + 1)Y 2 |X = i]P(X = i) = E[(i + 1)Y 2 |X = i]P(X = i)
i=0 i=0
1
= (i + 1)E[Y 2 |X = i]P(X = i).
i=0

Now, since given X = i, Y is conditionally Poisson(3(i + 1)),




E[Y 2 |X = i] = ( + 2 ) = 3(i + 1) + 9(i + 1)2 .
=3(i+1)

It now follows that


1
E[(X + 1)Y 2 ] = (i + 1)[3(i + 1) + 9(i + 1)2 ]P(X = i)
i=0
1
= (i + 1)2 [3 + 9(i + 1)]P(X = i)
i=0
= 12(1/3) + 84(2/3) = 4 + 56 = 60.

42. Write

E[XY ] = E[XY |X = n]P(X = n) = E[nY |X = n]P(X = n)
n=0 n=0

n e 1
= nE[Y |X = n]P(X = n) = nn+1
n!
n=0 n=0
     
X X +11 1
= E = E = 1E .
X +1 X +1 X +1

By a problem in the previous chapter, this last expectation is equal to (1 e )/ .


Hence,
1 e
E[XY ] = 1 .

43. Write

E[XY ] = E[XY |X = n]P(X = n) = E[nY |X = n]P(X = n)
n=1 n=1

n
= nE[Y |X = n]P(X = n) = n 1 q P(X = n)
n=1 n=1
1 1 h i
= E[X 2 ] = var(X) + (E[X])2
1q 1q
 
1 p 1 1+ p
= 2
+ 2
= .
1 q (1 p) (1 p) (1 q)(1 p)2
44 Chapter 3 Problem Solutions

44. Write

E[X 2 ] = E[X 2 |Y = k]P(Y = k) = (k + k2 )P(Y = k) = E[Y +Y 2 ]
k=1 n=1
= E[Y ] + E[Y 2 ] = m + (r + m2 ) = m + m2 + r.

45. Using probability generating functions, we see that

GV (z) = E[zX+Y ] = E[zX zY ] = E[zX ]E[zY ]


= [(1 p) + pz]n [(1 p) + pz]m = [(1 p) + pz]n+m .

Thus, V binomial(n + m, p). We next compute

P(V = 10|X = 4) = P(X +Y = 10|X = 4) = P(4 +Y = 10|X = 4)


 
m 6
= P(Y = 6|X = 4) = P(Y = 6) = p (1 p)m6 .
6

46. Write

GY (z) = E[zY ] = E[zY |X = k]P(X = k) = ek(z1) P(X = k)
k=1 k=1

(1 p)ez1
= (ez1 )k P(X = k) = GX (ez1 ) =
1 pez1
.
k=1
CHAPTER 4
Problem Solutions

1. Let Vi denote the input voltage at the ith sampling time. The problem tells us that the
Vi are independent and uniformly distributed on [0, 7]. The alarm sounds if Vi > 5 for
i = 1, 2, 3. The probability of this is
 3  3
\
P {Vi > 5} = P(Vi > 5).
i=1 i=1
R7
Now, P(Vi > 5) = 5 (1/7) dt = 2/7. Hence, the desired probability is (2/7)3 =
8/343 = 0.0233.
R
2. We must solve t f (x) dx = 1/2 for t. Now,
Z
1
3 1
2x dx = 2 = 2 .
t x t t

Solving 1/t 2 = 1/2, we find that t = 2.
R
3. To find c, we solve 01 cx1/2 dx = 1. The left-hand side of this equation is 2cx1/2 |10 =
R
2c. Solving 2c = 1 yields c = 1/2. For the median, we must solve t1 (1/2)x1/2 dx =
1/2 or x1/2 |t1 = 1/2. We find that t = 1/4.
R
4. (a) For t 0, P(X > t) = t e x dx = e x |t = e t .
(b) First, P(X > t + t|X > t) = P(X > t + t, X > t)/P(X > t). Next, observe that
{X > t + t} {X > t} = {X > t + t},
and so P(X > t + t|X > t) = P(X > t + t)/P(X > t) = e (t+t) /e t =
e t .
5. Let Xi denote the voltage output by regulator i. Then the Xi are i.i.d. exp( ) random
variables. Now put
10
Y := I(v,) (Xi )
i=1
so that Y counts the number of regulators that output more than v volts. We must
compute P(Y = 3). Now, the I(v,) (Xi ) are i.i.d. Bernoulli(p) random variables, where
Z
x

p = P(Xi > v) = e dx = e = e v .
x
v v

Next, we now from the previous chapter that a sum of n i.i.d. Bernoulli(p) random
variables is a binomial(n, p). Thus,
   
n 3 10 3 v
P(Y = 3) = p (1 p)n3 = e (1 e v )7 = 120e3 v (1 e v )7 .
3 3

45
46 Chapter 4 Problem Solutions

R
6. First note that P(Xi > 2) = e x dx = e x |
2 2 =e
2 .

 T 
(a) P min(X1 , . . . , Xn ) > 2 = P ni=1 {Xi > 2} = ni=1 P(Xi > 2) = e2n .
(b) Write
 
P max(X1 , . . . , Xn ) > 2 = 1 P max(X1 , . . . , Xn ) 2
 n 
\
= 1P {Xi 2}
i=1
n
= 1 P(Xi 2) = 1 [1 e2 ]n .
i=1

R2
7. (a) P(Y 2) = 0 e y dy = 1 e2 .
(b) P(X 12,Y 12) = P(X 12)P(Y 12) = (1 e12 )(1 e12 ).
(c) Write

P({X 12} {Y 12}) = 1 P(X > 12,Y > 12)


= 1 P(X > 12)P(Y > 12)
= 1 e12 e12 = 1 e12( + ) .

8. (a) Make the change of variable y = x p , dy = px p1 dx to get


Z Z
p
px p1 e x dx = ey dy = ey = 1.
0 0 0

(b) The same change of variables also yields


Z Z
p p
P(X > t) = px p1 e x dx = ey dy = e t .
t t p

(c) The probability that none of the Xi exceeds 3 is


 n  n
\ p
P {Xi 3} = P(Xi 3) = [1 P(X1 > 3)]n = [1 e 3 ]n .
i=1 i=1

The probability that at least one of them exceeds 3 is


 n   n 
[ \ p
P {Xi > 3} = 1 P {Xi 3} = 1 [1 e 3 ]n .
i=1 i=1

2 /2
9. (a) Since f 0 (x) = xex / 2 , f 0 (x) < 0 for x > 0 and f 0 (x) > 0 for x < 0.
2
(b) Since f 00 (x) = (x2 1)ex /2 / 2 , we see that f 00 (x) > 0 for |x| > 1 and f 00 (x) <
0 for |x| < 1.
2 /2 2 /2
(c) Rearrange ex x2 /2 to get ex 2/x2 0 as |x| .
Chapter 4 Problem Solutions 47

10. Following the hint, write f (x) = ((x m)/ )/ , where is the standard normal
density. Observe that f 0 (x) = 0 ((x m)/ )/ 2 and f 00 (x) = 00 ((x m)/ )/ 3 .
(a) Since the argument of 0 is positive for x > m and negative for x < m, f (x) is
decreasing for x > m and increasing for x < m. Hence, f has a global maximum
at x = m.
(b) Since the absolute value of the argument of 00 is greater than one if and only if
|x m| > , f (x) is concave for |x m| < and convex for |x m| > .
11. Since is bounded, lim ((x m)/ )/ = 0. Hence, lim f (x) = 0. For
x 6= m, we have
h  2 . i
exp xm 2 2 2
f (x) =  2 = 0
2 2 xm 2 (x m)2


as 0. Otherwise, since f (m) = [ 2 ]1 , lim 0 f (m) = .
12. (a) f (x) = n pn fn (x) is obviously nonnegative. Also,
Z Z Z


f (x) dx = pn fn (x) dx =
n
pn
fn (x) dx = pn 1 = pn = 1.
n n n

3/4 3/4

1/2 1/2
(b) (c)
1/4 1/4

0 x 0 x
0 1 2 3 0 1 2 3
R
13. Clearly, (g h)(x) = g(y)h(x y) dy
0 since g and h are nonnegative. Next,
Z Z Z 
(g h)(x) dx = g(y)h(x y) dy dx

Z Z 
= g(y) h(x y) dx dy

Z Z  Z
= g(y) h( ) d dy = g(y) dy = 1.

| {z }
=1
R
14. (a) Let p > 1. On (p) = 0 x p1 ex dx, use integration by parts with u = x p1 and
dv = ex dx. Then du = (p 1)x p2 dx, v = ex , and
Z

(p) = x p1 ex +(p 1) x(p1)1 ex dx = (p 1)(p 1).
0 0
| {z }
=0
48 Chapter 4 Problem Solutions

R
(b) On (1/2) = 0 x1/2 ex dx,make the change of variable x = y2 /2 or y = 2x.
Then dx = y dy and x1/2 = 2/y. Hence,
Z Z y2 /2 Z ey2 /2
2 y2 /2
(1/2) = e y dy = 2 e dy = 2 2 dy
0 y 0 0 2
1
= 2 2 = .
2
(c) By repeatedly using the recursion formula in part (a), we have
     
2n + 1 2n 1 2n 1 2n 1 2n 3 2n 3
= =
2 2 2 2 2 2
..
.
2n 1 2n 3 5 3 1
= (1/2)
2 2 2 2 2
2n 1 2n 3 5 3 1
=
2 2 2 2 2
(2n 1)!!
= .
2n
(d) First note that g p (y) = 0 for y 0, and similarly for gq (y). Hence, in order to
have g p (y)gq (x y) > 0, we need y > 0 and x y > 0, or equivalently, x > y > 0.
Of course, if x 0 this does not happen. Thus, (g p gq )(x) = 0 for x 0. For
x > 0, we follow the hint and write
Z
(g p gq )(x) = g p (y)gq (x y) dy

Z x
= g p (y)gq (x y) dy
0
Z x
1
= y p1 ey (x y)q1 e(xy) dy
(p)(q) 0
Z
xq1 ex x p1
= y (1 y/x)q1 dy
(p)(q) 0
Z 1
xq ex
= (x ) p1 (1 )q1 d , ch. of var. = y/x,
(p)(q) 0
Z 1
x p+q1 ex
= p1 (1 )q1 d . ()
(p)(q) 0

Now, the left-hand side is a convolution of densities, and is therefore a density


by Problem 13. In particular, this means R
that the left-hand side integrates to
one. On the right-hand side, note that 0 x p+q1 ex dx = (p + q). Hence,
integrating the above equation with respect to x from zero to infinity yields
Z 1
(p + q)
1 = p1 (1 )q1 d .
(p)(q) 0

Solving for the above integral and substituting the result into (), we find that
(g p gq )(x) = g p+q (x).
Chapter 4 Problem Solutions 49

R R
15. (a) In f (x) dx = f ( x) dx, make the change of variable y = x, dy =
dx to get Z Z
f (x) dx = f (y) dy = 1.

(b) Observe that


( x)0 e x
g1, (x) = = e x ,
0!
which we recognize as the exp( ) density.
(c) The desired probability is
Z
( x)m1 e x
Pm (t) := dx.
t (m 1)!
R
Note that P1 (t) = t e x dx = e t . For m > 1, apply integration by parts
with u = ( x)m1 /(m 1)! and dv = e x dx. Then

( t)m1 e t
Pm (t) = + Pm1 (t).
(m 1)!

Applying this result recursively, we find that

( t)m1 e t ( t)m2 e t
Pm (t) = + + + e t .
(m 1)! (m 2)!

(d) We have

1 ( 21 x)m1/2 ex/2 (1/2)m (1/2)1/2 xm1/2 ex/2


g 2m+1 , 1 (x) = 2 = (2m1)!!
2 2 ((2m + 1)/2) m 2
xm1/2 ex/2
= .
(2m 1) 5 3 1 2

16. (a) We see that b1,1(x) = 1 is the uniform(0, 1) density, b2,2 (x) = 6x(1 x), and
b1/2,1 (x) = 1/(2 x ).
2
b (x)
1/2,1
b (x)
2,2
1.5

b (x)
1,1
1

0.5

0
0 0.25 0.5 0.75 1
x
50 Chapter 4 Problem Solutions

(b) From Problem 14(d) and its hint, we have


Z 1
x p+q1 ex
g p+q (x) = (g p gq )(x) = p1 (1 )q1 d .
(p)(q) 0

Integrating the left and right-hand sides with respect to x from zero to infinity
yields
Z
(p + q) 1 p1
1 = (1 )q1 d ,
(p)(q) 0
which says that the beta density integrates to one.
17. Starting with
Z 1
(p) (q) = (p + q) u p1 (1 u)q1 du,
0
make the change of variable u = sin2 , du = 2 sin cos d . We obtain
Z 1
(p) (q) = (p + q) u p1 (1 u)q1 du
0
Z /2
= (p + q) (sin2 ) p1 (1 sin2 )q1 2 sin cos d
0
Z /2
= 2(p + q) (sin )2p1 (cos )2q1 d .
0

Setting p = q = 1/2 on both sides yields


Z /2
(1/2)2 = 2 1 d = ,
0

and it follows that (1/2) = .
18. Starting with
Z 1
(p) (q) = (p + q) u p1 (1 u)q1 du,
0
make the change of variable u = sin2 , du = 2 sin cos d . We obtain
Z 1
(p) (q)
= u p1 (1 u)q1 du
(p + q) 0
Z /2
= (sin2 ) p1 (1 sin2 )q1 2 sin cos d
0
Z /2
= 2 (sin )2p1 (cos )2q1 d .
0

Setting p = (n + 1)/2 and q = 1/2 on both sides yields


 
n+1
Z /2
2
  = 2 sinn d ,
n+2 0

2
and the desired result follows.
Chapter 4 Problem Solutions 51

19. Starting with the integral definition of B(p, q), make the change of variable u = 1
e , which implies both du = e d and 1 u = e . Hence,
Z 1 Z
B(p, q) = u p1 (1 u)q1 du = (1 e ) p1 (e )q1 e d
0 0
Z
= (1 e ) p1 eq d .
0

20. We first use the fact that the density is even and then makepthe change of variable
e = 1 + x2 / , which implies both e d = 2x/ dx and x = (e 1). Thus,
Z  Z 
x2 ( +1)/2 x2 ( +1)/2
1+ dx = 2 1+ dx
0
Z
1
= 2 (e )( +1)/2 2 e p d
0 (e 1)
Z /2 1/2 p
= (e ) (e ) / e 1 d
0
Z /2
= (e ) (1 e )1/2 d
0
Z
= (1 e )1/21 e /2 d .
0

By the preceding problem, this is equal to B(1/2, /2), and we see that Students t
density integrates to one.
21. (a) Using Stirlings formula,
1+   1 + (1+ )/21/2 (1+ )/2  1 +  /2
2 e e1/2
2  2 = 2
  /21/2 /2   /21/2
2 e
2 2 2
 1 +  /2 ( /2)1/2 1
= e1/2 = [(1 + 1/ ) ]1/2
2 e1/2
1 1
(e1 )1/2 = .
2e 1/2 2
(b) First write
  
x2 ( +1)/2 x2  1/2  x2 1/2 2 2
1+ = 1+ 1+ [ex ]1/2 11/2 = ex /2 .

It then follows that
 
2 ( +1)/2
1+ 
1 + x 1
f (x) = = 2 
B( 12 , 2 )  
2 ( +1)/2
1 + x
2
2
1 ex /2
2 /2 =
.
2 ex 2
52 Chapter 4 Problem Solutions

22. Making the change of variable t = 1/(1 + z) as suggested in the hint, note that it is
equivalent to 1 + z = t 1 , which implies dz = t 2 dt. Thus,
Z Z 1  p1 Z 1
z p1 1 p+q dt 1 t  p1 p+q2
p+q
dz = 1 t 2
= t dt
0 (1 + z) 0 t t 0 t
Z 1
= (1 t) p1t q1 dt = B(q, p) = B(p, q).
0

Hence, fZ (z) integrates to one.


Z Z
2 2
23. E[X] = x dx = 2x2 dx = = 2.
1 x3 1 x 1
24. If the input-output relation has n levels, then the distance from Vmax to +Vmax should
be n; i.e., n = 2Vmax , or = 2Vmax /n. Next, we have from the example in the text
that the performance is 2 /12, and we need 2 /12 < , or
1  2Vmax 2
< .
12 n

Solving this for n yields Vmax / 3 < n = 2b . Taking natural logarithms, we have
.
b > ln Vmax / 3 ln 2.

25. We use the change of variable x = z m as follows:


Z Z Z
E[Z] = z fZ (z) dz = z f (z m) dz = (x + m) f (x) dx

Z Z
= x f (x) dx + m f (x) dx = E[X] + m = 0 + m = m.

Z Z
2 2
26. E[X 2 ] = x2 dx = dx = 2 ln x = 2( 0) = .
1 x3 1 x 1
R
27. First note that since Students t density is even, E[|X|k ] = k
|x| f (x) dx is propor-
tional to
Z Z 1 Z
xk xk xk
dx = dx + dx
0 (1 + x2 / )( +1)/2 0 (1 + x2 / )( +1)/2 1 (1 + x2 / )( +1)/2
With regard to this last integral, observe that
Z Z Z
xk xk dx
dx dx = ( +1)/2 ,
1 (1 + x / )( +1)/2
2 1 (x / )( +1)/2
2 1 x +1k
which is finite if + 1 k > 1, or k < . Next, instead of breaking
the range of
integration at one, we break it at the solution of x2 / = 1, or x = . Then
Z Z Z
xk xk 
( +1)/2 dx
dx dx = ,
(1 + x / )( +1)/2
2 (x / + x2 / )( +1)/2
2 2
x +1k
which is infinite if + 1 k 1, or k .
Chapter 4 Problem Solutions 53

28. Begin with E[Y 4 ] = E[(Z + n)4 ] = E[Z 4 + 4Z 3 n + 6Z 2 n2 + 4Zn3 + n4 ]. The moments
of the standard normal were computed in an example in this chapter. Hence E[Y 4 ] =
3 + 4 0 n + 6 1 n2 + 4 0 n3 + n4 = 3 + 6n2 + n4 .
Z Z
x p1 ex 1 (n + p)
29. E[X n ] = xn dx = x(n+p)1 ex dx = .
0 (p) (p) 0 (p)
30. (a) First write
Z Z Z 2
x2 /2 1 2 x2 /2 2 2 ex /2
E[X] = x xe dx = x e dx = x dx,
0 2 2 2
where the last integral isthe second moment of a standard normal density, which
2 p
is one. Hence, E[X] = = /2.
2
(b) For higher-order moments, first write
Z Z
2 /2 2 /2
E[X n ] = xn xex dx = xn+1 ex dx.
0 0
2

Now make the change of variable t = x /2, which implies x = 2t, or dx =
dt/ 2t. Hence,
Z
dt
E[X n ] = [(2t)1/2 ]n+1 et
0
Z
21/2t 1/2
= 2n/2 t [(n/2)+1]1 et dt = 2n/2 (1 + n/2).
0

31. Let Xi denote the flow on link i, and put Yi := I( ,) (Xi ) so that Yi = 1 if the flow
on link i is greater than . Put Z := ni=1 Yi so that Z counts the number of links
with flows greater than . The buffer overflows if Z > 2. Since the Xi are i.i.d.,
so are the Yi . Furthermore, the Yi are Bernoulli(p), where p = P(Xi > ). Hence,
Z binomial(n, p). Thus,
P(Z > 2) = 1 P(Z 2)
      
n n n 2
= 1 (1 p)n + p(1 p)n1 + p (1 p)n2
0 1 2
= 1 (1 p)n2 [(1 p)2 + np(1 p) + 21 n(n 1)p2 ].
In remains to compute
Z
x2 /2

x2 /2 2 /2
p = P(Xi > ) = xe dx = e = e .

32. The key is to use the change of variable = x p , which implies both d = px p1 dx
and x = ( / )1/p . Hence,
Z Z
p
E[X n ] = xn px p1 e x dx = [( / )1/p ]n e d
0 0
Z 
= (1/ ) n/p
[(n/p)+1]1 e d = (1 + n/p) n/p .
0
54 Chapter 4 Problem Solutions

33. Write
Z Z
x1/2 ex dx =
E[Y ] = E[(X 1/4 )2 ] = E[X 1/2 ] = x3/21 ex dx
0 0

= (3/2) = (1/2)(1/2) = /2.

34. We have
 n   n 
[ \
P {Xi < /2} = 1 P {Xi /2}
i=1 i=1
n
= 1 P(Xi /2)
i=1
Z
n
x
= 1 e dx , with := 1/ ,
/2

= 1 (e /2 )n = 1 en/2 .

35. Let Xi exp( ) be i.i.d., where = 1/20. We must compute


 5   5 
[ \
P {Xi > 25} = 1 P {Xi 25}
i=1 i=1
5
= 1 P(Xi 25)
i=1
Z 25
5
x
= 1 e dx
0

= 1 (1 e25 )5 = 1 (1 e5/4 )5 = 0.815.

36. The first two calculations are


Z 2 Z 1/2
h(X) = (1/2) log 2 dx = log 2 and h(X) = 2 log(1/2) dx = log(1/2).
0 0

For the third calculation, note that ln f (x) = 21 [(x m)/ ]2 + 12 ln 2 2 . Then
Z  
1
h(X) = f (x) [(x m)/ ]2 + ln 2 2 dx
2

    
1 X m 2
= E + ln 2 2 = 21 {1 + ln 2 2 } = 1
2 ln 2 2 e.
2

37. The main difficulty is to compute


Z
x2n (1 + x2 / )( +1)/2 dx.

Chapter 4 Problem Solutions 55

First use the fact that the integrand is even and then makep the change of variable
e = 1 + x2 / , which implies both e d = 2x/ dx and x = (e 1). Thus,
Z Z
2x
x2n (1 + x2 / )( +1)/2 dx = x2n1 (1 + x2 / )( +1)/2 dx
0
Z p
= ( (e 1) )2n1 (e )( +1)/2 e d
0
Z
= n+1/2 (e 1)n1/2 e ( +1)/2 e d
0
Z
= n+1/2 (1 e )n1/2 e ( 2n)/2 d
0
Z
= n+1/2 (1 e )(n+1/2)1 e ( 2n)/2 d
0

= n+1/2 B n + 1/2, ( 2n)/2 , by Problem 19.

Hence,
 1
E[X 2n ] = n+1/2 B n + 1/2, ( 2n)/2
B( 21 , 2 )
2n
( 2n+1
2 )( 2 ) ( +1
2 )
2n+1
n ( 2 )( 2 )
2n
= n+1/2 = .
( +1
2 ) ( 12 )( 2 ) ( 12 )( 2 )

2 s2 /2
38. From MX (s) = e , we have MX0 (s) = MX (s) 2 s and then

MX00 (s) = MX (s) 4 s2 + MX (s) 2 .

Since MX (0) = 1, we have MX00 (1) = 2 .


2
39. Let M(s) := es /2 denote the moment generating function of the standard normal ran-
dom variable. For the N(m, 2 ) moment generating function, we use the change of
variable y = (x m)/ , dy = dx/ to write
Z Z 2 Z 2
exp[ 21 ( xm )2 ] ey /2 ey /2
esx dx = es( y+m) dy = esm es y dy
2 2 2
2 s2 /2
= esm M(s ) = esm+ .

Z 1
s x1s 1 1
40. E[esY ] = E[es ln(1/X) ] = E[eln X ] = E[X s ] = xs dx = = .
0 1 s 0 1s
41. First note that 
x, x 0,
|x| =
x, x < 0.
Then the Laplace( ) mgf is
Z
E[esX ] = esx 2 e |x| dx

56 Chapter 4 Problem Solutions

Z Z 0 
sx x sx x
= e e dx + e e dx
2 0
Z Z 0 
x( s) x( +s)
= e dx + e dx .
2 0

Of these last two integrals, the one on the left is finite if > Re s, while the second is
finite if Re s > . For both of them to be finite, we need < Re s < . For such
s both integrals are easy to evaluate. We get
 
sX 1 1 2 2
MX (s) := E[e ] = + = 2 2 = 2 2.
2 s +s 2 s s
Now, MX0 (s) = 2s 2 /( 2 s2 )2 , and so the mean is MX0 (0) = 0. We continue with
( 2 s2 )2 + 4s2 ( 2 s2 )
MX00 (s) = 2 2
( 2 s2 )4 .
Hence, the second moment is MX00 (0) = 2/ 2 . Since the mean is zero, the second
moment is also the variance.
42. Since X is a nonnegative random variable, for s 0, sX 0 and esX 1. Hence, for
s 0, MX (s) = E[esX ] E[1] = 1 < . For s > 0, we show that MX (s) = . We use
the fact that for z > 0,
n
z z3
ez = .
n=0 n! 3!
Then for s > 0, sX > 0, and we can write
  Z
(sX)3 s3 2s3 x3
MX (s) = E[esX ] E = E[X 3 ] = dx = .
3! 3! 3! 1 x3

43. We apply integration by parts with u = x p1 /(p) and dv = ex(1s) dx. Then du =
x p2 /(p 1) dx and v = ex(1s) /(1 s). Hence,
Z Z
x p1 ex x p1
Mp (s) = esx dx = ex(1s) dx
0 (p) 0 (p)
Z
x p1 ex(1s) 1 x p2 x(1s)
= + e dx.
(p) 1 s 0 1 s 0 (p 1)
The last term is Mp1 (s)/(1 s). The other term is zero if p > 1 and Re s < 1.
44. (a) In this case, we use the change of variable t = x(1 s), which implies x =
t/(1 s) and dx = dt/(1 s). Hence,
Z Z
x p1 ex 1
Mp (s) = esx dx = x p1 ex(1s) dx
0 (p) (p) 0
Z 
1 t  p1 t dt
= e
(p) 0 1 s 1s
 1 p 1 Z  1 p
p1 t
= t e dt = .
1s (p) 0 1s
| {z }
=1
Chapter 4 Problem Solutions 57

(b) From MX (s) = (1 s)p , we find MX0 (s) = p(1 s)p1 , MX00 (s) = p(p + 1)(1
s)p2 , and so on. The general result is

(n) (n + p)
MX (s) = p(p + 1) (p + [n 1])(1 s)pn = (1 s)pn .
(p)
Hence, the Taylor series is

sn (n) sn (n + p)
MX (s) = n! MX (0) = n! (p)
.
n=0 n=0

45. (a) Make the change of variable t = x or x = t/ , dx = dt/ . Thus,


Z Z
( x) p1 e x t p1 et
E[esX ] = esx dx = e(s/ )t dt,
0 (p) 0 (p)
which is the moment generating function of g p evaluated at s/ . Hence,
 1 p  p
E[esX ] = = ,
1 s/ s
and the characteristic function is
 1 p  p
E[e j X ] = = .
1 j / j
 m  m
(b) The Erlang(m, ) mgf is , and the chf is .
s j
 1 k/2
(c) The chi-squared with k degrees of freedom mgf is , and the chf is
 1 k/2 1 2s
.
1 2 j
46. First write
x /2 Z 2 Z x2 (12s)/2
2 2e e
MY (s) = E[esY ] = E[esX ] = esx dx = dx.
2 2

If we let (1 2s) = 1/ 2 ; i.e., = 1/ 1 2s, then
Z (x/ )2 /2
e 1
MY (s) = dx = = .
2 1 2s

47. First observe that


2 2 /2 2 2 2 2 2
esx e(xm) = e(x 2xm+m 2sx )/2 = e[x (12s)2xm]/2 em /2
2 2 2 2
= e(12s){x 2xm/(12s)+[m/(12s)] [m/(12s)] }/2 em /2
2 2 2
= e(12s){x[m/(12s)]} /2 em /[2(12s)] em /2
2 /2 2 /(12s)
= e(12s){x[m/(12s)]} esm .
58 Chapter 4 Problem Solutions


If we now let 1 2s = 1/ 2 , or = 1/ 1 2s, and = m/(1 2s), then
Z (xm)2 /2 Z [(x )/ ]2 /2
sY sX 2 sx2 e sm2 /(12s) e
E[e ] = E[e ] = e dx = e dx
2 2
2
2 /(12s) esm /(12s)
= esm = .
1 2s

48. Y ( ) = E[e jY ] = E[e j (aX+b) ] = E[e j( a)X ]e j b = X (a )e j b .


49. The key observation is that

, 0,
| | =
, < 0.

It then follows that


Z
1 | | j x
fX (x) = e e d
2
Z Z 
1 1 0 j x
= e e j x d + e e d
2 0 2
Z Z 0 
1
= e ( + jx) d + e ( jx) d
2 0
 0 
1 1 ( + jx) 1
( jx)
= e + e
2 + jx 0 jx
   
1 1 1 1 2 /
= + = 2 2
= 2 .
2 + jx jx 2 + x + x2

 2  2
d ex /2 ex /2
50. (a) = x = x f (x).
dx 2 2
Z Z Z
d j x
(b) X0 ( ) = e f (x) dx = j e j x x f (x) dx = j e j x f 0 (x) dx.
d
(c) In this last integral, let u = e j x and dv = f 0 (x) dx. Then du = j e j x dx, v =
f (x), and the last integral is equal to
Z
j x

e f (x) j e j x f (x) dx = jX ( ).

| {z }
=0

(d) Combining (b) and (c), we have X0 ( ) = j[ jX ( )] = X ( ).


2 /2
(e) If K( ) := X ( )e , then
2 /2 2 /2 2 /2 2 /2
K 0 ( ) = X0 ( )e + X ( ) e = X ( )e + X ( ) e = 0.
Chapter 4 Problem Solutions 59

(f) By the mean-value theorem of calculus, for every , there is a 0 between 0 and
such that K( ) K(0) = K 0 (0 )( 0). Since the derivative is zero, we have
2
K( ) = K(0) = X (0) = 1. It then follows that X ( ) = e /2 .
51. Following the hints, we first write

d d x p ex px p1 ex x p ex
xg p (x) = = = pg p (x) xg p (x) = (p x)g p (x).
dx dx (p) (p)
In Z Z
d
X0 ( ) = e j x g p (x) dx = j e j x xg p (x) dx,
d 0 0
apply integration by parts with u = xg p (x) and dv = e j x dx. Then du is given above,
v = e j x /( j ), and
 Z 
0 xg p (x)e j x 1 j x
X ( ) = j j 0 e (p x)g p (x) dx
j 0
| {z }
=0
 Z Z 
1 j x 1 j x
= p e g p (x) dx e ( jx)g p (x) dx
0 j 0
 
1 1
= pX ( ) X0 ( ) = (p/ )X ( ) + (1/ j )X0 ( ).
j
Rearrange this to get

X0 ( )(1 1/ j ) = (p/ )X ( ),

and multiply through by j to get

X0 ( )( j + 1) = j pX ( ).

Armed with this, the derivative of K( ) := X ( )(1 j ) p is

K 0 ( ) = X0 ( )(1 j ) p + X ( )p(1 j ) p1 ( j)
= (1 j ) p1 [X0 ( )(1 j ) j pX ( )] = 0.

By the mean-value theorem of calculus, for every , there is a 0 between 0 and


such that K( ) K(0) = K 0 (0 )( 0). Since the derivative is zero, we have
K( ) = K(0) = X (0) = 1. It then follows that X ( ) = 1/(1 j ) p .
52. We use the formula cov(X, Z) = E[XZ] E[X]E[Z]. The mean of an exp( ) random
variable is 1/ . Hence, E[X] = 1. Since Z := X + Y , E[Z] = E[X] + E[Y ]. Since
the Laplace random variable has zero mean, E[Y ] = 0. Hence, E[Z] = E[X] = 1.
Next, E[XZ] = E[X(X + Y )] = E[X 2 ] + E[XY ] = E[X 2 ] + E[X]E[Y ] by independence.
Since E[Y ] = 0, E[XZ] = E[X 2 ] = var(X) + (E[X])2 = 1 + 12 = 2, where we have used
the fact that the variance of an exp( ) random variable is 1/ 2 . We can now write
cov(X, Z) = 2 1 = 1. Since Z is the sum of independent, and therefore uncorrelated,
random variables, var(Z) = var(X + Y ) = var(X) + var(Y ) = 1 + 2 = 3, where we
have used the fact that the variance of a Laplace( ) random variable is 2/ 2 .
60 Chapter 4 Problem Solutions

53. Since Z = X +Y , where X N(0, 1) and Y Laplace(1) are independent, we have

var(Z) = var(X +Y ) = var(X) + var(Y ) = 1 + 2 = 3.

54. Write

MZ (s) = E[esZ ] = E[es(XY ) ] = E[esX esY ] = E[esX ]E[esY ] = MX (s)MY (s).

If MX (s) = MY (s) = /( s), then

2
MZ (s) = = = 2 2,
s (s) s +s s

which is the Laplace( ) mgf.


55. Because the Xi are independent, we can write
 n  n n
sYn s(X1 ++Xn ) sXi
MYn (s) := E[e ] = E[e ] = E e = E[esXi ] = MXi (s). ()
i=1 i=1 i=1

2 2
(a) For Xi N(mi , i2 ), MXi (s) = esmi +i s /2 . Hence,
n  n  n  
smi +i2 s2 /2 2 2 2 s2 /2
MYn (s) = e = exp s mi + i s /2 = e|sm+{z },
i=1 i=1 i=1
N(m, 2 ) mgf

provided we put m := m1 + + mn and 2 := 12 + + n2 .


(b) For Cauchy random variables, we must observe that the moment generating
function exists only for s = j . Equivalently, we must use characteristic func-
tions. In this case, () becomes
n
Yn ( ) := E[e jYn ] = Xi ( ).
i=1

Now, the Cauchy(i ) chf is Xi ( ) = ei | | . Hence,


n   n  
i | | | |
Yn ( ) = e = exp i | | = |e {z } ,
i=1 i=1
Cauchy( ) chf

provided we put := 1 + + n .
(c) For Xi gamma(pi , ), the mgf is MXi (s) = [ /( s)] pi . Hence,
n   pi   p1 ++pn  p

MYn (s) = s
=
s
=
s
,
i=1 | {z }
gamma(p, ) mgf

provided we put p := p1 + + pn .
Chapter 4 Problem Solutions 61

56. From part (c) of the preceding problem, Y gamma(rp, ). The table inside the back
cover of the text gives the nth moment of a gamma random variable. Hence,
(n + rp)
E[Y n ] = .
n (rp)

57. Let Ti denote the time to transmit packet i. Then the time to transmit n packets is T :=
T1 + + Tn . We need to find the density of T . Since the Ti are exponential, we can
apply the remark in the statement of Problem 55(c) to conclude that T Erlang(n, ).
Hence,
( t)n1 e t
fT (t) = , t 0.
(n 1)!
 n  n
1 1
58. Observe that Y = ln = ln . By Problem 40, each term is an exp(1)
i=1 X i i=1 Xi
random variable. Hence, by the remark in the statement of Problem 55(c), Y
Erlang(n, 1); i.e.,
yn1 ey
fY (y) = , y 0.
(n 1)!
59. Consider the characteristic function,
 n  n
j Y j (1 X1 ++n Xn ) j(i )Xi
Y ( ) = E[e ] = E[e ] = E e = E[e j(i )Xi ]
i=1 i=1
n n   n  
|i | i | |
= e = e = exp i | | .
i=1 i=1 i=1

This is the chf of a Cauchy random variable with parameter ni=1 i . Hence,

ni=1 i
fY (y) =  2 .
ni=1 i + y2

60. We need to compute P(|X Y | 2). If we put Z := X Y , then we need to compute


P(|Z| 2). We first find the density of Z using characteristic functions. Write

Z ( ) = E[e j (XY ) ] = E[e j X e jY ] = E[e j X ]E[e j( )Y ] = e| | e| | = e2| | ,


which is the chf of a Cauchy(2) random variable. Since the Cauchy density is even,
Z 2    1  2 2
1 1 z 2 1
P(|Z| 2) = 2 fZ (z) dz = 2 tan + = tan1 (1) = = .
0 2 2 0 4 2

61. Let X := U +V +W be the sum of the three voltages. The alarm sounds if X > x. To
find P(X > x), we need the density of X. Since U, V , and W are i.i.d. exp( ) random
variables, by the remark in the statement of Problem 55(c), X Erlang(3, ). By
Problem 15(c),
2
( x)k e x
P(X > x) = .
k=0 k!
62 Chapter 4 Problem Solutions

62. Let Xi Cauchy( ) be the i.i.d. line loads. Let Y := X1 + + Xn be the total load.
The substation shuts down if Y > `. To find P(Y > `), we need to find the density of
Y . By Problem 55(b), Y Cauchy(n ), and so
Z   y  1 
1
P(Y > `) = fY (y) dy = tan1 +
` n 2 `
     ` 
1 ` 1 1 1
= 1 tan1 + = tan1 .
n 2 2 n

63. Let the Ui uniform[0, 1] be the i.i.d. efficiencies of the extractors. Let Xi = 1 if
extractor i operates with efficiency less than 0.25; in symbols, Xi = I[0,0.25) (Ui ), which
is Bernoulli(p) with p = 0.25. Then Y := X1 + + X13 is the number of extractors
operating at less than 0.25 efficiency. The outpost operates normally if Y < 3. We
must compute P(Y < 3). Since Y is the sum of i.i.d. Bernoulli(p) random variables,
Y binomial(13, p). Thus,

P(Y < 3) = P(Y = 0) + P(Y = 1) + P(Y = 2)


     
13 0 13 13 12 13 2
= p (1 p) + p(1 p) + p (1 p)11
0 1 2
= (1 p)11 [(1 p)2 + 13p(1 p) + 78p2 ] = 0.3326.

64. By the remark in the statement of Problem 55(c), R = T + A is chi-squared with


k = 2 degrees of freedom. Since the number of degrees of freedom is even, R is
Erlang(k/2, 1/2) = Erlang(1, 1/2) = exp(1/2). Hence,
Z
P(R > r) = (1/2)ex/2 dx = er/2 .
r

65. (a) Since c2n+k is a density, it integrates to one. So,


Z Z 2
( 2 /2)n e /2
0
ck, 2 (x) dx =
0 n=0 n!
c2n+k (x) dx

2 /2 Z
( 2 /2)n e
= n!
c2n+k (x) dx
n=0 |0 {z }
=1
2
( 2 /2)n e /2
= n!
= 1.
n=0 | {z }
Poisson( 2 /2) pmf

(b) The mgf is


Z
Mk, 2 (s) = esx ck, 2 (x) dx
0
Z  2 
sx ( 2 /2)n e /2
= e c2n+k (x) dx
0 n=0 n!
Chapter 4 Problem Solutions 63

2 /2 Z
( 2 /2)n e
= esx c2n+k (x) dx
n=0 n! 0
2 /2 
(2n+k)/2
( 2 /2)n e 1
=
n=0 n! 1 2s
2  
e /2
1 2 /2 n
=
(1 2s)k/2 n=0 n! 1 2s
h  i
1
2
e /2 ( 2 /2)/(12s) exp ( 2 /2) 12s 1
= e =
(1 2s)k/2 (1 2s)k/2
h  i
2s
exp ( 2 /2) 12s exp[s 2 /(1 2s)]
= = .
(1 2s)k/2 (1 2s)k/2

(c) If we first note that


 
d s 2 (1 2s) 2 s 2 (2)
= = 2,
ds 1 2s s=0 1 2s
s=0

(s) + (s)
then it is easy to show that Mk,0 2 (s) has the general form , where
(1 2s)k
2 0 2
(0) = and (0) = k. Hence, E[X] = Mk, 2 (0) = + k.
(d) The usual mgf argument gives
n
MY (s) = E[esY ] = E[es(X1 ++Xn ) ] = Mki ,i2 (s)
i=1
n
exp[si2 /(1 2s)]
=
i=1 (1 2s)ki /2
exp[s(12 + + n2 )/(1 2s)]
= .
(1 2s)(k1 ++kn )/2

If we put k := k1 + + kn and 2 := 12 + + n2 , we see that Y is noncentral


chi-squared with k degrees of freedom and noncentrality parameter 2 .
(e) We first consider

e x + e x 1 ( x )n
2n xn
= [1 + (1)n ] =
2 2 n=0 n! n=0 (2n)!

2n xn
= 1 3 5 (2n 1) 2n n!
n=0

2n ( 2 /2)n (x/2)n
= 1 3 5 (2n 1) n!
n=0

( 2 /2)n (x/2)n
= ( 2n+1
, by Problem 14(c).
n=0 2 ) n!
64 Chapter 4 Problem Solutions

We can now write


2 2
e(x+ )/2 e x + e x e(x+ )/2 ( 2 /2)n (x/2)n

2 x

2
=
2 x n=0 ( 2n+1
2 ) n!
2 /2
( 2 /2)n e (1/2)(x/2)n1/2 ex/2
= n!

( 2n+1
n=0 2 )
2 /2
( 2 /2)n e
= n!
c2n+1 (x) = c1, 2 (x).
n=0

R
66. First, P(X a) = a 2x3 dx = 1/a2 , while, using the result of Problem 23, the
Markov bound is E[X]/a = 2/a. Thus, the true probability is 1/a2 , but the bound
is 2/a, which decays much more slowly for large a.
67. We begin by noting that P(X a) = ea , E[X] = 1, and E[X 2 ] = 2. Hence, the Markov
bound is 1/a, and the Chebyshev bound is 2/a2 . To find the Chernoff bound, we must
minimize h(s) := esa MX (s) = esa /(1 s) over 0 < s < 1. Now,

(1 s)(a)esa + esa
h0 (s) = .
(1 s)2

Solving h0 (s) = 0, we find s = (a 1)/a, which is positive only for a > 1. Hence, the
Chernoff bound is valid only for a > 1. For a > 1, the Chernoff bound is

ea(a1)/a
h((a 1)/a) = = ae1a .
1 (a 1)/a

(a) It is easy to see that the Markov bound is smaller than the Chebyshev bound
for 0 < a < 2. However, note that the Markov bound is greater than one for
0 < a < 1, and the Chebyshev bound is greater than one for 0 < a < 2.
(b) MATLAB.
2 0
10
Chebyshev 2/a2 Markov 1/a
2
1.5 10
2
Chebyshev 2/a
4
10
1 Chernoff ae1a
1a
Chernoff ae 6
10
0.5
Markov 1/a 8 P(X > a)
10
P(X > a)
0
1 2 3 4 5 6 6 8 10 12 14 16 18 20
a a

The Markov bound is the smallest on [1, 2]. The Chebyshev bound is the small-
est from a = 2 to a bit more than a = 5. Beyond that, the Chernoff bound is the
smallest.
CHAPTER 5
Problem Solutions

1. For x 0, x
Z x
t t x
F(x) = e dt = e = 1e .
0 0

For x < 0, F(x) = 0.


2. For x 0,
Z x x
t (t/ )2 /2

(t/ )2 /2 (x/ ) 2 /2
F(x) = e dt = e = 1e .
0 2 0

For x < 0, F(x) = 0.


3. For x 0,
Z x x
p p p
F(x) = pt p1 e t dt = e t = 1 e x .
0 0

For x < 0, F(x) = 0.


4. For x 0, first write
Z xr Z x/ r
2 t 2 (t/ )2 /2 2 2 /2
F(x) = e dt = 2 e d ,
0 3 0

where we have used the change of variable = t/ . Next use integration by parts
2
with u = and dv = e /2 d . Then
r  x/ Z x/ 
2
2 /2 2 /2
F(x) = e + 0 e d
0
r Z x/ 2 /2
2 x (x/ )2 /2 e
= e +2 d
0 2
r
2 x (x/ )2 /2
= e + 2[(x/ ) 1/2]

r
2 x (x/ )2 /2
= 2(x/ ) 1 e .

For x < 0, F(x) = 0.


5. For y > 0, F(y) = P(Y y) = P(eZ y) = P(Z ln y) = FZ (ln y). Then fY (y) =
fZ (ln y)/y for y > 0. Since Y := eZ > 0, fY (y) = 0 for y 0.

65
66 Chapter 5 Problem Solutions

6. To begin, write FY (y) = P(Y y) = P(1 X y) = P(1 y X) = 1 FX (1 y).


Thus, fY (y) = fX (1 y) (1) = fX (1 y). In the case of X uniform(0, 1),

fY (y) = fX (1 y) = I(0,1) (1 y) = I(0,1) (y),

since 0 < 1 y < 1 if and only if 0 < y < 1.


7. For y > 0,

FY (y) = P(Y y) = P(ln(1/X) y) = P(1/X ey ) = P(X ey ) = 1 FX (ey ).

Thus, fY (y) = fX (ey ) (ey ) = ey , since fX (ey ) = I(0,1) (ey ) = 1 for y > 0.
Since Y := ln(1/X) > 0, fY (y) = 0 for y 0.
8. For y 0,

FY (y) = P( X P y) = P(X p y/ ) = P(X (y/ )1/p ) = FX ((y/ )1/p ).

Thus, fY (y) = fX ((y/ )1/p ) 1p (y/ )(1/p)1 / . Using the formula for the Weibull
density, we find that fY (y) = ey for y 0. Since Y := X p 0, fY (y) = 0 for y < 0.
Thus, Y exp(1).
2
9. For y 0, write FY (y) = P( X y) = P(X y2 ) = FX (y2 ) = 1 ey . Thus,
2 y 2
fY (y) = ey (2y) = e(y/(1/ 2)) /2 ,
(1/ 2)2

which is the Rayleigh(1/ 2 ) density.
10. Recall that the moment generating function of X N(m, 2 ) is MX (s) = E[esX ] =
2 2
esm+s /2 . Thus,
2 2 /2
E[Y n ] = E[(eX )n ] = E[enX ] = MX (n) = enm+n .

11. For y > 0, FY (y) = P(X 2 y) = P( y X y) = FX ( y) FX ( y). Thus,

fY (y) = fX ( y)( 12 y1/2 ) fX ( y)( 12 y1/2 ).

Since fX is even,

ey/2
fY (y) = y1/2 fX ( y) = , y > 0.
2 y

12. For y > 0,

FY (y) = P((X + m)2 y)



= P( y X + m y)

= P( y m X y m)

= FX ( y m) FX ( y m).
Chapter 5 Problem Solutions 67

Thus,

fY (y) = fX ( y m)( 21 y1/2 ) fX ( y m)( 12 y1/2 )


1 2 2
= e( ym) /2 + e( 2m) /2 2
2 y
 
1 2 2
= e(y2 ym+m )/2 + e(y+2 ym+m )/2 2
2 y
2  
e(y+m )/2 em y + em y
= , y > 0.
2 y 2

13. Using the example mentioned in the hint, we have


n n
FXmax (z) = FXk (z) = F(z)n and FXmin (z) = 1 [1FXk (z)] = 1[1F(z)]n .
k=1 k=1

14. Let Z := max(X,Y ). Since X and Y are i.i.d., we have from the preceding problem
that FZ (z) = FX (z)2 . Hence,

fZ (z) = 2FX (z) fX (z) = 2(1 e z ) e z , z 0.

Next,
Z Z
E[Z] = z fZ (z) dz = 2 ze z (1 e z ) dz
0 0
Z Z
= 2 z e z dz z (2 )e(2 )z dz
0 0
1 1 3
= 2 = .
2 2

15. Use the laws of total probability and substitution and the fact that conditioned on
X = m, Y Erlang(m, ). In particular, E[Y |X = m] = m/ . We can now write

E[XY ] = E[XY |X = m]P(X = m) = E[mY |X = m]P(X = m)
m=0 m=0

= mE[Y |X = m]P(X = m) = m(m/ )P(X = m)
m=0 m=0
1 + 2
= E[X 2 ] = , since X Poisson( ).

16. The problem statement tells us that P(Y > y|X = n) = eny . Using the law of total
probability and the pgf of X Poisson( ), we have

P(Y > y) = P(Y > y|X = n)P(X = n) = eny P(X = n)
n=0 n=0
yX y (ey 1)
= E[e ] = GX (e ) = e .
68 Chapter 5 Problem Solutions

17. (a) Using the law of substitution, independence, and the fact that Y N(0, 1), write

FZ|X (z|i) = P(Z z|X = i) = P(X +Y z|X = i) = P(Y z i|X = i)


= P(Y z i) = (z i).

Next,
1
FZ (z) = FZ|X (z|i)P(X = i) = (1 p)(z) + p(z 1),
i=0

and so
2 /2 2 /2
(1 p)ez + pe(z1)
fZ (z) = .
2

(b) From part (a), it is easy to see that fZ|X (z|i) = exp[(z i)2 /2]/ 2 . Hence,

fZ|X (z|1) P(X = 0) exp[(z 1)2 /2] 1 p


becomes ,
fZ|X (z|0) P(X = 1) exp[z2 /2] p

or ez1/2 (1 p)/p. Taking logarithms, we can further simplify this to


1 1 p
z + ln .
2 p

18. Use substitution and independence to write

FZ|A,X (z|a, i) = P(Z z|A = a, X = i) = P(X/A +Y z|A = a, X = i)


= P(Y z i/a|A = a, X = i) = P(Y z i/a) = (z i/a).

19. (a) Write



FZn (z) = P(Zn z) = P( Yn z) = P(Yn z2 ) = FYn (z2 ).

For future reference, note that

fZn (z) = fYn (z2 ) (2z).

Since Yn is chi-squared with n degrees of freedom, i.e., gamma(n/2, 1/2),

1 (y/2)n/21 ey/2
fYn (y) = 2 , y > 0,
(n/2)
and so
2 /2
(z2 /2)n/21 ez
fZn (z) = z , z > 0.
(n/2)
(b) When n = 1, we obtain the folded normal density,
2 /2 2
(z2 /2)1/2 ez ez /2
fZ1 (z) = z = 2 , z > 0.
(1/2) 2
Chapter 5 Problem Solutions 69

(c) When n = 2, we obtain the Rayleigh(1) density,


2 /2
(z2 /2)0 ez 2 /2
fZ2 (z) = z = zez , z > 0.
(1)
(d) When n = 3, we obtain the Maxwell density,
2 /2 2 r
(z2 /2)1/2 ez z2 ez /2 2 2 z2 /2
fZ3 (z) = z = 1 1 = z e , z > 0.
(3/2) 2 2 ( 2 )

(e) When n = 2m, we obtain the Nakagami-m density,


2 /2
(z2 /2)m1 ez 2 2
fZ2m (z) = z = z2m1 ez /2 , z > 0.
(m) 2m (m)

20. Let Z := X1 +
+ Xn . By Problem 55(a) in Chapter 4, Z N(0, n), and it follows
that V := Z/ n N(0, 1). We can now write Y = Z 2 = nV 2 . By Problem 11, V 2 is
chi-squared with one degree of freedom. Hence,
FY (y) = P(nV 2 y) = P(V 2 y/n),
and
e(y/n)/2 1 e(y/n)/2
fY (y) = fV 2 (y/n)/n = p = , y > 0.
2 y/n n 2 ny

21. (a) Since FY (y) = P(Y y) = P(X 1/q y) = P(X yq ) = FX (yq ),


q q
q q1 q1 (yq ) p1 ey qyqp1 ey
fY (y) = fX (y ) (qy ) = qy = , y > 0.
(p) (p)
q
(b) Since q > 0, as y 0, yq 0, and ey 1. Hence, the behavior of fY (y)
as y 0 is determined by the behavior of y p1 . For p > 1, p 1 > 0, and
so y p1 0. For p = 1, y p1 = y0 = 1. For 0 < p < 1, p 1 < 0 and so
y p1 = 1/y1p . Thus,

0, p > 1,
lim fY (y) = q/(1/q), p = 1,
y0
, 0 < p < 1.

(c) We begin with the given formula


q
q( y) p1 e( y)
fY (y) = , y > 0.
(p/q)

(i) Taking q = p and replacing with 1/p yields


1/p y) p p
fY (y) = 1/p p( 1/p y) p1 e( = 1/p p 11/p y p1 e y
p
= py p1 e y ,
which is the Weibull(p, ) density.
70 Chapter 5 Problem Solutions


(ii) Taking p = q = 2 and replacing with 1/( 2 ) yields
2 2
fY (y) = 2/( 2 )[y/( 2 )]e[y/( 2 )] = (y/ 2 )e(y/ ) /2 ,
which is the required Rayleigh density.

(iii) Taking p = 3, q = 2, and replacing with 1/( 2 ) yields
2 2
2/( 2 )[y/( 2 )]2 e[y/( 2 )] (y2 / 3 )e(y/ ) /2
fY (y) = = 1 1
(3/2) 2 2 ( 2 )
r
2
2 y (y/ )2
= e ,
3
which is the required Maxwell density.
(d) In
Z q
q( y) p1 e( y)
E[Y n ] = yn dy,
0 (p/q)
make the change of variable t = ( y)q , dt = q( y)q1 dy. Then
Z
1 q
E[Y n ] = n
( y)n ( y) pq e( y) q( y)q1 dy
(p/q) 0
Z
1
= (t 1/q )n+pq et dt
(p/q) n 0
Z
1 ((n + p)/q)
= t (n+p)/q1 et dt = .
(p/q) n 0 (p/q) n
(e) We use the same change of variable as in part (d) to write
Z y q Z ( y)q
q( ) p1 e( ) 1
FY (y) = d = (t 1/q ) pq et dt
0 (p/q) (p/q) 0
Z ( y)q
1
= t (p/q)1 et dt = G p/q (( y)q ).
(p/q) 0

2
22. Following the hint, let u = t 1 and dv = tet /2 dt so that du = 1/t 2 dt and v =
2
et /2 . Then
Z Z t 2 /2
et /2
2
t 2 /2 e
e dt = dt
x t x x t2
2 Z t 2 /2
ex /2 e
= dt ()
x x t2
2
ex /2
< .
x
The next step is to write the integral in () as
Z t 2 /2 Z
e 1 2 /2
dt = tet dt
x t2 x t3
Chapter 5 Problem Solutions 71

2
and apply integration by parts with u = t 3 and dv = tet /2 dt so that du = 3/t 4 dt
2
and v = et /2 . Then
Z t 2 /2 Z t 2 /2
et /2
2
e e
2
dt = 3 3 dt
x t t x x t4
2 Z t 2 /2
ex /2 e
= 3 dt.
x3 x t4
Substituting this into (), we find that
Z 2 2 Z t 2 /2
t 2 /2 ex /2 ex /2 e
e dt = +3 dt
x x x3 x t4
2 2
ex /2 ex /2
.
x x3

23. (a) Write


Z
E[FXc (Z)] = FXc (z) fZ (z) dz

Z Z 
= fX (x) dx fZ (z) dz
z
Z Z 
= I(z,) (x) fX (x) dx fZ (z) dz

Z Z 
= fX (x) I(z,) (x) fZ (z) dz dx

Z Z 
= fX (x) I(,x) (z) fZ (z) dz dx

Z Z x 
= fX (x) fZ (z) dz dx

Z
= fX (x)FZ (x) dx = E[FZ (X)].

(b) From Problem 15(c) in Chapter 4, we have


m1
( z)k e z
FZ (z) = 1 k!
, for z 0,
k=0

and FZ (z) = 0 for z < 0. Hence,


  
m1
( X)k e X
E[FZ (X)] = E I[0,) (X) 1
k=0 k!
m1
k
= P(X 0) E[X k e X I[0,) (X)].
k=0 k!
72 Chapter 5 Problem Solutions

(c) Let X N(0, 1) so that E[Q(Z)] = E[FXc (Z)] = E[FZ (X)]. Next, since Z
exp( ) = Erlang(1, ), we can use the result of part (b) to write
E[FZ (X)] = P(X 0) E[e X I[0,) (X)]
Z x2 /2
1 x e
= e dx
2 0 2
2 Z
1 e /2 2 +2x + 2 )/2
= e(x dx
2 2 0
2 Z
1 e /2 2 /2
= e(x+ ) dx
2 2 0
2 Z
1 e /2 2 /2 1 2
= et dt = e /2 Q( ).
2 2 2

(d) Put Z := Y and make the following observations. First, for z 0,

FZ (z) = P( Y z) = P( 2Y z2 ) = P(Y (z/ )2 ) = FY ((z/ )2 ),
and FZ (z) = 0 for z < 0. Second, since Y is chi-squared with 2m-degrees of
freedom, Y Erlang(m, 1/2). Hence,
2 /2
m1
((z/ )2 /2)k e(z/ )
FZ (z) = 1 k!
, for z 0.
k=0

Third, with X N(0, 1),



E[Q( Y )] = E[Q(Z)] = E[FXc (Z)] = E[FZ (X)]
is equal to
  2 
m1
((X/ )2 /2)k e(X/ ) /2
E I[0,) (X) 1 ,
k=0 k!
which simplifies to
m1
1 2
P(X 0) E[X 2k e(X/ ) /2 I[0,) (X)].
k=0 2k 2k k!
e 2 := (1 + 1/ 2 )1 , we have
Now, P(X 0) = 1/2, and with
Z x2 /2
2k (X/ )2 /2 2k (x/ )2 /2 e
E[X e I[0,) (X)] = x e dx
0 2
Z
e 2
= x2k e(x/e ) /2 dx
2 e 0
Z
e 1 2
= x2k e(x/e ) /2 dx
2 2 e
e
= e 2k .
1 3 5 (2k 1)
2
Putting this all together, the desired result follows.
Chapter 5 Problem Solutions 73

(e) If m = 1 in part (d), then Z defined in solution of part (d) is Rayleigh( ) by the
same argument as in the solution of Problem 19(b). Hence, the desired result
follows by taking m = 1 in the result of part (d).
(f) Since the Vi are independent exp(i ), the mgf of Y := V1 + +Vm is
m m
k k
MY (s) = k s = ck k s ,
k=1 k=1

where the ck are the result of expansion by partial fractions. Inverse transform-
ing term by term, we find that
m
fY (y) = ck k ek y , y 0.
k=1

Using this, we can write


m
FY (y) = ck (1 ek y ), y 0.
k=1

Next, put Z := Y , and note that for z 0,

FZ (z) = P( Y z) = P(Y z2 ) = FY (z2 ).

Hence,
m
2
FZ (z) = ck (1 ek z ), z 0.
k=1

We can now write that E[Q(Z)] = E[FXc (Z)] = E[FZ (X)] is equal to
  m 
k X 2
E I[0,) (X) ck (1 e ) .
k=1

ek2 := (1 + 2k )1 ,
Now observe that with
Z x2 /2 Z x2 /2
k X 2 k x2 e 1 k x2 e
E[I[0,) (X)e ] = e dx = e dx
0 2 2 2
Z 2 Z (x/ek )2 /2
1 e(1+2k )x /2 ek e
= dx = dx
2 2 2 2
ek
ek 1
= = p .
2 2 1 + 2k
Putting this all together, the desired result follows.
24. Recall that the noncentral chi-squared density with k degrees of freedom and noncen-
trality parameter 2 is given by
2 /2
( 2 /2)n e
ck, 2 (t) := n!
c2n+k (t), t > 0,
n=0
74 Chapter 5 Problem Solutions

where c2n+k denotes the central chi-squared density with 2n + k degrees of freedom.
Hence,
Z x Z x 2
( 2 /2)n e /2
Ck, 2 (x) =
0
ck, 2 (t) dt =
0 n=0
n!
c2n+k (t) dt

2 /2 Z x 2 /2
( 2 /2)n e ( 2 /2)n e
= n! 0
c2n+k (t) dt = n!
C2n+k (x).
n=0 n=0

25. (a) Begin by writing



FZn (z) = P( Yn z) = P(Yn z2 ) = FYn (z2 ).

Then fZn (z) = fYn (z2 ) 2z. Next observe that



(mz/2)2`+n/21 (mz/2)2`+n/21
In/21 (mz) = `!(` + (n/2) 1 + 1) = .
`=0 `=0 `!(` + n/2)

From Problem 65 in Chapter 4,


2 /2
(m2 /2)` em
fYn (y) := `!
c2`+n (y)
`=0
2 /2
(m2 /2)` em (y/2)(2`+n)/21 ey/2
= `!
21
((2` + n)/2)
.
`=0

So,

2 +z2 )/2 (m2 /2)` (z2 /2)`+n/21
fZn (z) = fYn (z2 ) 2z = 2ze(m 21
`=0 `!(` + n/2)

2 +z2 )/2 (mz)2`+n/21 mn/2+1 (1/2)2`+n/21 zn/21
= ze(m
`=0 `!(` + n/2)
zn/2 (m2 +z2 )
= e In/21 (mz).
mn/21
(b) Obvious.
(c) Begin by writing

FYn (y) = P(Zn2 y) = P(Zn y ) = FZn (y1/2 ).

Then

fYn (y) = fZn (y1/2 ) y1/2 /2


(y1/2 )n/2 (m2 +y)/2
= 1
2 e In/21 (m y )y1/2
mn/21
 y n/21 2
= 12 e(m +y)/2 In/21 (m y ).
m
Chapter 5 Problem Solutions 75

(d) First write


Z
t n/2 (m2 +t 2 )/2
FZcn (z) = e In/21 (mt) dt
z mn/21
Z
(mt)n/2 (m2 +t 2 )/2
= e In/21 (mt) dt.
z mn1
Now apply integration by parts with
2 /2 2 /2
u = (mt)n/21 In/21 (mt) and dv = mtet em /mn1 dt.

Then
2 /2 2 /2
v = mem et /mn1 , and by the hint, du = (mt)n/21 In/22 (mt) m dt.

Thus,

e(m +t )/2
2 2

FZcn (z) = (mt)n/21 In/21 (mt)


mn2 z
Z
(mt)n/21 (m2 +t 2 )/2
+ e I n/22 (mt) dt
z mn3
 z n/21 2 +z2 )/2
= e(m In/21 (mz) + FZn2 (z).
m
(e) Using induction, this is immediate from part (d).
2 /2
e z) = ez
(f) It is easy to see that Q(0, z) = Q(0, . We then turn to

e (m2 +z2 )/2


m
Q(m, z) =
m
e (m/z)k Ik (mz)
k=0
 
(m2 +z2 )/2 k 2k k
= e m(m/z) Ik (mz) + z (mz) Ik1 (mz)z
k=0
 
(m2 +z2 )/2 k k
= e m(m/z) Ik (mz) + (m/z) Ik1 (mz)z
k=0
 
2 2
(m +z )/2 k
= ze (m/z) Ik1 (mz) (m/z)Ik (mz)
k=0

2 +z2 )/2
= ze(m (m/z)k Ik1 (mz) (m/z)k+1 Ik (mz)
k=0
(m2 +z2 )/2 2 +z2 )/2
= ze I1 (mz) = ze(m I1 (mz).

To conclude, we compute
Z
Q 2 +t 2 )/2
= te(m I0 (mt) dt
m m z
Z Z
2 +t 2 )/2 2 +t 2 )/2
= mte(m I0 (mt) dt + te(m I1 (mt) t dt.
z z
76 Chapter 5 Problem Solutions

Write this last integral as


Z
2 +t 2 )/2
(mt)I1 (mt) (t/m)e(m dt.
z

2 2
Now apply integration by parts with u = (mt)I1 (mt) and dv = te(t +m )/2 /m dt.
2 2
Then du = (mt)I0 (mt)m dt and v = e(m +t )/2 /m, and the above integral is
equal to Z
2 +z2 )/2 2 +t 2 )/2
ze(m I1 (mz) + mte(m I0 (mt) dt.
z
2 +z2 )/2
Putting this all together, we find that Q/ m = ze(m I1 (mz).

(x/2)2`+
26. Recall that I (x) := `!(` + + 1) .
`=0

(a) Write

I (x) 1 (x/2)2` 1

= + as x 0.
(x/2) ( + 1) `=1 `!(` + + 1) ( + 1)

Now write

zn/2 (m2 +z2 )/2


fZn (z) = e In/21 (mz)
mn/21
zn/2 2 2 In/21 (mz)
= n/21 e(m +z )/2 (mz/2)n/21
m (mz/2)n/21
zn1 2 2 In/21 (mz)
= n/21 e(m +z )/2 .
2 (mz/2)n/21

Thus,

2 0, n > 1,
em /2 2 p
lim fZn (z) = n/21 lim zn1 = em /2 2/ , n = 1,
z0 2 (n/2) z0
, 0 < n < 1.

(b) First note that



(x/2)2`+ 1 (` + )(x/2)2`+ 1
I 1 (x) = = .
`=0 `!(` + ) `=0 `!(` + + 1)

Second, with the change of index ` = k + 1,



(x/2)2k+ +1 (x/2)2`+ 1
I +1 (x) = k!(k + + 2) = (` 1)!(` + + 1)
k=0 `=1

`(x/2)2`+ 1 `(x/2)2`+ 1
= = `!(` + + 1) .
`=1 `!(` + + 1) `=0
Chapter 5 Problem Solutions 77

It is now easy to see that



(2` + )(x/2)2`+ 1
I 1 (x) + I +1 (x) = `!(` + + 1) = 2I0 (x).
`=0

It is similarly easy to see that



(x/2)2`+ 1
I 1 (x) I +1 (x) = `!(` + + 1) = 2( /x)I (x).
`=0
R
(c) To the integral In (x) := (2 )1 ex cos cos(n ) d , apply integration by parts
with u = ex cos and dv = cos n d . Then du = ex cos (x sin ) d and v =
sin(n )/n. We find that
 Z 
1 x cos sin n x x cos
In (x) = e + e sin n sin d .
2 n n
| {z }
=0

We next use the identity sin A sin B = 12 [cos(A B) cos(A + B)] to get
x
In (x) = [In1 (x) In+1 (x)].
2n
R x cos
(d) Since I0 (x) := (1/ ) 0 e d , make the change of variable t = /2.
Then
Z /2 Z /2
1 1
I0 (x) = ex cos(t+ /2) dt = ex sint dt
/2 /2
Z /2 Z /2
1 (x sint)k 1 (x)k
= dt = sink t dt
/2 k=0 k! k=0 k! /2
| {z }
= 0 for k odd
Z /2 Z
1 x2` 2 x2` /2
= (2`)! sin2` t dt = sin2` t dt
`=0 /2 `=0 (2`)! 0
 
2` + 1

2 x2` 2
= (2`)!
`=0

2` + 2
 , by Problem 18 in Ch. 4.
2
2
Now,
(2`)! = 1 3 5 (2` 1) 2 4 6 2` = 1 3 5 (2` 1) 2` `!,
and from Problem 14(c) in Chapter 4,
 
2` + 1 1 3 5 (2` 1)
= .
2 2`
Hence,

(x/2)2`
I0 (x) = `!(` + 1) =: I0 (x).
`=0
78 Chapter 5 Problem Solutions

(e) Begin by writing


Z
1 x cos
In1 (x) = e cos([n 1] ) d
2
Z
1
= ex cos [cos n cos sin n sin ] d .
2
Then Z
1
1
2 [In1 (x) + In+1 (x)] = ex cos cos n cos d .
2
Since
 Z  Z
1 x cos 1 x cos
In0 (x) = e cos(n ) d = e cos(n ) cos d ,
x 2 2
we see that 12 [In1 (x) + In+1 (x)] = In0 (x).
27. MATLAB.

0.5
0.4
0.3
0.2
0.1
0
0 1 2 3 4

28. See previous problem solution for graph. The probabilities are:

k P(X = k)
0 0.0039
1 0.0469
2 0.2109
3 0.4219
4 0.3164

29. (a) Sketch of f (t):


1

1/2

1/3

1/6

0 1
Chapter 5 Problem Solutions 79

R R
(b) P(X = 0) = {0} f (t) dt = 1/2, and P(X = 1) = {1} f (t) dt = 1/6.
(c) We have
Z 1 Z 1
1 t 1 1
P(0 < X < 1) = f (t) dt = 3 e u(t) + 2 (t) + 6 (t 1) dt
0+ 0+
Z 1 Z 1 1
1 t 1 t 1 e1
= 3 e dt = 3 e dt = 31 et =
0+ 0 0 3
and
Z Z
1 t
P(X > 1) = f (t) dt = 3 e dt = 31 et = e1 /3.
1+ 1 1

(d) Write

P(0 X 1) = P(X = 0) + P(0 < X < 1) + P(X = 1)


1 1 e1 1
= + + = 1 e1 /3
2 3 6
and
1 + 2e1
P(X > 1) = P(X = 1) + P(X > 1) = 1/6 + e1 /3 = .
6
(e) Write
Z Z
t t
E[X] = t f (t) dt = 3 e dt + 0 P(X = 0) + 1 P(X = 1)
0
1
= 3 E[exp RV w/ = 1] + 1/6 = 1/3 + 1/6 = 1/2.

x 1 R
30. For the first part of the problem, we have E[eX ] = e 2 [ (x) + I(0,1] (x)] dx =
R 1
1 0 1 1 x x
2 e + 2 0 e dx = 1/2 + e /2 0 = 1/2 + (e 1)/2 = e/2. For the second part, first
write
P({X = 0} {X 1/2}) P(X = 0)
P(X = 0|X 1/2) = = .
P(X 1/2) P(X 1/2)
Since P(X = 0) = 1/2 and
Z 1/2 Z 1/2
1
P(X 1/2) = 2 [ (x) + I(0,1] (x)] dx = 12 + 21 dx = 21 + 41 = 34 ,
0

1/2
we have P(X = 0|X 1/2) = 3/4 = 2/3.
R
31. The approach is to find the density and then compute E[X] = x fX (x) dx. The catch
is that the cdf has a jump at x = 1/2, and so the density has an impulse there. Put

2x, 0 < x < 1/2,
fX (x) := 1, 1/2 < x < 1,

0, otherwise.
80 Chapter 5 Problem Solutions

Then the density is fX (x) = fX (x) + 41 (x 1/2). Hence,


Z Z 1/2 Z 1
E[X] = x fX (x) dx = x 2x dx + x 1 dx + 41 12
0 1/2

2 1 3

1 2
= 3 2 + 12 [1 2 ]+ 8
1

1 3 1 11 1
= 12 + 8 + 8 = 24 + 8 = 7/12.
R
32. The approach is to find the density and then compute E[ X ] = x fX (x) dx. The
catch is that the cdf has a jump at x = 4, and so the density has an impulse there. Put

x1/2 /8, 0 < x < 4,
fX (x) := 1/20, 4 < x < 9,

0, otherwise.

Then the density is fX (x) = fX (x) + 14 (x 4). To begin, we compute


Z Z 4 Z 9 9

x1/2 fX (x) dx = 1/8 dx + x1/2 /20 dx = 1/2 + x3/2 /30
0 4 4
= 1/2 + (27 8)/30 = 1/2 + 19/30 = 17/15.
The complete answer is

E[ X ] = 17/15 + 4/4 = 17/15 + 1/2 = 34/30 + 15/30 = 49/30.

33. First note that for y < 0,


Z y Z y
e|t| dt = et dt = ey ,

and for y 0,
Z y Z 0 Z y y

e|t| dt = et dt + et dt = 1 + (et ) = 1 + 1 ey = 2 ey .
0 0

Hence,
y
e /4, y < 0,
FY (y) = (2 ey )/4 + 1/3, 0 y < 7,

(2 ey )/4 + 1/3 + 1/6, y 7,
y
e /4, y < 0,
= 5/6 ey /4, 0 y < 7,

1 ey /4, y 7.

1
0.75
0.5
0.25
0
2 0 2 4 6 8
Chapter 5 Problem Solutions 81

34. Let {Y = 0} = {loose connection}, and {Y = 1} = {Y = 0} c . Then P(Y = 0) =


P(Y = 1) = 1/2. Using the law of total probability,

1
FX (x) = P(X x) = P(X x|Y = i)P(Y = i)
i=0
 Z x

1
= 2 P(X x|Y = 0) + I(0,1] (t) dt .

Since P(X = 0|Y = 0) = 1, we see that



1, x 1,
1
FX (x) = (1 + x), 0 x < 1,
2
0, x < 0.

Since there is a jump at x = 0, we must be careful in computing the density. It is


1
fX (x) = 2 [I(0,1) (x) + (x)].

1 1

1/2 1/2

0 1 0 1
cdf density

35. (a) Recall that as x varies from +1 to 1, cos1 x varies from 0 to . Hence,
FX (x) = P(cos x) = P [ , x ] [x , ] , where x := cos1 x. Since
uniform[ , ],

x x ( ) x cos1 x
FX (x) = + = 2 = 1 .
2 2 2

(b) Recall that as y varies from +1 to 1, sin1 y varies from /2 to /2. For
y 0,

FY (y) = P(sin y) = P [ , y ] [ y , ]
y y ( ) + 2y 1 sin1 y
= + = = + ,
2 2 2 2
and for y < 0,

FY (y) = P(sin y) = P( [ y , y ])
2y + 1 sin1 y
= = + .
2 2
82 Chapter 5 Problem Solutions

(c) Write
1 1 1/
fX (x) = = ,
1 x2 1 x2
and
1 1 1/
fY (y) = p = p .
1y2 1 y2

(d) First write


Y + 1 
FZ (z) = P(Z z) = P z = P(Y 2z 1) = FY (2z 1).
2
Then differentiate to get

2 1
fZ (z) = fY (2z 1) 2 = p = p
1 (2z 1)2 z(1 z)
( 12 + 21 )
= z1/21 (1 z)1/21 ,
( 12 )( 21 )

which is the beta density with p = q = 1/2.


36. The cdf is



1, y 3,
Z 1+ 1+y
FY (y) = 1/4 dx, 1 y < 3,


1 1+y
0, y < 1,

1, y 3,
1
= 1 + y, 1 y < 3,
2
0, y < 1,

and the density is


1
, 1 < y < 3,
fY (y) = 4 1+y
0, otherwise.

37. The cdf is



1,
p y 2,


2(3 2/y )
FY (y) = , 1/2 y < 2,
6

1/3, 0 y < 1/2,


0, y < 0,
and the density is

2 3/2 1 1
fY (y) = y I(1/2,2) (y) + (y) + (y 2).
6 3 3
Chapter 5 Problem Solutions 83

38. For 0 y 1, we first compute the cdf


p
FY (y) = P(Y y) = P( 1 R2 y) = P(1 R2 y2 ) = P(1 y2 R2 )
p
p 2 1 y2 1
2
= P( 1 y R) = = 1 (1 y2 )1/2 .
2 2
We then differentiate to get the density

p y , 0 < y < 1,
fY (y) = 2(1 y2 )
0, otherwise.

39. The first thing to note is that for 0 R 2, 0 R2 2. It is then easyto see that
the minimum value of Z = [R2 (1 R2 /4)]1 occurs when R2 = 2 or R = 2. Hence,
the random variable Z takes values in the range [1, ). So, for z 1, we write
   
1 2 2
FZ (z) = P 2 z = P R (1 R /4) 1/z
R (1 R2 /4)
 
= P (R2 /4)(1 R2 /4) 1/(4z) .

Put y := 1/(4z) and observe that x(1 x) y if and only if 0 x2 x + y, with


equality if and only if
1 1 4y
x = .
2
Since we will have x = R2 /4 1/2, we need the negative root. Thus,
   
2 1 1 4y 2
p
FZ (z) = P (R /4) = P R 2[1 1 1/z ]
2
q p
 q  2 2[1 1 1/z ]
p
= P R 2[1 1 1/z ] =
2
q p
= 1 1 1 1/z.
Differentiating, we obtain
1 p 1/2 d p
fZ (z) = 1 1 1/z [1 1 1/z ]
2 dz
1 p 1/2 1 1
= 1 1 1/z (1 1/z)1/2 2
2 2 z
1 p
= 2 [(1 1 1/z )(1 1/z)]1/2 .
4z

40. First note that as R varies from 0 to 2, T varies from to 2 . For t 2 ,
write
   2 
2
FT (t) = P(T t) = P p t = P 2 1 R /4
1 R2 /4 t
84 Chapter 5 Problem Solutions

= P(R2 /4 1 2 /t 2 ) = P(R2 4[1 2 /t 2 ])


 q 
= P R 2 1 2 /t 2 = 2(1 2 /t 2 )1/2 .

Differentiating, we find that



2 2 2 2 1/2 2 2
fT (t) = (1 /t ) = .
t3 t2 t2 2

For the second part of the problem, observe that as R varies between 0 and 2, M
varies between 1 and e . For m in this range, note that ln m < 0, and write

2
FM (m) = P(M m) = P(e (R/2)/ 1R /4 m)
 .q 
= P (R/2) 1 R2 /4 ln m
 . 
= P 2 (R2 /4) (1 R2 /4) ( ln m)2
 
= P (R2 /4) 1/[1 + { /( ln m)}2 ]
 
= P R 2[1 + { /( ln m)}2 ]1/2

= 1 2[1 + { /( ln m)}2 ]1/2 ,

Differentiating, we find that


2 2
2 2 3/2 2
fM (m) = [1 + { /( ln m)} ] = .
m( ln m)3 m[( ln m)2 + 2 ]3/2

41. (a) For X uniform[1, 1],



1, y 1,

FY (y) = (y + y )/2, 0 y < 1,

0, y < 0,

and 
1
fY (y) = 2 [1 + 1/(2 y )], 0 < y < 1,
0, otherwise.
(b) For X uniform[1, 2],


1, y 2,

(y + 1)/3, 1 y < 2,
FY (y) =

(y + y )/3, 0 y < 1,

0, y < 0,

and
1/3, 1<y<2
1
fY (y) = 3 [1 + 1/(2 y )], 0 < y < 1,

0, otherwise.
Chapter 5 Problem Solutions 85

(c) For X uniform[2, 3],




1, y 2,

(y + 2)/5, 1 y < 2,
FY (y) =

(y + y )/5, 0 y < 1,

0, y < 0,
and
1
fY (y) = [I (y) + {1 + 1/(2 y )}I(0,1) (y) + (y 1) + (y 2)].
5 (1,2)
(d) For X exp( ),

1, y 2,
FY (y) = P(X y) + P(X 3), 0 y < 2,

0, y < 0,

1, y 2,
= (1 e y ) + e3 , 0 y < 2,

0, y < 0,
and
fY (y) = e y I(0,2) (y) + e3 (y) + (e2 e3 ) (y 2).
42. (a) If X uniform[1, 1], then Y = g(X) = 0, and so
FY (y) = u(y) (the unit step function), and fY (y) = (y).
(b) If X uniform[2, 2],

1, y 1,
1
FY (y) = 2 (y + 1), 0 y < 1,

0, y < 0,
and
1
fY (y) = [I (y) + (y)].
2 (0,1)
(c) We have
1, y 1,
FY (y) = (y + 1)/3, 0 y < 1,

0, y < 0,
and
1
fY (y) = 3 [I(0,1) (y) + (y) + (y 1)].
(d) If X Laplace( ),

R y+1 1,
y 1,
x
FY (y) = 2 0 2e dx, 0 y < 1,


0, y < 0,

1, y 1,
= 1 e (y+1) , 0 y < 1,

0, y < 0,
86 Chapter 5 Problem Solutions

and
fY (y) = e (y+1) I(0,1) (y) + (1 e ) (y) + e2 (y 1).
43. (a) If X uniform[3, 2],


1, y 1,
1 1/3
FY (y) = 5 (y + y + 2), 0 y < 1,
1

5 [y + 2 (y)1/2 ], 1 y < 0,

0, y < 1,
and
1 2/3
fY (y) = 5 [1 + 1/(3y )]I(0,1) (y) + 15 [1 + 1/(2 y )]I(1,0) (y) + 15 (y 1).

(b) If X uniform[3, 1],




1, y 1,
1 1/3
FY (y) = 4 (y + y + 2), 0 y < 1,
1

4 [y + 2 (y)1/2 ], 1 y < 0,

0, y < 1,
and
1 2/3
fY (y) = 4 [1 + 1/(3y )]I(0,1) (y) + 41 [1 + 1/(2 y )]I(1,0) (y).

(c) If X uniform[1, 1],




1, y 1,
1 1/3
FY (y) = 2 (y + 1), 0 y < 1,
1 1/2 ],

2 [1 (y) 1 y < 0,

0, y < 1,
and
1 1
fY (y) = I(0,1) (y) + I(1,0) (y).
6y2/3 4 y
44. We have
0, y < 1,
1 [y + 1 + y + 1],

1 y < 1,
6
FY (y) = 1

6 [3 + y + 1 ],
1 y < 8,
1, y 8,
and
fY (y) = 1 1
6 (y 1) + [ 6
1
+ 12 (y + 1)1/2 ]I(1,1) (y) + 12
1
(y + 1)1/2 I(1,8) (y).

45. We have
0, y < 0,

FY (y) = (y2 + y + y + 1)/4, 0 y < 1,

1, y 1,
and
1 1
fY (y) = 4 (y) + 4 [2y + 1/(2 y) + 1]I(0,1) (y).
Chapter 5 Problem Solutions 87

46. We have
1, y 1,

1 2
6 [4 + 3y y ], 0 y < 1,
FY (y) = 1 2

6 [3 + 2y y ], 1 y < 0,

0, y < 1,
and
fY (y) = [ 12 13 y]I(0,1) (y) + 31 [1 y]I(1,0) (y) + 61 (y).

47. We have
1,
p y 1,
FY (y) = 1
3 [1 + y + y/(2 y) ], 0 y < 1,
0, y < 0,
and  
 y 1/2 1
1
fY (y) = 3 (y) + 1 + .
2y (2 y)2

48. Observe that g is a periodic sawtooth function with period one. Also note that since
0 g(x) < 1, Y = g(X) takes values in [0, 1).

(a) We begin by writing



FY (y) = P(Y y) = P(g(X) y) = P(k X k+y) = FX (k+y)FX (k).
k=0 k=0

When X exp(1), we obtain, for 0 y < 1,



1 ey
FY (y) = (1 ek ey ) (1 ek ) = ek (1 ey ) =
1 e1
.
k=0 k=0

Differentiating, we get

ey
fY (y) = , 0 y < 1.
1 e1
We say that Y has a truncated exponential density.
(b) When X uniform[0, 1), we obtain Y = X uniform[0, 1).
(c) Suppose X uniform[ , + ), where = m + for some integer m 0 and
some 0 < < 1. For 0 y <

FY (y) = (m + 1 + y) (m + 1) = y,

and for y < 1,

FY (y) = [(m + 1 + ) (m + 1)] + [(m + y) (m + )] = y.

Since FY (y) = y for 0 y < 1, Y uniform[0, 1).


88 Chapter 5 Problem Solutions

49. In the derivation of Property (vii) of cdfs, we started with the formula

[
(, x0 ) = (, x0 n1 ].
n=1

However, we can also write



[
(, x0 ) = (, x0 n1 ).
n=1

Hence,
 
[
G(x0 ) = P(X < x0 ) = P {X < x0 n } = lim P(X < x0 N1 ) = lim G(x0 N1 ).
1
N N
n=1

Thus, G is left continuous. In a similar way, we can adapt the derivation of Prop-
erty (vi) of cdfs to write
 
\
P(X x0 ) = P {X < x0 + 1n } = lim P(X < x0 + 1n ) = lim G(x0 + 1n ) = G(x0 +).
N N
n=1

To conclude, write

P(X = x0 ) = P(X x0 ) P(X < x0 ) = G(x0 +) G(x0 ).

50. First note that since FY (t) is right continuous, so is 1FY (t) = P(Y > t). Next, we use
the assumption P(Y > t + t|T > t) = P(Y > t) to show that with h(t) := P(Y > t),
h(t + t) = h(t) + h(t). To this end, write
P({Y > t + t} {Y > t})
h(t) = ln P(Y > t) = ln P(Y > t + t|Y > t) = ln
P(Y > t)
P(Y > t + t)
= ln = ln P(Y > t + t) ln P(Y > t) = h(t + t) h(t).
P(Y > t)
Rewrite this result as h(t + t) = h(t) + h(t). Then with t = t, we have h(2t) =
2h(t). With t = 2t, we have h(3t) = h(t) + h(2t) = h(t) + 2h(t) = 3h(t). In general,
h(nt) = nh(t). In a similar manner we can show that
t t
h(t) = h ++ = mh(t/m),
m m
and so h(t/m) = h(t)/m. We now have that for rational a = n/m, h(at) = h(n(t/m)) =
nh(t/m) = (n/m)h(t) = ah(t). For general a 0, let ak a with ak rational. Then by
the right continuity of h,

h(at) = lim h(ak t) = lim ak h(t) = ah(t).


k k

We can now write


h(t) = h(t 1) = th(1).
Chapter 5 Problem Solutions 89

Thus,
t h(1) = h(t) = ln P(Y > t) = ln(1 FY (t)),
and we have 1 FY (t) = eh(1)t , which implies Y exp(h(1)). Of couse, h(1) =
ln P(Y > 1) = ln[1 FY (1)].
51. We begin with
  
1 n Xi m 1 n 
E[Yn ] = E = E[Xi ] m = 0.
n i=1 n i=1

For the variance, we use the fact that since independent random variables are uncor-
related, the variance of the sum is the sum of the variances. Thus,
n   n n
Xi m var(Xi ) 2
var(Yn ) = var = 2
= 2
= 1.
i=1 n i=1 n i=1 n

52. Let Xi denote the time to transmit the ith packet, where Xi has mean m and variance
2 . The total time to transmit n packets is Tn := X1 + + Xn . The expected total
time is E[Tn ] = nm. Since we do not know the distribution of the Xi , we cannot know
the distribution of Tn . However, we use the central limit theorem to approximate
P(Tn > 2nm). Note that the sample mean Mn = Tn /n. Write

P(Tn > 2nm) = P( 1n Tn > 2m) = P(Mn > 2m) = P(Mn m > m)
   
Mn m m m
= P > = P Yn >
/ n / n / n

= 1 FYn (m n/ ) 1 (m n/ ),

by the central limit theorem.


53. Let Xi = 1 if bit i is in error, and Xi = 0 otherwise. Then P(Xi = 1) = p. Although
the problem does not say so, let us assume that the Xi are independent. Then Mn =
1 n
n i=1 Xi is the fraction of bits in error. We cannot reliably decode of Mn > t. To
approximate the probability that we cannot reliably decode, write
     
Mn m t m t m t m
P(Mn > t) = P > = P Yn > = 1 FYn
/ n / n / n / n
   
t m tp
1 = 1 p ,
/ n p(1 p)/n

since m = E[Xi ] = p and 2 = var(Xi ) = p(1 p).


54. If the Xi are i.i.d. Poisson(1), then Tn := X1 + + Xn is Poisson(n). Thus,
   
nk en 1 1 kn1 2 1
P(Tn = k) = exp .
k! 2 2 1 n 1 n

Taking k = n, we obtain nn en n!/ 2 n or n! 2 nn+1/2 en .
90 Chapter 5 Problem Solutions

55. Recall that gn is the density of X1 + + Xn . If the Xi are i.i.d. uniform[1, 1], then
gn is the convolution of (1/2)I[1,1] (x) with itself n times. From graphical considera-
tions, it is clear that gn (x) = 0 for |x| > n; i.e., xmax = n.
56. To begin, write
 n  n h i
 
Yn ( ) = E e j (X1 ++Xn )/ n = E e j( / n )Xi
= 1 j / n
2 e + 1 j / n
2 e
i=1 i=1
 2 /2 n 2
= cosn ( / n ) 1 e /2 .
n

57. (a) MTTF = E[T ] = n (from Erlang in table).


(b) The desired probability is
Z n1
e
R(t) := P(T > t) = d .
t (n 1)!
R
Let Pn (t) denote the above integral. Then P1 (t) = t e d = e . For n > 1,
apply integration by parts with u = n1 /(n 1)! and dv = e d . Then

t n1 et
Pn (t) = + Pn1 (t).
(n 1)!

Applying this result recursively, we find that

t n1 et t n2 et
Pn (t) = + + + et ,
(n 1)! (n 2)!

which is the desired result.


(c) The failure rate is

fT (t) t n1 et /(n 1)! t n1


r(t) = = n1
= n1 k
.
R(t) t k t t
e (n 1)!
k=0 k! k=0 k!

For n = 2, r(t) = t/(1 + t):


1

0
0 2 4 6 8 10


58. (a) Let =1. For p = 1/2, r(t) = 1/(2 2). For p = 1, r(t) = 1. For p = 3/2,
r(t) = 3 t/2. For p = 2, r(t) = 2t. For p = 3, r(t) = 3t 2 .
Chapter 5 Problem Solutions 91

p=3
2 p=2
p = 3/2
1 p=1
p = 1/2
0
0 1
(b) We have from the text that
 Zt   Zt 
p
R(t) = exp r( ) d = exp p p1
d = e t .
0 0

(c) The MTTF is Z Z


p
E[T ] = R(t) dt = e t dt.
0 0

Now make the change of variable = t p , or t = ( / )1/p . Then


Z  1/p1 d Z
1
E[T ] = e = 1/p1 e d
0 p p 1/p 0
1 (1/p + 1)
= 1/p
(1/p) = .
p 1/p
(d) Using the result of part (b),
p
fT (t) = R0 (t) = t p1 e t , t > 0.

59. (a) Write


 Z
  Zt 
1
R(t) = exp r( ) d = exp p d
0 t0
= exp[p(lnt lnt0 )] = exp[ln(t0 /t) p ] = (t0 /t) p , t t0 .

(b) If t0 = 1 and p = 2, R(t) = 1/t 2 has the form

0
0 1 2 3 4 5
(c) For p > 1, the MTTF is
Z Z  t 1p 
t0
E[T ] = R(t) dt = (t0 /t) p dt = t0p = .
0 t0 1 p t0 p1
92 Chapter 5 Problem Solutions

(d) For the density, write

p  t0  p
p
pt0
fT (t) = r(t)R(t) = = p+1 , t t0 .
t t t

60. (a) A sketch of r(t) = t 2 2t + 2 for t 0 is:

6
4
2
0
0 1 2 3
(b) We first compute
Z t Z t
r( ) d = 2 2 + 2 d = 1 3 2
3 t t + 2t.
0 0

Then
Rt 1 3 t 2 +2t)
fT (t) = r(t)e 0 r( ) d = [t 2 2t + 2]e( 3 t , t 0.

61. (a) If T uniform[1, 2], then for 0 t < 1, R(t) = P(T > t) = 1, and for t 2,
R(t) = P(T > t) = 0. For 1 t < 2,
Z 2
R(t) = P(T > t) = 1 d = 2 t.
t

The complete formula and sketch are



1, 0 t < 1,
R(t) = 2 t, 1 t < 2,

0, t 2.

0
0 1 2 3
(b) The failure rate is
d d 1
r(t) = ln R(t) = ln(2 t) = , 1 < t < 2.
dt dt 2t
Chapter 5 Problem Solutions 93

(c) Since T uniform[1, 2], the MTTF is E[T ] = 1.5.


62. Write

R(t) := P(T > t) = P(T1 > t, T2 > t) = P(T1 > t)P(T2 > t) = R1 (t)R2 (t).

63. Write

R(t) := P(T > t) = P({T1 > t} {T2 > t}) = 1 P(T1 t, T2 t)


= 1 P(T1 t)P(T2 t) = 1 [1 R1 (t)][1 R2 (t)]
= 1 [1 R1 (t) R2 (t) + R1 (t)R2 (t)]
= R1 (t) + R2 (t) R1 (t)R2 (t).

64. We follow the hint and write


Z Z Z
E[Y n ] = E[T ] = P(T > t) dt = P(Y n > t) dt = P(Y > t 1/n ) dt.
0 0 0

We then make the change of variable y = t 1/n , or t = yn , dt = nyn1 dy, to get


Z
E[Y n ] = P(Y > y) nyn1 dy.
0
CHAPTER 6
Problem Solutions

1. Since the Xi are uncorrelated with common mean m and common variance 2 ,
  n  
1
E[Sn2 ] = E Xi2 nE[Mn ]
n1 i=1
1 h n oi
= n( 2 + m2 ) n var(Mn ) + (E[Mn ])2
n1
1  
= n( 2 + m2 ) n{ 2 /n + m2 }
n1
1  
= (n 1) 2 + nm2 nm2 = 2 .
n1
p
2. (a) The mean of a Rayleigh( ) random variable is /2. Consider
p
n := Mn / /2.

Then
p p p p
E[n ] = E[Mn / /2 ] = E[Mn ]/ /2 = /2 /2 = .

Thus, n is unbiased. Next,


p p p p
n = Mn / /2 E[Mn ] /2 = /2 /2 = ,

and we see that n is strongly consistent.


(b) MATLAB. Add the line of code lambdan=mean(X)/sqrt(pi/2).
p
(c) MATLAB. Since Mn /2, we solve for and put

n := 2(Mn / )2 .

Since = 3, add the line of code pin=2*(mean(X)/3) 2.


3. (a) The mean of a gamma(p, ) random variable is p/ . We put

pn := Mn .

Then E[pn ] = E[Mn ] = p/ = p. Also pn = Mn (p/ ) = p. Thus, pn


is unbiased and strongly consistent.
(b) MATLAB. In this problem = 1/2 and p = k/2, or k = 2p. We use kn := 2pn =
2( Mn ) = 2((1/2)Mn ) = Mn . We therefore add the line of code kn=mean(X).

94
Chapter 6 Problem Solutions 95

4. (a) Since the mean of a noncentral chi-squared random variable with k degrees of
freedom and noncentrality parameter 2 is k + 2 , we put

n2 := Mn k.

Then E[n2 ] = E[Mn k] = E[Mn ] k = (k + 2 ) k = 2 , and we see that


n2 is an unbiased estimator of 2 . Next, since n2 = Mn k E[Mn ] k =
(k + 2 ) k = 2 , the estimator is strongly consistent.
(b) MATLAB. Since k = 5, add the line of code lambda2n=mean(X)-5.
5. (a) Since the mean of a gamma(p, ) random variable is p/ , we put n := p/Mn .
Then n = p/Mn p/E[Mn ] = p/(p/ ) = , and we see that n is a strongly
consistent estimator of .
(b) MATLAB. Since p = 3, add the line of code lambdan=3/mean(X).
6. (a) Since the variance of a Laplace( ) random variable is 2/ 2 , we put
q
n := 2/Sn2 .
p
Since Sn2 converges to the variance, we have n 2/(2/ 2 ) = , and we see
that n is a strongly consistent estimator of .
(b) MATLAB. Add the line of code lambdan=sqrt(2/var(X)).
7. (a) The mean of a gamma(p, ) random variable is p/ . The second moment is
p(p + 1)/ 2 . Hence, the variance is

p(p + 1) p2 p
2
2 = 2.

Thus, Mn p/ and Sn2 p/ 2 . Solving for p and suggests that we put

Mn
n := and pn := n Mn .
Sn2

Now, Mn p/ and Sn2 p/ 2 . It follows that n (p/ ) (p/ 2 ) = and
then pn (p/ ) = p. Hence, n is a strongly consistent estimator of , and
pn is a strongly consistent estimator of p.
(b) MATLAB. Add the code
Mn = mean(X)
lambdan = Mn/var(X)
pn = lambdan*Mn

8. Using results from the problem referenced in the hint, we have



(q + p)/q (1 + p/q) p/q
E[X q ] = = = .
(p/q) q (p/q) q q
96 Chapter 6 Problem Solutions

This suggests that we put


 1/q
p/q
n := q .
1
n ni=1 Xi

Then  1/q
p/q
n = ,
(p/q)/ q
and we see that n is a strongly consistent estimator of .
9. In the preceding problem E[X q ] = (p/q)/ q . Now consider

2q (2q + p)/q (2 + p/q) (1 + p/q)(p/q)
E[X ] = 2q
= 2q
= .
(p/q) (p/q) 2q

In this equation, replace q by (p/q)/E[X q ] and solve for (p/q). Thus,

(E[X q ])2
p/q = .
var(X q )

This suggests that we first put

1 n q
Xnq := Xi
n i=1

and then
(Xnq )2
pn := q h  i
q q
1
n1 ni=1 (Xi )2 n(Xn )2

and 1/q
 1/q
pn /q Xnq
n := = h  i .
Xnq 1
n1 n
(X
i=1 i
q 2
) n(Xn
q 2
)

10. MATLAB. OMITTED.


11. MATLAB. The required script can be created using the code from Problem 2 followed
by the lines

global lambdan
lambdan = mean(X)/sqrt(pi/2)

followed by the script from Problem 10 modifed as follows: The chi-squared statistic
Z can be computed by inserting the lines

p = CDF(b) - CDF(a);
Z = sum((H-n*p).2./(n*p))
Chapter 6 Problem Solutions 97

after the creation of the right edge sequence in the script given in Problem 10, where
CDF is the function
function y = CDF(t) % Rayleigh CDF
global lambdan
y = zeros(size(t));
i = find(t>0);
y(i) = 1-exp(-(t(i)/lambdan).2/2);

In addition, the line defining y in the script from Problem 10 should be changed to
y=PDF(t), where PDF is the function
function y = PDF(t) % Rayleigh density
global lambdan
y = zeros(size(t));
i = find(t>0);
y(i) = (t(i)/lambdan2).*exp(-(t(i)/lambdan).2/2);

Finally, the chi-squared statistic Z should be compared with z = 22.362, since =


0.05 and since there are m = 15 bins and r = 1 estimated parameter, the degrees of
freedom parameter is k = m 1 r = 15 1 1 = 13 in the chi-squared table in the
text.
12. MATLAB. Similar to the solution of Problem 11 except that it is easier to use the
M ATLAB function chi2cdf or gamcdf to compute the required cdfs for evaluating
the chi-squared statistic Z. For the same reasons as in Problem 11, z = 22.362.
13. MATLAB. Similar to the solution of Problem 11 except that it is easier to use the
M ATLAB function ncx2cdf to compute the required cdfs for evaluating the chi-
squared statistic Z. For the same reasons as in Problem 11, z = 22.362.
14. MATLAB. Similar to the solution of Problem 11 except that it is easier to use the M AT-
LAB function gamcdf to compute the required cdfs for evaluating the chi-squared
statistic Z. For the same reasons as in Problem 11, z = 22.362.
15. MATLAB. Similar to the solution of Problem 11. For the same reasons as in Prob-
lem 11, z = 22.362.
16. Since
 n  n n
E[H j ] = E [e j ,e j+1 ) i =
I (X ) P(e j Xi < e j+1 ) = pj = np j ,
i=1 i=1 i=1
 
H j np j
E = 0.
np j
Since the Xi are i.i.d., the I[e j ,e j+1 ) (Xi ) are i.i.d. Bernoulli(p j ). Hence,
n 
E[(H j np j )2 ] = var(H j ) = var I[e j ,e j+1 ) (Xi )
i=1
n  
= var I[e j ,e j+1 ) (Xi ) = n p j (1 p j ),
i=1
98 Chapter 6 Problem Solutions

and so  2 
H j np j
E = 1 p j.
np j

17. If f is an even density, then


Z x Z x Z
F(x) = f (t) dt = f ( ) d = f ( ) d = 1 F(x).
x


18. The width of any confidence interval is w = 2 y/ n. If = 2 and n = 100,
2 2 2.576
w99% = = 1.03.
10
To make w99% < 1/4 requires
2 y
< 1/4 or n > (8 y)2 = (16 2.576)2 = 1699.
n

2
i ] = m, and var(Xi ) = var(Wi ) = 4. So, = 4.
19. First observe that with Xi = m+Wi , E[X
For 95% confidence interval, y /2 / n = 2 1.960/10 = 0.392, and so

m = 14.846 0.392 with 95% probability.

The corresponding confidence interval is [14.454, 15.238].


20. Write
 
1 n
P(|Mn m| ) = P( Mn m ) = P (m +Wi ) m
n i=1
 n   n 
1
= P Wi = P n Wi n
n i=1 i=1
| {z }
Cauchy(n)
2
= tan1 (n /n),


which is equal to 2/3 if and only if tan1 ( ) = /3, or = 3.
21. MATLAB. OMITTED.

22. We use the formula m = Mn y /2 Sn / n = 10.083 (1.960)(0.568)/10 to get

m = 10.083 0.111 with 95% probability,

and the confidence intervale is [9.972, 10.194].



23. We use the formula m = Mn y /2 Sn / n = 4.422 (1.812)(0.957)/10 to get

m = 4.422 0.173 with 93% probability,

and the confidence interval is [4.249, 4.595].


Chapter 6 Problem Solutions 99

24. We haveMn = number defective/n = 10/100 = 0.1. We use the formula m = Mn


y /2 Sn / n = 0.1 (1.645)(.302)/10 to get

m = 0.1 0.0497 with 90% probability.

The number of defectives is 10 000m, or

number of defectives = 1000 497 with 90% probability.

Thus, we are 90% sure that the number of defectives is between 503 and 1497 out of
a total of 10 000 units.

25. We have Mn = 1559/3000. We use the formula m = Mn y /2 Sn / n = 0.520

(1.645)(.5)/ 3000 to get

m = 0.520 0.015 with 90% probability,

and the confidence interval is [0.505, 0.535]. Hence, the probability is at least 90%
that more than 50.5% of the voters will vote for candiate A. So we are 90% sure that
candidate A will win. The 99% confidence interval is given by

m = 0.520 0.024 with 99% probability,

and the confidence interval is [0.496, 0.544]. Hence, we are not 99% sure that candi-
date A will win.
26. We have Mn = number defective/n = 48/500
= 0.0960. We use the formula m =
Mn y /2 Sn / n = 0.0960 (1.881)(.295)/ 500 to get

m = 0.0960 0.02482 with 94% probability.

The number of defectives is 100 000m, or

number of defectives = 9600 2482 with 94% probability.

Thus, we are 94% sure that the number of defectives is between 7118 and 12 082 out
of a total of 100 000 units.

27. (a) We have Mn = 6/100. We use the formula m = Mn y /2 Sn / n = 0.06
(2.170)(.239)/10 to get

m = 0.06 0.0519 with 97% probability.

We are thus 97% sure that p = m lies in the interval [0.0081, 0.1119]. Thus, we
are not 97% sure that p < 0.1.

(b) We have Mn = 71/1000. We use the formula m = Mn y /2 Sn / n = 0.071

(2.170)(.257)/ 1000 to get

m = 0.071 0.018 with 97% probability.

We are thus 97% sure that p = m lies in the interval [0.053, 0.089]. Thus, we are
97% sure that p < 0.089 < 0.1.
100 Chapter 6 Problem Solutions

28. (a) Let Ti denote the time to transmit the ith packet. Then we need to compute
 n   n 
[ \
P {Ti > t} = 1 P {Ti t} = 1 FT1 (t)n = 1 (1 et/ )n .
i=1 i=1

(b) Using the notation from part (a), T = T1 + + Tn . Since the Ti are i.i.d.
exp(1/ ), T is Erlang(n, 1/ ) by Problem 55(c) in Chapter 4 and the remark
following it. Hence,
(t/ )n1 et/
fT (t) = (1/ ) , t 0.
(n 1)!
(c) We have
y /2 Sn 1.960(1.798)
= Mn = 1.994 = 1.994 0.352
n 10
and confidence interval [1.642, 2.346] with 95% probability.
29. MATLAB. OMITTED.
30. By the hint,
 ni=1 Xi is Gaussian with mean nm and variance n 2 . Since Mn =
n
i=1 Xi /n, it is easy to see that Mn is still Gaussian, and its mean is (nm)/n = m.
Its variance is (n 2 )/n2 = 2 /n. Next, Mn m remains
p Gaussian but with mean zero
and the same variance 2 /n. Finally, (M p n m)/ /n remains Gaussian and with
mean zero, but its variance is ( 2 /n)/( 2 /n )2 = 1.

31. We use the fomula m = Mn y /2 Sn / n, where in this Gaussian case, y /2 is taken
from the tables using Students t distribution with n = 10. Thus,
y /2 Sn 2.262 1.904
m = Mn = 14.832 = 14.832 1.362,
n 10
and the confidence interval is [13.470, 16.194] with 95% probability.
32. We use [nVn2 /u, nVn2 /`], where u and ` are chosen from the appropriate table. For a
95% confidence interval, ` = 74.222 and u = 129.561. Thus,
 2   
nVn nVn2 100(4.413) 100(4.413)
, = , = [3.406, 5.946].
u ` 129.561 74.222

33. We use [(n 1)Sn2 /u, (n 1)Sn2 /`], where u and ` are chosen from the appropriate
table. For a 95% confidence interval, ` = 73.361 and u = 128.422. Thus,
   
(n 1)Sn2 (n 1)Sn2 99(4.736) 99(4.736)
, = , = [3.651, 6.391].
u ` 128.422 73.361

34. For the two-sided test at the 0.05 significance level, we compare |Zn | with y /2 =
1.960. Since |Zn | = 1.8 1.960 = y /2 , we accept the null hypothesis. For the
one-sided test of m > m0 at the 0.05 significance level, we compare Zn with y =
1.645. Since it is not the case that Zn = 1.80 > 1.645 = y , we do not accept
the null hypothesis.
Chapter 6 Problem Solutions 101

35. Suppose (y) = . Then by Problem 17, (y) = 1 (y), and so 1 (y) = ,
or (y) = 1 .
36. (a) Since Zn = 1.50 y = 1.555, the Internet service provider accepts the null
hypothesis.
(b) Since Zn = 1.50 > 1.555 = y , we accept the null hypothesis; i.e., we reject
the claim of the Internet service provider.
37. The computer vendor would take the null hypothesis to be m m0 . To give the vendor
the benefit of the doubt, the consumer group uses m m0 as the null hypothesis. To
accept the null hypothesis would require Zn y . Only by using the sigificance level
of 0.10, which has y = 1.282, can the consumer group give the benefit of the doubt
to the vendor and still reject the vendors claim.
38. Giving itself the benfit of the doubt, the company uses the null hypothesis m > m0 and
uses a 0.05 significance level. The null hypothesis will be accepted if Zn > y =
1.645. Since Zn = 1.6 > 1.645, the company believes it has justified its claim.
39. Write
n n
g) =
e(b |Yk (baxk + bb)|2 = |Yk (baxk + [Y abx])|2
k=1 k=1
n
= |(Yk Y ) + ab(xk x)|2
k=1
n h i
= (Yk Y )2 2b
a(xk x)(Yk Y ) + ab2 (xk x)2
k=1
aSxY + ab2 Sxx
= SYY 2b
S 
xY 2
= SYY 2b
aSxY + ab Sxx = SYY abSxY = SYY SxY /Sxx .
Sxx

40. Write

E[Y |X = x] = E[g(X) +W |X = x] = E[g(x) +W |X = x]


= g(x) + E[W |X = x] = g(x) + E[W ] = g(x).

41. MATLAB. OMITTED.


42. MATLAB. OMITTED.
43. MATLAB. If z = c/t q , then ln z = ln c q lnt. If y = ln z and x = lnt, then y = (q)x +
ln c. If y a(1)x + a(2), then q = a(1) and c = exp(a(2)). Hence, the two lines of
code that we need are

qhat = -a(1)
chat = exp(a(2))

44. Obvious.
102 Chapter 6 Problem Solutions

45. Write
ez /2 . s2 /2
2 2
e(zs) /2
fZe (z) = esz fZ (z)/MZ (s) = esz e = .
2 2
e = t, put s = t. Then Ze (t, 1).
To make E[Z]
46. Write

( z) p1 e z .  p
fZe (z) = esz fZ (z)/MZ (s) = esz , z > 0.
(p) s

Then
 p Z ( z) p1 ez( s) Z
( s) p z p1 ez( s)
e =
E[Z] z dz = z dz,
s 0 (p) 0 (p)

e = p/( s). To make


which is the mean of a gamma(p, s) density. Hence, E[Z]
e
E[Z] = t, we need p/( s) = t or s = p/t.
47. MATLAB. OMITTED.
48. First, .
pZe (zi ) = eszi pZ (zi ) [(1 p) + pes ].

Then

pZe (1) = es p/[(1 p) + pes ] and pZe (0) = (1 p)/[(1 p) + pes ].


CHAPTER 7
Problem Solutions

1. We have
FZ (z) = P(Z z) = P(Y X z) = P((X,Y ) Az ),
where
Az := {(x, y) : y x z} = {(x, y) : y x + z}.

z
z
x

2. We have
 h i
FZ (z) = P(Z z) = P(Y /X z) = P {Y /X z} {X < 0} {X > 0}
= P((X,Y ) D +
z Dz ) = P((X,Y ) Az ),

where
Az := D +
z Dz ,

and
D
z := {(x, y) : y/x z and x < 0} = {(x, y) : y zx and x < 0},

and
D+
z := {(x, y) : y/x z and x > 0} = {(x, y) : y zx and x > 0}.
z

y
pe
lo
fs
eo
lin

+
Dz Dz x

3. (a) R := (a, b] (c, d] is

103
104 Chapter 7 Problem Solutions

a b

(b) A := (, a] (, d] is
d

a b

(c) B := (, b] (, c] is

a b

(d) C := (a, b] (, c] is

a b

(e) D := (, a] (c, d] is
d

a b
Chapter 7 Problem Solutions 105

(f) A B is

a b

4. Following the hint and then observing that R and A B are disjoint, we have
  
P (X,Y ) (, b] (, d] = P (X,Y ) R + P (X,Y ) A B . ()

Next, by the inclusionexclusion formula,


   
P (X,Y ) A B = P (X,Y ) A + P (X,Y ) B P (X,Y ) A B
= FXY (a, d) + FXY (b, c) FXY (a, c).

Hence, () becomes

FXY (b, d) = P (X,Y ) R + FXY (a, d) + FXY (b, c) FXY (a, c),

which is easily rearranged to get the rectangle formula,



P (X,Y ) R = FXY (b, d) FXY (a, d) FXY (b, c) + FXY (a, c).

5. (a) {(x, y) : |x| y 1} is NOT a product set.


(b) {(x, y) : 2 < x 4, 1 y < 2} = (2, 4] [1, 2).
(c) {(x, y) : 2 < x 4, y = 1} = (2, 4] {1}.
(d) {(x, y) : 2 < x 4} = (2, 4] IR.
(e) {(x, y) : y = 1} = IR {1}.
(f) {(1, 1), (2, 1), (3, 1)} = {1, 2, 3} {1}.
(g) The union of {(1, 3), (2, 3), (3, 3)} and the set in (f) is equal to {1, 2, 3} {1, 3}.
(h) {(1, 0), (2, 0), (3, 0), (0, 1), (1, 1), (2, 1), (3, 1)} is NOT a product set.
6. We have
(
1, x 2, y e2y
1 e , y 0,
FX (x) = x 1, 1 x < 2, and FY (y) = y
0, y < 0,
0, x < 1,

where the quotient involving division by y is understood as taking its limiting value
of one when y = 0. Since FX (x)FY (y) 6= FXY (x, y) when 1 x 2 and y > 0, X and Y
are NOT independent.
106 Chapter 7 Problem Solutions

7. We have

1, x 3,  1 2y 5e3y ],
FX (x) = 2/7, 2 x < 3, and FY (y) = 7 [7 2e y 0,
0, y < 0.
0, x < 2,

8. Let y > 0. First compute


 
y + ex(y+1) (y + 1){1 + ex(y+1) (x)} {y + ex(y+1) }(1)
=
y y+1 (y + 1)2
(y + 1) x(y + 1)ex(y+1) y ex(y+1)
=
(y + 1)2
1 ex(y+1) {1 + x(y + 1)}
= .
(y + 1)2
Then compute
  
y + ex(y+1) {1 + x(y + 1)}ex(y+1) (y + 1) ex(y+1) (y + 1)
=
x y y+1 (y + 1)2
xex(y+1) (y + 1)2
= = xex(y+1) , x, y > 0.
(y + 1)2

9. The first step is to recognize that fXY (x, y) factors into


2
exp[|y x| x2 /2] ex /2 e|yx|
fXY (x, y) = = .
2 2 2 2
When integrating this last factor with respect to y, make the change of variable =
y x to get
Z 2 Z 2 Z
ex /2 1 |yx| ex /2 1 | |
fX (x) = fXY (x, y) dy = 2e dy = 2e d
2 2
2 Z 2
ex /2 ex /2
= e d = .
2 0 2
Thus, X N(0, 1).
10. The first step is to factor fXY (x, y) as
2 2
4e(xy) /2 4 e(xy) /2
fXY (x, y) = = 5 .
5
y 2 y 2
Regarding this last factor a function of x, it is an N(y, 1) density. In other words, when
integrated with respect to x, the result is one. In symbols,
Z Z (xy)2 /2
4 e 4
fY (y) = fXY (x, y) dx = dx = , y 1.
y5 2 y5
Chapter 7 Problem Solutions 107

11. We first analyze U := max(X,Y ). Then

FU (u) = P(max(X,Y ) u) = P(X u and Y u) = FXY (u, u),

and the density is



FXY (x, y) FXY (x, y)
fU (u) = + .
x x=u, y=u y x=u, y=u

If X and Y are independent, then

FU (u) = FX (u)FY (u), and fU (u) = fX (u)FY (u) + FX (u) fY (u).

If in addition X and Y have the same density, say f , (and therefore the same cdf, say
F), then
FU (u) = F(u)2 , and fU (u) = 2F(u) f (u).
We next analyze V := min(X,Y ). Using the inclusionexclusion formula,

FV (v) = P(min(X,Y ) v) = P(X v or Y v)


= P(X v) + P(Y v) P(X v and Y v)
= FX (v) + FY (v) FXY (v, v).

The density is

FXY (x, y) FXY (x, y)
fV (v) = fX (v) + fY (v) .
x x=v, y=v y x=v, y=v

If X and Y are independent, then

FV (v) = FX (v) + FY (v) FX (v)FY (v),

and
fV (v) = fX (v) + fY (v) fX (v)FY (v) FX (v) fY (v).
If in addition X and Y have the same density f and cdf F, then

FV (v) = 2F(v) F(v)2 , and fV (v) = 2[ f (v) F(v) f (v)].

12. Since X gamma(p, 1) and Y gamma(q, 1) are independent, we have from Prob-
lem 55(c) in Chapter 4 that Z gamma(p + q, 1). Since p = q = 1/2, we further have
Z gamma(1, 1) = exp(1). Hence, P(Z > 1) = e1 .
13. We have from Problem 55(b) in Chapter 4 that Z Cauchy( + ). Since = =
1/2, Z Cauchy(1). Thus,

1 1 1 1 3
P(Z 1) = tan1 (1) + = + = .
2 4 2 4
108 Chapter 7 Problem Solutions

14. First write Z  Z zy 


FZ (z) = fXY (x, y) dx dy.

It then follows that


Z Z Z
zy
fZ (z) = FZ (z) = fXY (x, y) dx dy = fXY (z y, y) dy.
z z

15. First write

FZ (z) = P((X,Y ) Az ) = P((X,Y ) B+


z Bz )
ZZ ZZ
= fXY (x, y) dx dy + fXY (x, y) dx dy
B+
z B
z
Z  Z z/y  Z 0
Z

= fXY (x, y) dx dy + fXY (x, y) dx dy.
0 z/y

Then
Z Z 0

fZ (z) = FZ (z) = fXY (z/y, y)/y dy fXY (z/y, y)/y dy
z 0
Z Z 0
= fXY (z/y, y)/|y| dy + fXY (z/y, y)/|y| dy
0
Z
= fXY (z/y, y)/|y| dy.

16. For the cdf, write Z  Z z+x 


FZ (z) = fXY (x, y) dy dx.

Then
Z Z Z
z+x
fZ (z) = FZ (z) = fXY (x, y) dy dx = fXY (x, z + x) dx.
z z

17. For the cdf, write

FZ (z) = P(Z z) = P(Y /X z)


= P(Y Xz, X > 0) + P(Y Xz, X < 0)
Z  Z xz  Z 0 Z 
= fXY (x, y) dy dx + fXY (x, y) dy dx.
0 xz

Then
Z Z 0 Z

fZ (z) = FZ (z) = fXY (x, xz)x dx fXY (x, xz)x dx = fXY (x, xz)|x| dx.
z 0

18. (a) The region D:


Chapter 7 Problem Solutions 109

1
D
0

1
1 0 1
(b) Since fXY (x, y) = Kxn ym for (x, y) D and since D contains negative values of
x, we must have n even in order that the density be nonnegative. In this case,
the integral of the density over the region D must be one. Hence, K must be
such that
ZZ Z 1 Z 1  Z 1  m+1 1
n m n y
1 = fXY (x, y) dy dx = Kx y dy dx = Kx dx
1 |x| 1 m + 1 |x|
D
Z 1 Z 1 
Kxn K
= [1 |x|m+1 ] dx = xn |x|n+m+1 dx
1 m + 1 m + 1 1
Z 1  
2K 2K 1 1
= xn xn+m+1 dx =
m+1 0 m+1 n+1 n+m+2
2K
= .
(n + 1)(n + m + 2)

Hence, K = (n + 1)(n + m + 2)/2.


(c) A sketch of Az with z = 0.3:

1 A
z

1 Az
1 0 1
(d) A sketch of Az D with z = 0.3:

1
1 0 1
(e)
Z z  Z 1  Z 1 Z 1 
P((X,Y ) Az ) = Kxn ym dy dx + Kxn ym dy dx
z z/x z x
110 Chapter 7 Problem Solutions

Z z Z 1
 Z 1 Z 1 
= Kxn ym dy dx + Kxn ym dy dx
z z/x z x
Z
z Z 1 
K
= x [1 (z/x)m+1 ] dx +
n
xn [1 xm+1 ] dx
m+1 z z
Z
z Z 1 
K n m+1 nm1 n n+m+1
= x z x dx + x x dx
m+1 z z
Z 1 Z
z Z 1 
K n m+1 nm1 n+m+1
= x dx z x dx x dx
m+1 z z z
 Z 
K 1 zn z 1 z(n+m+2)/2
= zm+1 xnm1 dx .
m+1 n+1 z n+m+2

If n 6= m, the remaining integral is equal to



( z )n+m+2 zn+1
.
nm
Otherwise, the integral is equal to

zm+1
ln z.
2

19. Let X uniform[0, w] and Y uniform[0, h]. We need to compute P(XY wh).
Before proceeding, we make a few observations. First, since X 0, we can write for
z > 0,
Z Z Z Z
P(XY z) = fXY (x, y) dy dx = fX (x) fY (y) dy dx.
0 z/x 0 z/x

Since Y uniform[0, h], the inner integral will be zero if z/x > h. Since z/x h if
and only if x z/h,
Z Z h Z 
1 z
P(XY z) = fX (x) dy dx = fX (x) 1 dx.
z/h z/x h z/h xh

We can now write


Z  Z
wh   w
P(XY wh) = fX (x) 1 dx = fX (x) 1 dx
wh/h xh w x
Z w 
1 w w
= 1 dx = (1 ) ln = (1 ) + ln .
w w x w

20. We first compute


Z
1 1
E[XY ] = E[cos sin ] = 2 E[sin 2] = sin 2 d
4
cos(2 ) cos(2 )
= = 0.
8
Chapter 7 Problem Solutions 111

Similarly,
Z
1 sin( ) sin( ) 00
E[X] = E[cos ] = cos d = = = 0,
2 2 2
and
Z
1 cos( ) cos( ) (1) (1)
E[Y ] = E[sin ] = sin d = = = 0.
2 2 2
To prove that X and Y are not independent, we argue by contradiction. However,
before we begin, observe that since (X,Y ) satisfies

X 2 +Y 2 = cos2 + sin2 = 1,

(X,Y ) always lies on the unit circle. Now consider the square of side one centered at
the origin,
S := {(x, y) : |x| 1/2, and |y| 1/2}.
Since this region lies strictly inside the unit circle, P((X,Y ) S) = 0. Now, to obtain
a contradiction suppose that X and Y are independent. Then fXY (x, y) = fX (x) fY (y),
where fX and fY are both arcsine densities by Problem 35 in Chapter 5. Hence, for
|x| < 1 and |y| < 1, fXY (x, y) fXY (0, 0) = 1/ 2 . We can now write
ZZ ZZ
P((X,Y ) S) = fXY (x, y) dx dy 1/ 2 dx dy = 1/ 2 > 0,
S S

which is a contradiction.
21. If E[h(X)k(Y )] = E[h(X)]E[k(Y )] for all bounded continuous functions h and k, then
we may specialize this equation to the functions h(x) = e j1 x and k(y) = e j2 y to show
that the joint characteristic function satisfies

XY (1 , 2 ) := E[e j(1 X+2Y ) ] = E[e j1 X e j2Y ] = E[e j1 X ]E[e j2Y ] = X (1 )Y (2 ).

Since the joint characteristic function is the product of the marginal characteristic
functions, X and Y are independent.
22. (a) Following the hint, let D denote the half-plane D := {(x, y) : x > x0 },
distance of point from origin is x0 / cos
y

D
x
x0
112 Chapter 7 Problem Solutions

and write

P(X > x0 ) = P(X > x0 ,Y IR) = P((X,Y ) D),

where X and Y are both N(0, 1). Then


ZZ 2 2 )/2
e(x +y
P((X,Y ) D) = dx dy.
2
D

Now convert to polar coordinates and write


Z /2 Z 2
er /2
P((X,Y ) D) = r dr d
/2 x0 / cos 2
Z
1 /2 2
= er /2 d
2 /2 x0 / cos
Z /2
1
= exp[(x0 / cos )2 /2] d
2 /2
Z /2  
1 x02
= exp d .
0 2 cos2

(b) In the preceding integral, make the change of variable = /2 t, d = dt.


Then cos becomes cos( /2 t) = sint, and the preceding integral becomes
Z  
1 /2 x02
exp dt.
0 2 sin2 t

23. Write
Z Z Z
fXY (x, y) 1 fX (x)
fY |X (y|x) dy = dy = fXY (x, y) dy = = 1.
fX (x) fX (x) fX (x)

24. First write


  
P (X,Y ) A x < X x + x
P((X,Y ) A|x < X x + x) =
P(x < X x + x)
  
P (X,Y ) A (X,Y ) (x, x + x] IR
=
P(x < X x + x)
n h io
P (X,Y ) A (x, x + x] IR
=
P(x < X x + x)
ZZ
fXY (t, y) dy dt
 
A (x,x+x]IR
= Z x+x
fX ( ) d
x
Chapter 7 Problem Solutions 113

ZZ
IA (t, y)I(x,x+x]IR (t, y) fXY (t, y) dy dt
= Z x+x
fX ( ) d
x
Z x+x Z
IA (t, y) fXY (t, y) dy dt
x
= Z x+x
fX ( ) d
x
Z Z
1 x+x
IA (t, y) fXY (t, y) dy dt
x x
= Z x+x
1
fX ( ) d .
x x
It now follows that
Z
IA (x, y) fXY (x, y) dy

lim P((X,Y ) A|x < X x + x) =
t fX (x)
Z
= IA (x, y) fY |X (y|x) dy.

25. We first compute, for x > 0,


xex(y+1)
fY |X (y|x) = = xexy , y > 0.
ex
As a function of y, this is an exponential density with parameter x. This is very
different from fY (y) = 1/(y + 1)2 . We next compute, for y > 0,
xex(y+1)
fX|Y (x|y) = = (y + 1)2 xe(y+1)x , x > 0.
1/(y + 1)2
As a function of x this is an Erlang(2, y + 1) density, which is not the same as fX
exp(1).
26. We first compute, for x > 0,
xex(y+1)
fY |X (y|x) = = xexy , y > 0.
ex
As a function of y, this is an exponential density with parameter x. Hence,
Z
E[Y |X = x] = y fY |X (y|x) dy = 1/x.
0
We next compute, for y > 0,
xex(y+1)
fX|Y (x|y) = = (y + 1)2 xe(y+1)x , x > 0.
1/(y + 1)2
As a function of x this is an Erlang(2, y + 1) density. Hence,
Z
E[X|Y = y] = x fX|Y (x|y) dx = 2/(y + 1).
0
114 Chapter 7 Problem Solutions

27. Write
Z Z Z 
P(X B|Y = y) fY (y) dy = fX|Y (x|y) dx fY (y) dy
B
Z Z 
= IB (x) fX|Y (x|y) dx fY (y) dy

Z Z 
= IB (x) fXY (x, y) dy dx

Z
= fX (x)dx = P(X B).
B

28. For z 0,
2 )/(2 2 )
Z z e(zy
2 2) Z z
2 e(y/ ) /2 ez/(2 1
fZ (z) = p dy = p dy
z z y2 2 2 2 z z y2
2) Z z 2) Z 1
ez/(2 1 ez/(2 1
= p dy = dt
2 0 z y2 2 0 1 t2
ez/(2 2 ) ez/(2 )
2

= [sin1 (1) sin1 (0)] = ,


2 2 2

which is an exponential density with parameter 1/(2 2 ).


29. For z 0,
Z Z
fZ (z) = x e xz e x dx + y e yz e y dy
0 0
Z
= 2 x e xz e x dx
0
Z
2 2
= x [ z + ]ex[ z+ ] dx.
z+ 0

Now, this last integral is the expectation of an exponential density with parameter
z + . Hence,

2 2 1 2
fZ (z) = = , z 0.
z+ z+ (z + 1)2

30. Using the law of total probability, substitution, and independence, we have
Z Z
P(X Y ) = P(X Y |X = x) fX (x) dx = P(x Y |X = x) fX (x) dx
Z0 Z 0

= P(Y x) fX (x) dx = e x e x dx
0 0
Z
 ( + ) 
= ex( + ) dx = e = .
0 + 0 +
Chapter 7 Problem Solutions 115

31. Using the law of total probability, substitution, and independence, we have
Z
P(Y / ln(1 + X 2 ) > 1) = P(Y / ln(1 + X 2 ) > 1|X = x) fX (x) dx

Z 2
= P(Y > ln(1 + x2 )|X = x) 1 dx
1
Z 2 Z 2
2
= 2
P(Y > ln(1 + x )) dx = e ln(1+x ) dx
1 1
Z 2 Z 2
2 )1 1
= eln(1+x dx = dx
1 1 1 + x2
= tan1 (2) tan1 (1).

32. First find the cdf using the law of total probability and substitution. Then differentiate
to obtain the density.
(a) For Z = eX Y ,
Z
FZ (z) = P(Z z) = P(eX Y z) = P(eX Y z|X = x) fX (x) dx

Z Z
= P(Y zex |X = x) fX (x) dx = FY |X (zex |x) fX (x) dx.

Then Z
fZ (z) = fY |X (zex |x)ex fX (x) dx,

and so Z
fZ (z) = fXY (x, zex )ex dx.

(b) Since Z = |X + Y | 0, we know that FZ (z) and fZ (z) are zero for z < 0. For
z 0, write
Z
FZ (z) = P(Z z) = P(|X +Y | z) = P(|X +Y | z|X = x) fX (x) dx

Z
= P(|x +Y | z|X = x) fX (x) dx

Z
= P(z x +Y z|X = x) fX (x) dx

Z
= P(z x Y z x|X = x) fX (x) dx

Z  
= FY |X (z x|x) FY |X (z x|x) fX (x) dx.

Then
Z  
fZ (z) = fY |X (z x|x) fY |X (z x|x)(1) fX (x) dx

Z  
= fY |X (z x|x) + fY |X (z x|x) fX (x) dx

Z
= fXY (x, z x) + fXY (x, z x) dx.

116 Chapter 7 Problem Solutions

33. (a) First find the cdf of Z using the law of total probability, substitution, and inde-
pendence. Then differentiate to obtain the density. Write
Z
FZ (z) = P(Z z) = P(Y /X z) = P(Y /X z|X = x) fX (x) dx

Z Z
= P(Y /x z|X = x) fX (x) dx = P(Y /x z) fX (x) dx

Z 0 Z
= P(Y /x z) fX (x) dx + P(Y /x z) fX (x) dx
0
Z 0 Z
= P(Y zx) fX (x) dx + P(Y zx) fX (x) dx
0
Z 0 Z
= [1 FY (zx)] fX (x) dx + FY (zx) fX (x) dx.
0

Then
Z 0 Z
fZ (z) = fY (zx)x fX (x) dx + fY (zx)x fX (x) dx
0
Z 0 Z
= fY (zx)|x| fX (x) dx + fY (zx)|x| fX (x) dx
0
Z
= fY (zx) fX (x)|x| dx.

(b) Using the result of part (a) and the fact that the integrand is even, write
Z |zx|2 /(2 2 ) (x/ )2 /2 Z
e e 2 x (x/ )2 [1+z2 ]/2
fZ (z) = |x| dx = e dx.
2 2 2 0 2

Now make the change of variable = (x/ ) 1 + z2 to get
Z
1/ 2 /2 1/  2 /2  1/
fZ (z) = e d = e = ,
1 + z2 0 1+z 2 0 1 + z2
which is the Cauchy(1) density.
(c) Using the result of part (a) and the fact that the integrand is even, write
Z Z
|zx| |x| 2
fZ (z) = 2e 2e |x| dx = xex (|z|+1) dx
2 0
Z
2 /2
= x (|z| + 1)e (|z|+1)x dx,
(|z| + 1) 0

where this last integral is the mean of an exponential density with parameter
(|z| + 1). Hence,

2 /2 1 1
fZ (z) = = .
(|z| + 1) (|z| + 1) 2(|z| + 1)2
Chapter 7 Problem Solutions 117

(d) For z > 0, use the result of part (a) and the fact that the integrand is even, write
Z 2 Z 1/z x2 /2
1 ex /2 e
fZ (z) = 2 I[1,1] (zx)
|x| dx = x dx
2 0 2
2
1 2 1/z 1 e1/(2z )
= (ex /2 ) 0 = .
2 2

The same formula holds for z < 0, and it is easy to check that fZ (0) = 1/ 2 .
(e) Since Z 0, for z 0, we use the result from part (a) to write
Z Z 
zx (zx/ )2 /2 x (x/ )2 /2 x 3 (x/ )2 (z2 +1)/2 dx
fZ (z) = 2
e 2
e x dx = z e
0 0
Z Z  3
2 (z2 +1)/2 t 2 /2 dt
= z 3 e d = z et
0 0 z2 + 1 z2 + 1
Z Z
z 2 /2 2z
= t 3 et dt = ses ds.
(z + 1)2
2
0 (z + 1)2
2
0

This last integral is the mean of an exp(1) density, which is one. Hence fZ (z) =
2z/(z2 + 1)2 .
34. For the cdf, use the law of total probability, substitution, and independence to write
Z
FZ (z) = P(Z z) = P(Y / ln X z) = P(Y / ln X z|X = x) fX (x) dx
0
Z Z
= P(Y / ln x z|X = x) fX (x) dx = P(Y / ln x z) fX (x) dx
0 0
Z 1 Z
= P(Y z ln x) fX (x) dx + P(Y z ln x) fX (x) dx.
0 1

Then
Z 1 Z
fZ (z) = fY (z ln x)(ln x) fX (x) dx + fY (z ln x)(ln x) fX (x) dx
Z0 1

= fX (x) fY (z ln x)| ln x| dx.


0

35. Use the law of total probability, substitution, and independence to write
Z 1/2 Z 1/2
E[e(X+Z)U ] = E[e(X+Z)U |U = u] du = E[e(X+Z)u |U = u] du
1/2 1/2
Z 1/2 Z 1/2 Z 1/2
1
= E[e(X+Z)u ] du = E[eXu ]E[eZu ] du = 2
du
1/2 1/2 1/2 (1 u)

1 1/2 1 1 2 4
= = = 2 = .
1u 1/2 1 1/2 1 + 1/2 3 3
118 Chapter 7 Problem Solutions

36. Use the law of total probability and substitution to write


Z 2 Z 2 Z 2
E[X 2Y ] = E[X 2Y |Y = y] dy = E[X 2 y|Y = y] dy = yE[X 2 |Y = y] dy
1 1 1
Z 2 Z 2
= y (2/y2 ) dy = 2 1/y dy = 2 ln 2.
1 1

37. Use the law of total probability and substitution to write


Z Z
E[X nY r ] = E[X nY r |Y = y] fY (y) dy = E[X n yr |Y = y] fY (y) dy
0 0
Z Z
(n + p)
= yr E[X n |Y = y] fY (y) dy =fY (y) dy yr
0 yn (p) 0
Z
(n + p) rn (n + p) (n + p) (r n)!
= y fY (y) dy = E[Y rn ] = rn .
(p) 0 (p) (p)

38. (a) Use the law of total probability, substitution, and independence to find the cdf.
Write
Z
FY (y) = P(Y y) = P(eVU y) = P(eVU y|V = v) fV (v) dv
0
Z Z
vU
= P(e y|V = v) fV (v) dv = P(evU y) fV (v) dv
0 0
Z
= P(U 1v ln y) fV (v) dv.
0

Then Z
1 1
fY (y) = vy fU ( v ln y) fV (v) dv.
0

To determine when fU ( 1v ln y) is nonzero, we consider the cases y > 1 and y < 1


separately. For y > 1, 1v ln y 0 for all v 0, and 1v ln y 1/2 for v 2 ln y.
Thus, fU ( 1v ln y) = 1 for v 2 ln y, and we can write
Z
e2 ln y 1
fY (y) = 1
yv vev dv = = 3.
2 ln y y y

For y < 1, fU ( 1v ln y) = 1 for 1/2 1v ln y, or v 2 ln y. Thus,


Z
1 2 ln y
fY (y) = 1
yv vev dv = e = y.
2 ln y y

Putting this all together, we have



1/y3 , y 1,
fY (y) = y, 0 y < 1,

0, y < 0.
Chapter 7 Problem Solutions 119

(b) Using the density of part (a),


Z 1 Z
1 1 4
E[Y ] = y2 dy + dy = +1 = .
0 1 y2 3 3
(c) Using the law of total probability, substitution, and independence, we have
Z 1/2 Z 1/2
VU VU
E[e ] = E[e |U = u] du = E[eVu |U = u] du
1/2 1/2
Z 1/2 Z 1/2
1 1 1/2 4
= E[eVu ] du = du = = .
1/2 1/2 (1 u)
2 1 u 1/2 3

39. (a) This problem is interesting because the answer does not depend on the random
variable X. Assuming X has a density fX (x), first write
Z
E[cos(X +Y )] = E[cos(X +Y )|X = x] fX (x) dx

Z
= E[cos(x +Y )|X = x] fX (x) dx.

Now use the conditional density of Y given X = x to write


Z x+ Z 2x+
dy d
E[cos(x +Y )|X = x] = cos(x + y) = cos = 0,
x 2 2x 2
since we are integrating cos over an interval of length 2 . Thus, E[cos(X +
Y )] = 0 as well.
(b) Write
Z Z 2
P(Y > y) = P(Y > y|X = x) fX (x) dx = P(Y > y|X = x) dx
1
Z 2
ey e2y
= exy dx = .
1 y
(c) Begin in the usual way by writing
Z Z
E[XeY ] = E[XeY |X = x] fX (x) dx = E[xeY |X = x] fX (x) dx

Z
= xE[eY |X = x] fX (x) dx.

Now observe that



2 x2 /2 2 /2
E[eY |X = x] = E[esY |X = x] = es = ex .
s=1 s=1

Then continue with


Z Z 7
2 /2 1 2 /2 1 x2 /2  7
E[XeY ] = xex fX (x) dx = xex dx = e 3
4 3 4
e49/2 e9/2
= .
4
120 Chapter 7 Problem Solutions

(d) Write
Z Z 2
E[cos(XY )] = E[cos(XY )|X = x] fX (x) dx = E[cos(xY )|X = x] dx
1
Z 2 Z 2
= E[Re(e jxY )|X = x] dx = Re E[e jxY |X = x] dx
1 1
Z 2 Z 2
2 (1/x)/2
= Re ex dx = ex/2 dx = 2(e1/2 e1 ).
1 1

40. Using the law of total probability, substitution, and independence,


Z
MY (s) = E[esY ] = E[esZX ] = E[esZX |Z = z] fZ (z) dz
0
Z Z
= E[eszX |Z = z] fZ (z) dz = E[eszX ] fZ (z) dz
0 0
Z Z
2 2 /2 2 2 /2 2 /2
= e(sz) fZ (z) dz = e(sz) zez dz
0 0
Z
2 2 )z2 /2
= ze(1s dz.
0

Now make the change of variable t = z 1 s2 2 to get
Z
t 2 dt 1 1/ 2
MY (s) = et /2 = 2 2
= .
0 1 s2 2 1 s2 2 1s 1/ 2 s2

Hence, Y Laplace(1/ ).
41. Using the law of total probability and substitution,
Z Z
n m n m
E[X Y ] = E[X Y |Y = y] fY (y) dy = E[X n ym |Y = y] fY (y) dy
0
Z Z 0
= ym E[X n |Y = y] fY (y) dy = ym 2n/2 yn (1 + n/2) fY (y) dy
0 0
(n + m)!
= 2n/2 (1 + n/2)E[Y n+m ] = 2n/2 (1 + n/2) .
n+m

42. (a) We use the law of total probability, substitution, and independence to write
Z
FZ (z) = P(Z z) = P(X/Y z) = P(X/Y z|Y = y) fY (y) dy
0
Z Z
= P(X/y z|Y = y) fY (y) dy = P(X zy|Y = y) fY (y) dy
Z0 Z 0

= P(X zy) fY (y) dy = FX (zy) fY (y) dy.


0 0

Differentiating, we have
Z Z
( zy) p1 e zy ( y)q1 e y
fZ (z) = fX (zy)y fY (y) dy = y dy.
0 0 (p) (q)
Chapter 7 Problem Solutions 121

Making the change of variable w = y, we obtain


Z
(zw) p1 ezw wq1 ew
fZ (z) = w dw
0 (p) (q)
Z
z p1
= w p+q1 ew(1+z) dw.
(p)(q) 0

Now make the change of variable = w(1 + z) so that d = (1 + z) dw and


w = /(1 + z). Then
Z   p+q1
z p1 d
fZ (z) = e
(p)(q) 0 1 + z (1 + z)
Z
z p1
= p+q1 e d
(p)(q)(1 + z) p+q 0
| {z }
= (p+q)

z p1
= .
B(p, q)(1 + z) p+q
(b) Starting with V := Z/(1 + Z), we first write
 
Z
FV (v) = P v = P(Z v + vZ) = P(Z(1 v) v)
1+Z
= P(Z v/(1 v)).
Differentiating, we have
 v  (1 v) + v  v  1
fV (v) = fZ 2
= fZ .
1 v (1 v) 1 v (1 v)2
Now apply the formula derived in part (a) and use the fact that
v 1
1+ =
1v 1v
to get
[v/(1 v)] p1 1 v p1 (1 v)q1
fV (v) =
1 p+q (1 v)2
= ,
B(p, q)( 1v ) B(p, q)
which is the beta density with parameters p and q.
43. Put q := (n 1)p and Zi := j6=i X j , which is gamma(q, ) by Problem 55(c) in Chap-
ter 4. Now observe that Yi = Xi /(Xi + Zi ), which has a beta density with parameters p
and q := (n 1)p by Problem 42(b).
44. Using the law of total probability, substitution, and independence,
p p
FZ (z) = P(Z z) = P(X/ Y /k z) = P(X z Y /k )
Z p Z p
= P(X z Y /k |Y = y) fY (y) dy = P(X z y/k |Y = y) fY (y) dy
Z0 p
0

= P(X z y/k ) fY (y) dy.


0
122 Chapter 7 Problem Solutions

Then
Z p p
fZ (z) = fX (z y/k ) y/k fY (y) dy
0
Z (z2 y/k)/2 p
e 1
(y/2)k/21 ey/2
= y/k 2 dy
0 2 (k/2)
Z
1 2
= (y/2)k/21/2 ey(1+z /k)/2 dy.
2 k (k/2) 0

Now make the change of variable = y(1 + z2 /k)/2, d = (1 + z2 /k)/2 dy to get


Z  k/21/2
1 d
fZ (z) = e
k (k/2) 0 1 + z2 /k 1 + z2 /k
Z
1
= k/2+1/21 e d
(1/2) k (k/2)(1 + z2 /k)k/2+1/2 0

1 (1 + z2 /k)(k+1)/2
= (k/2 + 1/2) = ,
(1/2) k (k/2)(1 + z2 /k)k/2+1/2 k B(1/2, k/2)

which is the required Students t density with k degrees of freedom.


45. We use the law of total probability, substitution, and independence to write
Z
FZ (z) = P(Z z) = P(X/Y z) = P(X/Y z|Y = y) fY (y) dy
0
Z Z
= P(X/y z|Y = y) fY (y) dy = P(X zy|Y = y) fY (y) dy
Z0 Z 0

= P(X zy) fY (y) dy = FX (zy) fY (y) dy.


0 0

Differentiating, we have
Z Z r r
( zy) p1 e( zy) ( y)q1 e( y)
fZ (z) = fX (zy)y fY (y) dy = r r y dy.
0 0 (p/r) (q/r)

Making the change of variable w = y, we obtain


Z r r
(zw) p1 e(zw) wq1 ew
fZ (z) = r r w dw
0 (p/r) (q/r)
Z
r2 z p1 r r
= w p+q1 ew (1+z ) dw.
(p/r)(q/r) 0

Now make the change of variable = wr (1 + zr ) so that d = rwr1 (1 + zr ) dw and


w = ( /[1 + zr ])1/r . Then
Z  (p+q1)/r
r2 z p1 d
fZ (z) = e (r1)/r
(p/r)(q/r) 0 1 + zr r(1 + zr )
1+zr
Chapter 7 Problem Solutions 123

Z
rz p1
= r (p+q)/r
(p+q)/r1 e d
(p/r)(q/r)(1 + z ) | 0
{z }
= ((p+q)/r)

rz p1
= .
B(p/r, q/r)(1 + zr )(p+q)/r

46. For 0 < z 1,


Z z Z z
1 1 1
fZ (z) = y1/2 (z y)1/2 dy = p dy.
4 0 4z 0 (y/z)(1 (y/z))

Now make the change of variable t 2 = y/z, 2t dt = dy/z to get


Z 1
1 1 1 1 1
fZ (z) = dt = sin t = /4.
2 0 1 t2 2 0

Next, for 1 < z 2,


Z 1 Z
1 1 1 1/ z 1
fZ (z) = p dy = dt
4z
z1 (y/z)(1 (y/z)) 2 11/z 1 t2

1 1 1/ z 1  1 p 
= sin t = sin (1/ z ) sin1 ( 1 1/z ) .
2 11/z 2
Putting this all together yields

 /4, p  0 < z 1,
1 1 1
fZ (z) = sin (1/ z ) sin ( 1 1/z ) , 1 < z 2,
2
0, otherwise.

47. Let denote the N(0, 1) density. Using


 
1 v u
fUV (u, v) = (u, v) = (u) p p ,
1 2 1 2
| {z }
N( u, 1 2 ) density in v

we see that
Z Z  
v u 1
fU (u) = fUV (u, v) dv = (u) p p dv
1 2 1 2
Z  
1 v u
= (u) p p dv = (u).
1 2 1 2
| {z }
density in v integrates to one

Similarly writing
 
u v 1
fUV (u, v) = (u, v) = p p (v),
1 2 1 2
124 Chapter 7 Problem Solutions

we have
Z Z  
u v
1
fV (v) = fUV (u, v) du = p p (v) du
1 2 1 2
Z  
1 v u
= (v) p p du = (v).
1 2 1 2

48. Using  
1 v u
(u, v) = (u) p p ,
1 2 1 2
we can write
 
1 x mX y mY
fXY (x, y) = ,
X Y X Y
  ! ym xmX
!
Y X
Y
1 x mX 1
= p p . ()
X Y X 1 2 1 2
R
Then in f XY (x, y) dy, make the change of variable v = (y mY )/Y , dv = dy/Y
to get
Z  Z !
1 x mX 1 v xm
X
X
fX (x) = fXY (x, y) dy = p p dv
X X 1 2 1 2
| {z }
density in v integrates to one
 
1 x mX
= .
X X

Thus, fX N(mX , X2 ). Using this along with (), we obtain


ymY
!
fXY (x, y) 1 Y xm
X
X
fY |X (y|x) = = p p
fX (x) Y 1 2 1 2
!
1 y [mY + YX (x mX )]
= p p .
Y 1 2 Y 1 2
 
Thus, fY |X ( |x) N mY + YX (x mX ), Y2 (1 2 ) . Proceeding in an analogous
way, using  
1 u v
(u, v) = p p (v),
1 2 1 2
we can write
 
1 x mX y mY
fXY (x, y) = ,
X Y X Y
!  
1
xmX
X ym
Y
Y
y mY
= p p . ()
X Y 1 2 1 2 Y
Chapter 7 Problem Solutions 125

R
Then in f XY (x, y) dx, make the change of variable u = (x mX )/X , du = dx/X
to get
Z  Z !
1 y mY 1 u ym
Y
Y
fY (y) = fXY (x, y) dx = p p du
Y Y 1 2 1 2
| {z }
density in u integrates to one
 
1 y mY
= .
Y Y

Thus, fY N(mY , Y2 ). Using this along with (), we obtain


!
fXY (x, y) 1
xmX
X ym
Y
Y
fX|Y (x|y) = = p p
fY (y) X 1 2 1 2
!
1 x [mX + YX (y mY )]
= p p .
X 1 2 X 1 2
 
Thus, fX|Y ( |y) N mX + YX (y mY ), X2 (1 2 ) .

49. From the solution of Problem 48, we have that


 Y 
fY |X ( |x) N mY + (x mX ), Y2 (1 2 )
X
and  
X
fX|Y ( |y) N mX + (y mY ), X2 (1 2 ) .
Y
Hence, Z
Y
E[Y |X = x] = y fY |X (y|x) dx = mY + (x mX ),
X
and Z
X
E[X|Y = y] = x fX|Y (x|y) dx = mX + (y mY ).
Y
50. From Problem 48, we know that fX N(mX , X2 ). Hence, E[X] = mX and E[X 2 ] =
var(X) + m2X = X2 + m2X . To compute cov(X,Y ), we use the law of total probability
and substitution to write
Z
cov(X,Y ) = E[(X mX )(Y mY )] = E[(X mX )(Y mY )|Y = y] fY (y) dy

Z
= E[(X mX )(y mY )|Y = y] fY (y) dy

Z
= (y mY )E[(X mX )|Y = y] fY (y) dy

Z 
= (y mY ) E[X|Y = y] mX fY (y) dy

Z n o
X
= (y mY ) (y mY ) fY (y) dy
Y
126 Chapter 7 Problem Solutions

Z
X X
= (y mY )2 fY (y) dy = E[(Y mY )2 ]
Y Y
X
= Y2 = X Y .
Y
It then follows that
cov(X,Y )
= .
X Y
51. (a) Using the results of Problem 47, we have
Z Z
1
fU (u) = fUV (u, v) dv = 1 (u, v) + 2 (u, v) dv
2
1
= [ (u) + (u)] = (u),
2
and
Z Z
1
fV (v) = fUV (u, v) du = 1 (u, v) + 2 (u, v) du
2
1
= [ (v) + (v)] = (v).
2
Thus, fU and fV are N(0, 1) densities.
(b) Write
Z Z
:= E[UV ] = uv fUV (u, v) du dv

Z Z
1
= uv [1 (u, v) + 2 (u, v)] du dv
2
Z Z Z Z 
1
= uv1 (u, v) du dv + uv2 (u, v) du dv
2
1 + 2
= .
2
(c) If indeed
1
fUV (u, v) = [ (u, v) + 2 (u, v)]
2 1
is a bivariate normal density, then
 
1
exp 2(1 2 )
[u2 2 uv + v2 ]
fUV (u, v) = p .
2 1 2

In particular then,
1
fUV (u, u) = [ (u, u) + 2 (u, u)],
2 1
or " #
2 2 2
eu /(1+ ) 1 eu /(1+1 ) eu /(1+2 )
p = q + q .
2 1 2 2 2 1 2 2 1 2
1 2
Chapter 7 Problem Solutions 127

Since t := u2 0 is arbitrary, part (iii) of the hint tells us that


1 1 1
p = q = q = 0,
2 1 2 2
4 1 1 4 1 22

which is false.
(d) First observe that
 
fUV (u, v) 1 1 (u, v) 2 (u, v)
fV |U (v|u) = = + .
fU (u) 2 (u) (u)
| {z } | {z }
N(1 u,112 ) N(2 u,122 )

Hence,
Z
1
v2 fV |U (v|u) dv = [(1 12 ) + (1 u)2 + (1 22 ) + (2 u)2 ]
2
1
= [2 12 22 + (12 + 22 )u2 ],
2
which depends on u unless 1 = 2 = 0.
52. For u0 , v0 0, let D := {(u, v) : u u0 , v v0 }. Then
ZZ
P(U > u0 ,V > v0 ) = (u, v) du dv
D
Z tan1 (v0 /u0 ) Z
= (r cos , r sin )r dr d (#)
0 v0 / sin
Z /2 Z
+ (r cos , r sin )r dr d . (##)
tan1 (v0 /u0 ) u0 / cos

Now, since  
r2
exp 2(1 2 )
[1 sin 2 ]
(r cos , r sin ) = p ,
2 1 2
we can express the anti-derivative of r (r cos , r sin ) with respect to r in closed
form as p  
1 2 r2
exp [1 sin 2 ] .
2 (1 sin 2 ) 2(1 2 )
Hence, the double integral in (#) reduces to
Z tan1 (v0 /u0 )
h(v20 , ) d .
0

The double integral in (##) reduces to


Z /2 p  
1 2 u20
exp [1 sin 2 ] d .
tan1 (v0 /u0 ) 2 (1 sin 2 ) 2(1 2 ) cos2
128 Chapter 7 Problem Solutions

Applying the change of variable t = /2 , dt = d , we obtain


Z /2tan1 (v0 /u0 )
h (u20 ,t) dt.
0

53. If = 0 in Problem 52, then U and V are independent. Taking u0 = v0 = x0 ,


Z /4 Z /4
Q(x0 )2 = h0 (x02 , ) d + h0 (x02 , ) d
0 0
Z /4 2 Z /4  
exp[x02 /(2 sin )] 1 x02
= 2 d = exp d .
0 2 0 2 sin2
54. Factor
2 exp[|x y| (y z)2 /2]
fXY Z (x, y, z) = , z 1,
z5 2
as
2
4 e(yz) /2 1 |xy|
fXY Z (x, y, z) = 2e .
z5 2
Now, the second factor on the right is an N(z, 1) density in the variable y, and the
third factor is a Laplace(1) density that has been shifted to have mean y. Hence, the
integral of the third factor with respect to x yields one, and we have by inspection that
2
4 e(yz) /2
fY Z (y, z) = , z 1.
z5 2
We then easily see that
fXY Z (x, y, z) 1 |xy|
fX|Y Z (x|y, z) := = 2e , z 1.
fY Z (y, z)
Next, since the right-hand factor in the formula for fY Z (y, z) is an N(z, 1) density in y,
if we integrate this factor with respect to y, we get one. Thus,
4
fZ (z) = 5 , z 1.
z
We can now see that
2
fY Z (y, z) e(yz) /2
fY |Z (y|z) := = , z 1.
fZ (z) 2
55. To find fXY (x, y), first write
2 /2 2 /2 2 /2 2 /2 2 2 /4
e(xy) e(yz) ez e(xy) e(zy/2) ey
fXY Z (x, y, z) := =
(2 )3/2 (2 )3/2
2 /2
2 2
e(xy) e(y/ 2 ) /2 e(zy/2)
=
2 2
2 /2
2 2
e(xy) e(y/ 2 ) /2 e(zy/2)
=
2 2 2 / 2
2 /2
2 2 /2
e(xy) e(y/ 2 ) /2 e[(zy/2)/(1/ 2 )]
= .
2 2 2 / 2
Chapter 7 Problem Solutions 129

Now the right-hand factor is an N(y/2, 1/2) density in the variable z. Hence, its
integral with respect to z is one. We thus have
2 /2
2 2 2
e(xy) e(y/ 2 ) /2 e(y/ 2 ) /2 e(xy) /2
fXY (x, y) = = ,
2 2 2 2 2
which shows that Y N(0, 2), and given Y = y, X is conditionally N(y, 1). Thus,
E[Y ] = 0 and var(Y ) = 2.
Next,
Z Z
E[X] = E[X|Y = y] fY (y) dy = y fY (y) dy = E[Y ] = 0,

and
Z Z
var(X) = E[X 2 ] = E[X 2 |Y = y] fY (y) dy = (1 + y2 ) fY (y) dy

= 1 + E[Y 2 ] = 1 + var(Y ) = 1 + 2 = 3.
Finally,
Z Z
E[XY ] = E[XY |Y = y] fY (y) dy = E[Xy|Y = y] fY (y) dy

Z Z
= yE[X|Y = y] fY (y) dy = y2 fY (y) dy = E[Y 2 ] = var(Y ) = 2.

56. First write


Z Z
E[XY ] = E[XY |Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= E[Xy|Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= yE[X|Y = y, Z = z] fY Z (y, z) dy dz.

Since fX|Y Z ( |y, z) N(y, z2 ), the preceding conditional expectation is just y. Hence,
Z Z Z
E[XY ] = y2 fY Z (y, z) dy dz = E[Y 2 ] = E[Y 2 |Z = z] fZ (z) dz.

Since fY |Z ( |z) exp(z), the preceding conditional expectation is just 2/z2 . Thus,
Z Z 2
2 2 3 6
E[XY ] = fZ (z) dz = z2 dz = .
z2 1 z2 7 7
A similar analysis yields
Z Z
E[Y Z] = E[Y Z|Z = z] fZ (z) dz = E[Y z|Z = z] fZ (z) dz

Z Z Z
= zE[Y |Z = z] fZ (z) dz = z(1/z) fZ (z) dz = fZ (z) dz = 1.

130 Chapter 7 Problem Solutions

57. Write
Z Z
E[XY Z] = E[XY Z|Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= E[Xyz|Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= yzE[X|Y = y, Z = z] fY Z (y, z) dy dz.

Since fX|Y Z ( |y, z) is a shifted Laplace density with mean y, the preceding conditional
expectation is just y. Hence,
Z Z Z
2 2
E[XY Z] = y z fY Z (y, z) dy dz = E[Y Z] = E[Y 2 Z|Z = z] fZ (z) dy

Z Z
2
= E[Y z|Z = z] fZ (z) dy = zE[Y 2 |Z = z] fZ (z) dy.

Since fY |Z ( |z) N(z, 1), the preceding conditional expectation is just 1 + z2 . Thus,
Z Z Z
E[XY Z] = z(1 + z2 ) fZ (z) dz = [z + z3 ] 4/z5 dz = 4 z4 + z2 dz
1 1
= 4(1/3 + 1) = 16/3.

58. Write
Z Z
E[XY Z] = E[XY Z|X = x,Y = y] fXY (x, y) dx dy

Z Z
= E[xyZ|X = x,Y = y] fXY (x, y) dx dy

Z Z
= xyE[Z|X = x,Y = y] fXY (x, y) dx dy

Z Z Z
= x2 y fXY (x, y) dx dy = E[X 2Y ] = E[X 2Y |X = x] fX (x) dx

Z Z
2
= E[x Y |X = x] fX (x) dx = x2 E[Y |X = x] fX (x) dx

Z Z 2
= x3 fX (x) dx = x3 dx = 15/4.
1

59. We use the law of total probability, substitution, and independence to write

N N
Y ( ) = E[e jY ] = E[e j i=1 Xi ] = E[e j i=1 Xi |N = n]P(N = n)
n=1

n
= E[e j i=1 Xi |N = n]P(N = n)
n=1

n
= E[e j i=1 Xi ]P(N = n)
n=1
Chapter 7 Problem Solutions 131

 n 
j Xi
= E e P(N = n)
n=1 i=1
 n 
j Xi
= E[e ] P(N = n)
n=1 i=1

= X ( )n P(N = n) = GN (X ( )).
n=1

Now, if N geometric1 (p), GN (z) = [(1 p)z]/[1 pz], and if X exp( ), X ( ) =


/( j ). Then

(1 p)X ( ) (1 p) /( j ) (1 p) (1 p)
Y ( ) = = = = ,
1 pX ( ) 1 p /( j ) ( j ) p (1 p) j

which is the exp((1 p) ) characteristic function. Thus, Y exp((1 p) ).


CHAPTER 8
Problem Solutions

1. We have
10 40   230 280 330
20 50 7 8 9 = 340 410 480
4 5 6
30 60 450 540 630
and
230 280 330
tr 340 410 480 = 1270.
450 540 630

2. MATLAB. See the answer to the previous problem.


3. MATLAB. We have
7 4
A0 = 8 5 .
9 6

4. Write
r r  n  n  r  n
tr(AB) = (AB)ii = Aik Bki = Bki Aik = (BA)kk = tr(BA).
i=1 i=1 k=1 k=1 i=1 k=1

5. (a) Write
r r  n  r n
0 0 0
tr(AB ) = (AB )ii = Aik (B )ki = Aik Bik .
i=1 i=1 k=1 i=1 k=1

(b) If tr(AB0 ) = 0 for all B, then in particular, it is true for B = A; i.e.,


r n
0 = tr(AA0 ) = A2ik ,
i=1 k=1

which implies Aik = 0 for all i and k. In other words, A is the zero matrix of size
r n.
6. Following the hint, we first write

0 kx yk2 = hx y, x yi = kxk2 2 hx, yi + 2 kyk2 .

Taking = hx, yi/kyk2 yields

|hx, yi|2 |hx, yi|2 |hx, yi|2


0 kxk2 2 2
+ 4
kyk2 = kxk2 ,
kyk kyk kyk2

132
Chapter 8 Problem Solutions 133

which can be rearranged to get |hx, yi|2 kxk2 kyk2 . Conversely, suppose |hx, yi|2 =
kxk2 kyk2 . There are two cases to consider. If y 6= 0, then reversing the above sequence
of observations implies 0 = kx yk2 , which implies x = y. On the other hand, if
y = 0 and if
|hx, yi| = kxk kyk,
then we must have kxk = 0; i.e., x = 0 and y = 0, and in this case x = y for all .
7. Consider the i j component of E[XB]. Since
 
(E[XB])i j = E[(XB)i j ] = E Xik Bk j = E[Xik ]Bk j = (E[X])ik Bk j
k k k
= (E[X]B)i j
holds for all i j, E[XB] = E[X]B.
 
8. tr(E[X]) = (E[X])ii = E[Xii ] = E Xii = E[tr(X)].
i i i

9. Write
E[kX E[X]k2 ] = E[(X E[X])0 (X E[X])], which is a scalar,
n o
= tr E[(X E[X])0 (X E[X])]
h n oi
= E tr (X E[X])0 (X E[X]) , by Problem 8,
h n oi
= E tr (X E[X])(X E[X])0 , by Problem 4,
n o
= tr E[(X E[X])(X E[X])0 ] , by Problem 8,
n n
= tr(C) = Cii = var(Xi ).
i=1 i=1

   0
10. Since E [X,Y, Z]0 = E[X], E[Y ], E[Z] , and

E[X 2 ] E[XY ] E[XZ] E[X]2 E[X]E[Y ] E[X]E[Z]
cov([X,Y, Z]0 ) = E[Y X] E[Y 2 ] E[Y Z] E[Y ]E[X] E[Y ]2 E[Y ]E[Z] ,
E[ZX] E[ZY ] E[Z 2 ] E[Z]E[X] E[Z]E[Y ] E[Z]2
we begin by computing all the entries of these matrices. For the mean vector,
Z 2 Z 2 2
3 2
E[Z] = 3
z 7 z dz = 7 z3 dz = 37 41 z4 = 3 15/28 = 45/28.
1 1 1

Next,
Z 2 Z 2 Z 2 2
3 2
E[Y ] = E[Y |Z = z] fZ (z) dz = 1
z 7 z dz = 3
7 z dz = 3
7 12 z2 = 9/14.
1 1 1 1

Since E[U] = 0 and since U and Z are independent,


E[X] = E[ZU +Y ] = E[Z]E[U] + E[Y ] = E[Y ] = 9/14.
134 Chapter 8 Problem Solutions

Thus, the desired mean vector is


 
E [X,Y, Z]0 = [9/14, 9/14, 45/28]0 .

We next compute the correlations. First,


Z 2 Z 2
E[Y Z] = E[Y Z|Z = z] fZ (z) dz = E[Y z|Z = z] fZ (z) dz
1 1
Z 2 Z 2 Z 2
= zE[Y |Z = z] fZ (z) dz = z(1/z) fZ (z) dz = fZ (z) dz = 1.
1 1 1

Next, Z 2 Z 2 2

E[Z 2 ] = z2 37 z2 dz = 3
7 z4 dz = 3
7 51 z5 = 93/35.
1 1 1

Again using the fact that E[U] = 0 and independence,

E[XZ] = E[(ZU +Y )Z] = E[Z 2 ]E[U] + E[Y Z] = E[Y Z] = 1.

Now, Z 2 Z 2
E[Y 2 ] = E[Y 2 |Z = z] fZ (z) dz = (2/z2 ) 73 z2 dz = 6/7.
1 1
We can now compute

E[XY ] = E[(ZU +Y )Y ] = E[ZY ]E[U] + E[Y 2 ] = 6/7,

and

E[X 2 ] = E[(ZU +Y )2 ] = E[Z 2 ]E[U 2 ] + 2E[U]E[ZY ] + E[Y 2 ]


= E[Z 2 ] + E[Y 2 ] = 93/35 + 6/7 = 123/35.

We now have that



123/35 6/7 1 81/196 81/196 405/392
cov([X,Y, Z]0 ) = 6/7 6/7 1 81/196 81/196 405/392
1 1 93/35 405/392 405/392 2025/784

3.1010 0.4439 0.0332
= 0.4439 0.4439 0.0332 .
0.0332 0.0332 0.0742
   0
11. Since E [X,Y, Z]0 = E[X], E[Y ], E[Z] , and

E[X 2 ] E[XY ] E[XZ] E[X]2 E[X]E[Y ] E[X]E[Z]
cov([X,Y, Z]0 ) = E[Y X] E[Y 2 ] E[Y Z] E[Y ]E[X] E[Y ]2 E[Y ]E[Z] ,
E[ZX] E[ZY ] E[Z 2 ] E[Z]E[X] E[Z]E[Y ] E[Z]2

we compute all the entries of these matrices. To make this job easier, we first factor

2 exp[|x y| (y z)2 /2]


fXY Z (x, y, z) = , z 1,
z5 2
Chapter 8 Problem Solutions 135

as fX|Y Z (x|y, z) fY |Z (y|z) fZ (z) by writing


2
e(yz) /2 4
1 |xy|
fXY Z (x, y, z) = 2e
5 , z 1.
2 z
We then see that as a function of x, fX|Y Z (x|y, z) is a shifted Laplace(1) density. Sim-
ilarly, as a function of y, fY |Z (y|z) is an N(z, 1) density. Thus,
Z Z Z Z
E[X] = E[X|Y = y, Z = z] fY Z (y, z) dy dz = y fY Z (y, z) dy dz

Z Z
= E[Y ] = E[Y |Z = z] fZ (z) dz = z fZ (z) dz = E[Z]

Z 4 Z
4
= z 5 dz = 4
dz = 4/3.
1 z 1 z
 
Thus, E [X,Y, Z]0 = [4/3, 4/3, 4/3]0 . We next compute
Z Z
E[X 2 ] = E[X 2 |Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= (2 + y2 ) fY Z (y, z) dy dz = 2 + E[Y 2 ],

where
Z Z
E[Y 2 ] = E[Y 2 |Z = z] fZ (z) dz = (1 + z2 ) fZ (z) dz = 1 + E[Z 2 ].

Now, Z Z
4 4
E[Z 2 ] = dz = z2 dz = 2.
1 z5 1 z3
Thus, E[Y 2 ] = 3 and E[X 2 ] = 5. We next turn to
Z Z
E[XY ] = E[XY |Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= E[Xy|Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= yE[X|Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= y2 fY Z (y, z) dy dz = E[Y 2 ] = 3.

We also have
Z Z
E[XZ] = E[XZ|Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= E[Xz|Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= yz fY Z (y, z) dy dz = E[Y Z]

Z Z
= E[Y Z|Z = z] fZ (z) dz = zE[Y |Z = z] fZ (z) dz

Z
= z2 fZ (z) dz = E[Z 2 ] = 2.

136 Chapter 8 Problem Solutions

We now have that



5 3 2 1 1 1
16
cov([X,Y, Z]0 ) = 3 3 2 1 1 1
9
2 2 2 1 1 1

29 11 2 3.2222 1.2222 0.2222
1
= 11 11 2 = 1.2222 1.2222 0.2222 .
9
2 2 2 0.2222 0.2222 0.2222
   0
12. Since E [X,Y, Z]0 = E[X], E[Y ], E[Z] , and

E[X 2 ] E[XY ] E[XZ] E[X]2 E[X]E[Y ] E[X]E[Z]
cov([X,Y, Z]0 ) = E[Y X] E[Y 2 ] E[Y Z] E[Y ]E[X] E[Y ]2 E[Y ]E[Z] ,
E[ZX] E[ZY ] E[Z 2 ] E[Z]E[X] E[Z]E[Y ] E[Z]2
we compute all the entries of these matrices. We begin with
Z Z
E[Z] = E[Z|Y = y, X = x] fXY (x, y) dy dx

Z Z
= x fXY (x, y) dy dx = E[X] = 3/2.

Next,
Z Z
E[Y ] = E[Y |X = x] fX (x) dx = x fX (x) dx = E[X] = 3/2.

We now compute
Z Z
E[XY ] = E[XY |X = x] fX (x) dx = E[xY |X = x] fX (x) dx

Z Z
= xE[Y |X = x] fX (x) dx = x2 fX (x) dx = E[X 2 ]

= var(X) + E[X]2 = 1/12 + (3/2) = 7/3. 2

Then
Z Z
E[XZ] = E[XZ|Y = y, X = x] fXY (x, y) dy dx

Z Z
= xE[Z|Y = y, X = x] fXY (x, y) dy dx

Z Z
= x2 fXY (x, y) dy dx = E[X 2 ] = 7/3,

and
Z Z
E[Y Z] = E[Y Z|Y = y, X = x] fXY (x, y) dy dx

Z Z
= yE[Z|Y = y, X = x] fXY (x, y) dy dx

Z Z
= xy fXY (x, y) dy dx = E[XY ] = 7/3.

Chapter 8 Problem Solutions 137

Next,
Z Z
E[Y 2 ] = E[Y 2 |X = x] fX (x) dx = 2x2 fX (x) dx = 2E[X 2 ] = 14/3,

and
Z Z
E[Z 2 ] = E[Z 2 |Y = y, X = x] fXY (x, y) dy dx

Z Z
= (1 + x2 ) fXY (x, y) dy dx = 1 + E[X 2 ] = 1 + 7/3 = 10/3.

We now have that



7 7 7  3 2 1 1 1
1
cov([X,Y, Z]0 ) = 7 14 7 1 1 1
3 2
7 7 10 1 1 1

1 1 1 0.0833 0.0833 0.0833
1
= 1 29 1 = 0.0833 2.4167 0.0833 .
12
1 1 13 0.0833 0.0833 1.0833
   0
13. Since E [X,Y, Z]0 = E[X], E[Y ], E[Z] , and

E[X 2 ] E[XY ] E[XZ] E[X]2 E[X]E[Y ] E[X]E[Z]
cov([X,Y, Z]0 ) = E[Y X] E[Y 2 ] E[Y Z] E[Y ]E[X] E[Y ]2 E[Y ]E[Z] ,
E[ZX] E[ZY ] E[Z 2 ] E[Z]E[X] E[Z]E[Y ] E[Z]2
we compute all the entries of these matrices. In order to do this, we first note that
Z N(0, 1). Next, as a function of y, fY |Z (y|z) is an N(z, 1) density. Similarly, as a
function of x, fX|Y Z (x|y, z) is an N(y, 1) density. Hence, E[Z] = 0,
Z Z
E[Y ] = E[Y |Z = z] fZ (z) dz = z fZ (z) dz = E[Z] = 0,

and
Z Z
E[X] = E[X|Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= y fY Z (y, z) dy dz = E[Y ] = 0.

We next compute
Z Z
E[XY ] = E[XY |Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= E[Xy|Y = y, Z = z] fY Z (y, z) dy dz

Z Z Z Z
= yE[X|Y = y, Z = z] fY Z (y, z) dy dz = y2 fY Z (y, z) dy dz

Z Z
= E[Y 2 ] = E[Y 2 |Z = z] fZ (z) dz = (1 + z2 ) fZ (z) dz

= 1 + E[Z 2 ] = 2.
138 Chapter 8 Problem Solutions

Then
Z Z
E[Y Z] = E[Y Z|Z = z] fZ (z) dz = zE[Y |Z = z] fZ (z) dz

Z
= z2 fZ (z) dz = E[Z 2 ] = 1,

and
Z Z
E[XZ] = E[XZ|Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= zE[X|Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= yz fY Z (y, z) dy dz = E[Y Z] = 1.

Next,
Z Z
E[Y 2 ] = E[Y 2 |Z = z] fZ (z) dz = (1 + z2 ) fZ (z) dz = 1 + E[Z 2 ] = 2,

and
Z Z
E[X 2 ] = E[X 2 |Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= (1 + y2 ) fY Z (y, z) dy dz = 1 + E[Y 2 ] = 3.

Since E[Z 2 ] = 1, we now have that



3 2 1
cov([X,Y, Z]0 ) = 2 2 1 .
1 1 1

14. We first note that Z N(0, 1). Next, as a function of y, fY |Z (y|z) is an N(z, 1) density.
Similarly, as a function of x, fX|Y Z (x|y, z) is an N(y, 1) density. Hence,
Z Z
E[e j(1 X+2Y +3 Z) ] = E[e j(1 X+2Y +3 Z) |Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= E[e j(1 X+2 y+3 z) |Y = y, Z = z] fY Z (y, z) dy dz

Z Z
= E[e j1 X |Y = y, Z = z]e j2 y e j3 z fY Z (y, z) dy dz

Z Z
2
= e j1 y1 /2 e j2 y e j3 z fY Z (y, z) dy dz

Z Z
2
= e1 /2 e j(1 +2 )y e j3 z fY Z (y, z) dy dz

2
= e1 /2 E[e j(1 +2 )Y e j3 Z ]
Chapter 8 Problem Solutions 139

Z
2
= e1 /2 E[e j(1 +2 )Y e j3 Z |Z = z] fZ (z) dz

Z
2
= e1 /2 e j3 z E[e j(1 +2 )Y |Z = z] fZ (z) dz

Z
2 2 /2
= e1 /2 e j3 z e j(1 +2 )z(1 +2 ) fZ (z) dz

Z
2 2 /2
= e1 /2 e(1 +2 ) e j(1 +2 +3 )z fZ (z) dz

2 2 /2
= e1 /2 e(1 +2 ) E[e j(1 +2 +3 )z ]
2 2 2 /2
= e1 /2 e(1 +2 ) /2 e(1 +2 +3 )
2 2 2
= e[1 +(1 +2 ) +(1 +2 +3 ) ]/2 .

15. RY := E[YY 0 ] = E[(AX)(AX)0 ] = E[AXX 0 A0 ] = AE[XX 0 ]A0 = ARX A0 .


16. First, since R = E[XX 0 ], we see that R0 = E[XX 0 ]0 = E[(XX 0 )0 ] = E[XX 0 ] = R. Thus,
R is symmetric. Next, define the scalar Y := c0 X. Then

0 E[Y 2 ] = E[YY 0 ] = E[(c0 X)(c0 X)0 ] = E[c0 XX 0 c] = c0 E[XX 0 ]c = c0 Rc,


and we see that R is positive semidefinite.
17. Use the CauchySchwarz inequality to write

(CXY )i j = E[(Xi mX,i )(Y j mY, j )]
q q
E[(Xi mX,i )2 ]E[(Y j mY, j )2 ] = (CX )ii (CY ) j j .

18. If  
cos sin
P = ,
sin cos
then
        
U X cos sin X X cos +Y sin
:= P0 = = .
V Y sin cos Y X sin +Y cos

We now have to find such that E[UV ] = 0. Write

E[UV ] = E[(X cos +Y sin )(X sin +Y cos )]


= E[(Y 2 X 2 ) sin cos + XY (cos2 sin2 )]
Y2 X2
= sin 2 + E[XY ] cos 2 ,
2
which is zero if and only if
2E[XY ]
tan 2 = .
X2 Y2
Hence,  
1 1 2E[XY ]
= tan .
2 X2 Y2
140 Chapter 8 Problem Solutions

19. (a) Write (ei e0i )mn = (ei )m (e0i )n , which equals one if and only if m = i and n = i.
Hence, ei e0i must be all zeros except at position i i where it is one.
(b) Write
  0
e1 e4 e5 e1
E 0E = e04
e5
= e1 e01 + e4 e04 + e5 e05
= diag(1, 0, 0, 0, 0) + diag(0, 0, 0, 1, 0) + diag(0, 0, 0, 0, 1)
= diag(1, 0, 0, 1, 1).

20. (a) We must solve P0CX P = diagonal. With X := U +V , we have

CX = E[XX 0 ] = E[(U +V )(U +V )0 ] = CU + E[UV 0 ] + E[VU 0 ] +CV


= QMQ0 + I = Q(M + I)Q0 .

Hence, Q0CX Q = M + I, which is diagonal. The point here is that we may take
P = Q.
(b) We now put Y := P0 X = QX = Q(U +V ). Then

CY = E[YY 0 ] = E[Q(U +V )(U 0 +V 0 )Q0 ]


= Q{CU + E[UV 0 ] + E[VU 0 ] +CV }Q0 = QCU Q0 + QIQ0 = M + I.

21. Starting with u = x + y and v = x y, we have


u+v uv
x = and y = .
2 2
We can now write
" x x
# " #
u v 1/2 1/2
dH = y y
= , det dH = 1/2,
u v
1/2 1/2

and
| det dH | = 1/2,
and so u+v uv 1
fUV (u, v) = fXY , .
2 2 2

22. Starting with u = xy and v = y/x, write y = xv. Then u = x2 v, and x = (u/v)1/2 . We
also have y = (u/v)1/2 v = (uv)1/2 . Then
x x
u = (1/2)/ uv, v = (1/2) u/v3/2 ,
y
p y
p
u = (1/2) v/u, v = (1/2) u/v.
Chapter 8 Problem Solutions 141

In other words, " #


(1/2)/ uv (1/2) u/v3/2
dH = p p ,
(1/2) v/u (1/2) u/v
and so
1 1 1
| det dH | = + = .
4v 4v 2|v|
Thus, p 1
fUV (u, v) = fXY u/v, uv , u, v > 0.
2v
For fU , write
Z Z p 1
fU (u) = fUV (u, v) dv = fXY u/v, uv dv
0 0 2v

Now make the change of variable y = uv, or y2 = uv. Then 2y dy = u dv, and
Z u 1
fU (u) = fXY , y dy.
0 y y
For fV , write
Z Z p 1
fV (v) = fUV (u, v) du = fXY u/v, uv du.
0 0 2v
p
This time make the change of variable x = u/v or x2 = u/v. Then 2x dx = du/v,
and Z
fV (v) = fXY (x, vx)x dx.
0

23. Starting with u = x and v = y/x, we find that y = xv = uv. Hence,


" x x # " #
u v 1 0
dH = y y = , det dH = u, and | det dH | = |u|.
u v
v u

Then
|u| |uv|
fUV (u, v) = fXY (u, uv)|u| = 2e 2e |u|,
and
Z Z
2 (1+|v|)|u| 2
fV (v) = 4 |u|e du = 2 ue (1+|v|)u du
0
Z

= u (1 + |v|)e (1+|v|)u du.
2(1 + |v|) 0

Now, this last integral is the mean of an exponential density with parameter (1+|v|).
Hence,
1 1
fV (v) = = .
2(1 + |v|) (1 + |v|) 2(1 + |v|)2
142 Chapter 8 Problem Solutions


24. Starting with u = 2 ln x cos(2 y) and v = 2 ln x sin(2 y), we have

u2 + v2 = (2 ln x)[cos2 (2 y) + sin2 (2 y)] = 2 ln x.


2 +v2 )/2
Hence, x = e(u . We also have
v 1
= tan(2 y) or y = tan1 (v/u).
u 2
We can now write
x 2 +v2 )/2 x 2 +v2 )/2
u = ue(u , v = ve(u ,
y 1 1 v y 1 1 1
u = 2 1+(u/v)2 u2
, v = 2 1+(u/v)2 u.

In other words,
2 +v2 )/2 2 +v2 )/2
ue(u ve(u
dH = ,
1 1
2 1+(u/v)2 v
u2
1 1
2 1+(u/v)2 1u

and so

1 e(u2 +v2 )/2 2 2
1 e(u +v )/2 v2

| det dH| =
2 1 + (v/u)2 2 1 + (v/u)2 u2
2 2
e(u +v )/2 1 (v/u)2
= +
1 + (v/u)2 1 + (v/u)2
2
2 2
eu /2 ev /2
= .
2 2
2 2
We next use the formula fUV (u, v) = fXY (x, y) | det dH|, where x = e(u +v )/2 and
y = tan1 (v/u)/(2 ). Fortunately, since these formulas for x and y lie in (0, 1],
fXY (x, y) = I(0,1] (x)I(0,1] (y) = 1, and we see that
2 2
eu /2 ev /2
fUV (u, v) = 1 | det dH| = .
2 2
2
Integrating out v shows that fU (u) = eu /2 / 2 , and integrating out v shows that
2
fV (v) = ev /2 / 2 . It now follows that fUV (u, v) = fU (u) fV (v), and we see that U
and V are independent.
25. Starting with u = x + y and v = x/(x + y) = x/u, we see that v = x/u, or x = uv. Next,
from y = u x = u uv, we get y = u(1 v). We can now write
x x
u = v, v = u,
y y
u = 1 v, v = u.

In other words,  
v u
dH = ,
1 v u
Chapter 8 Problem Solutions 143

and so
| det dH| = | uv u(1 v)| = |u|.
We next write
fUV (u, v) = fXY (uv, u(1 v))|u|.
If X and Y are independent gamma RVs, then for u > 0 and 0 < v < 1,

( uv) p1 e uv ( u(1 v))q1 e u(1v)


fUV (u, v) = u
(p) (q)
( u) p+q1 e u (p + q) p1
= v (1 v)q1 ,
(p + q) (p)(q)

which we recognize as the product of a gamma(p + q, ) and a beta(p, q) density.


Hence, it is easy to integrate out either u or v and show that fUV (u, v) = fU (u) fV (v),
where fU gamma(p + q, ) and fV beta(p, q). Thus, U and V are independent.
26. Starting with u = x + y and v = x/y, write x = yv and then u = yv + y = y(v + 1). Solve
for y = u/(v + 1) and x = u y = u u/(v + 1) = uv/(v + 1). Next,

x x
u = v/(v + 1), v = u/(v + 1)2 ,
y y
u = 1/(v + 1), v = u/(v + 1)2 .

In other words,  
v/(v + 1) u/(v + 1)2
dH = ,
1/(v + 1) u/(v + 1)2
and so

uv u u(v + 1) |u|

| det dH | =
= = .
(v + 1)3 (v + 1)3 (v + 1)3 (v + 1)2

We next write  
uv u |u|
fUV (u, v) = fXY , .
v + 1 v + 1 (v + 1)2
When X gamma(p, ) and Y gamma(q, ) are independent, then U and V are
nonnegative, and

[ uv/(v + 1)] p1 e uv/(v+1) [ u/(v + 1)]q1 e u/(v+1) u


fUV (u, v) =
(p) (q) (v + 1)2
( u) p+q1 e u (p + q) v p1
=
(p + q) (p)(q) (v + 1) p+q
( u) p+q1 e u v p1
= ,
(p + q) B(p, q)(v + 1) p+q

which shows that U and V are independent with the required marginal densities.
144 Chapter 8 Problem Solutions

27. From solution of the Example at the end of the section,

fR, (r, ) = fXY (r cos , r sin )r.

What is different in this problem is that X and Y are correlated Gaussian random
variables; i.e.,
2 2 2
e(x 2 xy+y )/[2(1 )]
fXY (x, y) = p .
2 1 2
Hence,
2 cos2 2 (r cos )(r sin )+r2 sin2 )/[2(1 2 )]
re(r
fR, (r, ) = p
2 1 2
2 (1 sin 2 )/[2(1 2 )]
rer
= p .
2 1 2
To find the density of , we must integrate this with respect to r. Notice that the
2 2
integrand is proportional to re r /2 , whose anti-derivative is e r /2 / . Here we
2
have = (1 sin 2 )/(1 ). We can now write
Z Z
1 2 /2 1 1
f ( ) = fR, (r, ) dr = p re r dr = p
0 2 1 2 0 2 1 2
p
1 2 1 2
= p = .
2 1 2 (1 sin 2 ) 2 (1 sin 2 )

28. Since X N(0, 1), we have E[X] = E[X 3 ] = 0, E[X 4 ] = 3, and E[X 6 ] = 15. Since
W N(0, 1), E[W ] = 0 too. Hence, E[Y ] = E[X 3 + W ] = 0. It then follows that mY ,
mX and b = mX Amy are zero. We next compute

CXY = E[XY ] = E[X(X 3 +W )] = E[X 4 ] + E[X]E[W ] = 3 + 0 = 3,

and
CY = E[Y 2 ] = E[X 6 + 2X 3W +W 2 ] = 15 + 0 + 1 = 16.
Hence, ACY = CXY implies A = 3/16, and then Xb = A(Y mY ) + mX = (3/16)Y .
29. First note that since X and W are zero mean, so is Y . Next,

CXY = E[XY ] = E[X(X +W )] = E[X 2 ] + E[XW ] = E[X 2 ] + E[X]E[W ] = E[X 2 ] = 1,

and

CY = E[Y 2 ] = E[(X +W )2 ] = E[X 2 + 2XW +W 2 ]


= E[X 2 ] + 2E[X]E[W ] + E[W 2 ] = 1 + 0 + 2/ 2 = 1 + 2/ 2 .

Then A = CXY /CY = 1/[1 + 2/ 2 ], and

2
Xb = Y.
2+2
Chapter 8 Problem Solutions 145

30. We first have E[Y ] = E[GX +W ] = GmX + 0 = GmX . Next,

CXY = E[(X mX )(Y mY )0 ] = E[(X mX )(GX +W GmX )0 ]


= E[(X mX )(G{X mX } +W )0 ] = CX G0 +CXW = CX G0 ,

and

CY = E[(Y mY )(Y mY )0 ] = E[(G{X mX } +W )(G{X mX } +W )0 ]


= GCX G0 + GCXW +CW X G0 +CW = GCX G0 +CW ,

since X and W are uncorrelated. Solving ACY = CXY implies

A = CX G0 (GCX G0 +CW )1 and Xb = CX G0 (GCX G0 +CW )1 (Y GmX ) + mX .

31. We begin with the result of Problem 30 that

A = CX G0 (GCX G0 +CW )1 .

Following the hint, we make the identifications = CW , = CX , = G, and = G0 .


Then

A = CX G0 ( + )1
= CX G0 [CW
1 1
CW G(CX1 + G0CW
1
G)1 G0CW
1
]
= CX G0CW
1
CX G0CW
1
G(CX1 + G0CW
1
G)1 G0CW
1

= [CX CX G0CW
1
G(CX1 + G0CW
1
G)1 ]G0CW
1

= [CX (CX1 + G0CW


1
G) CX G0CW
1
G](CX1 + G0CW
1
G)1 G0CW
1

= [I +CX G0CW
1
G CX G0CW
1
G](CX1 + G0CW
1
G)1 G0CW
1

= (CX1 + G0CW
1
G)1 G0CW
1
.

32. We begin with


Xb = A(Y mY ) + mX , where ACY = CXY .
Next, with Z := BX, we have mZ = BmX and

CZY = E[(Z mZ )(Y mY )0 ] = E[B(X mX )(Y mY )0 ] = BCXY .

e Y = CZY . Starting with ACY = CXY , multiply this equation by B to


We must solve AC
e := BA solves the required equation. Hence,
get (BA)CY = BCXY = CZY . We see that A
the linear MMSE estimate of X based on Z is

b
(BA)(Y mY ) + mZ = (BA)(Y mY ) + BmX = B{A(Y mY ) + mX } = BX.

33. We first show that the orthogonality condition

E[(CY )0 (X AY )] = 0, for all C,


146 Chapter 8 Problem Solutions

implies A is optimal. Write

E[kX BY k2 ] = E[k(X AY ) + (AY BY )k2 ]


= E[k(X AY ) + (A B)Y )k2 ]
= E[kX AY k2 ] + 2E[{(A B)Y }0 (X AY )] + E[k(A B)Y k2 ]
= E[kX AY k2 ] + E[k(A B)Y k2 ]
E[kX AY k2 ],

where the cross terms vanish by taking C = A B in the orthogonality condition.


Next, rewrite the orthogonality condition as

E[(CY )0 (X AY )] = tr{E[(CY )0 (X AY )]} = E[tr{(CY )0 (X AY )}]


= E[tr{(X AY )(CY )0 }] = tr{E[(X AY )(CY )0 ]}
= tr{(RXY ARY )C0 }.

Now, this expression must be zero for all C, including C = RXY ARY . However,
since tr(DD0 ) = 0 implies D = 0, we conclude that the optimal A solves ARY = RXY .
Next, the best constant estimator is easily found by writing

E[kX bk2 ] = E[k(X mX ) + (mX b)k2 ] = E[kX mX k2 ] + kmX bk2 .

Hence, the optimal value of b is b = mX .


34. To begin, write

b
E[(X X)(X b 0 ] = E[{(X mX ) A(Y mY )}{(X mX ) A(Y mY )}0 ]
X)
= CX ACY X CXY A0 + ACY A0 . ()

We now use the fact that ACY = CXY . If we multiply ACY = CXY on the right by A0 ,
we obtain
ACY A0 = CXY A0 .
Furthermore, since ACY A0 is symmetric, we can take the transpose of the above ex-
pression and obtain
ACY A0 = ACY X .
By making appropriate substitutions in (), we find that the error covariance is also
given by
CX ACY X , CX CXY A0 , and CX ACY A0 .

35. Write
n o
E[kX Xk b 0 (X X)]
b 2 ] = E[(X X) b 0 (X X)]
b = tr E[(X X) b
b 0 (X X)}]
= E[tr{(X X) b b
= E[tr{(X X)(X b 0 }]
X)
b
= tr{E[(X X)(X b 0 ]} = tr{CX ACY X }.
X)
Chapter 8 Problem Solutions 147

36. MATLAB. We found



0.0622 0.0467 0.0136 0.1007
0.0489 0.0908 0.0359 0.1812
A =
0.0269 0.0070 0.0166 0.0921

0.0619 0.0205 0.0067 0.0403

and MSE = 0.0806.


37. Write X in the form X = [Y 0 , Z 0 ]0 , where Y := [X1 , . . . , Xm ]0 . Then
   0 0      
Y Y Z CY CY Z C1 C2
CX = E = = ,
Z CZY CZ C20 C3

and       
Y 0 CY C1
CXY = E Y = = .
Z CZY C20
Solving ACY = CXY becomes
   
C1 I
AC1 = , or A = .
C20 C20 C11

The linear MMSE estimate of X = [Y 0 , Z 0 ]0 is


   
I Y
AY = Y = .
C20 C11 C20 C11Y

In other words, Yb = Y and Zb = C20 C11Y . Note that the matrix required for the linear
MMSE estimate of Z based on Y is the solution of BCY = CZY or BC1 = C20 ; i.e.,
B = C20 C11 . Next, the error covariance for estimating X based on Y is
    
b b 0 ] = CX ACY X = C10 C2 I C1 C2
E[(X X)(X X)
C2 C3 C20 C11
   
C1 C2 C1 C2
=
C20 C3 C20 C20 C11C2
 
0 0
= ,
0 C3 C20 C11C2

and the MSE is


b 2 ] = tr(CX ACY X ) = tr(C3 C20 C1C2 ).
E[kX Xk 1

38. Since P0 decorrelates Y , the covariance matrix CZ of Z := P0Y is diagonal. Writing


ACZ = CXZ in component form, and using the fact that CZ is diagonal,

Aik (CZ )k j = (CXZ )i j


k

becomes
Ai j (CZ ) j j = (CXZ )i j
148 Chapter 8 Problem Solutions

If (CZ ) j j 6= 0, then Ai j = (CXZ )i j /(CZ ) j j . If (CZ ) j j = 0, then Ai j (CZ ) j j = (CXZ )i j


can be solved only if (CXZ )i j = 0, which we now show to be the case by using the
CauchySchwarz inequality. Write

|(CXZ )i j | = E[(Xi (mX )i )(Z j (mZ ) j )]
q
E[(Xi (mX )i )2 ]E[(Z j (mZ ) j )2 ]
q
= (CX )ii (CZ ) j j .

Hence, if (CZ ) j j = 0 then (CXZ )i j = 0, and any value of Ai j solves Ai j (CZ ) j j =


(CXZ )i j . Now that we have shown that we can always solve ACZ = CXZ , observe
that this equation is equivalent to

A(P0CY P) = CXY P or (AP0 )CY = CXY .

Thus, A = AP0 solves the original problem.


39. Since X has the form X = [Y 0 , Z 0 ]0 , if we take G = [I, 0] and W 0, then
  
I 0 Y
GX +W = = Y.
Z

40. Write  
1 n 2 1 n 1 n
E Xk = E[Xk2 ] = 2 = 2 .
n k=1 n k=1 n k=1

41. Write  
1 n 1 n 1 n
E Xk Xk0
n k=1
=
n k=1
E[Xk Xk0 ] = C = C.
n k=1

42. MATLAB. Additional code:

Mn = mean(X,2)
MnMAT = kron(ones(1,n),Mn);
Chat = (X-MnMAT)*(X-MnMAT)/(n-1)

43. We must first find fY |X (y|x). To this end, use substitution and independence to write

P(Y y|X = x) = P(X +W y|X = x) = P(x +W y|X = x) = P(W y x).

Then fY |X (y|x) = fW (y x) = ( /2)e |yx| . For fixed y, the maximizing value of x


is x = y. Hence, gML (y) = y.
44. By the same argument as in the solution of Problem 43, fY |X (y|x) = ( /2)e |yx| .
When X exp( ) and we maximize over x, we must impose the constraint x 0.
Hence, 
y, y 0,
gML (y) = argmax 2 e |yx| =
x0 0, y < 0.
Chapter 8 Problem Solutions 149

When X uniform[0, 1],



y, 0 y 1,
gML (y) = argmax 2 e |yx| = 1, y > 1,
0x1
0, y < 0.

45. By the same argument as in the solution of Problem 43, fY |X (y|x) = ( /2)e |yx| .
For the MAP estimator, we must maximize
|yx|
fY |X (y|x) fX (x) = 2e e x , x 0.
By considering separately the cases x y and x > y,
(
y e( )x , 0 x y,
2 e
fY |X (y|x) fX (x) =
y ( + )x
2 e e , x > y, x 0.
When y 0, observe that the two formulas agree at x = y and have the common value
( /2)e y ; in fact, if > , the first formula is maximized at x = y, while the
second formula is always maximized at x = y. If y < 0, then only the second formula
is valid, and its region of validity is x 0. This formula is maximized at x = 0. Hence,
for > , 
y, y 0,
gMAP (y) =
0, y < 0.
We now consider the case . As before, if y < 0, the maximizing value of x is
zero. If y 0, then the maximum value of fY |X (y|x) fX (x) for 0 x y occurs at
x = 0 with a maximum value of ( /2)e y . The maximum value of fY |X (y|x) fX (x)
for x y occurs at x = y with a maximum value of ( /2)e y . For < ,

max{( /2)e y , ( /2)e y } = ( /2)e y ,


which corresponds to x = 0. Hence, for < ,
gMAP (y) = 0, < y < .

46. From the formula


2 /2
fXY (x, y) = (x/y2 )e(x/y) e y , x, y > 0,
we see that Y exp( ), and
p that given Y = y, X Rayleigh(y). Hence, the MMSE
estimator is E[X|Y = y] = /2 y. To compute the MAP estimator, we must solve
2 /2
argmax(x/y2 )e(x/y) .
x0

We do this by differentiating with respect to x and setting the derivative equal to zero.
Write  
2
2 (x/y)2 /2 e(x/y) /2 x2
(x/y )e = 1 2 .
x y2 y
Hence,
gMAP (y) = y.
150 Chapter 8 Problem Solutions

47. Suppose that

E[(X g1 (Y ))h(Y )] = 0 and E[(X g2 (Y ))h(Y )] = 0.

Subtracting the second equation from the first yields

E[{g2 (Y ) g1 (Y )}h(Y )] = 0. ()

Since h is an arbitrary bounded function, put h(y) := sgn[g2 (y) g1 (y)], where

1, x > 0,
sgn(x) := 0, x = 0,

1, x < 0.

Note also that x sgn(x) = |x|. Then () becomes E[|g2 (Y ) g1 (Y )|] = 0.


CHAPTER 9
Problem Solutions

1. We first compute
 
12 1 2
detC = det = 12 22 (1 2 )2 = 12 22 (1 2 ),
1 2 22
p
and detC = 1 2 1 2 . Next,

1
 
1 22 1 2 2 (1 2 ) 1 2 (1 2 )
C1 = =

1
1
,

detC 1 2 12
2 2
1 2 (1 ) 2 (1 )2

and
x y
 
  x   (1 2 ) 1 2 (1 )
2 2
x y C1 = x y 1 x y
y + 2
2
1 2 (1 ) 1 (1 )
2

x2 xy xy y2
= +
12 (1 2 ) 1 2 (1 2 ) 1 2 (1 2 ) 12 (1 2 )
1 h x 2 xy  y 2 i
= 2 + ,
1 2 1 1 2 2
and the result follows.
2. Here is the plot:

n=4
1 n=1

0
4 2 0 2 4
3. (a) First write c1 X + c2Y = c1 X + c2 (3X) = (c1 + 3c2 )X, which is easily seen to be
N(0, (c1 + 3c2 )2 ). Thus, X and Y are jointly Gaussian.
(b) Observe that E[XY ] = E[X(3X)] = 3E[X 2 ] = 3 and E[Y 2 ] = E[(3X)2 ] = 9E[X 2 ] =
9. Since X and Y have zero means,
   
E[X 2 ] E[XY ] 1 3
cov([X,Y ]0 ) = = .
E[Y X] E[Y 2 ] 3 9

151
152 Chapter 9 Problem Solutions

(c) The conditional cdf of Y given X = x is

FY |X (y|x) = P(Y y|X = x) = P(3X y|X = x).

By substitution, this last conditional probability is P(3x y|X = x). The event
{3x y} is deterministic and therefore independent of X. Hence, we can drop
the conditioning and get

FY |X (y|x) = P(3x y).

If 3x y, then {3x y} = , and {3x y} = otherwise. Hence, the above


probability is just u(y 3x).
4. If Y = ni=1 ci Xi , its characteristic function is
 n  n n
j (ni=1 ci Xi ) j ci Xi
Y ( ) = E[e ] = E e = E[e j ci Xi ] = Xi ( ci )
i=1 i=1 i=1
n
2 2 n 2 ( n c2 2 )/2
= e j( ci )mi ( ci ) i /2 = e j (i=1 ci mi ) i=1 i i ,
i=1

which is the characteristic function of an N(ni=1 ci mi , ni=1 c2i i2 ) random variable.


5. First, E[Y ] = E[AX + b] = AE[X] + b = Am + b. Second,

E[{Y (Am + b)}{Y (Am + b)}0 ] = E[A(X m)(X m)0 A0 ] = ACA0 .

6. Write out

Y1 = X1
Y2 = X1 + X2
Y3 = X1 + X2 + X3
..
.

In general, Yn = Yn1 + Xn , or Xn = Yn Yn1 , which we can write in matrix-vector


notation as

X1 1 0 0 Y1
X2 1 1 0 0
Y2
X3 0 1 1 0 0 Y3


.. = .. .. .
. . .

Xn1 0 0 1 1 0 Yn1
Xn 0 0 1 1 Yn
| {z }
=: A

Since X = AY and Y is Gaussian, so is X.


7. Since X N(0,C), the scalar Y := 0 X is also Gaussian and has zero mean. Hence,
E[( 0 XX 0 )k ] = E[(Y 2 )k ] = E[Y 2k ] = (2k 1) 5 3 1 (E[Y 2 ])k . Now observe that
E[Y 2 ] = E[ 0 XX 0 ] = 0C and the result follows.
Chapter 9 Problem Solutions 153

8. We first have
 
1 n 1 n
E[Y j ] = E[X j X] = m E Xi
n i=1
= m m = 0.
n i=1

Next, since E[XY j ] = E[X(X j X)], we first compute


 n 
1 n n 1 2 2 2
E[ X X ] = 2 E[Xi X j ] = 2 ( + m ) + m
n i=1 j=1 n i=1 i6= j

1 n o 1 n o 2
= 2 n( 2 + m2 ) + n(n 1)m2 = 2 n 2 + n2 m2 = + m2 ,
n n n
and
1 n 1n 2 2 2
o 2
E[ X X j ] = E[X X
i j ] = ( + m ) + (n 1)m = + m2 .
n i=1 n n

It now follows that E[ X Y j ] = 0.


9. Following the hint, in the expansion of E[( 0 X)2k ], the sum of all the coefficients of
i1 i2k is (2k)!E[Xi1 Xi2k ]. The corresponding sum of coeficients in the expan-
sion of
(2k 1)(2k 3) 5 3 1 ( 0C )k
is
(2k 1)(2k 3) 5 3 1 2k k! C j1 j2 C j2k1 j2k ,
j1 ,..., j2k

where the sum is over all j1 , . . . , j2k that are permutations of i1 , . . . , i2k and such that
the product C j1 j2 C j2k1 j2k is distinct. Since

(2k 1)(2k 3) 5 3 1 2k k! = (2k 1)(2k 3) 5 3 1 (2k)(2k 2) 4 2


= (2k)!,

Wicks Theorem follows.


10. Write

E[X1 X2 X3 X4 ] = C j1 j2 C j3 j4 = C12C34 +C13C24 +C14C23 .


j 1 , j2 , j3 , j4

E[X1 X32 X4 ] = C13C34 +C14C33 .


E[X12 X22 ] = C11C22 +C12
2
.

11. Put a := [a1 , . . . , an ]0 . Then Y = a0 X, and


0 0
Y ( ) = E[e jY ] = E[e j (a X) ] = E[e j( a) X ] = X ( a)
0 0 0 0 2 /2
= e j( a) m( a) C( a)/2 = e j (a m)(a Ca) ,

which is the characteristic function of a scalar N(a0 m, a0Ca) random variable.


154 Chapter 9 Problem Solutions

0 0
12. With X = [U 0 ,W 0 ]0 and = [ 0 , 0 ]0 , we have X ( ) = e j m C /2 , where
 
  mU
0m = 0 0 = 0 mU + 0 mW ,
mW

and
    
0
  S 0  0 0  S
C = 0 0 = = 0 S + 0 T .
0 T T

Thus,
0 0 m )( 0 S + 0 T )/2 0 0 0 m 0 T /2
X ( ) = e j( mU + W = e j mU S /2 e j W ,

which has the required form U ( )W ( ) of a product of Gaussian characteristic


functions.
13. (a) Since X is N(0,C) and Y := C1/2 X, Y is also normal. It remains to find the
mean and covariance of Y . We have E[Y ] = E[C1/2 X] = C1/2 E[X] = 0 and
E[YY 0 ] = E[C1/2 XX 0C1/2 ] = C1/2 E[XX 0 ]C1/2 = C1/2CC1/2 = I. Hence,
Y N(0, I).
(b) Since the covariance matrix of Y is diagonal, the components of Y are uncorre-
lated. Since Y is also Gaussian, the components of Y are independent. Since the
covariance matrix of Y is the indentity, each Yk N(0, 1). Hence, each Yk2 is chi-
squared with one degree of freedom by Problem 46 in Chapter 4 or Problem 11
in Chapter 5.
(c) By the Remark in Problem 55(c) in Chapter 4, V is chi-squared with n degrees
of freedom.
14. Since  
X Y
Z := det = X 2 +Y 2 ,
Y X

where X and Y are independent N(0, 1), observe that X 2 and Y 2 are chi-squared with
one degree of freedom. Hence, Z is chi-squared with two degrees of freedom, which
is the same as exp(1/2).
15. Begin with
Z Z
1 0 0 0 1 0 0
fX (x) = e j x e j m C /2 d = e j(xm) e C /2 d .
(2 )n IRn (2 )n IRn

Now make the multivariate change of variable = C1/2 , d = detC1/2 d . Then


Z
1 0 1/2 0 d
fX (x) = e j(xm) C e /2
(2 )n IRn detC1/2
Z
1 1/2 0 0 d
= e j{C (xm)} e /2 .
(2 )n IRn detC
Chapter 9 Problem Solutions 155

Put t = C1/2 (x m) so that


Z n  Z 
1 0 0 /2 d 1 1 jti i 2 /2
fX (x) = e jt e = e e i d i .
(2 )n IRn detC detC i=1 2
2
Observe that ei /2 is the characteristic function of a scalar N(0, 1) random variable.
Hence,
n ti2 /2
1 e 1
fX (x) = = exp[t 0t/2].
detC i=1 2 (2 )n/2 detC
Recalling that t = C1/2 (x m) yields

exp[ 12 (x m)0C1 (x m)]


fX (x) = .
(2 )n/2 detC

16. First observe that  


X Y
Z := det = XV YU.
U V
Then consider the conditional cumulative distribution function,
FZ|UV (z|u, v) = P(Z z|U = u,V = v) = P(XV YU z|U = u,V = v)
= P(Xv Yu z|U = u,V = v).
Since [X,Y ]0 and [U,V ]0 are jointly Gaussian and uncorrelated, they are independent.
Hence, we can drop the conditioning and get
FZ|UV (z|u, v) = P(Xv Yu z).

Next, since X and Y are independent and N(0, 1), Xv Yu N(0, u2 + v2 ). Hence,
fZ|UV ( |u, v) N(0, u2 + v2 ).

17. We first use the fact that A solves ACY = CXY to show that (X mX ) A(Y mY ) and
Y are uncorrelated. Write
E[{(X mX ) A(Y mY )}Y 0 ] = CXY ACY = 0.
We next show that (X mX ) A(Y mY ) and Y are jointly Gaussian by writing them
as an affine transformation of the Gaussian vector [X 0 ,Y 0 ]0 ; i.e.,
      
(X mX ) A(Y mY ) I A X AmY mX
= + .
Y 0 I Y 0
It now follows that (X mX ) A(Y mY ) and Y are independent. Using the hints on
substitution and independence, we compute the conditional characteristic function,
0  0 0 
E[e X |Y = y] = E e j [(XmX )A(Y mY )] e j [mX +A(Y mY )] Y = y
0  0 
= e j [mX +A(ymY )] E e j [(XmX )A(Y mY )] Y = y
0  0 
= e j [mX +A(ymY )] E e j [(XmX )A(Y mY )] .
156 Chapter 9 Problem Solutions

This last expectation is the characteristic function of the zero-mean Gaussian random
vector (X mX ) A(Y mY ). To compute its covariance matrix first observe that
since ACY = CXY , we have ACY A0 = CXY A0 . Then
E[{(X mX ) A(Y mY )}{(X mX ) A(Y mY )}0 ]
= CX CXY A0 ACY X + ACY A0
= CX ACY X .
We now have
0 0 0
E[e X |Y = y] = e j [mX +A(ymY )] e [CX ACY X ] /2 .
$ 
Thus, given Y = y, X is conditionally N mX + A(y mY ),CX ACY X .
18. First observe that  
X Y
Z := det = XV YU.
U V
Then consider the conditional cumulative distribution function,
FZ|UV (z|u, v) = P(Z z|U = u,V = v) = P(XV YU z|U = u,V = v)
= P(Xv Yu z|U = u,V = v).
Since [X,Y,U,V ]0 is Gaussian, given U = u and V = v, [X,Y ]0 is conditionally
   
u
N A ,C[X,Y ]0 AC[U,V ]0 ,[X,Y ]0 ,
v
where A solves AC[U,V ]0 = C[X,Y ]0 ,[U,V ]0 . We now turn to the conditional distribution of
Xv Yu. Since the conditional distribution of [X,Y ]0 is Gaussian, so is the conditional
distribution of the linear combination Xv Yu. Hence, all we need to find are the
conditional mean and the conditional variance of Xv Yu; i.e.,
   
  X
E[Xv Yu|U = u,V = v] = E v u U = u,V = v
Y
  
  X
= v u E U = u,V = v
Y
 
  u
= v u A ,
v
and
    2 
  X u
A
E v u
Y v U = u,V = v
    2 
  X U
= E v u A U = u,V = v
Y V
      0  
  X U X U
= v u E A A U = u,V = v v
Y V Y V u
 
  
v
= v u C[X,Y ]0 AC[U,V ]0 ,[X,Y ]0 ,
u
Chapter 9 Problem Solutions 157

where the last step uses the fact that [X,Y ]0 A[U,V ]0 is independent of [U,V ]0 . If
[X,Y ]0 and [U,V ]0 are uncorrelated, i.e., C[X,Y ]0 ,[U,V ]0 = 0, then A = 0 solves AC[U,V ]0 =
0; in this case, the conditional mean is zero, and the conditional variance simplifies to
   
  v   v
v u C[X,Y ]0 = v u I = v2 + u2 .
u u

19. First write

Z E[Z] = (X + jY ) (mX + jmY ) = (X mX ) + j(Y mY ).

Then

cov(Z) := E[(Z E[Z])(Z E[Z]) ]


= E[{(X mX ) + j(Y mY )}{(X mX ) + j(Y mY )} ]
= E[{(X mX ) + j(Y mY )}{(X mX ) j(Y mY )}]
= var(X) j cov(X,Y ) + j cov(Y, X) + var(Y ) = var(X) + var(Y ).

20. (a) Write

K := E[(Z E[Z])(Z E[Z])H ]


= E[{(X mX ) + j(Y mY )}{(X mX ) + j(Y mY )}H ]
= E[{(X mX ) + j(Y mY )}{(X mX )H j(Y mY )H }]
= CX jCXY + jCY X +CY = (CX +CY ) + j(CY X CXY ).

(b) If CXY = CY X , the (CXY )ii = (CY X )ii implies

E[(Xi (mX )i )(Yi (mY )i )] = E[(Yi (mY )i )(Xi (mX )i )]


= E[(Xi (mX )i )(Yi (mY )i )].

Hence, E[(Xi (mX )i )(Yi (mY )i )] = 0, and we see that Xi and Yi are uncorre-
lated.
(c) By part (a), if K is real, then CY X = CXY . Circular symmetry implies CY X =
CXY . It follows that CXY = CXY , and then CXY = 0; i.e., X and Y are uncor-
related.
21. First,
0 0
ex (2I)x/2 ey (2I)y/2
fX (x) = and fY (y) = .
(2 )n/2 (1/2)n/2 (2 )n/2 (1/2)n/2
Then
0 0 y) H
e(x x+y e(x+ jy) (x+ jy)
fXY (x, y) = fX (x) fY (y) = = .
n n
22. (a) Immediate from Problem 20(a).
(b) Since 0 Q is a scalar, ( 0 Q )0 = 0 Q . Since Q0 = Q, ( 0 Q )0 = 0 Q0 =
0 Q . Thus, 0 Q = 0 Q , and it follows that 0 Q = 0.
158 Chapter 9 Problem Solutions

(c) Begin by observing that if K = R + jQ and w = + j , then


Kw = (R + jQ)( + j ) = R + jR + jQ Q .
Next,
wH (Kw) = ( 0 j 0 )(R + jR + jQ Q )
= 0 R + j 0 R + j 0 Q 0 Q j 0 R + 0 R + 0 Q j 0 Q
= 0 R + j 0 R + j 0 Q + 0 Q j 0 R + 0 R + 0 Q j 0 Q
= 0 R + 0 R + 2 0 Q ,
where we have used the result of parts (a) and (b).
23. First,
AZ = ( + j )(X + jY ) = ( X Y ) + j( X + Y ).
Second,     
X X Y
= .
Y X + Y
Now assume that circular symmetry holds; i.e., CX = CY and CXY = CY X . Put
U := X Y and V := X + Y . Assuming zero means to simplify the notation,
CU = E[( X Y )( X Y )0 ] = CX 0 CY X 0 CXY 0 + 0CY
= CX 0 CY X 0 + CY X 0 + 0CX .
Similarly,
CV = E[( X + Y )( X + Y )0 ] = CX 0 + CY X 0 + CXY 0 + CY 0
= CX 0 + CY X 0 CY X 0 + CX 0 .
Hence, CU = CV . It remains to compute
CUV = E[( X Y )( X + Y )0 ] = CX 0 + CXY 0 CY X 0 CY 0
= CX 0 CY X 0 CY X 0 CX 0
and
CVU = E[( X + Y )( X Y )0 ] = CX 0 CXY 0 + CY X 0 CY 0
= CX 0 + CY X 0 + CY X 0 CX 0 ,
which shows that CU = CV . Thus, if Z is circularly symmetric, so is AZ.
24. To begin, note that with R = [X 0 ,U 0 ]0 and I = [Y 0 ,V 0 ]0 ,
   
CX CXU CY CYV
CR = , CI = ,
CUX CU CVY CV
and    
CXY CXV CY X CYU
CRI = , CIR = .
CUY CUV CV X CVU
Also, is circularly symmetric means CR = CI and CRI = CIR .
Chapter 9 Problem Solutions 159

(a) We assume zero means to simplify the notation. First,

KZW = E[ZW H ] = E[(X + jY )(U + jV )H ] = E[(X + jY )(U H jV H )]


= CXU jCXV + jCYU +CYV
= 2(CXU jCXV ), since is circularly symmetric.

Second,
   
CXU CXV CXU CXV
CZeWe = = , since is circularly symmetric.
CYU CYV CXV CXU

It is now clear that KZW = 0 if and only if CZeWe = 0.


(b) Assuming zero means again, we compute

KW = E[WW H ] = E[(U + jV )(U + jV )H ] = E[(U + jV )(U H jV H )]


= CU jCUV + jCVU +CV = 2(CU jCUV ).

We now see that AKW = KZW becomes

2( + j )(CU jCUV ) = 2(CXU jCXV )

or
( CU + CUV ) + j( CU CUV ) = CXU jCXV . ()
We also have  
CU CUV
CWe =
CVU CV
e e = C e e becomes
so that ACW ZW
  
CU CUV
= CZeWe
CVU CV
or  
CU CVU CUV CV
= CZeWe
CU + CVU CUV + CV
or    
CU + CUV CUV CU CXU CXV
= ,
CU CUV CUV + CU CXV CXU
which is equivalent to ().
(c) If A solves AKW = KZW , then by part (b), A e solves AC
e e = C e e . Hence, by
W ZW
e = w,
Problem 17, given W e
 
ew
Ze N mZe + A( e ee .
e mWe ),CZe ACWZ

e e e is equivalent to
Next, CZe ACWZ
    
CX CXY CUX CUY
,
CY X CY CV X CVY
160 Chapter 9 Problem Solutions

which, by the circular symmetry of , becomes


    
CX CXY CUX CUY
,
CXY CX CUY CUX
or    
CX CXY CUX + CUY CUY CUX
,
CXY CX CUX CUY CUY + CUX
which is equivalent to
2(CX jCXY ) ( + j ) 2(CUX jCUY ),
which is exactly KZ AKW Z . Thus, given W = w,
$ 
Z N mZ + A(w mW ), KZ AKW Z .
25. Let Z = X + jY with X and Y independent N(0, 1/2) as in the text.
(a) Since X and Y are zero mean,
cov(Z) = E[ZZ ] = E[X 2 +Y 2 ] = 12 + 12 = 1.

(b) First write 2|Z|2 = 2(X 2 +Y 2 ) = ( 2 X)2 + ( 2Y )2 . Now, 2 X and 2Y are
both N(0, 1). Hence, their squares are chi-squared with one degree of freedom
by Problem 46 in Chapter 4 or Problem 11 in Chapter 5. Hence, by Prob-
lem 55(c) in Chapter 4 and the remark following it, 2|Z|2 is chi-squared with
two degrees of freedom.
26. With X N(mr , 1) and Y N(mi , 1), it follows either from Problem 47 in Chapter 4
or from Problem 12 in Chapter 5 that X 2 and Y 2 are noncentral chi-squared with one
degree of freedom and respective noncentrality parameters m2r and m2i . Since X and Y
are independent, it follows from Problem 65 in Chapter 4 that X 2 +Y 2 is noncentral
chi-squared with two degrees of freedom and noncentralityparameter m2r + m2i . It is
now immediate from Problem 26 in Chapter 5 that |Z| = X 2 +Y 2 has the orginal
Rice density.
27. (a) The covariance matrix of W is
E[WW H ] = E[K 1/2 ZZ H K 1/2 ] = K 1/2 E[ZZ H ]K 1/2 = K 1/2 KK 1/2 = I.
Hence,
H 2
ew w n
e|wk |
fW (w) =
n
= .
k=1

(b) By part (a), the Wk = Uk + jVk are i.i.d. N(0, 1) with


2 2 2
2
2
e|wk | e(uk +vk ) e[(uk /(1/ 2 )) +(vk /(1/ 2 )) ]/2
fWk (w) = = $p 2 = $p 2
2 /2 2 /2
2 /2
2 /2
e[uk /(1/ 2 )] e[vk /(1/ 2 )]
= p p = fUkVk (uk , vk ).
2 /2 2 /2
Hence, Uk and Vk are independent N(0, 1/2).
Chapter 9 Problem Solutions 161

(c) Write
n
2kW k2 = ( 2Uk )2 + ( 2Vk )2 .
k=1

Since 2Uk and 2Vk are independent N(0, 1), their squares are chi-squared
with one degree
of freedom
by Problem 46 in Chapter 4 or Problem 11 in Chap-
ter 5. Next, ( 2Uk )2 + ( 2Vk )2 is chi-squared with two degrees of freedom by
Problem 55(c) in Chapter 4 and the remark following it. Similarly, since the Wk
are indepdendent, 2kW k2 is chi-squared with 2n degrees of freedom.
28. (a) Write

0 = (u + v)0 M(u + v) = u0 Mu + v0 Mu + u0 Mv + v0 Mv = 2v0 Mu,

since M 0 = M. Hence, v0 Mu = 0.
(b) By part (a) with v = Mu we have

0 = v0 Mu = (Mu)0 Mu = kMuk2 .

Hence, Mu = 0 for all u, and it follows that M must be the zero matrix.
29. We have from the text that
  
  CX CXY
0 0
CY X CY
is equal to
0CX + 0CXY + 0CY X + 0CY ,
which, upon noting that 0CXY is a scalar and therefore equal to its transpose, sim-
plifies to
0CX + 2 0CY X + 0CY . ()
We also have from the text (via Problem 22) that

wH Kw = 0 (CX +CY ) + 0 (CX +CY ) + 2 0 (CY X CXY ) .

If () is equal to wH Kw/2 for all and all , then in particular, this must hold for all
when = 0. This implies
CX +CY CX CY
0CX = 0 or 0 = 0.
2 2
Since is arbitrary, (CX CY )/2 = 0, or CX = CY . This means that we can now write

wH Kw/2 = 0CX + 0CY + 0 (CY X CXY ) .

Comparing this with () shows that

2 0CY X = 0 (CY X CXY ) or 0 (CY X +CXY ) = 0.

Taking = arbitrary and noting that CY X +CXY is symmetric, it follows that CY X +


CXY = 0, and so CXY = CY X .
162 Chapter 9 Problem Solutions

30. (a) Since is 2n 2n, det(2) = 22n det . From the hint it follows that det =
(det K)2 /22n .
(b) Write
VV 1 = (A + BCD)[A1 A1 B(C1 + DA1 B)1 DA1 ]
= (A + BCD)A1 [I B(C1 + DA1 B)1 DA1 ]
= (I + BCDA1 )[I B(C1 + DA1 B)1 DA1 ]
= I + BCDA1 B(C1 + DA1 B)1 DA1
BCDA1 B(C1 + DA1 B)1 DA1
= I + BCDA1
B[I +CDA1 B](C1 + DA1 B)1 DA1
= I + BCDA1
BC[C1 + DA1 B](C1 + DA1 B)1 DA1
= I + BCDA1 BCDA1 = I.

(c) To begin, write


  
CX CY X 1 CX1CY X 1
1 = 1 1
CY X CX CY X CX 1
 
CX 1 +CY X 1CY X CX1 CY X 1 CY X 1
= 1 1 1
CY X CX CY X CX CY X CX1CY X 1 +CX 1
 
CX 1 +CY X 1CY X CX1 0
=
CY X 1 CX 1CY X CX1 (CY X CX1CY X +CX )1
 
CX 1 +CY X 1CY X CX1 0
= .
CY X 1 CX 1CY X CX1 I
Using the hint that
1 = CX1 CX1CY X 1CY X CX1 ,
we easily obtain
CX 1 = I CY X 1CY X CX1 ,
from which it follows that
 
1 I 0
= .
CY X 1 CX 1CY X CX1 I
To show that the lower-left block is also zero, use the hint to write
CY X 1 CX 1CY X CX1
= CY X [CX1 CX1CY X 1CY X CX1 ] CX 1CY X CX1
= CY X CX1 CY X CX1CY X 1CY X CX1 CX 1CY X CX1
= CY X CX1 [CY X CX1CY X +CX ]1CY X CX1
= CY X CX1 1CY X CX1 = 0.
Chapter 9 Problem Solutions 163

(d) Write
KK 1 = 2(CX + jCY X )(1 jCX1CY X 1 )/2
= CX 1 +CY X CX1CY X 1 + j(CY X 1 CY X 1 )
= 1 = I.

(e) We begin with


    
  1 x  0 0 1 CX1CY X 1 x
x0 y0 = x y
y 1CY X CX1 1 y
 1 1

  x +CX CY X 1 y
= x0 y0
1CY X CX1 x + 1 y
= x 0 1 x + x 0CX1CY X 1 y y 0 1CY X CX1 x + y 0 1 y.
Now, since each of the above terms on the third line is a scalar, each term is
equal to its transpose. In particular,
y 0 1CY X CX1 x = x 0CX1CXY 1 y = x 0CX1CY X 1 y.
Hence,
 
  x
1
2 x 0 y 0 1 = 1 0 1 0 1 1 0 1
2 (x x + 2x CX CY X y + y y). ()
y
We next compute
zH K 1 z = 1 0
2 (x jy 0 )(1 jCX1CY X 1 )(x + jy)
1 0
= 2 (x jy 0 )[(1 x +CX1CY X 1 y) + j(1 y CX1CY X 1 x)]
1 0
= 2 (x jy 0 )[(1 x +CX1CY X 1 y) + j(1 y 1CY X CX1 x)]

by the hint that CX1CY X 1 = 1CY X CX1 . We continue with

zH K 1 z = 1 0 1 1 1 0 1
2 [{x ( x +CX CY X y) + y ( y CY X CX x)}
1 1

+ j{x 0 (1 y 1CY X CX1 x) y 0 (1 x +CX1CY X 1 y)}]


1 0 1 0 1 1 0 1 0 1 1
= 2 [{x x + x CX CY X y + y y y CY X CX x}
+ j{x 0 1 y x 0 1CY X CX1 x y 0 1 x y 0CX1CY X 1 y}].
We now use the fact that since each of the terms in the last line is a scalar, it is
equal to its transpose. Also CXY = CY X . Hence,
zH K 1 z = 1 0 1 0 1 1
2 [{x x + 2x CX CY X y + y y}
0 1

j{x 0 1CY X CX1 x + y 0CX1CY X 1 y}].


Since
x 0 1CY X CX1 x = (x 0 1CY X CX1 x)0 = x 0CX1CY X 1 x = x 0 1CY X CX1 x,
and similarly for y 0CX1CY X 1 y, the two imaginary terms above are zero.
CHAPTER 10
Problem Solutions

1. Write

mX (t) := E[Xt ] = E[g(t, Z)]


= g(t, 1)P(Z = 1) + g(t, 2)P(Z = 2) + g(t, 3)P(Z = 3)
= p1 a(t) + p2 b(t) + p3 c(t),

and

RX (t, s) := E[Xt Xs ] = E[g(t, Z)g(s, Z)]


= g(t, 1)g(s, 1)p1 + g(t, 2)g(s, 2)p2 + g(t, 3)g(s, 3)p3
= a(t)a(s)p1 + b(t)b(s)p2 + c(t)c(s)p3 .

2. Imitating the derivation of the CauchySchwarz inequality for random variables in


Chapter 2 of the text, write
Z
0 |g( ) h( )|2 d

Z Z
= 2
|g( )| d h( )g( ) d

Z Z
g( )h( ) d + | |2 |h( )|2 d .

Then put R
g( )h( ) d
= R
2
|h( )| d
to get
R 2

Z g( )h( ) d
0 |g( )|2 d R
2
|h( )| d
R 2 R 2

g( )h( ) d g( )h( ) d Z
R 2
+ R 2 |h( )|2 d
|h( )| d 2
|h( )| d
R 2

Z g( )h( ) d
2
= |g( )| d R 2
.
|h( )| d

Rearranging yields the desired result.

164
Chapter 10 Problem Solutions 165

3. Write

CX (t1 ,t2 ) = E[(Xt1 mX (t1 ))(Xt2 mX (t2 ))]


= E[Xt1 Xt2 ] mX (t1 )E[Xt2 ] E[Xt1 ]mX (t2 ) + mX (t1 )mX (t2 )
= E[Xt1 Xt2 ] mX (t1 )mX (t2 ) mX (t1 )mX (t2 ) + mX (t1 )mX (t2 )
= RX (t1 ,t2 ) mX (t1 )mX (t2 ).

Similarly,

CXY (t1 ,t2 ) = E[(Xt1 mX (t1 ))(Yt2 mY (t2 ))]


= E[Xt1 Yt2 ] mX (t1 )E[Yt2 ] E[Xt1 ]mY (t2 ) + mX (t1 )mY (t2 )
= E[Xt1 Xt2 ] mX (t1 )mY (t2 ) mX (t1 )mY (t2 ) + mX (t1 )mY (t2 )
= RXY (t1 ,t2 ) mX (t1 )mY (t2 ).

4. Write
 n 2   n  n   n n

0 E ci Xti = E ci Xti k tk
c X = ci E[Xti Xtk ]ck
i=1 i=1 k=1 i=1 k=1
n n
= ci RX (ti ,tk )ck .
i=1 k=1

5. Since Xt has zero mean, var(Xt ) = RX (t,t) = t. Thus, Xt N(0,t), and


2
ex /(2t)
fXt (x) = .
2 t

6. First note that by making the change of variable k = n i, we have



Yn = h(k)Xnk .
k=

(a) Write
 
mY (n) = E[Yn ] = E h(k)Xnk = h(k)E[Xnk ]
k= k=

= h(k)mX (n k).
k=

(b) Write
  
E[XnYm ] = E Xn h(k)Xmk = h(k)E[Xn Xmk ]
k= k=

= h(k)RX (n, m k).
k=
166 Chapter 10 Problem Solutions

(c) Write
  
E[YnYm ] = E h(l)Xnl Ym = h(l)E[Xnl Ym ]
l= l=
 
= h(l)RXY (n l, m) = h(l) h(k)RX (n l, m k) .
l= l= k=

7. Let Xt = cos(2 f t + ), where uniform[ , ].


(a) Consider choices t1 = 0 and t2 = ( /2)/(2 f ). Then Xt1 = cos() and Xt2 =
sin(), which are not jointly continuous.
(b) Write
Z
d
E[g(Xt )] = E[g(cos(2 f t + ))] = g(cos(2 f t + ))
2
Z +2 f t Z
d d
= g(cos( )) = g(cos( )) ,
+2 f t 2 2
since the integrand has period 2 and the range of integration has length 2 .
Thus, E[g(Xt )] does not depend on t.
8. (a) Using independence and a trigonometric identity, write
E[Yt1 Yt2 ] = E[Xt1 Xt2 cos(2 f t1 + ) cos(2 f t2 + )]
= E[Xt1 Xt2 ]E[cos(2 f t1 + ) cos(2 f t2 + )]
 
= 21 RX (t1 t2 )E cos(2 f [t1 t2 ]) + cos(2 f [t1 + t2 ] + 2)
  
= 21 RX (t1 t2 ) cos(2 f [t1 t2 ]) + E cos(2 f [t1 + t2 ] + 2)
| {z }
=0
1
= 2 RX (t1 t2 ) cos(2 f [t1 t2 ]).

(b) A similar argument yields


E[Xt1 Yt2 ] = E[Xt1 Xt2 cos(2 f t2 + )] = E[Xt1 Xt2 ] E[cos(2 f t2 + )] = 0.
| {z }
=0

(c) It is clear that Yt is zero mean. Together with part (a) it follows that Yt is WSS.
9. By Problem 7(b), FXt (x) = P(Xt x) = E[I(,x] (Xt )] does not depend on t, and so
we can restrict attention to the case t = 0. Since
X0 = cos() has the arcsine density
of Problem 35 in Chapter 5, f (x) = (1/ )/ 1 x2 for |x| < 1.
10. Second-order strict stationarity means that for every two-dimensional set B, for every
t1 ,t2 , and t, P((Xt1 +t , Xt2 +t ) B) does not depend on t. In particular, this is true
whenever B has the form B = A IR for any one-dimensional set A; i.e.,
P((Xt1 +t , Xt2 +t ) B) = P((Xt1 +t , Xt2 +t ) A IR) = P(Xt1 A, Xt2 IR)
= P(Xt1 A).
does not depend on t. Hence Xt is first-order strictly stationary.
Chapter 10 Problem Solutions 167

11. (a) If p1 = p2 = 0 and p3 = 1, then Xt = c(t) = 1 with probability one. Then


E[Xt ] = 1 does not depend on t, and E[Xt1 Xt2 ] = (1)2 = 1 depends on t1
and t2 only through their difference t1 t2 ; in fact the correlation function is a
constant function of t1 t2 . Thus, Xt is WSS.
(b) If p1 = 1 and p2 = p3 = 0, then Xt = e|t| with probability one. Then E[Xt ] =
e|t| depends on t. Hence, Xt is not WSS.
(c) First, the only way to have X0 = 1 is to have Xt = a(t) = e|t| , which requires
Z = 1. Hence,
P(X0 = 1) = P(Z = 1) = p1 .

Second, the only way to have Xt 0 for 0 t 0.5 is to have Xt = c(t) = 1,


which requires Z = 3. Hence,

P(Xt 0, 0 t 0.5) = P(Z = 3) = p3 .

Third, the only way to have Xt 0 for 0.5 t 1 is to have Xt = b(t) = sin(2 t)
or Xt = c(t) = 1. Hence,

P(Xt 0, 0.5 t 1) = P(Z = 2 or Z = 3) = p2 + p3 .

12. With Yk := q(Xk , Xk+1 , . . . , Xk+L1 ), write

E[e j(1Y1+m ++nYn+m ) ] = E[e j{1 q(X1+m ,...,Xm+L )++n q(Xn+m ,...,Xn+m+L1 )} ].

The exponential on the right is just a function of X1+m , . . . , Xn+L1+m . Since the Xk
process is strictly stationary, the expectation on the right is unchanged if we replace
X1+m , . . . , Xn+L1+m by X1 , . . . , Xn+L1 ; i.e., the above right-hand side is equal to

E[e j{1 q(X1 ,...,XL )++n q(Xn ,...,Xn+L1 )} ] = E[e j(1Y1 ++nYn ) ].

Hence, the Yk process is strictly stationary.


13. We begin with
Z Z
|x|
E[g(X0 )] = E[X0 I[0,) (X0 )] = xI[0,) (x) 2e dx = 1
2 x e x dx,
0

which is just 1/2 times the expectation of an exp( ) random variable. We thus see
that E[g(X0 )] = 1/(2 ). Next, for n 6= 0, we compute
Z 2 Z
ex /2 1 2
E[g(Xn )] = xI[0,) (x) dx = xex /2 dx
2 2 0
1
2  1
= ex /2 = .
2 0 2

Hence, E[g(X0 )] 6= E[g(Xn )] for n 6= 0, and it follows that Xk is not strictly stationary.
168 Chapter 10 Problem Solutions

14. First consider the mean function,


Z T0 Z t+T0 Z T0
1 1 1
E[q(t + T )] = q(t + ) d = q( ) d = q( ) d ,
T0 0 T0 t T0 0

where we have used the fact that since q has period T0 , the integral of q over any inter-
val of length T0 yields the same result. The second thing to consider is the correlation
function. Write
Z
1 T0
E[q(t1 + T )q(t2 + T )] = q(t1 + )q(t2 + ) d
T0 0
Z
1 t2 +T0
= q(t1 + t2 )q( ) d
T0 t2
Z
1 T0
= q([t1 t2 ] + )q( ) d ,
T0 0
where we have used the fact that as a function of , the product q([t1 t2 ] + )q( )
has period T0 . Since the mean function does not depend on t, and since the correlation
function depends on t1 and t2 only through their difference, Xt is WSS.
15. For arbitrary functions h, write
 
E[h(Xt1 +t , . . . , Xtn +t )] = E h q(t1 + t + T ), . . . , q(tn + t + T )
Z 
1 T0
= h q(t1 + t + ), . . . , q(tn + t + ) d
T0 0
Z
1 t+T0 
= h q(t1 + ), . . . , q(tn + ) d
T0 t
Z 
1 T0
= h q(t1 + ), . . . , q(tn + ) d ,
T0 0
where the last step follows because we are integrating a function of period T0 over an
interval of length T0 . Hence,

E[h(Xt1 +t , . . . , Xtn +t )] = E[h(Xt1 , . . . , Xtn )],

and we see that Xt is strictly stationary.


16. First write E[Yn ] = E[Xn Xn1 ] = E[Xn ] E[Xn1 ] = 0 since E[Xn ] does not depend
on n. Next, write

E[YnYm ] = E[(Xn Xn1 )(Xm Xm1 )]


= RX (n m) RX ([n 1] m) RX (n [m 1]) + RX (n m)
= 2RX (n m) RX (n m 1) RX (n m + 1),

which depends on n and m only through their difference. Hence, Yn is WSS.


2
17. From the Fourier transform table, SX ( f ) = 2 e(2 f ) /2 .
18. From the Fourier transform table, SX ( f ) = e2 | f | .
Chapter 10 Problem Solutions 169

19. (a) Since correlation functions are real and even, we can write
Z Z
SX ( f ) = RX ( )e j2 f d = RX ( ) cos(2 f ) d

Z Z
= 2 RX ( ) cos(2 f ) d = 2 Re RX ( )e j2 f d .
0 0

(b) OMITTED.
(c) The requested plot is at the left; at the right the plot is focused closer to f = 0.

2 2

1 1

0 0
6 3 0 3 6 1 0.5 0 0.5 1
(d) The requested plot is at the left; at the right the plot is focused closer to f = 0.

4 4

2 2

0 0
6 3 0 3 6 1 0.5 0 0.5 1

20. (a) RX (n) = E[Xk+n Xk ] = E[Xk Xk+n ] = RX (n).


(b) Since RX (n) is real and even, we can write

SX ( f ) = RX (n)e j2 f n = RX (n)[cos(2 f n) j sin(2 f n)]
n= n=
odd function of n
z }| {
= RX (n) cos(2 f n) j RX (n) sin(2 f n)
n= n=
| {z }
=0

= RX (n) cos(2 f n),
n=

which is a real and even function of f .


21. (a) Since correlation functions are real and even, we can write

SX ( f ) = RX (n)e j2 f n
n=
170 Chapter 10 Problem Solutions

1
= RX (0) + RX (n)e j2 f n + RX (n)e j2 f n
n=1 n=

= RX (0) + RX (n)e j2 f n + RX (k)e j2 f k
n=1 k=1

= RX (0) + 2 RX (n) cos(2 f n)
n=1

= RX (0) + 2 Re RX (n)e j2 f n .
n=1

(b) OMITTED.
(c) Here is the plot:

0
0.5 0 0.5
(d) Here is the plot:

0
0.5 0 0.5
22. Write
Z Z
h(t)e j2 f t dt = h( )e j2 f ( ) d

Z
= h( )e j2 f d

Z

= h( ) e j2 f d

Z 
j2 f
= h( )e d , since h is real,


= H( f ) .

23. We begin with SXY ( f ) = H( f ) SX ( f ) = ( j2 f ) SX ( f ) = j2 f SX ( f ). It then fol-


lows that
d d 2 2
RXY ( ) = RX ( ) = e /2 = e /2 .
d d
Chapter 10 Problem Solutions 171

Similarly, since SY ( f ) = |H( f )|2 SX ( f ) = ( j2 f )( j2 f )SX ( f ),

d2 d d 2 /2 2
RY ( ) = 2
RX ( ) = RXY ( ) = e = e /2 (1 2 ).
d d d

24. Since RX ( ) = 1/(1 + 2 ), we have from the transform table that SX ( f ) = e2 | f | .


Similarly, since h(t) = 3 sin( t)/( t), we have from the transform table that H( f ) =
3I[1/2,1/2] ( f ). We can now write

SY ( f ) = |H( f )|2 SX ( f ) = 9I[1/2,1/2] ( f ) e2 | f | = 9 e2 | f | I[1/2,1/2] ( f ).

2 /2 2
25. First note that since RX ( ) = e 2 e(2 f ) /2 .
, SX ( f ) =
 2 2 2
(a) SXY ( f ) = H( f ) SX ( f ) = e(2 f ) /2 ] 2 e(2 f ) /2 = 2 e(2 f ) .
(b) Writing
1 2 2
SXY ( f ) = 2 2e( 2) (2 f ) /2 ,
2
we have from the transform table that
1 2 1 2
RXY ( ) = e( / 2) /2 = e /4 .
2 2

(c) Write
1 2
E[Xt1 Yt2 ] = RXY (t1 t2 ) = e(t1 t2 ) /4 .
2
2 2 2
(d) SY ( f ) = |H( f )|2 SX ( f ) = e(2 f ) 2 e(2 f ) /2 = 2 e3(2 f ) /2 .
(e) Writing
1 2 2
SY ( f ) = 2 3e( 3) (2 f ) /2 ,
3
we have from the transform table that
1 2 1 2
RY ( ) = e( / 3) /2 = e /6 .
3 3

26. We have from the transform table that SX ( f ) = [sin( f )/( f )]2 . The goal is to
choose a filter H( f ) so that RY ( ) = sin( )/( ); i.e., so that SY ( f ) = I[1/2,1/2] ( f ).
Thus, the formula SY ( f ) = |H( f )|2 SX ( f ) becomes
 2
sin( f )
I[1/2,1/2] ( f ) = |H( f )|2 .
f

We therefore take
f
, | f | 1/2,
H( f ) = sin( f )
0, | f | > 1/2.
172 Chapter 10 Problem Solutions

27. Since Yt and Zt are responses of LTI systems to a WSS input, Yt and Zt are individually
WSS. If we can show that E[Yt1 Zt2 ] depends on t1 and t2 only throught their difference,
then Yt and Zt will be J-WSS. We show this to be the case. Write
Z Z 
E[Yt1 Zt2 ] = E h( )Xt1 d g( )Xt2 d

Z Z
= h( )g( )E[Xt1 Xt2 ] d d

Z Z
= h( )g( )RX ([t1 ] [t2 ]) d d ,

which depends on t1 t2 as claimed.


R
28. Observe that if h(t) := (t) (t 1), then h( )Xt d = Xt Xt1 .
(a) Since Yt := Xt Xt1 is the response of an LTI system to a WSS input, Xt and Yt
is J-WSS.
(b) Since H( f ) = 1 e j2 f ,

|H( f )|2 = (1 e j2 f )(1 e j2 f ) = 2 e j2 f e j2 f


e j2 f + e j2 f
= 22 = 2[1 cos(2 f )],
2
we have
2 4[1 cos(2 f )]
SY ( f ) = |H( f )|2 SX ( f ) = 2[1 cos(2 f )] = .
1 + (2 f )2 1 + (2 f )2
Rt
29. In Yt = t3 X d , make the change of variable = t , d = d to get
Z 3 Z
Yt = Xt d = I[0,3] ( )Xt d .
0

This shows that Yt is the response to Xt of the LTI system with impulse h(t) = I[0,3] (t).
Hence, Yt is WSS.
30. Apply the formula Z 

H(0) = h(t)e j2 f t dt
f =0

with h(t) = (1/ )[(sint)/t]2 . Then from the table, H( f ) = (1 | f |)I[1/ ,1/ ] ( f ),
and we find that
Z  Z 
1 sint 2 sint 2
1 = dt, which is equivalent to = dt.
t t

31. (a) Following the hint, we have



E[XnYm ] = h(k)RX (n, m k) = h(k)RX (n m + k).
k= k=
Chapter 10 Problem Solutions 173

(b) Similarly,
 
E[YnYm ] = h(l) h(k)RX (n l, m k)
l= k=
 
= h(l) h(k)RX (n l [m k])
l= k=
 
= h(l) h(k)RX ([n m] [l k]) .
l= k=

(c) By part (a),



RX (n) = h(k)RX (n + k),
k=
and so
 
j2 f n
SXY ( f ) = RX (n)e = h(k)RX (n + k) e j2 f n
n= n= k=
 
j2 f n
= h(k) RX (n + k)e
k= n=
 
= h(k) RX (m)e j2 f (mk)
k= m=
  
j2 f k j2 f m
= h(k)e RX (m)e
k= m=
 
= h(k)e j2 f k SX ( f ), since h(k) is real,
k=

= H( f ) SX ( f ).

(d) By part (b),


 
RY (n) = h(l) h(k)RX (n [l k]) ,
l= k=

and so
  
SY ( f ) = h(l) h(k)RX (n [l k]) e j2 f n
n= l= k=

= h(l) h(k) RX (n [l k])e j2 f n
l= k= n=

= h(l) h(k) RX (m)e j2 f (m+[lk])
l= k= m=

= h(l)e j2 f l h(k)e j2 f k RX (m)e j2 f m
l= k= m=

= H( f )H( f ) SX ( f ) = |H( f )|2 SX ( f ).


174 Chapter 10 Problem Solutions

32. By the hint,


Z Z
1 T nE0 1
|x(t)|2 dt = + |x(t)|2 dt.
2T 0 2T 2T 0
We first observe that since
T T0
T0 T0 + = T0 + T0 ,
n n n
T /n T0 . It follows that
nE0 E0 E0
= .
2T 2(T /n) 2T0
We next show that the integral on the right goes to zero. Write
Z Z T0
1 1 E0
|x(t)|2 dt |x(t)|2 dt = 0.
2T 0 2T 0 2T
A similar argument shows that
Z 0
1
|x(t)|2 dt E0 /(2T0 ).
2T T
RT
Putting this all together shows that (1/2T ) T |x(t)|2 dt E0 /T0 .
33. Write Z  Z Z

E Xt2 dt = E[Xt2 ] dt = RX (0) d = .

R
34. If the function q(W ) := 0W S1 ( f ) S2 ( f ) d f is identically zero, then so is its deriva-
tive, q0 (W ) = S1 ( f ) S2 ( f ). But then S1 ( f ) = S2 ( f ) for all f 0.
35. If h(t) = I[T,T ] (t), and white noise is applied to the corresponding system, the cross
power spectral density of the input and output is
sin(2 T f ) N0
SXY ( f ) = H( f ) N0 /2 = 2T ,
2 T f 2
which is real and even, but not nonnegative. Similarly, if h(t) = et I[0,) (t),
N0 /2
SXY ( f ) = ,
1 + j2 f
which is complex valued.
36. First write

2 (1 f 2 )2 N0 /2, | f | 1,
SY ( f ) = |H( f )| N0 /2 =
0, | f | > 1.
Then
Z 1 Z
N0
PY = SY ( f ) d f =2 (1 2 f 2 + f 4 ) d f
2 0
h i 1

= N0 f 32 f 3 + 15 f 5 = N0 [1 23 + 51 ] = 8N0 /15.
0
Chapter 10 Problem Solutions 175

2 2 /2
37. First note that RX ( ) = e(2 ) p
/2 has power spectral density S ( f ) = e f
X / 2 .
Using the definition of H( f ) = | f | I[1,1] ( f ),
Z Z Z 1
E[Yt2 ] = RY (0) = SY ( f ) d f = |H( f )|2 SX ( f ) d f = | f |SX ( f ) d f
1
Z 1 r Z r
2 1 f 2 /2 2  f 2 /2  1
= 2 f SX ( f ) d f = fe df = e
0 0 0
r
2
= (1 e1/2 ) .

38. First observe that


 2
2 sin f
SY ( f ) = |H( f )| SX ( f ) = N0 /2.
f
This is the Fourier transform of

RY ( ) = (1 | |)I[1,1] ( ) N20 .

Then PY = RY (0) = N0 /2.


39. (a) First note that
1/(RC) 1
H( f ) = = .
1/(RC) + j2 f 1 + j(2 f )RC
Then
N0 /2
SXY ( f ) = H( f ) SX ( f ) = .
1 j(2 f )RC
(b) The inverse Fourier transform of SXY ( f ) = H( f ) N0 /2 is

RXY ( ) = h( ) N0 /2 = h( )N0 /2,

where the last step follows because h is real. Hence,


N0 /(RC)
RXY ( ) = e u( ).
2RC

(c) E[Xt1 Yt2 ] = RXY (t1 t2 ) = (N0 /(2RC))e(t1 t2 )/(RC) u(t2 t1 ).


N0 /2
(d) SY ( f ) = |H( f )|2 SX ( f ) = .
1 + (2 f RC)2
(e) Since
 
N0 /2 N0 2(1/(RC)) RC
SY ( f ) = =  ,
1 + (2 f RC)2 2(RC)2 1 2 2
RC + (2 f )2

we have that
N0 | |/(RC)
RY ( ) = e .
4RC
176 Chapter 10 Problem Solutions

(f) PY = RY (0) = N0 /(4RC).


40. To begin, write E[Yt+1/2Yt ] = RY (1/2). Next, since the input has power spectral den-
sity N0 /2 and since h(t) = 1/(1 + t 2 ) has transform H( f ) = e2 | f | , we can write
N0
SY ( f ) = |H( f )|2 SX ( f ) = | e2 | f | |2 N20 = 2 e4 | f | N20 = 2 e2 (2)| f | .

From the transform table, we conclude that


N0 2 N0
RY ( ) = = ,
2 4 + 2 4 + 2
and so E[Yt+1/2Yt ] = RY (1/2) = 4 N0 /17.
41. Since H( f ) = sin( T f )/( T f ), we can write
 
N0 sin T f 2 N0
SY ( f ) = |H( f )|2 = T .
2 T f 2T

Then
N0
RY ( ) = (1 | |/T )I[T,T ] ( ).
2T
42. To begin, write
Z t Z t Z
(t )
Yt = e t
e X d = e X d = e(t ) u(t )X d ,

where u is the unit-step function. We then see that Yt is the response to Xt of the
LTI system with impulse response h(t) := et u(t). Hence, we know from the text
that Xt and Yt are jointly wide-sense stationary. Next, since SX ( f ) = N0 /2, RX ( ) =
(N0 /2) ( ). We then compute in the time domain,
Z Z
N0 N0
RXY ( ) = h( )RX ( ) d = h( ) ( ) d = h( )
2 2
N0
= e u( ).
2
Next,
N0 /2
SXY ( f ) = H( f ) SX ( f ) = ,
1 j2 f
and
N0 /2
SY ( f ) = |H( f )|2 SX ( f ) = .
1 + (2 f )2
It then follows that RY ( ) = (N0 /4)e| | .
43. Consider the impulse response

h( ) := hn ( n).
n=
Chapter 10 Problem Solutions 177

Then
Z Z  


h( )Xt d =
n=
hn ( n) Xt d
Z
= hn

Xt ( n) d = hn Xtn =: Yt .
n= n=

(a) Since Yt is the response of the LTI system with impulse response h(t) to the
WSS input Xt , Xt and Yt are J-WSS.
(b) Since Yt is the response of the LTI system with impulse response h(t) to the
WSS input Xt , SY ( f ) = |H( f )|2 SX ( f ), where
Z Z  
H( f ) = h( )e j2 f d = n
h ( n) e j2 f d
n=
Z
= hn ( n)e j2 f d = hn e j2 f n
n= n=

has period one. Hence, P( f ) = |H( f )|2 is real, nonnegative, and has period one.

44. When the input power spectral density is SW ( f ) = 3, the output power spectral density
2
is |H( f )|2 3. We are also told that this output power spectral density is equal to e f .
2 2 2
Hence, |H( f )|2 3 = e f , or |H( f )|2 = e f /3. Next, if SX ( f ) = e f I[1,1] ( f ), then
2 2
SY ( f ) = |H( f )|2 SX ( f ) = (e f /3) e f I[1,1] ( f ) = (1/3)I[1,1] ( f ). It then follows
that
1 sin(2 ) 2 sin(2 )
RY ( ) = 2 = .
3 2 3 2
45. Since H( f ) = GI[B,B] ( f ) and Yt is the response to white noise, the output power
spectral density is SY ( f ) = G2 I[B,B] ( f ) N0 /2, and so

G2 N0 sin(2 B ) sin(2 B )
RY ( ) = 2B = G2 BN0 .
2 2 B 2 B
Note that
sin(2 Bk/(2B)) sin( k)
RY (kt) = RY (k/(2B)) = G2 BN0 = G2 BN0 ,
2 Bk/(2B) k

which is G2 BN0 for k = 0 and zero otherwise. It is obvious that the Xi are zero mean.
Since E[Xi X j ] = RY (i j), and the Xi are uncorrelated with variance E[Xi2 ] = RY (0) =
G2 BN0 .
46. (a) First write RX ( ) = E[Xt+ Xt ]. Then RX ( ) = E[Xt Xt ] = (E[Xt Xt
]) =


RX ( ) .
(b) Since Z
SX ( f ) = RX ( )e j2 f d ,

178 Chapter 10 Problem Solutions

we can write
Z Z
SX ( f ) = RX ( ) e j2 f d = RX (t) e j2 f t dt

Z
= RX (t)e j2 f t dt, by part (a),

= SX ( f ).
Since SX ( f ) is equal to its complex conjugate, SX ( f ) is real.
(c) Write
 Z
  Z
E[Xt1 Yt2 ] = E Xt1 h( )Xt2 d = h( ) E[Xt1 Xt2 ] d

Z Z
= h( ) RX ([t1 t2 ] + ) d = h( ) RX ([t1 t2 ] ) d .

(d) By part (c), Z


RXY ( ) = h( ) RX ( ) d ,

which is the convolution of h() and
RX . Hence, the transform of this equa-
tion is the product of the transform of h() and SX . We just have to observe
that
Z Z Z 
h( ) e j2 f d = h(t) e j2 f t dt = h(t)e j2 f t dt = H( f ) .

Hence, SXY ( f ) = H( f ) SX ( f ). Next, since


 Z  

RY ( ) = E[Yt+ Yt ] = E h( )Xt+ d Yt

Z Z
= h( )E[Xt+ Yt ] d = h( )RXY ( ) d ,

is a convolution, its transform is


SY ( f ) = H( f )SXY ( f ) = H( f )H( f ) SX ( f ) = |H( f )|2 SX ( f ).
Z Z
47. (a) RX ( ) = SX ( f )e j2 f d f = ( f )e j2 f d f = e j2 0 = 1.

(b) Write
Z
RX ( ) = [ ( f f0 ) + ( f + f0 )]e j2 f d f = e j2 f0 + e j2 f0

= 2 cos(2 f0 ).

(c) First write


2 /2
h 1 2 2 1 i 2
SX ( f ) = e f = e( 2 ) (2 f ) /2 2
2 2
h 1 2 2 1 i
= e( 2 ) (2 f ) /2 2 2 .
2
Chapter 10 Problem Solutions 179

From the Fourier transform table with = 1/(2 ),


2 /2
RX ( ) = 2 e(2 ) .

(d) From the transform table with = 1/(2 ),

1 1/(2 ) 2
RX ( ) = = .
(1/(2 ))2 + 2 1 + (2 )2

48. Write
Z  Z Z W

E[Xt2 ] = RX (0) = SX ( f )e j2 f d f = SX ( f ) d f = 1 d f = 2W.
=0 W

49. (a) e f u( f ) is not even.


2
(b) e f cos( f ) is not nonnegative.
(c) (1 f 2 )/(1 + f 4 ) is not nonnegative.
(d) 1/(1 + j f 2 ) is not real valued.
50. (a) Since sin is odd, it is NOT a valid correlation function.
(b) Since the Fourier transform of cos is [ ( f 1) + ( f + 1)]/2, which is real,
even, and nonnegative, cos IS a valid correlation function.
2 2
(c) Since the Fourier transform of e /2 is 2 e(2 f ) /2 , which is real, even, and
2
nonnegative, e /2 IS a valid correlation function.
(d) Since the Fourier transform of e| | is 2/[1 + (2 f )2 ], which is real, even, and
nonnegative, e| | IS a valid correlation function.
(e) Since the value of 2 e| | at = 0 is less than the value for other values of ,
2 e| | is NOT a valid correlation function.
(f) Since the Fourier transform of I[T,T ] ( ) is (2T ) sin(2 T f )/(2 T f ) is not non-
negative, I[T,T ] ( ) is NOT a valid correlation function.
51. Since R0 ( ) is a correlation function, S0 ( f ) is real, even, and nonnegative. Since
R( ) = R0 ( ) cos(2 f0 ),
1
S( f ) = 2 [S0 ( f f0 ) + S0 ( f + f0 )],

which is obviously real and nonnegative. It is also even since


1
S( f ) = 2 [S0 ( f f 0 ) + S0 ( f + f 0 )]
1
= 2 [S0 ( f + f 0 ) + S0 ( f f 0 )], since S0 is even,
= S( f ).

Since S( f ) is real, even, and nonnegative, R( ) is a valid correlation function.


180 Chapter 10 Problem Solutions

52. First observe that the Fourier transform of R( ) = R( 0 ) + R( + 0 ) is S( f ) =


2S( f ) cos(2 f 0 ). Hence, the answer cannot be (a) because it is possible to have
S( f ) > 0 and cos(2 f 0 ) < 0 for some values of f . Let S( f ) = I[1/(40 ),1/(40 )] ( f ),
which is real, even, and nonnegative. Hence, its inverse transform, which we denote
by R( ), is a correlation function. In this case, S( f ) = 2S( f ) cos(2 f 0 ) 0 for all f ,
and is real and even too. Hence, for this choice of R( ), R( ) is a correlation function.
Therefore, the answer is (b).
53. To begin, write
Z Z
j2 f
R( ) = S( f )e df = S( f )[cos(2 f ) j sin(2 f )] d f .

Since S is real and even, the integral of S( f ) sin(2 f ) is zero, and we have
Z
R( ) = S( f ) cos(2 f ) d f ,

which is a real and even function of . Finally,
Z Z

|R( )| = S( f )e j2 f d f S( f )e j2 f d f

Z Z
= |S( f )| d f = S( f )d f = R(0).

54. Let S0 ( f ) denote the Fourier transform of R0 ( ), and let S( f ) denote the Fourier
transform of R( ).
(a) The derivation in the text showing that the transform of a correlation function
is real and even uses only the fact that correlation functions are real and even.
Hence, S0 ( f ) is real and even. Furthermore, since R is the convolution of R0
with itself, S( f ) = S0 ( f )2 , which is real, even, and nonnegative. Hence, R( ) is
a correlation function.
(b) If R0 ( ) = I[T,T ] ( ), then
 2
sin(2 T f ) sin(2 T f )
S0 ( f ) = 2T and S( f ) = 2T 2T .
2 T f 2 T f

Hence, R( ) = 2T 1 | |/(2T ) I[2T,2T ] ( ).
55. Taking = N0 /2 as in the text,
h(t) = v(t0 t) = sin(t0 t)I[0, ] (t0 t).
Then h is causal for t0 .
2
2 ) /2 , V ( f ) = 2 2e2(2 f )2 /2
2
56. Since v(t) = e(t/ = 2 e(2 f ) . Then
2
V ( f ) e j2 f t0 2 e(2 f ) e j2 f t0
H( f ) = = 2
SX ( f ) e(2 f ) /2
2 2
= 2 e(2 f ) /2 e j2 f t0 = 2 2 e(2 f ) /2 e j2 f t0 ,
2
and it follows that h(t) = 2 e(tt0 ) /2 .
Chapter 10 Problem Solutions 181

57. Let v0 (n) := k h(n k)v(k) and Yn := k h(n k)Xk . The SNR is v0 (n0 )2 /E[Yn20 ]. We
have Z Z
1/2 1/2
E[Yn20 ] = SY ( f ) d f = |H( f )|2 SX ( f ) d f .
1/2 1/2

Let V ( f ) := k v(k)e j2 f k . Then


Z 1/2 2
j2 f n0

2
|v0 (n0 )| = H( f )V ( f )e d f
1/2
Z 1/2
p V ( f )e j2 f n0 2
= H( f ) SX ( f ) p d f
1/2 SX ( f )
Z 1/2 Z 1/2
|V ( f ) e j2 f n0 |2
|H( f )|2 SX ( f ) d f d f,
1/2 1/2 SX ( f )

with equality if and only if


p V ( f ) e j2 f n0
H( f ) SX ( f ) = p (#)
SX ( f )

for some constant . It is now clear that the SNR is upper bounded by
Z 1/2
|V ( f ) e j2 f n0 |2
df
1/2 SX ( f )

and that the SNR equals the bound if and only if (#) holds with equality for some
constant . Hence, the matched filter transfer function is

V ( f ) e j2 f n0
H( f ) = .
SX ( f )

58. We begin by writing

E[Ut ] = E[Vt + Xt ] = E[Vt ] + E[Xt ],

which does not depend on t since Vt and Xt are each individually WSS. Next write

E[Ut1 Vt2 ] = E[(Vt1 + Xt1 )Vt2 ] = RV (t1 t2 ) + RXV (t1 t2 ). ()

Now write

E[Ut1 Ut2 ] = E[Ut1 (Vt2 + Xt2 ] = E[Ut1 Vt2 ] + E[Ut1 Xt2 ].

By (), the term E[Ut1 Vt2 ] depends on t1 and t2 only through their difference. Since

E[Ut1 Xt2 ] = E[(Vt1 + Xt1 )Xt2 ] = RV X (t1 t2 ) + RX (t1 t2 ),

it follows that E[Ut1 Ut2 ] depends on t1 and t2 only through their difference. Hence, Ut
and Vt are J-WSS.
182 Chapter 10 Problem Solutions

59. The assumptions in the problem imply that Vt and Xt are J-WSS, and by the preceding
problem, it follows that Ut and Vt are J-WSS. We can therefore apply the formulas
for the Wiener filter derived in the text. It just remains to compute the quantities used
in the formulas. First,
RVU ( ) = E[Vt+ Ut ] = E[Vt+ (Vt + Xt )] = RV ( ) + RXV ( ) = RV ( ),
which implies SVU ( f ) = SV ( f ). Similarly,
RU ( ) = E[Ut+ Ut ] = E[(Vt+ + Xt+ )Ut ]
= RVU ( ) + E[Xt+ (Vt + Xt )]
= RV ( ) + RXV ( ) + RX ( ) = RV ( ) + RX ( ),
and so SU ( f ) = SV ( f ) + SX ( f ). We then have
SVU ( f ) SV ( f )
H( f ) = = .
SU ( f ) SV ( f ) + SX ( f )

60. The formula for RV ( ) implies SV ( f ) = (1 | f |)I[1,1] ( f ). We then have


SV ( f ) (1 | f |)I[1,1] ( f )
H( f ) = =
SV ( f ) + SX ( f ) (1 | f |)I[1,1] ( f ) + 1 I[1,1] ( f )
(1 | f |)I[1,1] ( f )
= = I[1,1] ( f ),
1 | f |I[1,1] ( f )
and so
sin(2 t)
h(t) = 2 .
2 t
61. To begin, write
E[|Vt Vbt |2 ] = E[(Vt Vbt )(Vt Vbt )] = E[(Vt Vbt )Vt ] E[(Vt Vbt )Vbt ]
= E[(Vt Vbt )Vt ], by the orthogonality principle,
Z
= E[Vt2 ] E[Vbt Vt ] = RV (0) E[Vbt Vt ] = SV ( f ) d f E[Vbt Vt ].

Next observe that


 Z
  Z
E[Vbt Vt ] = E h( )Ut d Vt = h( )E[Vt Ut ] d

Z Z
= h( )RVU ( ) d = h( )RVU ( ) d , since RVU ( ) is real,

Z
= H( f )SVU ( f ) d f , by Parsevals formula,

Z Z
SVU ( f ) |SVU ( f )|2
= SVU ( f ) d f = d f.
SU ( f ) SU ( f )
Putting these two observations together yields
Z Z Z  
|SVU ( f )|2 |SVU ( f )|2
E[|Vt Vbt |2 ] = SV ( f ) d f df = SV ( f ) d f.
SU ( f ) SU ( f )
Chapter 10 Problem Solutions 183

62. Denote the optimal estimator by Vbn = k= h(k)Unk , and denote any other estima-
tor by Ven =
k= h(k)Unk . The discrete-time orthogonality principle says that if
 
E (Vn Vbn ) h(k)Unk = 0 ()
k=

for every h, then h is optimal in that E[|Vn Vbn |2 ] E[|Vn Ven |2 ] for every h. To es-
tablish the orthogonality principle, assume the above equation holds for every choice
of h. Then we can write

E[|Vn Ven |2 ] = E[|(Vn Vbn ) + (Vbn Ven )|2 ]


= E[|Vn Vbn |2 + 2(Vn Vbn )(Vbn Ven ) + |Vbn Ven |2 ]
= E[|Vn Vbn |2 ] + 2E[(Vn Vbn )(Vbn Ven )] + E[|Vbn Ven |2 ]. ()

Now observe that


  
E[(Vn Vbn )(Vbn Ven )] = E (Vn Vbn ) h(k)Unk h(k)Unk
k= k=
  

= E (Vn Vbn ) h(k) h(k) Unk = 0, by ().
k=

We can now continue () writing

E[|Vn Ven |2 ] = E[|Vn Vbn |2 ] + E[|Vbn Ven |2 ] E[|Vn Vbn |2 ],

and thus h is the filter that minimizes the mean-squared error.


The next task is to characterize the filter h that satisfies the orthogonality condition
for every choice of h. Write the orthogonality condition as
   
0 = E (Vn Vbn ) h(k)Utk = E h(k)(Vn Vbn )Utk
k= k=

= E[h(k)(Vn Vbn )Utk ] = h(k)E[(Vn Vbn )Utk ]
k= k=

= b (k)].
h(k)[RVU (k) RVU
k=

Since this must hold for all h, take h(k) = RVU (k) RVU
b (k) to get


RVU (k) R b (k) 2 = 0.
VU
k=

b (k) for
Thus, the orthogonality condition holds for all h if and only if RVU (k) = RVU
all k.
184 Chapter 10 Problem Solutions

b
b . Recall that Vn is the response of an LTI system to
The next task is to analyze RVU
input Un . Applying the result of Problem 31(a) with X replaced by U and Y replaced
by Vb , we have, also using the fact that RU is even,

b (m) = RU Vb (m) =
RVU h(k)RU (m k).
k=

Taking discrete-time Fourier transforms of



b (m) =
RVU (m) = RVU h(k)RU (m k)
k=

yields
SVU ( f )
SVU ( f ) = H( f )SU ( f ), and so H( f ) = .
SU ( f )

63. We have

SV ( f ) 2 /[ 2 + (2 f )2 ] 2
H( f ) = = =
SV ( f ) + SX ( f ) 2 /[ 2 + (2 f )2 ] + 1 2 + 2 + (2 f )2
2A
= ,
A A2 + (2 f )2

where A := 2 + 2 . Hence, h(t) = ( /A)eA|t| .
64. To begin, write

+ j2 f 1
K( f ) = = + j2 f .
A + j2 f A + j2 f 1 + j2 f

Then
d At
k(t) = eAt u(t) + e u(t)
dt
= eAt u(t) AeAt u(t) + eAt (t)
= ( A)eAt u(t) + (t),

since eAt (t) = (t) for both t = 0 and for t 6= 0. This is a causal impulse response.
65. Let Zt := Vt+t . Then the causal Wiener filter for Zt yields the prediction or smoothing
filter for Vt+t . The WienerHopf equation for Zt is
Z
RZU ( ) = ht ( )RU ( ) d , 0.
0

Now, RZU ( ) = E[Zt+ Ut ] = E[Vt+ +t Ut ] = RVU ( + t), and so we must solve


Z
RVU ( + t) = ht ( )RU ( ) d , 0.
0
Chapter 10 Problem Solutions 185

For white noise with RU ( ) = ( ), this reduces to


Z
RVU ( + t) = ht ( ) ( ) d = ht ( ), 0.
0

If h(t) = RVU (t)u(t) denotes the causal Wiener filter, then for t 0 (prediction), we
can write
ht ( ) = RVU ( + t) = h( + t), 0.
If t < 0 (smoothing), we can write ht ( ) = h( + t) only for t. For 0
< t, h( + t) = 0 while ht ( ) = RVU ( + t).
66. By the hint, the limit of the double sums is the desired double integral. If we can show
that each of these double sums is nonnegative, then the limit will also be nonnegative.
To this end put Zi := Xti e j2 f ti t i . Then
 n 2   n  n   n n


0 E Zi = E Zi Zk = E[Zi Zk ]
i=1 i=1 k=1 i=1 k=1
n n
= E[Xti Xtk ]e j2 f ti e j2 f tk t i t k
i=1 k=1
n n
= RX (ti tk )e j2 f ti e j2 f tk t i t k .
i=1 k=1

67. (a) The Fourier transform of CY ( ) = e| | is 2/[1 + (2 f )2 ], which is continuous


1 RT
at f = 0. Hence, we have convergence in mean square of 2T T t dt to E[Yt ].
Y
(b) The Fourier transform of CY ( ) = sin( )/( ) is I[1/2,1/2] ( f ), which is con-
1 RT
tinuous at f = 0. Hence, we have convergence in mean square of 2T T Yt dt to
E[Yt ].
68. We first point out that this is not a question about mean-square convergence. Write
Z T
1 sin(2 T + ) sin(2 (T ) + )
cos(2 t + ) dt = .
2T T 2T 2
Since | sin x | 1, we can write

1 ZT 2

2T T cos(2 t + ) dt 4 T 0,

and so the limit in question exists and is equal to zero.


69. As suggested by the hint, put Yt := Xt+ Xt . It will be sufficient if Yt is WSS and if the
Fourier transform of the covariance function of Yt is continuous at the origin. First,
since Xt is WSS, the mean of Yt is
E[Yt ] = E[Xt+ Xt ] = RX ( ),
which does not depend on t. Before examining the correlation function of Yt , we
assume that Xt is fourth-order strictly stationary so that
E[Yt1 Yt2 ] = E[Xt1 + Xt1 Xt2 + Xt2 ]
186 Chapter 10 Problem Solutions

must be unchanged if on the right-hand side we subtract t2 from every subscript ex-
pression to get
E[Xt1 + t2 Xt1 t2 X X0 ].
Since this depends on t1 and t2 only through their difference, we see that Yt is WSS if
Xt is fourth-order strictly stationary. Now, the covariance function of Yt is
C( ) = E[X + X X X0 ] RX ( )2 .
If the Fourier transform of this function of is continuous at the origin, then
Z T
1
Xt+ Xt dt RX ( ).
2T T

70. As suggested by the hint, put Yt := IB (Xt ). It will be sufficient if Yt is WSS and if
the Fourier transform of the covariance function of Yt is continuous at the origin. We
assume at the outset that Xt is second-order strictly stationary. Then the mean of Yt is
E[Yt ] = E[IB (Xt )] = P(Xt B),
which does not depend on t. Similarly,
E[Yt1 Yt2 ] = E[IB (Xt1 )IB (Xt2 )] = P(Xt1 B, Xt2 B)
must be unchanged if on the right-hand side we subtract t2 from every subscript to get
P(Xt1 t2 B, X0 B).
Since this depends on t1 and t2 only through their difference, we see that Yt is WSS if
Xt is second-order strictly stationary. Now, the covariance function of Yt is
C( ) = P(X B, X0 B) P(Xt B)2 .
If the Fourier transform of this function of is continuous at the origin, then
Z T
1
IB (Xt ) dt P(Xt B).
2T T

71. We make the following definition and apply the hints:


Z
1 T
RXY ( ) := lim RXY ( + , ) d
T 2T T
Z  Z 
1 T
= lim h( )RX ( + , ) d d
T 2T T
Z  Z 
1 T
= h( ) lim RX ( + , ) d d
T 2T T
Z  Z 
1 T
= h( ) lim RX ( + + , ) d d
T 2T T
Z  Z 
1 T
= h( ) lim RX ( + + , ) d d .
T 2T T
| {z }
= RX ( + )
Chapter 10 Problem Solutions 187

72. Write
Z
1 T
RY ( ) := lim RY ( + , ) d
T 2T T
Z  Z 
1 T
= lim h( )RXY ( + , ) d d
T 2T T
Z  Z 
1 T
= h( ) lim RXY ([ ] + , ) d d .
T 2T T
| {z }
= RXY ( )

73. Let SXY ( f ) denote the Fourier transform of RXY ( ), and let SY ( f ) denote the Fourier
transform of RY ( ). Then by the preceding two problems,

SY ( f ) = H( f )SXY ( f ) = H( f )H( f )SX ( f ) = |H( f )|2 SX ( f ),

where, since h is real, H( f ) = H( f ) .


74. This is an instance of Problem 32.
CHAPTER 11
Problem Solutions

1. With = 3 and t = 10,

P(Nt = 0) = e t = e310 = e30 = 9.358 1014 .

2. With = 12 per minute and t = 20 seconds, t = 4. Thus,


 
t ( t)2 ( t)3
P(Nt > 3) = 1 P(Nt 3) = 1 e 1 + t + +
2! 3!
 2 3

4 4
= 1 e4 1 + 4 + + = 1 e4 (5 + 8 + 32/3)
2 6
= 1 e4 (39/3 + 32/3) = 1 e4 (71/3) = 0.5665.

(2 5)10 e25
3. (a) P(N5 = 10) = = 0.125.
10!
(b) We have
 5   
\ 5 5
22 e2
P {Ni Ni1 = 2} = P(Ni Ni1 = 2) = 2!
i=1 i=1 i=1
2 5 10
= (2e ) = 32e = 1.453 103 .

4. Let Nt denote the number of crates sold through time t (in days). Then Ni Ni1 is
the number of crates sold on day i, and so
 5  5 5 h i
\
P {Ni Ni1 3} = P(Ni Ni1 3) = 1 P(Ni Ni1 2)
i=1 i=1 i=1
5 h i
= 1 e 1 + + 2 /2!
i=1
h i5
= 1 e3 1 + 3 + 9/2 = 0.06385.

5. Let Nt denote the number of fishing rods sold through time t (in days). Then Ni Ni1
is the number of crates sold on day i, and so
 5   5 
[ \
P {Ni Ni1 3} = 1 P {Ni Ni1 2}
i=1 i=1
5
= 1 P(Ni Ni1 2)
i=1

188
Chapter 11 Problem Solutions 189

5  
= 1 e (1 + + 2 /2!)
i=1
 5
= 1 e2 (1 + 2 + 4/2!) = 1 e10 55 = 0.858.

6. Since the average time between hit songs is 7 months, the rate is = 1/7 per month.
(a) Since a year is 12 months, we write

P(N12 > 2) = 1 P(N12 2) = 1 e12 [1 + 12 + (12 )2 /2!]


= 1 e12/7 [1 + 12/7 + (12/7)2 /2] = 0.247.

(b) Let Tn denote the time of the nth hit song. Since Tn = X1 + + Xn , E[Tn ] =
nE[X1 ] = 7n. For n = 10, E[T10 ] = 70 months.
7. (a) Since N0 0, Nt = Nt N0 . Since (0,t] (t,t + t] = , Nt N0 and Nt+t Nt
are independent.
(b) Write

P(Nt+t = k + `|Nt = k) = P(Nt+t Nt = `|Nt = k), by substitution,


= P(Nt+t Nt = `|Nt N0 = k), since N0 0,
= P(Nt+t Nt = `), by independent increments.

(c) Write
P(Nt+t = k + `|Nt = k)P(Nt = k)
P(Nt = k|Nt+t = k + `) =
P(Nt+t = k + `)
P(Nt+t Nt = `)P(Nt = k)
= , by part (b),
P(Nt+t = k + `)
( t)` e t ( t)k e t
`! k!
=
[ (t+t)]k+` e (t+t)
(k+`)!
  k  `
k+` t t
= .
k t + t t + t

(d) In part (c), put ` = n k and put p = t/(t + t). Then


 
n k
P(Nt = k|Nt+t = n) = p (1 p)nk , k = 0, . . . , n.
k

8. The nth customer arrives at time Tn Erlang(n, ). Hence,


(1 + n) n(n) n
E[Tn ] = = = .
(n) (n)
Alternatively, since Tn = X1 + +Xn , where the Xi are i.i.d. exp( ), E[Tn ] = nE[Xi ] =
n/ .
190 Chapter 11 Problem Solutions

9. (a) E[Xi ] = 1/ = 0.5 weeks.


(b) P(N2 = 0) = e 2 = e4 = 0.0183.
(c) E[N12 ] = 12 = 24 snowstorms.
(d) Write
 12   12 
[ \
P {Ni Ni1 5} = 1 P {Ni Ni1 4}
i=1 i=1
= 1 [e (1 + + 2 /2 + 3 /6 + 4 /24)]12
= 1 [e2 (1 + 2 + 2 + 4/3 + 2/3)]12
= 1 [7e2 ]12 = 1 0.523 = 0.477.

10. First observe that since 1/ = 2 months, = 1/2 per month.

(a) P(N4 = 0) = e 4 = e4/2 = e2 = 0.135.


(b) Write
 4   4 
[ \
P {Ni Ni1 2} = 1 P {Ni Ni1 1}
i=1 i=1
4
= 1 P(Ni Ni1 1)
i=1
4
= 1 [e 1 + e 1 ( 1)]
i=1
= 1 [e (1 + )]4 = 1 [ 32 e1/2 ]4
81 2
= 1 16 e = 0.315.

11. We need to find var(Tn ) = var(X1 + + Xn ). Since the Xi are independent, they
are uncorrelated, and so the variance of the sum is the sum of the variances. Since
Xi exp( ), var(Tn ) = n var(Xi ) = n/ 2 . An alternative approach is to use the fact
that Tn Erlang(n, ). Since the moments of Tn are available,

(1 + n)n  n 2 n
var(Tn ) = E[Tn2 ] (E[Tn ])2 = 2
= 2.

12. To begin, use the law of total probability, substitution, and independence to write
Z 1 Z 1
Nln(1+tU) Nln(1+tU)
Yt
E[z ] = E[z ] = E[z |U = u] du = E[zNln(1+tu) |U = u] du
0 0
Z 1 Z 1 Z 1
z1
= E[zNln(1+tu) ] du = exp[(z 1) ln(1 + tu)] du = eln(1+tu) du
0 0 0
Z 1
= (1 + tu)z1 du.
0
Chapter 11 Problem Solutions 191

To compute this integral, we need to treat the cases z = 0 and z 6= 0 separately. We


find that
ln(1 + t)
, z = 0,
Yt
E[z ] = t
z
(1 + t) 1 , z 6= 0.

tz
13. Denote the arrival times of Nt by T1 , T2 , . . . , and let Xk := Tk Tk1 denote the inter-
arrival times. Similarly, denote the arrival times of Mt by S1 , S2 , . . . . (As it turns out,
we do not need the interarrival times of Mt .) Then for arbitrary k > 1, we use the law
of total probability, substitution, and independence to compute
P(MTk MTk1 = m)
Z Z
= P(MTk MTk1 = m|Tk = t, Tk1 = ) fTk Tk1 (t, ) dt d
Z0 Z
= P(Mt M = m|Tk = t, Tk1 = ) fTk Tk1 (t, ) dt d
Z0 Z
= P(Mt M = m) fTk Tk1 (t, ) dt d
0
Z Z
[ (t )]m e (t )
= fTk Tk1 (t, ) dt d
0 m!
 
[ (Tk Tk1 )]m e (Tk Tk1 )
= E
m!
  Z
( Xk )m e Xk ( x)m e x
= E = e x dx
m! 0 m!
Z
m
= xm ( + )e( + )x dx
( + )m! 0
| {z }
mth moment of exp( + ) density
 m
m m!
= m
= ,
( + )m! ( + ) + +
which is a geometric0 ( /( + )) pmf in m.
14. It suffices to show that the probability generating function GMt (z) has the form e (z1)
for some > 0. We use the law of total probability, substitution, and independence
to write
Nt Nt
GMt (z) = E[zMt ] = E[zi=1 Yi ] = E[zi=1 Yi |Nt = n]P(Nt = n)
n=0

n n
= E[zi=1 Yi |Nt = n]P(Nt = n) = E[zi=1 Yi ]P(Nt = n)
n=0 n=0
 n 
Yi
= E[z ] P(Nt = n) = [pz + (1 p)]n P(Nt = n)
n=0 i=1 n=0
 t([pz+(1p)]1)
= GNt pz + (1 p) = e = e p t(z1) .
Thus, Mt Poisson(p t).
192 Chapter 11 Problem Solutions

15. Let Mt = Ni=1


t
Vi denote the total energy throught time t, with Mt = 0 for Nt = 0. We
use the law of total probability, substitution, and independence to write
 Nt


E[Mt ] = E[Mt |Nt = n]P(Nt = n) = E Vi Nt = n P(Nt = n)
n=0 n=1 i=1

 n  n 


= E Vi Nt = n P(Nt = n) = E[Vi ] P(Nt = n)
n=1 i=1 n=1 i=1

= nE[V1 ]P(Nt = n) = E[V1 ] nP(Nt = n) = E[V1 ]E[Nt ] = E[V1 ]( t).
n=1 n=1

The average time between lightning strikes is 1/ minutes.


16. We have
2.132(1.96)
= 5.170 = 5.170 0.418 with 95% probability.
10
In other words, [4.752, 5.588] with 95% probability.
17. To begin, observe that
n n n
Y = g(Ti ) = gk I(tk1 ,tk ] (Ti ) = gk I(tk1 ,tk ] (Ti ) = gk (Ntk Ntk1 )
i=1 i=1 k=1 k=1 i=1 k=1

is a sum of independent random variables. Then


n n Z
E[Y ] = gk E[Ntk Ntk1 ] = gk (tk tk1 ) =
0
g( ) d ,
k=1 k=1

since g is piecewise constant. Next,


n
n
Y ( ) = E[e jY ] = E[e j k=1 gk (Ntk Ntk1 ) ] = E[e j gk (Ntk Ntk1 ) ]
k=1
n h i  n 
= exp (tk tk1 )(e j gk 1) = exp (tk tk1 )(e j gk 1)
k=1 k=1
Z

j g( )

= exp e 1 d ,
0

since e j g( ) 1 is piecewise constant with values e j gk 1 (or the value zero if


does not lie in any (tk1 ,tk ]). We now compute the correlation,
 n  n 
E[Y Z] = E gk (Ntk Ntk1 ) hl (Ntl Ntl1 )
k=1 l=1
= gk hl E[(Ntk Ntk1 )(Ntl Ntl1 )] + gk hk E[(Ntk Ntk1 )2 ]
k6=l k
2
= gk hl (tk tk1 )(tl tl1 ) + gk hk [ (tk tk1 ) + 2 (tk tk1 )2 ]
k6=l k
Chapter 11 Problem Solutions 193

= gk hl 2 (tk tk1 )(tl tl1 ) + gk hk (tk tk1 )


k,l k
  
= gk (tk tk1 ) hl (tl tl1 ) + gk hk (tk tk1 )
k l k
Z Z Z
= g( ) d h( ) d + g( )h( ) d .
0 0 0

Hence, Z
cov(Y, Z) = E[Y Z] E[Y ]E[Z] = g( )h( ) d .
0

18. The key is to use g( ) = h(t ) in the preceding problem. It then immediately
follows that
Z Z 
j h(t )

E[Yt ] = h(t ) d , Yt ( ) = exp e 1 d ,
0 0

and Z
cov(Yt ,Ys ) = h(t )h(s ) d .
0

19. MATLAB. Replace the line


X = -log(rand(1))/lambda; % Generate exp(lambda) RV
with
X = randn(1)2; % Generate chi-squared RV
20. If Nt is a Poisson process of rate , then FX is the exp( ) cdf. Hence,
Z Z
P(X1 < Y1 ) = FX (y) fY (y) dy = [1 e y ] fY (y) dy = 1 MY ( ).
0 0

If Mt is a Poisson process of rate , then MY is the exp( ) mgf, and


P(X1 < Y1 ) = 1 = 1 = .
( ) + +

21. Since Tn := X1 + + Xn is the sum of i.i.d. random variables, var(Tn ) = n var(X1 ).


Since X1 uniform[0, 1], var(X1 ) = 1/12, and var(Tn ) = n/12.
22. In the case of a Poisson process, Tk is Erlang(k, ). Hence,

   
k1
( t)l e t ( t)l e t
Fk (t) = 1 l! =
l!
k=1 k=1 l=0 k=1 l=k
   
l
( t) e t ( t) e t
l
= I[k,) (l) = I[k,) (l)
k=1 l=0 l! l=0 l! k=1

( t)l e t
= l l!
= E[Nt ] = t.
l=0
194 Chapter 11 Problem Solutions

23. To begin, write




 

E[Nt |X1 = x] = E I[0,t] (Tn ) X1 = x = E I[0,t] (X1 + + Xn ) X1 = x
n=1 n=1
 
= E I[0,t] (x + X2 + + Xn ) X1 = x .
n=1

(a) If x > t, then x + X2 + + Xn > t, and so I[0,t] (x + X2 + + Xn ) = 0. Thus,


E[Nt |X1 = x] = 0 for x > t.
(b) First, for n = 1 and x t, I[0,t] (x) = 1. Next, for n 2 and x t, I[0,t] (x + X2 +
+ Xn ) = I[0,tx] (X2 + + Xn ). So,
 
E[Nt |X1 = x] = 1 + E I[0,tx] (X2 + + Xn )
n=2
 
= 1 + E I[0,tx] (X1 + + Xn )
n=1
= 1 + E[Ntx ],

where we have used fact that the Xi are i.i.d.


(c) By the law of total probability,
Z
E[Nt ] = E[Nt |X1 = x] f (x) dx
0
Z t Z
= E[Nt |X1 = x] f (x) dx + E[Nt |X1 = x] f (x) dx
0 t | {z }
= 0 for x>t
Z t  Z t
= 1 + E[Ntx ] f (x) dx = F(t) + E[Ntx ] f (x) dx.
0 0

24. With the understanding that m(t) = 0 for t < 0, we can write the renewal equation as
Z
m(t) = F(t) + m(t x) f (x) dx,
0

where the last term is a convolution. Hence, taking the Laplace transform of the
renewal equation yields
Z Z
M(s) := m(t)est dt = F(t)est dt + M(s)MX (s).
0 0

Using integration by parts, we have


Z Z
F(t)est 1
F(t)est dt = f (t)est dt.
0 s 0 s 0

Thus,
1
M(s) = MX (s) + M(s)MX (s),
s
Chapter 11 Problem Solutions 195

which we can rearrange as


1
M(s)[1 MX (s)] = MX (s),
s
or
1 MX (s) 1 /( s) 1
M(s) = = = = 2.
s 1 MX (s) s 1 /( s) s s s
It follows that m(t) = tu(t).
25. For 0 s < t < , write
Z t Z s  Z s Z t 
E[Vt Vs ] = E X d X d = E[X X ] d d
0 0 0 0

Z s Z t  Z s Z t 
2
= RX ( ) d d = ( ) d d
0 0 0 0
Z s
2 2
= d = s.
0

26. For 0 s < t < , write

E[Wt Ws ] = E[(Wt Ws )Ws ] + E[Ws2 ]


= E[(Wt Ws )(Ws W0 )] + 2 s
= E[Wt Ws ]E[Ws W0 ] + 2 s = 0 0 + 2 s = 2 s.

27. For t2 > t1 , write

E[Yt1 Yt2 ] = E[eWt1 eWt2 ] = E[eWt2 Wt1 e2Wt1 ] = E[eWt2 Wt1 e2(Wt1 W0 ) ]
2 (t 2t 2 (t
= E[eWt2 Wt1 ]E[e2(Wt1 W0 ) ] = e 2 t1 )/2 e4 1 /2 = e 2 +3t1 )/2 .

28. Since cov(Wti ,Wt j ) = 2 min(ti ,t j ),



t1 t1 t1 t1
t1 t2 t2 t2

t3
cov(X) = 2 t1 t2 t3 .
.. .. .. .. ..
. . . . .
t1 t2 t3 tn

29. Let 0 t1 < < tn+1 < and 0 s1 < < sm+1 < , and suppose that
n m
g( ) = gi I(ti ,ti+1 ] ( ) and h( ) = h j I(s j ,s j+1 ] ( ).
i=1 j=1

Denote the distinct points of {ti } {s j } in increasing order by 1 < < p . Then
p p
g( ) = gk I(k ,k+1 ] ( ) and h( ) = hk I(k ,k+1 ] ( ),
k=1 k=1
196 Chapter 11 Problem Solutions

where the gk are taken from the gi , and the hk are taken from the h j . We can now
write
Z Z p p

0
g( ) dW +
0
h( ) dW = gk (Wk+1 Wk ) + hk (Wk+1 Wk )
k=1 k=1
p
= [gk + hk ](Wk+1 Wk )
k=1
Z
= [g( ) + h( )] dW .
0

30. Following the hint, we first write


 Z Z 2   Z 2 
E g( ) dW h( ) dW = E g( ) dW
0 0 0
Z Z 
2E g( ) dW h( ) dW
0 0
 Z
2 
+E h( ) dW
0
Z Z
= 2 g( )2 d + 2 h( )2 d
0 0
Z Z

2E g( ) dW h( ) dW .
0 0

Second, we write
 Z Z
2   Z
2 
E g( ) dW h( ) dW = E [g( ) h( )] dW
0 0 0
Z
= 2 [g( ) h( )]2 d
0
Z Z
= 2 g( )2 d + 2 h( )2 d
0 0
Z
2
2 g( )h( ) d .
0

Comparing these two formulas, we see that


Z Z  Z
E g( ) dW h( ) dW = 2 g( )h( ) d .
0 0 0

31. (a) Following the hint,


Z t Z
Yt := g( ) dW = g( )I[0,t] ( ) dW ,
0 0

it follows that
Z Z t
E[Yt2 ] = 2 2
[g( )I[0,t] ( )] d = 2
g( )2 d .
0 0
Chapter 11 Problem Solutions 197

(b) For t1 ,t2 0, write


Z Z 
E[Yt1 Yt2 ] = E g( )I[0,t1 ] ( ) dW g( )I[0,t2 ] ( ) dW
0 0
Z  
= 2 g( )I[0,t1 ] ( ) g( )I[0,t2 ] ( ) d
0
Z Z min(t1 ,t2 )
= 2 g( )2 I[0,t1 ] ( )I[0,t2 ] ( ) d = 2 g( )2 d .
0 0

32. By independence of V and the Wiener process along with the result of the previous
problem,
Z min(t1 ,t2 )
RY (t1 ,t2 ) = e (t1 +t2 ) E[V 2 ] + 2 e (t1 +t2 ) e2 d .
0

If t1 t2 , this last integral is equal to


Z t1
1 2 t1
e2 d = (e 1),
0 2
and so
 2  2 (t2 t1 )
E[Yt1 Yt2 ] = e (t1 +t2 ) q2 + e .
2 2
Similarly, if t2 < t1 ,
 2  2 (t1 t2 )
E[Yt1 Yt2 ] = e (t1 +t2 ) q2 + e .
2 2
In either case, we can write
 2  2 |t1 t2 |
E[Yt1 Yt2 ] = e (t1 +t2 ) q2 + e .
2 2

33. Write

e (t1 +t2 ) e (t1 +t2 ) 2


E[Yt1 Yt2 ] = E[We2 t1 We2 t2 ] = min(e2 t1 , e2 t2 ).
2 2
For t1 t2 , this reduces to

e (t1 +t2 ) 2 2 t1 2 (t2 t1 )


e = e .
2 2
If t2 < t1 , we have
e (t1 +t2 ) 2 2 t2 2 (t1 t2 )
e = e .
2 2
We conclude that
2 |t1 t2 |
E[Yt1 Yt2 ] = e .
2
198 Chapter 11 Problem Solutions

Z t
2  Z t
34. (a) P(t) := E[Yt2 ] = E g( ) dW = g( )2 d .
0 0
(b) If g(t) is never zero, then for 0 t1 < t2 < ,
Z t2
P(t2 ) P(t1 ) = g( )2 d > 0.
t1

Thus, P(t1 ) < P(t2 ).


(c) First,
Z P1 (t)

E[Xt ] = E[YP1 (t) ] = E g( ) dW = 0
0
since Wiener integrals have zero mean. Second,
Z P1 (t) 2  Z P1 (t)
2
E[Xt ] = E g( ) dW = g( )2 d = P(P1 (t)) = t.
0 0

35. (a) For t > 0, E[Wt2 ] = E[(Wt W0 )2 ] = t.


(b) For s < 0, E[Ws2 ] = E[(W0 Ws )2 ] = s.
(c) From parts (a) and (b) we see that no matter what the sign of t, E[Wt2 ] = |t|.
Whether t > s or s < t, we can write

|t s| = E[(Wt Ws )2 ] = E[Wt2 ] 2E[Wt Ws ] + E[Ws2 ],

and so
E[Wt2 ] + E[Ws2 ] |t s| |t| + |s| |t s|
E[Wt Ws ] = = .
2 2

36. (a) Write


=1
Z z }| {
P(X = xk ) = P((X,Y ) {xk } IR) =
I{xk } (xi ) IIR (y) fXY (xi , y) dy
i
Z   Z
=

I{xk } (xi ) fXY (xi , y) dy =
fXY (xk , y) dy.
i

(b) Write
 Z
P(Y C) = P (X,Y ) IR C =
IIR (xi )IC (y) fXY (xi , y) dy
i
Z   Z  
= IC (y) fXY (xi , y) dy = fXY (xi , y) dy.
i C i

(c) First write


P(X = xk ,Y C) P((X,Y ) {xk } C)
P(Y C|X = xk ) = = .
P(X = xk ) P(X = xk )
Chapter 11 Problem Solutions 199

Then since
Z Z
P((X,Y ) {xk } C) =
I{xk } (xi )IC (y) fXY (xi , y) dy =
C
fXY (xk , y) dy,
i

we have
Z
fXY (xk , y) dy Z  
C fXY (xk , y)
P(Y C|X = xk ) = = dy.
pX (xk ) C pX (xk )
(d) Write
Z Z  


P(X B|Y = y) fY (y) dy =

B i X|Y i fY (y) dy
I (x )p (x |y)
i
Z
fXY (xi , y)
= IB (xi ) fY (y)
fY (y) dy
i
Z
= IB (xi )
fXY (xi , y) dy
i
Z
= IB (xi )
IIR (y) fXY (xi , y) dy
i
Z
=
IBIR (xi , y) fXY (xi , y) dy
i
= P((X,Y ) B IR)
= P(X B,Y IR) = P(X B).

37. The key observations are that since F is nondecreasing,


F 1 (U) x U F(x),
and since F 1 is nondecreasing,
U F(x) F 1 (U) x.
Hence, {F 1 (U) x} = {U F(x)}, and with X := F 1 (U) we can write
Z F(x)
P(X x) = P(F 1 (U) x) = P(U F(x)) = 1 du = F(x).
0

38. (a) The cdf is


F(x)
1
3/4 x/2

1/2
1/4
x
0 1 1 2
2
200 Chapter 11 Problem Solutions

(b) For 1/2 u < 1, Bu = [2u, ), and G(u) = 2u. For 1/4 < u < 1/2, Bu = [1, ),
= 1. For u = 1/4,
and G(u) Bu = [1/2, ), and G(u) = 1/2. For 0 u < 1/4,
Bu = [ u, ), and G(u) = u. Hence,

G(u)
2
2u
1
1/2
u
0 1 1 3 1
4 2 4

39. Suppose G(u) x. Since F is nondecreasing, F(G(u)) F(x). By the definition of


G(u), F(G(u)) u. Thus, F(x) F(G(u)) u. Now suppose u F(x). Then by the
definition of G(u), G(u) x.

40. With X := G(U), write


Z F(x)
P(X x) = P(G(U) x) = P(U F(x)) = 1 du = F(x).
0

41. To compute G(u), we used the M ATLAB function

function x = G(u)
i1 = find(u <= .25);
i2 = find(.25 < u & u <= .5);
i3 = find(.5 < u & u <= 1);
x(i1) = sqrt(u(i1));
x(i2) = 1;
x(i3) = 2*u(i3);

We obtained the histogram

0
0 0.5 1 2

This is explained by noting that the density corresponding to F(x) is impulsive:


Chapter 11 Problem Solutions 201

f (x)
1
3/4
1/2
1/4
x
0 1 1 2
2

42. For an integer-valued process,

m,n (B) = IB ( jm , . . . , jn )pm,n ( jm , . . . , jn ).


jm ,..., jn

If B = {im } {in }, then

m,n+1 (B IR) = IBIR ( jm , . . . , jn , j)pm,n+1 ( jm , . . . , jn , j)


jm ,..., jn , j

= pm,n+1 (im , . . . , in , j),


j

m1,n (IR B) = IIRB ( j, jm , . . . , jn )pm1,n ( j, jm , . . . , jn )


j, jm ,..., jn

= pm1,n ( j, im , . . . , in ),
j

and m,n (B) = pm,n (im , . . . , in ). Hence, if the conditions

m,n+1 (B IR) = m,n (B) and m1,n (IR B) = m,n (B)

hold for all B and we take B = {im } {in }, then we obtain

pm,n+1 (im , . . . , in , j) = pm,n (im , . . . , in )


j

and
pm1,n ( j, im , . . . , in ) = pm,n (im , . . . , in ).
j

Conversely, suppose these two formulas hold. Fix an arbitrary B IRnm+1 . If we


multiply both formulass by IB (im , . . . , in ) and sum over all im , . . . , in , we obtain the
original consistency conditions.
43. For the first condition, write

pm,n+1 (im , . . . , in , j) = q(im )r(im+1 |im ) r(in |in1 )r( j|in )


j j

= q(im )r(im+1 |im ) r(in |in1 )


| {z } r( j|in ) .
j
= pm,n (im ,...,in ) | {z }
=1
202 Chapter 11 Problem Solutions

For the second condition, write

pm1,n ( j, im , . . . , in ) = q( j)r(im | j)r(im+1 |im ) r(in |in1 )


j j
 
= q( j)r(im | j) r(im+1 |im ) r(in |in1 ) .
j
| {z }
= q(im )
| {z }
= pm,n (im ,...,in )

44. If

n (Bn ) = n+1 (Bn IR)


Z Z
= IBn (x1 , . . . , xn )IIR (y) fn+1 (x1 , . . . , xn , y) dy dxn dx1

Z Z Z 
= fn+1 (x1 , . . . , xn , y) dy dxn dx1 ,

Bn

then necessarily the quantity in square brackets is the joint density fn (x1 , . . . , xn ).
Conversely, if the quantity in square brackets is equal to fn (x1 , . . . , xn ), we can write
Z Z
n+1 (Bn IR) = IBn (x1 , . . . , xn )IIR (y) fn+1 (x1 , . . . , xn , y) dy dxn dx1

Z Z Z 
= fn+1 (x1 , . . . , xn , y) dy dxn dx1

Bn
Z Z
= fn (x1 , . . . , xn )dxn dx1 = n (Bn ).
Bn

45. The key is to write

t1 ,...,tn+1 (Bn,k )
Z Z
= IBn (x1 , . . . , xk1 , xk+1 , . . . , xn )IIR (xk ) ft1 ,...,tn+1 (x1 , . . . , xn+1 ) dxn+1 dx1

Z Z Z 
= ft1 ,...,tn+1 (x1 , . . . , xn+1 ) dxk dxn+1 dxk+1 dxk1 dx1 .

Bn

Then if t1 ,...,tn+1 (Bn,k ) = n (Bn ), we must have


Z
ft1 ,...,tk1 ,tk+1 ,...,tn (x1 , . . . , xk1 , xk+1 , . . . , xn ) = ft1 ,...,tn+1 (x1 , . . . , xn+1 ) dxk

and conversely.
46. By the hint, [Wt1 , . . . ,Wtn ]0 is a linear transformation of a vector of independent Gaus-
sian increments, which is a Gaussian vector. Hence, [Wt1 , . . . ,Wtn ]0 is a Gaussian
vector. Since n and the times t1 < < tn are arbitrary, Wt is a Gaussian process.
Chapter 11 Problem Solutions 203

47. Let X = [Wt1 W0 , . . . ,Wtn Wtn1 ]0 , and let Y = [Wt1 , . . . ,Wtn ]0 . Then Y = AX, where
A denotes the matrix given in the statement of Problem 46. Since the components of
X are independent,
2
n
exi /[2(ti ti1 )]
fX (x) = p ,
i=1 2 (ti ti1 )
where it is understood that t0 := 0. Using the example suggested in the hint, we have

fX (x)
fY (y) = .
| det A| x=A1 y

Fortunately, det A = 1, and by definition, Xi = Wti Wti1 . Hence,


2
e(wi wi1 ) /[2(ti ti1 )]
n
ft1 ,...,tn (w1 , . . . , wn ) = p2 (t t ) ,
i=1 i i1

where it is understood that w0 := 0.


48. By the hint, it suffices to show that C is positive semidefinite. Write
n n n n
a0Ca = ai akCik = ai ak R(ti tk )
i=1 k=1 i=1 k=1
n n Z
= ai ak S( f )e j2 f (ti tk ) d f
i=1 k=1
Z n 2

= S( f ) ai e j2 f ti d f 0.
i=1
CHAPTER 12
Problem Solutions

1. If P(A|X = i,Y = j, Z = k) = h(i), then

P(A, X = i,Y = j, Z = k) = P(A|X = i,Y = j, Z = k)P(X = i,Y = j, Z = k)


= h(i)P(X = i,Y = j, Z = k).

Summing over j yields

P(A, X = i, Z = k) = h(i)P(X = i, Z = k), ()

and thus h(i) = P(A|X = i, Z = k). Next, summing () over k yields

P(A, X = i) = h(i)P(X = i),

and then h(i) = P(A|X = i) as well.


2. Write

X1 = g(X0 , Z1 )
X2 = g(X1 , Z2 ) = g(g(X0 , Z1 ), Z2 )
X3 = g(X2 , Z3 ) = g(g(g(X0 , Z1 ), Z2 ), Z3 )
..
.

In general, Xn is a function of X0 , Z1 , . . . , Zn , and thus (X0 , . . . , Xn ) is a function of


(X0 , Z1 , . . . , Zn ), which is independent of Zn+1 . Now observe that

P(Xn+1 = in+1 |Xn = in , . . . , X0 = i0 ) = P(g(Xn , Zn+1 ) = in+1 |Xn = in , . . . , X0 = i0 )


= P(g(in , Zn+1 ) = in+1 |Xn = in , . . . , X0 = i0 )
= P(g(in , Zn+1 ) = in+1 ),

where we have used substitution and independence. Since

P(Xn+1 = in+1 |Xn = in , . . . , X0 = i0 )

does not depend on in1 , . . . , i0 , Xn is a Markov chain.


3. Write

P(A B C) P(A B C) P(B C)


P(A B|C) = = = P(A|B C)P(B|C).
P(C) P(B C) P(C)

204
Chapter 12 Problem Solutions 205

4. Write

P(X0 = i, X1 = j, X2 = k, X3 = l)
= P(X3 = l|X2 = k, X1 = j, X0 = i)
P(X2 = k|X1 = j, X0 = i)P(X1 = j|X0 = i)P(X0 = i)
= pkl p jk pi j i .

5. The first equation we write is

0 = k pk0 as 0 = 0 p00 + 1 p10 = 0 (1 a) + 1 b.


k

This tells us that 1 = (a/b)0 . Then we use the fact that 0 + 1 = 1 to write 0 +
(a/b)0 = 1, or 0 = 1/(1 + a/b) = b/(a + b). We also have 1 = (a/b)0 = a/(a +
b).
6. The state transition diagram is

1/2

1/2 1/2
1/4

1/2
1 2

3/4

The stationary distribution is 0 = 5/12, 1 = 1/3, 2 = 1/4.


7. The state transition diagram is

1/2
1/4
1/4
1/2

1 2

3/4 3/4

The stationary distribution is 0 = 1/5, 1 = 7/10, 2 = 1/10.


206 Chapter 12 Problem Solutions

8. The state transition diagram is

1/2
1/2

0 9/10 1

1/10 1/10
1/2
3 2
9/10
1/2

The stationary distribution is 0 = 9/28, 1 = 5/28, 2 = 5/28, 3 = 9/28.


9. The first equation is

0 = 0 p00 + 1 p10 = 0 (1 a) + 1 b,

which tells us that 1 = (a/b)0 . The next equation is

1 = 0 p01 + 1 p11 + 2 p21 = 0 a + 1 (1 [a + b]) + 2 b,

which tells us that

(a + b)1 = 0 a + 2 b or (a + b)(a/b)0 = 0 a + 2 b,

from which it follows that 2 = (a/b)2 0 . Now suppose that i = (a/b)i 0 holds for
i = 0, . . . , j < N. Then from

j = j1 p j1, j + j p j j + j+1 p j+1, j = j1 a + j (1 [a + b]) + j+1 b,

we obtain

(a + b) j = j1 a + j+1 b or (a + b)(a/b) j 0 = (a/b) j1 a + j+1 b,

from which it follows that j+1 = (a/b) j+1 0 . To find 0 , we use the fact that 0 +
+ N = 1. Using the finite geometric series formula, we have
N
1 (a/b)N+1
1 = (a/b) j 0 = 0
1 a/b
.
j=0

Hence,
1 a/b  a  j
j = , j = 0, . . . , N, a 6= b.
1 (a/b)N+1 b
If a = b, then j = 0 for j = 0, . . . , N implies j = 1/(N + 1).
Chapter 12 Problem Solutions 207

10. Let and b denote the two solutions in the example. For 0 1, put

ej := j + (1 )bj 0.

Then
 
ej = j + (1 )bj = j + (1 ) bj = 1 + (1 ) 1 = 1.
j j j j

Hence, e is a valid pmf. Finally, note that

ei pi j = [ i + (1 )bi ]pi j = i pi j + (1 ) bi pi j
i i i i
= j + (1 )bj = ej .

11. MATLAB. OMITTED.


12. Write

E[T1 ( j)|X0 = i] = kP(T1 ( j) = k|X0 = i) + P(T1 ( j) = |X0 = i)
k=1

(k)
= k fi j + (1 fi j ).
k=1

13. (a) First note that {T1 = 2} = {X2 = j, X1 6= j} and that


h
{T2 = 5} = {X5 = j} {X4 = j, X3 6= j, X2 6= j, X1 6= j}
{X4 6= j, X3 = j, X2 6= j, X1 6= j}
{X4 6= j, X3 6= j, X2 = j, X1 =6 j} i
{X4 6= j, X3 6= j, X2 6= j, X1 = j} .

Then

{T2 = 5} {T1 = 2} = {X5 = j, X4 6= j, X3 6= j, X2 = j, X1 6= j}.

It now follows that

P(T2 = 5|T1 = 2, X0 = i) = P(X5 = j, X4 6= j, X3 6= j|X2 = j, X1 6= j, X0 = i).

(b) Write
P(X5 = j, X4 6= j, X3 6= j, X2 = j, X1 6= j, X0 = i)
as  
[
P {X5 = j, X4 6= j, X3 6= j, X2 = j, X1 = l, X0 = i} ,
l6= j

which is equal to

P(X5 = j, X4 6= j, X3 6= j|X2 = j, X1 = l, X0 = i)P(X2 = j, X1 = l, X0 = i).


l6= j
208 Chapter 12 Problem Solutions

By the Markov property, this simplifies to

P(X5 = j, X4 6= j, X3 6= j|X2 = j)P(X2 = j, X1 = l, X0 = i),


l6= j

which is just
P(X5 = j, X4 6= j, X3 6= j|X2 = j)P(X2 = j, X1 6= j, X0 = i).
It now follows that P(X5 = j, X4 6= j, X3 6= j|X2 = j, X1 6= j, X0 = i) is equal to
P(X5 = j, X4 6= j, X3 6= j|X2 = j).
(c) Write
P(X5 = j, X4 6= j, X3 6= j|X2 = j) = P(X5 = j, X4 = k, X3 = l|X2 = j)
k6= j,l6= j

= P(X3 = j, X2 = k, X1 = l|X0 = j)
k6= j,l6= j
= P(X3 = j, X2 6= j, X1 6= j|X0 = j).

14. Write
 
\
P(V ( j) = |X0 = i) = P {V ( j) L} X0 = i
L=1
 M 
\
= lim P
{V ( j) L} X0 = i , limit property of P,
M
L=1
= lim P(V ( j) M|X0 = i), decreasing events.
M

15. First write






E[V ( j)|X0 = i] = E I{ j} (Xn ) X0 = i = E[I{ j} (Xn )|X0 = i]
n=1 n=1

(n)
= P(Xn = j|X0 = i) = pi j . ()
n=1 n=1

Next,
P(V ( j) = |X0 = i) = lim P(V ( j) L|X0 = i), from preceding problem,
L
= lim fi j ( f j j )L1 , from the text.
L

Now, if j is recurrent ( f j j = 1), the above limit is fi j . If fi j > 0, then V ( j) = with


positive probability, and so the expectation on the left in () must be infinite, and
hence, so is the sum on the far right in (). If fi j = 0, then for any n = 1, 2, . . . , we
can write
 
[
0 = fi j := P(T1 ( j) < |Xi = 0) P
{Xm = j} Xi = 0
m=1
(n)
P(Xn = j|X0 = i) = pi j ,
Chapter 12 Problem Solutions 209

(n)
and it follows that
n=1 pi j = 0.
In the transient case ( f j j < 1), we have from the text that

P(V ( j) L|X0 = i) = fi j ( f j j )L1 .

Using the preceding problem along with f j j < 1, we have

P(V ( j) = |X0 = i) = lim fi j ( f j j )L1 = 0.


L

Now, we also have from the text that



fi j ( f j j )L1 (1 f j j ), L 1,
P(V ( j) = L|X0 = i) =
1 fi j , L = 0,
where we use the fact that the number of visits to state j is zero if and only if we never
visit state j, and this happens if and only if T1 ( j) = . We can therefore write

fi j
E[V ( j)|X0 = i] = L P(V ( j) = L|X0 = i) =
1 fjj
< ,
L=0

and it follows that the last sum in () is finite too.


16. From m 1 bm c m ,
1 1 1
.
m bm c m 1
Then
1 m 1 1
.
bm c 1/m
17. (a) Write h( j) = IS ( j) = nl=1 I{sl } ( j). Then
 
1 m 1 m n n
1 m
lim
m m
h(Xk ) = m lim
m k=1 I{sl } (Xk ) = m
lim
m {sl }
I (Xk )
k=1 l=1 l=1 k=1
n n
= m
lim Vm (sl )/m = s l , by Theorems 3 and 4,
l=1 l=1
= IS ( j) j = h( j) j .
j j

(b) With h( j) = nl=1 cl ISl ( j), where each Sl is a finite subset of states, we can write
 
1 m 1 m n n
1 m
lim
m m
h(Xk ) = lim
m m
cl ISl (Xk ) = cl lim
m m
ISl (Xk )
k=1 k=1 l=1 l=1 k=1
n  
= cl ISl ( j) j , by part (a)
l=1 j
 n 
= l Sl j =
c I ( j) h( j) j .
j l=1 j
210 Chapter 12 Problem Solutions

(c) If h( j) = 0 for all but finitely many states, then there are at most finitely many
distinct nonzero values that h( j) can take, say c1 , . . . , cn . Put Sl := { j : h( j) =
cl }. By the assumption about h, each Sl is a finite set. Furthermore, we can
write
n
h( j) = cl ISl ( j).
l=1

Hence, the desired result is immediate by part (b).


18. MATLAB. OMITTED.
19. MATLAB. OMITTED.
20. Suppose k Ai A j . Since k Ai , i k. Since k A j , j k. Hence, i j. Now,
if l Ai , l i j and we see that l A j . Similarly, if l A j , l j i and we see
that l Ai . Thus, Ai = A j .
21. Yes. As pointed out in the discussion at the end of the section, by combining Theo-
rem 8 and Theorem 6, the chain has a unique stationary distribution. Now, by The-
orem 4, all states are positive recurrent, which by definition means that the expected
time to return to a state is finite.
22. Write

( t)n e t
gi,i+n = lim = lim ( t)n1 e t = 0, for n 2.
t0 t n! n! t0

23. The state transition diagram is

1 5
2
4
1 2

The stationary distribution is 0 = 11/15, 1 = 1/5, 2 = 1/15.


24. MATLAB. Change the line A=P-In; to A=P;.
25. The forward equation is

j+1
p0i j (t) = pik (t)gk j = pik gk j
k k= j1
= pi, j1 (t) j1 pi j (t)[ j + j ] + pi, j+1 (t) j+1 .
Chapter 12 Problem Solutions 211

The backward equation is


i+1
p0i j (t) = gik pk j (t) = gik pi j (t)
k k=i1
= i pi1, j (t) (i + i )pii (t) + i pi+1, j (t).
The chain is conservative because gi,i1 +gi,i+1 = i + i = [(i + i )] = gii < .
26. We use the formula 0 = k k gk j . For j = 0, we have
0 = k gk0 = 0 (0 ) + 1 1 ,
k

which implies 1 = (0 /1 )0 . For j = 1, we have


0 = k gk1 = 0 0 + 1 [(1 + 1 )] + 2 2
k
= 0 0 + (0 /1 )0 [(1 + 1 )] + 2 2 = 0 (0 1 /1 ) + 2 2 ,
which implies 2 = 0 0 1 /(1 2 ). Now suppose that i = 0 0 i1 /(1 i )
for i = 1, . . . , j. Then from
0 = k gk j = j1 j1 j ( j + j ) + j+1 j+1
k
0 j2 0 j1
= 0 j1 0 ( j + j ) + j+1 j+1 ,
1 j1 1 j
and it follows that
0 j
j+1 = 0 .
1 j+1
Now, let

0 j1
B := < .
j=1 1 j

From the condition j=0 j = 1, we have



0 j1
1 = 0 + 1 + = 0 + 0 = 0 (1 + B).
j=1 1 j

It then follows that 0 = 1/(1 + B), and


 
1 0 j1
j = , j 1.
1 + B 1 j

If i = and i = , then B = j=1 ( / ) j , which is finite if and only if < . In


in this case,
 k
1
1+B = = ,
k=0 1 + /
and
j = (1 / )( / ) j geometric0 ( / ).
212 Chapter 12 Problem Solutions

27. The solution is very similar the that of the preceding problem. In this case, put
N
0 j1
BN := < ,
j=1 1 j

and then 0 = 1/(1 + BN ), and


 
1 0 j1
j = , j = 1, . . . , N.
1 + BN 1 j

If i = and i = , then
N
1 ( / )N+1
1 + BN = ( / )k =
1 /
,
k=0

and
1 /
j = ( / ) j , j = 0, . . . , N.
1 ( / )N+1
If = , then j = 1/(N + 1).
28. To begin, write

mi (t) := E[Xt |X0 = i] = jP(Xt = j|X0 = i) = j pi j (t).


j j

Then
   
m0i (t) = j p0i j (t) = j pik (t)gk j = pik (t) jgk j
j j k k j
h i
= pik (t) (k 1)gk,k1 + kgkk + (k + 1)gk,k+1
k
h i
= pik (t) (k 1)(k ) k(k + + k ) + (k + 1)(k + )
k
= pik (t)[k + k + ] = mi (t) + mi (t) + .
k

We must solve
m0i (t) + ( )mi (t) = , mi (0) = i.
If 6= , it is readily verified that
 

mi (t) = i e( )t +

solves the equation. If = , then mi (t) = t + i solves m0i (t) = with mi (0) = i.
29. To begin, write

p0i j (t) = pik (t)gk j = pi, j1 (t)g j1, j + pi j (t)g j j = pi, j1 (t) pi j (t). ()
k
Chapter 12 Problem Solutions 213

Observe that for j = i, pi,i1 (t) = 0 since the chain can not go from state i to a lower
state. Thus,
p0ii (t) = pii (t).
Also, pii (0) = P(Xt = i|X0 = i) = 1. Thus, pii (t) = e t , and it is easily verified that

( t)n e t
pi,i+n (t) = , n = 0, 1, 2, . . . ,
n!
solves (). Thus, Xt is a Poisson process of rate with X0 = i instead of X0 = 0.
30. (a) The assumption that D > implies that the k are not all zero, in which case,
the k would not sum to one and would not be a pmf. Aside from this, it is clear
that k 0 for all k and that k k = D/D = 1. Next, since p j j = 0, write

1
gk j
k pk j = [k gkk /D] gkk k gk j
D k6= j
=
k k6= j
  
1
=
D
k kj
g j j j = j g j j /D = j .
g
k
| {z }
=0

(b) The assumption that D > implies that the k are not all zero. Aside from
this, it is clear that k 0 for all k and that k k = D/D = 1. Next, write

1 gk j
k gk j = [(k /gkk )/D]gk j =
D k
k
g kk
k k
  
1 gk j
=
D
k gkk j
k6= j
  
1
=
D
k pk j j , since p j j = 0,
k
| {z }
= j
= 0.

(c) Since k and k are pmfs, they sum to one. Hence, if gii = g for all i, D = g
and D = 1/g. In (a), k = k gkk /D = k g/g = k . In (b), k = (k /gkk )/D =
(k /g)/(1/g) = k .
31. Following the hints, write

P(T > t + t|T > t, X0 = i) = P(Xs = i, 0 s t + t|Xs = i, 0 s t)


= P(Xs = i,t s t + t|Xs = i, 0 s t)
= P(Xs = i,t s t + t|Xt = i)
= P(Xs = i, 0 s t|X0 = i)
= P(T > t|X0 = i).
214 Chapter 12 Problem Solutions

The exponential parameter is


1 P(T > t|X0 = i) 1 P(Xs = i, 0 s t|X0 = i)
lim = lim
t0 t t0 t
1 P(Xt = i|X0 = i)
= lim =: gii .
t0 t
32. Using substitution in reverse, write
P(Wt y|Ws = x,Wsn1 = xn1 , . . . ,Ws0 = x0 )
= P(Wt x y x|Ws = x,Wsn1 = xn1 , . . . ,Ws0 = x0 )
= P(Wt Ws y x|Ws = x,Wsn1 = xn1 , . . . ,Ws0 = x0 ).
Now, using the fact that W0 0, this last conditional probability is equal to
P(Wt Ws y x|Ws Wsn1 = x xn1 , . . . ,Ws1 Ws0 = x1 x0 ,Ws0 W0 = x0 ).
Since the Wiener process has independent increments that are Gaussian, this last ex-
pression reduces to
Z yx
exp[{ /[ t s ]}2 /2]
P(Wt Ws y x) = p d .
2 2 (t s)
Since this depends on x but not on xn1 , . . . , x0 ,
P(Wt y|Ws = x,Wsn1 = xn1 , . . . ,Ws0 = x0 ) = P(Wt y|Ws = x).
Hence, Wt is a Markov process.
33. Write
P(X = x|Y = y, Z = z)P(Y = y|Z = z)
y
P(X = x,Y = y, Z = z) P(Y = y, Z = z)
= P(Y = y, Z = z)

P(Z = z)
y
1
P(Z = z)
= P(X = x,Y = y, Z = z)
y
P(X = x, Z = z)
= = P(X = x|Z = z).
P(Z = z)
34. Using the law of total conditional probability, write
Pt+s (B) = P(Xt+s B|X0 = x)
Z
= P(Xt+s B|Xs = z, X0 = x) fXs |X0 (z|x) dz

Z
= P(Xt+s B|Xs = z) fs (x, z) dz

Z
= P(Xt B|X0 = z) fs (x, z) dz

Z
= Pt (z, B) fs (x, z) dz.

CHAPTER 13
Problem Solutions

1. First observe that E[|Xn | p ] = n p/2 P(U 1/n) = n p/2 (1/n) = n p/21 , which goes to
zero if and only if p < 2. Thus, 1 p < 2.
 2 
Nt 1 var(Nt ) t
2. E = 2 E[|Nt t|2 ] = = 2 = 0.
t t t2 t t
3. Given > 0, let n N imply |C(k)| /2. Then for such n,

1 n1 1 N1 1 n

n k=0
C(k) =
n k=0
C(k) + C(k)
n k=N

and n1
1 N1 n N1
C(k) 1 |C(k)| + 1 /2 1 |C(k)| + /2.
n n k=0 n k=N n k=0
k=0

For large enough n, the right-hand side will be less that .


4. Applying the hint followed by the CauchySchwarz inequality shows that
n1 q
1
C(k) E[(X1 m)2 ]E[(Mn m)2 ].
n
k=0

Hence, if Mn converges in mean square to m, the left-hand side goes to zero as n .


5. Starting with the hint, we write

E[ZIA ] = E[(Z Zn )IA ] + E[Zn IA ]


E[Z Zn ] + nE[IA ], since Zn Z and Zn n,
= E[|Z Zn |] + nP(A).

Since Zn converges in mean to Z, given > 0, we can let n N imply E[|Z Zn |]


/2. Then in particular we have

E[ZIA ] < /2 + NP(A).

Hence, if 0 < < /(2N),

P(A) < implies E[ZIA ] < /2 + N /(2N) = .

6. Following the hint, we find that given > 0, there exists a > 0 such that
Z x
P(U x) = x < implies E[ f (x +U)I{Ux} ] = f (x + t) dt < .
0

215
216 Chapter 13 Problem Solutions

Now make the change of variable = x + t to get


Z x Z x+x
f (x + t) dt = f ( ) d = F(x + x) F(x).
0 x

For 0 < x < , take A = {U > 1 + x} and Z = f (x 1 + U). Then P(U >
1 + x) = x implies
Z 1
E[ f (x + 1 +U)I{U>1+Dx} ] = f (x 1 + t) dt < .
1+x

In this last integral make the change of variable = x 1 + t to get


Z x
f ( ) d = F(x) F(x + x).
x+x

Thus, F(x + x) F(x) > . We can now write that given > 0, there exists a
such that |x| < implies |F(x + x) F(x)| < .
7. Following the hint, we can write
h1 1 i 1 p 1 q
exp ln(|X|/ ) p + ln(|Y |/ )q eln(|X|/ ) + eln(|Y |/ )
p q p q
or h i    
1 |X| p 1 |Y | q
exp ln(|X|/ ) + ln(|Y |/ ) +
p q
or
|XY | 1 |X| p 1 |Y |q
+ .
p p q q
Hence,
E[|XY |] 1 E[|X| p ] 1 E[|Y |q ] 1 p 1 q
+ = + = 1.
p p q q p p q q
The hint assumes neither nor is zero or infinity. However, if either or is zero,
then both sides of the inequality are zero. If neither is zero and one of them is infinity,
then the right-hand side is infinity and the inequality is trivial.
8. Following the hint, with X = |Z| , Y = 1, and p = / , we have

E[|Z| ] = E[|XY |] E[|X| p ]1/p E[|Y |q ]1/q = E[(|Z| ) / ] / 1 = E[|Z| ] / .

Raising both sides to the 1/ yields E[|Z| ]1/ E[|Z| ]1/ .


9. By Lyapunovs inequality, E[|Xn X| ]1/ E[|Xn X| ]1/ . Raising both sides to
the power yields E[|Xn X| ] E[|Xn X| ] / . Hence, if E[|Xn X| ] 0, then
E[|Xn X| ] 0 too.
10. Following the hint, write

E[|X +Y | p ] = E[|X +Y | |X +Y | p1 ]
E[|X| |X +Y | p1 ] + E[|Y | |X +Y | p1 ]
E[|X| p ]1/p E[(|X +Y | p1 )q ]1/q + E[|Y | p ]1/p E[(|X +Y | p1 )q ]1/q ,
Chapter 13 Problem Solutions 217

where 1/q := 1 1/p. Hence, 1/q = (p 1)/p and q = p/(p 1). Now divide the
above inequality by E[|X +Y | p ](p1)/p to get

E[|X +Y | p ]1/p E[|X| p ]1/p + E[|Y | p ]1/p .

11. For a Wiener process, Wt Wt0 N(0, 2 |t t0 |). To simplify the notation, put 2 :=
2 |t t0 |. Then
Z 2 Z (x/ )2 /2
e(x/ ) /2 e
E[|Wt Wt0 |] = |x| dx = 2x dx
2 2
0
Z r
2 t 2 /2 2  t 2 /2  |t t0 |
= te dt = e = 2 ,
2 0 2 0 2

which goes to zero as |t t0 | 0.


12. Let t0 > 0 be arbitrary. Since E[|Nt Nt0 |2 ] = |t t0 | + 2 |t t0 |2 , it is clear that as
t t0 , E[|Nt Nt0 |2 ] 0. Hence, Nt is continuous in mean square.
13. First write

R(t, s) R( , ) = R(t, s) R(t, ) + R(t, ) R( , )


= E[Xt (Xs X )] + E[(Xt X )X ].

Then by the CauchySchwarz inequality,


q q
|R(t, s) R( , )| E[Xt2 ]E[(Xs X )2 ] + E[(Xt X )2 ]E[X2 ],

which goes to zero as (t, s) ( , ). Note that we need the boundedness of E[Xt2 ] for
t near .
14. (a) E[|Xt+T Xt |2 ] = R(0) 2R(T ) + R(0) = 0.
(b) Write
q
|R(t + T ) R(t)| = E[(Xt+T Xt )X0 ] E[|Xt+T Xt |2 ]E[X02 ],

which is zero by part (a). Hence, R(t + T ) = R(t) for all t.


15. Write

k(Xn +Yn ) (X +Y )k p = k(Xn X) + (Yn Y )k p kXn Xk p + kYn Y k p 0.

16. Let t0 be arbitrary Since

kZt Zt0 k p = k(Xt +Yt ) (Xt0 +Yt0 )k p


= k(Xt Xt0 ) + (Yt Yt0 )k p
kXt Xt0 k p + kYt Yt0 k p ,

it is clear that if Xt and Yt are continuous in mean of order p, then so is Zt := Xt +Yt .


218 Chapter 13 Problem Solutions

17. Let > 0 be given. Since kXn Xk p 0, there is an N such that for n N, kXn
Xk p < /2. Thus, for n, m N,
kXn Xm k p = k(Xn X) + (X Xm )k p kXn Xk p + kX Xm k p < /2 + /2 = .

18. Suppose Xn is Cauchy in L p . With = 1, there is an N such that for all n, m N,


kXn Xm k p < 1. In particular, with m = N we have from

kXn k p kXN k p kXn XN k p

that
kXn k p kXn XN k p + kXN k p < 1 + kXN k p , for n N.
To get a bound that also holds for n = 1, . . . , N 1, write
kXn k p max(kX1 k p , . . . , kXN1 k p , 1 + kXN k p ).

19. Since Xn converges, it is Cauchy and therefore bounded by the preceding two prob-
lems. Hence, we can write kXn k p B < for some constant B. Given > 0, let
n N imply kXn Xk p < /(2kY kq ) and kYn Y kq < /(2B). Then
kXnYn XY k1 = E[|XnYn XnY + XnY XY |]
E[|Xn (Yn Y )|] + E[|(Xn X)Y |]
kXn k p kYn Y kq + kXn Xk p kY kq , by Holders inequality,

< B + kY kq = .
2B 2kY kq

20. If Xn converges in mean of order p to both X and Y , write


kX Y k p = k(X Xn ) + (Xn Y )k p kX Xn k p + kXn Y k p 0.
Since kX Y k p = 0, E[|X Y | p ] = 0.
21. For the rightmost inequality, observe that
kX Y k p = kX + (Y )k p kXk p + k Y k p = kXk p + kY k p .
For the remaining inequality, first write
kXk p = k(X Y ) +Y k p kX Y k p + kY k p ,
from which it follows that
kXk p kY k p kX Y k p . ()
Similarly, from
kY k p = k(Y X) + Xk p kY Xk p + kXk p ,
it follows that
kY k p kXk p kY Xk p = kX Y k p . ()
From () and () it follows that

kXk p kY k p kX Y k p .
Chapter 13 Problem Solutions 219

22. By taking pth roots, we see that


lim E[|Xn | p ] = E[|X| p ] (#)
n

is equivalent to kXn k p kXk p . Since



kXk p kY k p kX Y k p ,

we see that convergence in mean of order p implies (#).


23. Write
kX +Y k2p + kX Y k2p = hX +Y, X +Y i + hX Y, X Y i
= hX, Xi + 2hX,Y i + hY,Y i + hX, Xi 2hX,Y i + hY,Y i
= 2(kXk2p + kY k2p ).

24. Write
|hXn ,Yn i hX,Y i| = |hXn ,Yn i hX,Yn i + hX,Yn i hX,Y i|
khXn X,Y i| + |hX,Yn Y i|
kXn Xk2 kYn k2 + kXk2 kYn Y k2 0.
Here we have used the CauchySchwarz inequality and the fact that since Yn con-
verges, it is bounded.
25. As in the example, for n > m, we can write
n n
kYn Ym k22 |hk | |hl | |hXk , Xl i|.
k=m+1 l=m+1

In this problem, hXk , Xl i = 0 for k 6= l. Hence,


n n
kYn Ym k22 B|hk |2 B |hk |2 0
k=m+1 k=m+1

as n > m on account of the assumption that 2


k=1 |hk | < .

26. Put Yn := nk=1 hk Xk . It suffices to show that Yn is Cauchy in L p . Write, for n > m,
n n n

kYn Ym k p = hk Xk

|hk | kXk k p = B1/p
|hk | 0
k=m+1 p k=m+1 k=m+1

as n > m on account of the assumption that


k=1 |hk | < .

27. Write
 n  
E[Y Z] = E i i i1 j j j1
X (t t ) X (s s )
i=1 j=1
n
= R(i , j )(ti ti1 )(s j s j1 ).
i=1 j=1
220 Chapter 13 Problem Solutions

28. First note that if


n
Y := g(i )Xi (ti ti1 ) and Z := g( j )X j (s j s j1 ),
i=1 j=1

then
n
E[Y Z] = g(i )R(i , j )g( j )(ti ti1 )(s j s j1 ).
i=1 j=1

Hence, given finer and finer partitions, with Ym defined analogously to Y above, we
see that

E[|Ym Yk |2 ] = E[Ym2 ] 2E[YmYk ] + E[Yk2 ]


Z bZ b Z bZ b
g(t)R(t, s)g(s) dt ds 2 g(t)R(t, s)g(s) dt ds
a a a a
Z bZ b
+ g(t)R(t, s)g(s) dt ds = 0.
a a

Thus, Ym is Cauchy in L2 , and there exists a Y L2 with kYm Y k2 0. Furthermore,


since Ym converges in mean square to Y , E[Ym2 ] E[Y 2 ], and it is clear that
Z bZ b
E[Ym2 ] g(t)R(t, s)g(s) dt ds.
a a

29. Consider the formula


Z T
R(t s) (s) ds = (t), 0 t T. ()
0

Since R is defined for all t, we can extend the definition of on the right-hand side in
the obvious way. Furthermore, since R has period T , so will the extended definition
of . Hence, both R and have Fourier series representations, say

R(t) = rn e j2 nt/T and (s) = n e j2 ns/T .


n n

Substituting these into () yields


Z T
rn e j2 nt/T 0
e j2 ns/T (s) ds = n e j2 nt/T .
n | {z } n
= T n

It follows that Trn n = n . Now, if is an eigenfunction, cannot be the zero


function. Hence, there is at least one value of n with n 6= 0. For all n with n 6= 0,
= Trn . Thus,
(t) = n e j2 nt/T .
n:rn = /T
Chapter 13 Problem Solutions 221

Rb
30. If a R(t, s) (s) ds = (t) then
Z bZ b Z b Z b 
0 R(t, s) (t) (s) dt ds = (t) R(t, s) (s) ds dt
a a a a
Z b Z b
= (t)[ (t)] dt = (t)2 dt.
a a

Hence, 0.
31. To begin, write
Z b Z b Z b Z b 
k k (t)m (t) dt = k k (t)m (t) dt = R(t, s)k (s) ds m (t) dt
a a a a
Z b Z b 
= k (s) R(s,t)m (t) dt ds, since R(t, s) = R(s,t),
a a
Z b Z b
= k (s) m m (s) ds = m k (s)m (s) ds.
a a

We can now write Z b


(k m ) k (t)m (t) dt = 0.
a
Rb
If k 6= m , we must have a k (t)m (t) dt = 0.
32. (a) Write
Z T Z T 

0
R(t, s)g(s) ds =
0
k k (t)k (s) g(s) ds
k=1
Z T
= k k (t) 0
g(s)k (s) ds = k gk k (t).
k=1 k=1

(b) Write
Z T Z T  
R(t, s) (s) ds = R(t, s) g(s) gk k (s) ds
0 0 k=1
Z T Z T
= R(t, s)g(s) ds gk R(t, s)k (s) ds
0 k=1 |0 {z }
= k k (t)

= k gk k (t) k gk k (t) = 0.
k=1 k=1

(c) Write
Z TZ T Z T Z T

E[Z 2 ] = (t)R(t, s) (s) dt ds = (t) R(t, s) (s) ds dt = 0.
0 0 0 0
222 Chapter 13 Problem Solutions

33. Consider the equation


Z T
(t) = e|ts| (s) ds (#)
0
Z t Z T
= e(ts) (s) ds + e(st) (s) ds
0 t
Z t Z T
t
= e s
e (s) ds + e t
es (s) ds.
0 t

Differentiating yields
Z t Z T
0 (t) = et es (s) ds + et et (t) + et es (s) ds + et (et (t))
0 t
Z t Z T
= et es (s) ds + et es (s) ds. (##)
0 t

Differentiating again yields


Z t Z T
00 (t) = et es (s) ds et et (t) + et es (s) ds + et (et (t))
0 t
Z T
= e|ts| (s) ds 2 (t)
0
= (t) 2 (t), by (#),
= ( 2) (t).

We can now write

00 (t) = (1 2/ ) (t) = (2/ 1) (t).


p
For 0 < < 2, put := 2/ 1. Then 00 (t) = 2 (t). Hence, we must have

(t) = A cos t + B sin t

for some constants A and B.


34. The required orthogonality principle is that
 L 
E (Xt Xbt ) ci Ai = 0
i=1

for all constants ci , where


L
Xbt = cbj A j .
j=1

In particular, we must have E[(Xt Xbt )Ai ] = 0. Now, we know from the text that
E[Xt Ai ] = i i (t). We also have
L
E[Xbt Ai ] = cbj E[A j Ai ] = cbi i .
j=1
Chapter 13 Problem Solutions 223

Hence, cbi = i (t), and we have


L
Xbt = Ai i (t).
i=1

35. Since kYn Y k2 0, by the hint, E[Yn ] E[Y ] and E[Yn2 ] E[Y 2 ]. Since gn is
piecewise constant, we know that E[Yn ] = 0, and so E[Y ] = 0 too. Next, an argument
analogous to the one in Problem 21 tells us that if kgn gk 0, then kgn k kgk.
Hence,
Z Z
2
E[Y ] = lim E[Yn2 ] = lim 2 2
gn (t) dt = 2
g(t)2 dt.
n n 0 0

36. First write


kY Ye k2 = kY Yn k2 + kYn Yen k2 + kYen Ye k2 ,
where the first and last terms on the right go to zero. As for the middle term, write
 Z Z 2 
kYn Yen k22 = E gn (t) dWt gen (t) dWt
0 0
Z
2  Z
= E [gn (t) gen (t)] dWt = 2 [gn (t) gen (t)]2 dt.
0 0

We can now write


kYn Yen k2 = kgn gen k kgn gk + kg gen k 0.

37. We know from our earlier work that the Wiener integralR
is linear on piecewise-con-
R
stant functions. To analyze theRgeneral case, let Y = 0 g(t) dWt and Z = 0 h(t) dWt .
We must show that aY + bZ = 0 ag(t) + bh(t) dWt . Let gn (t) and hn (t) be piecewise-
constant
R
functions suchR that kgn gk 0 and khn hk 0 and such that Yn :=

0 g n (t) dW t and Zn := 0 hn (t) dWt converge in mean square to Y and Z, respectively.
Now observe that
Z Z Z
aYn + bZn = a gn (t) dWt + b hn (t) dWt = agn (t) + bhn (t) dWt , ()
0 0 0

since gn and hn are piecewise constant. Next, since


k(agn + bhn ) (ag + bh)k |a| kgn gk + |b| khn hk 0,
R
it follows that the right-hand side of () converges in mean square to 0 ag(t) +
bh(t) dWt . Since the left-hand side of () converges in mean square to aY + bZ, the
desired result follows.
38. We must find all values of for which E[|Xt /t|2 ] 0. First compute
Z t 2  Z t
t 2 +1
2
E[Xt ] = E dW = 2 d = .
0 0 2 + 1

Then t 2 +1 /t 2 = t 2 1 0 if and only if 2 1 < 0; i.e., 0 < < 1/2.


224 Chapter 13 Problem Solutions

39. Using the law of total probability, substitution, and independence, write
 Z T 2  Z  Z T 2 

2
E[YT ] = E n
dW = E n
dW T = t fT (t) dt
0 0 0
Z  Z t 2  Z  Z t 2 

= E dW T = t fT (t) dt =
n
E n dW fT (t) dt.
0 0 0 0

Now use properties of the Wiener process to write


Z  Z t  Z 2n+1
2 2n t E[T 2n+1 ]
E[YT ] = d fT (t) dt = fT (t) dt =
0 0 0 2n + 1 2n + 1
(2n + 1)!/ 2n+1 (2n)!
= = 2n+1 .
2n + 1
40. (a) Write
g(t + t) g(t) = E[ f (Wt+t )] E[ f (Wt )]
= E[ f (Wt+t ) f (Wt )]
E[ f 0 (Wt )(Wt+t Wt )] + 21 E[ f 00 (Wt )(Wt+t Wt )2 ]
= E[ f 0 (Wt W0 )(Wt+t Wt )]
+ 12 E[ f 00 (Wt W0 )(Wt+t Wt )2 ]
= E[ f 0 (Wt W0 )] 0 + 21 E[ f 00 (Wt W0 )] t.

It then follows that g0 (t) = 12 E[ f 00 (Wt )].


(b) If f (x) = ex , then g0 (t) = 21 E[eWt ] = 12 g(t). In this case, g(t) = et/2 , since g(0) =
E[eW0 ] = 1.
2
(c) We have by direct calculation that g(t) = E[eWt ] = es t/2 s=1 = et/2 .
41. Let C be the ball of radius r, C := {Y L p : kY k p r}. For X
/ C, i.e., kXk p > r, we
show that
r
Xb = X.
kXk p
To begin, note that the proposed formula for Xb satisfies kXk
b p = r so that Xb C as
required. Now observe that


b p = X r X = 1 r kXk p = kXk p r.
kX Xk kXk p p kXk p
Next, for any Y C,

kX Y k p kXk p kY k p
= kXk p kY k p
kXk p r
b p.
= kX Xk
b
Thus, no Y C is closer to X than X.
Chapter 13 Problem Solutions 225

42. Suppose that Xb and Xe are both elements of a subspace M and that hX X,Y
b i = 0 for
e
all Y M and hX X,Y i = 0 for all Y M. Then write
kXb Xk
e 22 = hXb X,
e Xb Xi
e = h(Xb X) + (X X),
e Xb Xei = 0 + 0 = 0.
| {z }
M

43. For XbN is to be the projection of XbM onto N, it is sufficient that the orthogonality
principle be satisfied. In other words, it suffices to show that
hXbM XbN ,Y i = 0, for all Y N.
Observe that
hXbM XbN ,Y i = h(XbM X) + (X XbN ),Y i = hX XbM ,Y i + hX XbN ,Y i.
Now, the last term on the right is zero by the orthogonality principle for the projection
of X onto N, since Y N. To show that hX XbM ,Y i = 0, observe that since N M,
Y N implies Y M. By the orthogonality principle for the projection of X onto M,
hX XbM ,Y i = 0 for Y M.
44. In the diagram, M is the disk and N is the horizontal line segment. The is XbM , the
projection of X onto the disk M. The  is XbN , the projection of X onto line segment
N. The is ([XbM )N , the projection of the circle XbM onto the line segment N. We see
[
that (Xb ) 6= Xb .
M N N

X
XbM

XbN
d
(XbM )N

45. Suppose that gn (Y ) M and gn (Y ) converges in mean square to some X. We must


show that X M. Since gn (Y ) converges, it is Cauchy. Writing
kgn (Y ) gm (Y )k22 = E[|gn (Y ) gm (Y )|2 ]
Z
= |gn (y) gm (y)|2 fY (y) dy = kgn gm kY ,

we see that gn is Cauchy in G, which is complete. Hence, there exists a g G with
kgn gkY 0. We claim X = g(Y ). Write
kg(Y ) Xk2 = kg(Y ) gn (Y ) + gn (Y ) Xk2
kg(Y ) gn (Y )k2 + kgn (Y ) Xk2 ,
where the last term goes to zero by assumption. Now observe that
Z
kg(Y ) gn (Y )k22 = E[|g(Y ) gn (Y )|2 ] = |g(y) gn (y)|2 fY (y) dy

= kg gn kY2 0.
Thus, X = g(Y ) M as required.
226 Chapter 13 Problem Solutions

R
46. We claim that the required projection is 01 f (t) dWt . Note that this is an element of M
since Z 1 Z
f (t)2 dt f (t)2 dt < .
0 0
Consider the orthogonality condition
Z Z 1 Z 1  Z
Z 1

E f (t) dWt f (t) dWt g(t) dWt = E f (t) dWt g(t) dWt .
0 0 0 1 0

Now put
 
f (t), t > 1, 0, t > 1,
f(t) := and g(t) :=
0, 0 t 1, g(t), 0 t 1,

so that the orthgonality condition becomes


Z Z  Z
E f(t) dWt g(t) dWt = f(t)g(t) dt,
0 0 0

which is zero since f(t)g(t) = 0 for all t 0.


47. The function g(t) will be optimal if the orthogonality condition
 Z Z 
E X g( ) dW g( ) dW = 0
0 0
R
holds for all g with 0 g( )2 d . In particular, this must be true for g( ) = I[0,t] ( ). In
this case, the above expectation reduces to
 Z  Z Z 
E X I[0,t] ( ) dW E g( ) dW I[0,t] ( ) dW .
0 0 0

Now, since Z Z t
I[0,t] ( ) dW = dW = Wt ,
0 0
we have the further simplification
Z Z t
E[XWt ] 2 g( )I[0,t] ( ) d = E[XWt ] 2 g( ) d .
0 0

Since this must be equal to zero for all t 0, we can differentiate and obtain
1 d
g(t) = E[XWt ].
2 dt

48. (a) Using the methods of Chapter 5, it is not too hard to show that


1, y 1/4,

[2 1 4y ]/2, 0 y < 1/4,
FY (y) =

[3/2 (1/2) 1 4y ]/2, 2 y < 0,

0, y < 2.
Chapter 13 Problem Solutions 227

It then follows that



1/ 1 4y, 0 < y < 1/4,
fY (y) = 1/(2 1 4y ), 2 < y < 0,

0, otherwise.
(b) We must find a gb(y) such that
E[v(X)g(Y )] = E[b
g(Y )g(Y )], for all bounded g.
For future reference, note that
Z 1
1
E[v(X)g(Y )] = E[v(X)g(X(1 X))] = v(x)g(x(1 x)) dx.
2 1

Now, by considering the problem of solving g(x) = y for x in the two cases
0 y 1/4 and 2 y < 0 suggests that we try
   
1 v 1+ 14y + 1 v 1 14y , 0 y 1/4,
2 2 2 2
gb(y) =  
v 1 14y
, 2 y < 0.
2

To check, we compute
g(Y )g(Y )] = E[b
E[b g(X(1 X))g(X(1 X))]
 Z 0 Z 1/2 Z 1 
1
= + + gb(x(1 x))g(x(1 x)) dx
2 1 0 1/2
Z 0
1
= v(x)g(x(1 x)) dx
2 1
Z 1/2
v(1 x) + v(x)
+ g(x(1 x)) dx
0 2
Z 1 
v(x) + v(1 x)
+ g(x(1 x)) dx
1/2 2
Z 0 Z 1 
1
= v(x)g(x(1 x)) dx + v(x)g(x(1 x)) dx
2 1 0
Z 1
= v(x)g(x(1 x)) dx.
1

49. (a) Using the methods of Chapter 5, it is not too hard to show that
(
F (sin1 (y)) + 1 F ( sin1 (y)), 0 y 1,
FY (y) =
F (sin1 (y)) F ( sin1 (y)), 1 y < 0.
Hence,


f (sin1 (y)) + f ( sin1 (y))

p , 0 y < 1,
1 y2
fY (y) =
f (sin1 (y)) + f ( sin1 (y))


p , 1 < y < 0.
1 y2
228 Chapter 13 Problem Solutions

(b) Consider
Z
E[v(X)g(Y )] = E[v(cos )g(sin )] = v(cos )g(sin ) f ( ) d .

Next, write
Z /2 Z 1 p f (sin1 (y))
v(cos )g(sin ) f ( ) d = v( 1 y2 )g(y) p dy
0 0 1 y2

and
Z 0 Z 0 p f (sin1 (y))
v(cos )g(sin ) f ( ) d = v( 1 y2 )g(y) p dy.
/2 1 1 y2

Similarly, we can write with a bit more work


Z Z /2
v(cos )g(sin ) f ( ) d = v(cos( t))g(sin( t)) f ( t) dt
/2 0
Z /2
= v( cost)g(sint) f ( t) dt
0
Z 1 p f ( sin1 (y))
= v( 1 y2 )g(y) p dy,
0 1 y2

and
Z /2
v(cos )g(sin ) f ( ) d

Z 0
= v(cos( t))g(sin( t)) f ( t) dt
/2
Z 0
= v( cost)g(sint) f ( t) dt
/2
Z 0 p f ( sin1 (y))
= v( 1 y2 )g(y) p dy.
1 1 y2

Putting all this together, we see that

E[v(X)g(Y )]
Z 1 p p
v( 1 y2 ) f (sin1 (y)) + v( 1 y2 ) f ( sin1 (y))
=
0 f (sin1 (y)) + f ( sin1 (y))
g(y) fY (y) dy
Z 0 p p
v( 1 y2 ) f (sin1 (y)) + v( 1 y2 ) f ( sin1 (y))
+
1 f (sin1 (y)) + f ( sin1 (y))
g(y) fY (y) dy.
Chapter 13 Problem Solutions 229

We conclude that

E[v(X)|Y = y]

1 1
2 2
v( 1y ) f (sin 1(y))+v( 1y1) f ( sin (y)) , 0 < y < 1,
f (sin (y))+ f ( sin (y))
= 1
1
2 2
v( 1y ) f (sin 1(y))+v( 1y )1f ( sin (y)) , 1 < y < 0.
f (sin (y))+ f ( sin (y))

50. Since X 0 and g(y) = IB (E[X|Y ]) 0, E[Xg(Y )] 0. On the other hand,

1
E[X|Y ]I(,1/n) (E[X|Y ]) I (E[X|Y ]),
n (,1/n)
and so
1
0 E[Xg(Y )] = E[E[X|Y ]g(Y )] P(E[X|Y ] < 1/n) < 0,
n
which is a contradiction. Hence, P(E[X|Y ] < 0) = 0.
51. Write

E[Xg(Y )] = E[(X + X )g(Y )]


= E[X + g(Y )] E[X g(Y )]
   
= E E[X + |Y ]g(Y ) E E[X |Y ]g(Y )
  
= E E[X + |Y ] E[X |Y ] g(Y ) .

By uniqueness, we conclude that E[X|Y ] = E[X + |Y ] E[X |Y ].


52. Following the hint, we begin with

E[X|Y ] = E[X + |Y ] E[X |Y ] E[X + |Y ] + E[X |Y ] = E[X + |Y ] + E[X |Y ].

Then
     
E E[X|Y ] E E[X + |Y ] + E E[X |Y ] = E[X + ] + E[X ] < .

53. To show that E[h(Y )X|Y ] = h(Y )E[X|Y ], we have to show that the right-hand side
satisfies the characterizing equation of the left-hand side. Since the characterizing
equation for the left-hand side is
 
E[{h(Y )X}g(Y )] = E E[h(Y )X|Y ]g(Y ) , for all bounded g,

we must show that


 
E[{h(Y )X}g(Y )] = E {h(Y )E[X|Y ]}g(Y ) , for all bounded g. ()

The only other thing we know is the characterizing equation for E[X|Y ], which is
 
E[Xg(Y )] = E E[X|Y ]g(Y ) , for all bounded g.
230 Chapter 13 Problem Solutions

Since g in the above formula is an arbitrary and bounded function, and since h is
also bounded, we can rewrite the above formula by replacing g(Y ) with g(Y )h(Y ) for
arbitrary bounded g. We thus have
 
E[X{g(Y )h(Y )}] = E E[X|Y ]{g(Y )h(Y )} , for all bounded g,

which is equivalent to ().


 
54. We must show that E[X|q(Y )] satisfies the characterizing
  of E E[X|Y ] q(Y ) .
equation
To write down the characterizing equation for E E[X|Y ] q(Y ) , it is convenient to use
the notation Z := E[X|Y ]. Then the characterizing equation for E[Z|q(Y )] is
 
E[Zg(q(Y ))] = E E[Z|q(Y )]g(q(Y )) , for all bounded g.

We must show that this equation holds when E[Z|q(Y )] is replaced by E[X|q(Y )]; i.e.,
we must show that
 
E[Zg(q(Y ))] = E E[X|q(Y )]g(q(Y )) , for all bounded g.

Replacing Z by its definition, we must show that


 
E[E[X|Y ]g(q(Y ))] = E E[X|q(Y )]g(q(Y )) , for all bounded g. ()

We begin with the characterizing equation for E[X|Y ], which is


 
E[Xh(Y )] = E E[X|Y ]h(Y ) , for all bounded h.

Since h is bounded and arbitrary, we can replace h(Y ) by g(q(Y )) for arbitrary bound-
ed g. Thus,
 
E[Xg(q(Y ))] = E E[X|Y ]g(q(Y )) , for all bounded g.

We next consider the characterizing equation of E[X|q(Y )], which is


 
E[Xg(q(Y ))] = E E[X|q(Y )]g(q(Y )) , for all bounded g.

Combining these last two equations yields ().


55. The desired result,

E[{h(Y )X}g(Y )] = E[{h(Y )E[X|Y ]}g(Y )],

can be rewritten as
E[(X E[X|Y ])g(Y )h(Y )] = 0,

where g is a bounded function and h(Y ) L2 . But then g(Y )h(Y ) L2 , and there-
fore this last equation must hold by the orthogonality principle since E[X|Y ] is the
projection of X onto M = {v(Y ) : E[v(Y )2 ] < }.
Chapter 13 Problem Solutions 231

56. Let hn (Y ) be bounded and converge to h(Y ). Then for bounded g,


h i
E[{h(Y )X}g(Y )] = E lim hn (Y )Xg(Y )
n
= lim E[X{hn (Y )g(Y )}]
n
 
= lim E E[X|Y ]{hn (Y )g(Y )}
n
h i
= E lim hn (Y )E[X|Y ]g(Y )
n
 
= E {h(Y )E[X|Y ]}g(Y ) .

By uniqueness, E[h(Y )X|Y ] = h(Y )E[X|Y ].


57. First write
n
Y = E[Y |Y ] = E[X1 + + Xn |Y ] = E[Xi |Y ].
i=1
By symmetry, we must have E[Xi |Y ] = E[X1 |Y ] for all i. Then Y = nE[X1 |Y ], or
E[X1 |Y ] = Y /n.
58. Write

E[Xn+1 |Y1 , . . . ,Yn ] = E[Yn+1 +Yn + +Y1 |Y1 , . . . ,Yn ]


= E[Yn+1 |Y1 , . . . ,Yn ] +Yn + +Y1
= E[Yn+1 ] + Xn , by indep. & def. of Xn ,
= Xn , since E[Yn+1 ] = 0.

59. For n 1,  
E[Xn+1 ] = E E[Xn+1 |Yn , . . . ,Y1 ] = E[Xn ],
where the second equality uses the definition of a martingale. Hence, E[Xn ] = E[X1 ]
for n 1.
60. For n 1,  
E[Xn+1 ] = E E[Xn+1 |Yn , . . . ,Y1 ] E[Xn ],
where the inequality uses the definition of a supermartingale. Since Xn 0, E[Xn ] 0.
Hence, 0 E[Xn ] E[X1 ] for n 1.
61. Since Xn+1 := E[Z|Yn+1 , . . . ,Y1 ],
 
E[Xn+1 |Yn , . . . ,Y1 ] = E E[Z|Yn+1 , . . . ,Y1 ] Yn , . . . ,Y1
= E[Z|Yn , . . . ,Y1 ], by the smoothing property,
=: Xn .

62. Since Xn := w(Yn ) w(Y1 ), observe that Xn is a function of Y1 , . . . ,Yn and that Xn+1 =
w(Yn+1 )Xn . Then

E[Xn+1 |Yn , . . . ,Y1 ] = E[w(Yn+1 )Xn |Yn , . . . ,Y1 ] = Xn E[w(Yn+1 )|Yn , . . . ,Y1 ]
= Xn E[w(Yn+1 )], by independence.
232 Chapter 13 Problem Solutions

It remains to compute
Z Z Z
f (y)
E[w(Yn+1 )] = w(y) f (y) dy = f (y) dy = f(y) dy = 1.
f (y)

Hence, E[Xn+1 |Yn , . . . ,Y1 ] = Xn ; i.e., Xn is a martingale with respect to Yn .


63. To begin, write

fYn+1 Y1 (yn+1 , . . . , y1 )
wn+1 (y1 , . . . , yn+1 ) =
fYn+1 Y1 (yn+1 , . . . , y1 )
fY |Y Y (yn+1 |yn , . . . , y1 ) fYn Y1 (yn , . . . , y1 )
= n+1 n 1 .
fYn+1 |Yn Y1 (yn+1 |yn , . . . , y1 ) fYn Y1 (yn , . . . , y1 )

If we put
fYn+1 |Yn Y1 (yn+1 |yn , . . . , y1 )
wn+1 (yn+1 , . . . , y1 ) := ,
fYn+1 |Yn Y1 (yn+1 |yn , . . . , y1 )
then Xn+1 := wn+1 (Y1 , . . . ,Yn+1 ) = wn+1 (Yn+1 , . . . ,Y1 )Xn , where Xn is a function of
Y1 , . . . ,Yn . We can now write

E[Xn+1 |Yn , . . . ,Y1 ] = E[wn+1 (Yn+1 , . . . ,Y1 )Xn |Yn , . . . ,Y1 ]


= Xn E[wn+1 (Yn+1 , . . . ,Y1 )|Yn , . . . ,Y1 ].

We now show that this last factor is equal to one. Write

E[wn+1 (Yn+1 ,Yn , . . . ,Y1 )|Yn = yn , . . . ,Y1 = y1 ]


= E[wn+1 (Yn+1 , yn . . . , y1 )|Yn = yn , . . . ,Y1 = y1 ]
Z
= wn+1 (y, yn , . . . , y1 ) fYn+1 |Yn Y1 (y|yn , . . . , y1 ) dy

Z
= fYn+1 |Yn Y1 (y|yn , . . . , y1 ) dy = 1.

64. Since Wk depends on Yk and Yk1 , and since Xn = W1 + + Wn , Xn depends on


Y0 , . . . ,Yn . Note also that Xn+1 = Xn +Wn+1 . Now write

E[Xn+1 |Yn , . . . ,Y0 ] = Xn + E[Wn+1 |Yn , . . . ,Y0 ].

Next,
 Z 

E[Wn+1 |Yn = yn , . . . ,Y0 = y0 ] = E Yn+1
zp(z|Yn ) dz Yn = yn , . . . ,Y0 = y0

 Z 

= E Yn+1
zp(z|yn ) dz Yn = yn , . . . ,Y0 = y0

Z
= E[Yn+1 |Yn = yn , . . . ,Y0 = y0 ] zp(z|yn ) dz

Z Z
= z fYn+1 |Yn Y0 (z|yn , . . . , y0 ) dz zp(z|yn ) dz

Chapter 13 Problem Solutions 233

Z Z
= z fYn+1 |Yn (z|yn ) dz zp(z|yn ) dz

Z Z
= zp(z|yn ) dz zp(z|yn ) dz = 0.

Hence, E[Xn+1 |Yn , . . . ,Y0 ] = Xn ; i.e., Xn is a martingale with respect to Yn .


65. First write
 
E[Yn+1 |Xn = in , . . . , X0 = i0 ] = E Xn+1 Xn = in , . . . , X0 = i0
= j P(Xn+1 = j|Xn = in , . . . , X0 = i0 )
j

= j P(Xn+1 = j|Xn = in ).
j

Next, for 0 < i < N,

j P(Xn+1 = j|Xn = i) = i1 (1 a) + i+1 a


j
 1 a i1  1 a i+1
= (1 a) + a
a a
 1 a i1 h  1 a 2 i
= 1a+ a
a a
 1 a i1 h (1 a)2 i
= 1a+
a a
(1 a)i h 1ai
= 1+ = i.
ai1 a
If Xn = 0 or Xn = 1, then Xn+1 = Xn , and so

E[ Xn+1 |Xn , . . . , X0 ] = E[ Xn |Xn , . . . , X0 ] = Xn = Yn .

Hence, in all cases, E[Yn+1 |Xn , . . . , X0 ] = Yn , and so Yn is a martingale with respect to


Xn .
66. First, since Xn is a submartingale, it is clear that

An+1 := An + (E[Xn+1 |Yn , . . . ,Y1 ] Xn ) 0

and is a function of Y1 , . . . ,Yn . To show that Mn is a martingale, write

E[Mn+1 |Yn , . . . ,Y1 ] = E[Xn+1 An+1 |Yn , . . . ,Y1 ]


= E[Xn+1 |Yn , . . . ,Y1 ] An+1
h i
= E[Xn+1 |Yn , . . . ,Y1 ] An + (E[Xn+1 |Yn , . . . ,Y1 ] Xn )
= Xn An = Mn .

67. Without loss of generality, assume f1 f2 . Then

E[Z f1 Z f2 ] = E[Z f1 (Z f2 Z f1 ) ] + E[|Z f1 |2 ]


234 Chapter 13 Problem Solutions

= E[(Z f1 Z1/2 )(Z f2 Z f1 ) ] + E[|Z f1 |2 ]


= E[Z f1 Z1/2 ]E[(Z f2 Z f1 ) ] + E[|Z f1 |2 ]
Z f1
= 0 0 + E[|Z f1 |2 ] = S( ) d .
1/2

68. Using the geometric series formula,


n+1
1 n j2 f n 1 e j2 f e j2 f e j2 f 1 e j2 f n

n k=1
e =
n 1 e j2 f
=
n

1 e j2 f
e j2 f e j f n e j f n e j f n e j2 f e j f n sin( f n)
= j f j f = j f .
n e e e j f n e sin( f )

69. Following the hint, write


 N  N  
E[|Y |2 ] = E[YY ] = E dn Xn dk Xk
n=N k=N
N N N N
= dn dk E[Xn Xk ] = dn dk R(n k)
n=N k=N n=N k=N
N N Z 1/2
= dn dk S( f )e j2 f (nk) d f
n=N k=N 1/2
Z 1/2 N
2

j2 f n
= S( f ) dn e d f.
1/2 n=N

Hence, if Nn=N dn e j2 f n = 0, then E[|Y |2 ] = 0.


70. Write
 N  N   N N
E[T (G0 )T (H0 )] = E gn Xn hk Xk = gn hn R(n k)
n=N k=N n=N k=N
N N Z 1/2
= gn hn S( f )e j2 f (nk) d f
n=N k=N 1/2
Z 1/2  N  N 
j2 f n j2 f k
=
1/2
S( f ) gn e hk e df
n=N k=N
Z 1/2
= S( f )G0 ( f )H0 ( f ) d f .
1/2

71. (a) Write

en ) + T (G
kT (G) Y k2 = kT (G) T (Gn ) + T (Gn ) T (G en ) Y k2
kT (G) T (Gn )k2 + kT (Gn ) T (G en )k2 + kT (G
en ) Y k2 .
Chapter 13 Problem Solutions 235

On the right-hand side, the first and third terms go to zero. To analyze the middle
term, write
en )k2 = kT (Gn G
kT (Gn ) T (G en )k2 = kGn G
en k
kGn Gk + kG Gen k 0.

(b) To show T is norm preserving on L2 (S), let Gn G with T (Gn ) T (G). Then
by Problem 21, kT (Gn )k2 kT (G)k2 , and similarly, kGn k kGk. Now write

kT (G)k2 = lim kT (Gn )k2 = lim kGn k2 , since Gn is a trig. polynomial,


n n
= kGk.

(c) To show T is linear on L2 (S), fix G, H L2 (S), and let Gn and Hn be trigono-
metric polynomials converging to G and H, respectively. Then

Gn + Hn G + H, ()

and we can write

kT ( G + H) { T (G) + T (H)}k2
= kT ( G + H) T ( Gn + Hn )k2
+ kT ( Gn + Hn ) { T (Gn ) + T (Hn )}k2
+ k{ T (Gn ) + T (Hn )} { T (G) + T (H)}k2 .

Now, the first term on the right goes to zero on account of () and the defintion
of T . The second term on the right is equal to zero because T is linear on
trigonometric polynomials. The third term goes to zero upon observing that

k{ T (Gn ) + T (Hn )} { T (G) + T (H)}k2 | | kT (Gn ) T (G)k2


+ | | kT (Hn ) T (H)k2 .

(d) Using parts (b) and (c), kT (G) T (H)k2 = kT (G H)k2 = kG Hk implies
that T is actually uniformly continuous.
72. It suffices to show that I[1/2, f ] L2 (S). Write
Z 1/2 Z f Z 1/2
I[1/2, f ] ( ) 2 S( ) d = S( ) d S( ) d = R(0) = E[Xn2 ] < .
1/2 1/2 1/2

73. (a) We know that for trigonometric polynomials, E[T (G)] = 0. Hence, if Gn is a
sequence of trigonometric polynomials converging to G in L2 (S), then T (Gn )
T (G) in L2 , and then E[T (Gn )] E[T (G)].
(b) For trigonometric polynomials G and H, we have
Z 1/2
hT (G), T (H)i := E[T (G)T (H) ] = G( f )H( f ) S( f ) d f .
1/2
236 Chapter 13 Problem Solutions

If Gn and Hn are sequences of trigonometric polynomials converging to G and


H in L2 (S), then we can use the result of Problem 24 to write
Z 1/2
hT (G), T (H)i = lim hT (Gn ), T (Hn )i = lim Gn ( f )Hn ( f ) S( f ) d f
n n 1/2
Z 1/2
= G( f )H( f ) S( f ) d f .
1/2

(c) For 1/2 f1 < f2 f3 < f4 1/2, write

E[(Z f2 Z f1 )(Z f4 Z f3 ) ] = E[T (I( f1 , f2 ] )T (I( f3 , f4 ] ) ]


Z 1/2
= I( f1 , f2 ] ( f )I( f3 , f4 ] ( f )S( f ) d f = 0.
1/2

74. (a) We take


 Z 1/2 
L2 (SY ) := G: |G( f )|2 SY ( f ) d f < .
1/2

(b) Write
N N
TY (G0 ) = gnYn = gn hk Xnk
n=N n=N k=
N Z 1/2
= gn hk e j2 f (nk) dZ f
n=N k= 1/2
Z 1/2
= G0 ( f )H( f ) dZ f .
1/2

(c) For G L2 (SY ), let Gn be a sequence of trigonometric polynomials converging


to G in L2 (SY ). Since TY is continuous,
Z 1/2
TY (G) = lim TY (Gn ) = lim Gn ( f )H( f ) dZ f
n n 1/2

= lim T (Gn H) = T (GH), since T is continuous,


n
Z 1/2
= G( f )H( f ) dZ f .
1/2

(d) Using part (c),


Z 1/2
V f := TY (I[1/2, f ] ) = I[1/2, f ] ( )H( ) dZ ()
1/2
Z f
= H( ) dZ .
1/2
Chapter 13 Problem Solutions 237

(e) A slight generalization of () establishes the result for piecewise-constant func-


tions. For general G L2 (SY ), approximate G by a sequence of piecewise-
constant functions Gn and write
Z 1/2 Z 1/2 Z 1/2
G( f ) dV f = lim Gn ( f ) dV f = lim Gn ( f )H( f ) dZ f
1/2 n 1/2 n 1/2
Z 1/2
= lim T (Gn H) = T (GH) = G( f )H( f ) dZ f .
n 1/2
CHAPTER 14
Problem Solutions

1. We must show that P(|Xn | ) 0. Since Xn Cauchy(1/n) has an even density,


we can write
Z Z
1/(n ) 1/
P(|Xn | ) = 2 dx = 2 dy 0.
(1/n)2 + x2 n 1 + y2

2. Since |cn c| 0, given > 0, there is an N such that for n N, |cn c| < . For
such n,
{ : |cn c| } = ,
and so P(|cn c| ) = 0.
3. We show that Xn converges in probability to zero. Observe that Xn takes only the
values n and zero. Hence, for 0 < < 1, |Xn | if and only if Xn = n, which
happens if and only if U 1/ n. We can now write

P(|Xn | ) = P(U 1/ n ) = 1/ n 0.

4. Write Z
P(|Xn | ) = P(|V | cn ) = 2 fV (v) dv 0,
cn
since cn .
5. Using the hint,
 
[
P(X 6= Y ) = P {|X Y | 1/k} = lim P(|X Y | 1/K).
K
k=1

We now show that the above limit is zero. To begin, observe that

|Xn X| < 1/(2K) and |Xn Y | < 1/(2K)

imply

|X Y | = |X Xn + Xn Y | |X Xn | + |Xn Y | < 1/(2K) + 1/(2K) = 1/K.

Hence, |X Y | 1/K implies |Xn X| 1/(2K) or |Xn Y | 1/(2K), and we can


write

P(|X Y | 1/K) P {|Xn X| 1/(2K)} {|Xn Y | 1/(2K)}
 
P {|Xn X| 1/(2K)} + P {|Xn Y | 1/(2K)} ,

which goes to zero as n

238
Chapter 14 Problem Solutions 239

6. Let > 0 be given. We must show that for every > 0, for all sufficiently large n,
P(|Xn X| ) < . Without loss of generality, assume < . Let n be so large that
P(|Xn X| ) < . Then since < ,

P(|Xn X| ) P(|Xn X| ) < .

7. (a) Observe that

P(|X| > ) = 1 P(|X| ) = 1 P( X )


= 1 [FX ( ) FX (( ))] = 1 FX ( ) + FX (( ))
1 FX ( ) + FX ( ).

Now, since FX (x) 1 as x and FX (x) 0 as x , for large , 1


FX ( ) < /8 and FX ( ) < /8, and then P(|X| > ) < /4. Similarly, for
large , P(|Y | > ) < /4.
(b) Observe that if the four conditions hold, then

|Xn | = |Xn X + X| |Xn X| + |X| < + < 2 ,

and similarly, |Yn | 2 . Now that both (Xn ,Yn ) and (X,Y ) lie in the rectangle,

|g(Xn ,Yn ) g(X,Y )| < .

(c) By part (b), observe that

{|g(Xn ,Yn ) g(X,Y )| } {|Xn X| } {|Yn Y | }


{|X| > } {|Y | > }.

Hence,

P(|g(Xn ,Yn ) g(X,Y )| ) P(|Xn X| ) + P(|Yn Y | )


+ P(|X| > ) + P(|Y | > )
< /4 + /4 + /4 + /4 = .

8. (a) Since the Xi are i.i.d., the Xi2 are i.i.d. and therefore uncorrelated and have com-
mon mean E[Xi2 ] = 2 + m2 and common variance
 2  2
E Xi2 E[Xi2 ] = E[Xi4 ] E[Xi2 ] = 4 ( 2 + m2 )2 .

By the weak law of large numbers, Vn converges in probability to E[Xi2 ] = 2 +


m2 .
(b) Observe that Sn2 = g(Mn ,Vn )n/(n 1), where g(m, v) := v m2 is continuous.
Hence, by the preceding problem, g(Mn ,Vn ) converges in probability to g(m, v)
= ( 2 + m2 ) m2 = 2 . Next, by Problem 2, n/(n 1) converges in probability
to 1, and then the product [n/(n 1)]g(Mn ,Vn ) converges in probability to 1
g(m, v) = 2 .
240 Chapter 14 Problem Solutions

9. (a) Since Xn converges in probability to X, with = 1 we have P(|Xn X| 1) 0


as n . Now, if |X Xn | < 1, then |X| |Xn | < 1, and it follows that
|X| < 1 + |Xn | 1 + |Y |.
Equivalently,
|X| |Y | + 1 implies |Xn X| 1.
Hence,
P(|X| |Y | + 1) P(|Xn X| 1) 0.
(b) Following the hint, write
E[|Xn X|] = E[|Xn X|IAn ] + E[|Xn X|IAnc ]
  
E |Xn | + |X| IAn + P(Anc )
  
E Y + |X| IAn +
+ ,
  
since for large n, P(An ) < implies E Y + |X| IAn < . Hence, E[|Xn X|] <
2 .
10. (a) Suppose g(x) is bounded, nonnegative, and g(x) 0 as x 0. Then given
> 0, there exists a > 0 such that g(x) < /2 for all |x| < . For |x| ,
we use the fact that g is bounded to write g(x) G for some positive, finite G.
Since Xn converges to zero in probability, for large n, P(|Xn | ) < /(2G).
Now write
E[g(Xn )] = E[g(Xn )I[0, ) (|Xn |)] + E[g(Xn )I[ ,) (|Xn |)]
E[( /2)I[0, ) (|Xn |)] + E[GI[ ,) (|Xn |)]

= P(|Xn | < ) + G P(|Xn | )
2

< +G = .
2 2G
(b) By applying part (a) to the function g(x) = x/(1 + x), it follows that if Xn con-
verges in probability to zero, then
 
|Xn |
lim E = 0.
n 1 + |Xn |
Now we show that if the above limit holds, then Xn must converge in probability
to zero. Following the hint, we use the fact that g(x) = 1/(1 + x) is an increasing
function for x 0. Write
E[g(Xn )] = E[g(Xn )I[ ,) (|Xn |)] + E[g(Xn )I[0, ) (|Xn |)]
E[g(Xn )I[ ,) (|Xn |)]
E[g( )I[ ,) (|Xn |)]
= g( )P(|Xn | ).
Thus, if g(x) is nonnegative and nondecreasing, if E[g(Xn )] 0, then Xn con-
verges in distribution to zero.
Chapter 14 Problem Solutions 241

11. First note that for the constant random variable Y c, FY (y) = I[c,) (y). Similarly,
for Yn cn , FYn (y) = I[cn ,) (y). Since the only point at which FY is not continuous is
y = c, we must show that I[cn ,) (y) I[c,) (y) for all y 6= c. Consider a y with c < y.
For all sufficiently large n, cn will be very close to c so close that cn < y, which
implies FYn (y) = 1 = FY (y). Now consider y < c. For all sufficiently large n, cn will
be very close to c so close that y < cn , which implies FYn (y) = 0 = FY (y).
12. For 0 < c < , FY (y) = P(cX y) = FX (y/c), and FYn (y) = FX (y/cn ). Now, y is a
continuity point of FY if and only if y/c is a continuity point of FX . For such y, since
y/cn y/c, FX (y/cn ) FX (y/c). For c = 0, Y 0, and FY (y) = I[0,) (y). For y 6= 0,

+, y > 0,
y/cn
, y < 0,
and so 
1, y > 0,
FYn (y) = FX (y/cn )
0, y < 0,
which is exactly FY (y) for y 6= 0.
13. Since Xt Yt Zt ,
{Zt y} {Yt y} {Xt y},
we can write
P(Zt y) P(Yt y) P(Xt y),
or FZt (y) FYt (y) FXt (y), and it follows that

lim FZt (y) lim FYt (y) lim FXt (y) .


t t t
| {z } | {z }
= F(y) = F(y)

Thus, FYt (y) F(y).


14. Since Xn converges in mean to X, the inequality

E[Xn ] E[X] = E[Xn X] E[|Xn X|]

shows that E[Xn ] E[X]. We now need the following implications:

conv. in mean conv. in probability conv. in distribution Xn ( ) X ( ).

Since Xn is exponential, we also have


1/E[Xn ] 1/E[X]
Xn ( ) = .
1/E[Xn ] j 1/E[X] j
Since limits are unique, the above right-hand side must be X ( ), which implies X is
an exponential random variable.
15. Since Xn and Yn each converge in distribution to constants x and y, respectively, they
also converge in probability. Hence, as noted in the text, Xn +Yn converges in proba-
bility to x + y. Since convergence in probability implies convergence in distribution,
Xn +Yn converges in distribution to x + y.
242 Chapter 14 Problem Solutions

16. Following the hint, we note that each Yn is a finite linear combination of independent
Gaussian increments. Hence, each Yn is Gaussian. Since Y is the mean-square limit
of the Gaussian Yn , the distribution of RY is also Gaussian by the example cited in
theRhint. Furthermore, since each Yn = 0 gn ( ) dW , Yn has zero mean and variance
2 0 gn ( )2 d = 2 kgn k2 . By the cited example, Y has zero mean. Also, since

kgn k kgk kgn gk 0,

we have
var(Y ) = E[Y 2 ] = lim E[Yn2 ] = lim var(Yn )
n n
Z
= lim 2 kgn k2 = 2 kgk2 = 2 g( )2 d .
n 0

17. For any constants c1 , . . . , cn , write


n n Z Z  n 
ci Xti = ci 0
g(ti , ) dW =
0
ci g(ti , ) dW ,
i=1 i=1 i=1
| {z }
=: g( )

which is normal by the previous problem.


18. The plan is to show that the increments are Gaussian and uncorrelated. It will then
follow that the increments are independent. For 0 u < v s < t < , write

    Xu
Xv Xu 1 1 0 0
Xv .
=
Xt Xs 0 0 1 1 Xs
Xt
By writing
Z t Z Z
Xt := g( ) dW = g( )I[0,t] ( ) dW = h(t, ) dW ,
0 0 0

where h(t, ) := g( )I[0,t] ( ), we see that by the preceding problem, the vector on the
right above is Gaussian, and hence, so are the increments in the vector on the left.
Next,
Z t Z v 
E[(Xt Xs )(Xv Xu )] = E g( ) dW g( ) dW
s u
Z Z 
= E g( )I(s,t] ( ) dW g( )I(u,v] ( ) dW
0 0
Z
2 2
= g( ) I(s,t] ( )I(u,v] ( ) d = 0.
0

19. Given a linear combination nk=1 ck Ak , put g( ) := nk=1 ck k ( ). Then


n n Z b  Zb  n  Z b
ck Ak = ck X k ( ) d = X ck k ( ) d = g( )X d ,
a a a
k=1 k=1 k=1
Chapter 14 Problem Solutions 243

which is a mean-square limit of sums of the form

g(i )Xi (ti ti1 ).


i

Since X is a Gaussian process, these sums are Gaussian, and hence, so is their mean-
square limit.
20. If MXn (s) MX (s), then this holds when s = j ; i.e., Xn ( ) X ( ). But this
implies that Xn converges in distribution to X. Similarly, if GXn (z) GX (z), then
this holds when z = e j ; i.e., Xn ( ) X ( ). But this implies Xn converges in
distribution to X.
21. (a) Write

pn (k) = FXn (k + 1/2) FXn (k 1/2)


FX (k + 1/2) FX (k 1/2) = p(k),

where we have used the fact that since X is integer valued, k 1/2 is a continuity
point of FX .
(b) The continuity points of FX are the noninteger values of x. For such x > 0,
suppose k < x < k + 1. Then
k k
FXn (x) = P(Xn x) = P(Xn k) = pn (i) p(i) = P(X k) = FX (x).
i=0 i=0

22. Let
 
n k n!
pn (k) := P(Xn = k) = p (1 pn )nk = pk (1 pn )nk
k n k!(n k)! n

and p(k) := P(X = k) = k e /k!. Next, by Stirlings formula, as n ,



2 nn+1/2 en (n k)!
qn := 1 and rn (k) := 1,
n! 2 (n k)nk+1/2 e(nk)

and so qn rn (k) 1 as well. If we can show that pn (k)qn rn (k) p(k), then

pn (k)qn rn (k) lim pn (k)qn rn (k) p(k)


lim pn (k) = lim = n = = p(k).
n n qn rn (k) lim qn rn (k) 1
n

Now write

2 nn+1/2 en (n k)!
lim pn (k)qn rn (k) = lim pn (k)
n n n! 2 (n k)nk+1/2 e(nk)
ek nn+1/2
= lim pkn (1 pn )nk
k! n (n k)nk+1/2
244 Chapter 14 Problem Solutions

ek nn+1/2
= lim pkn (1 pn )nk nk+1/2
k! n n (1 k/n)nk+1/2
ek (npn )k (1 pn )nk
= lim
k! n (1 k/n)k+1/2 (1 k/n)n
ek (npn )k [1 (npn )/n]n (1 pn )k
= lim
k! n (1 k/n)k+1/2 (1 k/n)n
ek k e 1 k e
= k = = p(k).
k! 1 e k!

Note that since npn , pn 0, and so (1 pn )k 1.


23. First write
 
npn (z 1) n
GXn (z) = [(1 pn ) + pn z]n = [1 + pn (z 1)]n = 1 + .
n

Since npn (z 1) (z 1), GXn (z) e (z1) = GX (z).


24. (a) Here are the sketches:
I(,a] (t) ga,b (t) I(,b] (t)
1 1 1
t t t
a b a b a b
(b) From part (a) we can write

I(,a] (Y ) ga,b (Y ) I(,b] (Y ),

from which it follows that

E[I(,a] (Y )] E[ga,b (Y )] E[I(,b] (Y )],


| {z } | {z }
= P(Y a) = P(Y b)

or
FY (a) E[ga,b (Y )] FY (b).

(c) Since FXn (x) E[gx,x+x (Xn )],

lim FXn (x) lim E[gx,x+x (Xn )] = E[gx,x+x (X)] FX (x + x).


n n

(d) Similarly,

FX (x x) E[gxx,x (X)] = lim E[gxx,x (Xn )] lim FXn (x).


n n

(e) If x is a continuity point of FX , then given any > 0, for sufficiently small x,

FX (x) < FX (x x) and FX (x + x) < FX (x) + .


Chapter 14 Problem Solutions 245

Combining this with parts (c) and (d), we obtain

FX (x) lim FXn (x) lim FXn (x) < FX (x) + .


n n

Since > 0 is arbitrary, the liminf and the limsup are equal, limn FXn (x) exists
and is equal to FX (x).
25. If Xn converges in distribution to zero, then Xn converges in probability to zero, and
by Problem 10,  
|Xn |
E = 0.
1 + |Xn |
Conversely, if the above limit holds, then by Problem 10, Xn converges in probability
to zero, which implies convergence in distribution to zero.
26. Observe that fn (x) = n f (nx) implies

1, x > 0,
Fn (x) = F(nx) F(0), x = 0,

0, x < 0.

So, for x 6= 0, Fn (x) I[0,) (x), which is the cdf of X 0. In other words, Xn con-
verges in distribution to zero, which implies convergence in probability to zero.
27. Since Xn converges in mean square to X, Xn converges in distribution to X. Since
2
g(x) := x2 ex is a bounded continuous function, E[g(Xn )] E[g(X)].
28. Let 0 < < 1 be given. For large t, we have |u(t) 1| < , or

< u(t) 1 < or 1 < u(t) < 1 + .

Hence, for z > 0,

P(Zt z(1 )) P(Zt zu(t)) P(Zt z(1 + )).

Rewrite this as

FZt (z(1 )) P(Zt zu(t)) FZt (z(1 + )).

Then
F(z(1 )) = lim FZt (z(1 )) lim P(Zt zu(t))
t t

and
lim P(Zt zu(t)) lim FZt (z(1 + )) = FZ (z(1 + )).
t t

Since 0 < < 1 is arbitrary, and since F is continuous, we must have

lim P(Zt zu(t)) = F(z).


t

The case for z < 0 is similar.


246 Chapter 14 Problem Solutions

29. Rewrite P(c(t)Zt z) as P(Zt z/c(t)) = P(Zt (z/c)(c/c(t)). Then if we put


u(t) := c/c(t), we have by the preceding problem that P(c(t)Zt z) F(z/c).
30. First write FXt (x) = P(Zt + s(t) x) = P(Zt x s(t)). Let > 0 be given. Then
s(t) 0 implies that for large t, |s(t)| < , or

< s(t) < or < s(t) < .

Then

FZt (x ) = P(Zt x ) FXt (x) P(Zt x + ) = FZt (x + ).

It follows that

F(x ) lim FXt (x) lim FXt (x) F(x + ).


t t

Since F is continuous and > 0 is arbitrary,

lim FXt (x) = F(x).


t

31. Starting with Nbtc Nt Ndte , it is easy to see that


Nbtc Ndte
t t
Xt := p Yt p =: Zt .
/t /t
According to Problem 13, it suffices to show that Xt and Zt converge in distribution
to N(0, 1) random variables. By the preceding two problems, the distribution limit of
Zt is the same as that of c(t)Zt + s(t) if c(t) 1 and s(t) 0. We take
 q
t dte
c(t) := q  = t dte 1
t dte

and   
t dte t dte 1 q
s(t) := q  q  = dte 0.
dte dte

Finally, observe that


Ndte
dte

c(t)Zt + s(t) = q  ,
dte
goes through the values of Yn and therefore converges in distribution to an N(0, 1)
random variable. It is similar to show that the distribution limit of Xt is also N(0, 1).
32. (a) Let G := {Xn X}. For G,
1 1
.
1 + Xn ( )2 1 + X( )2

Since P(G c ) = 0, 1/(1 + Xn2 ) converges almost surely to 1/(1 + X 2 ).


Chapter 14 Problem Solutions 247

(b) Since almost sure convergence implies convergence in probability, which im-
plies convergence in distribution, we have 1/(1 + Xn2 ) converging in distribution
to 1/(1 + X 2 ). Since g(x) = 1/(1 + x2 ) is bounded and continuous, E[g(Xn )]
E[g(X)]; i.e.,    
1 1
lim E = E .
n 1 + Xn2 1 + X2

33. Let GX := {Xn X} and GY := {Yn Y }, where P(GXc ) = P(GYc ) = 0. Let G :=


{g(Xn ,Yn ) g(X,Y )}. We must show that P(G c ) = 0. Our plan is to show that
GX GY G. It follows that G c GXc GYc , and we can then write

P(G c ) P(GXc ) + P(GYc ) = 0.

For GX GY , (Xn ( ),Yn ( )) (X( ),Y ( )). Since g(x, y) is continuous, for
such , g(Xn ( ),Yn ( )) g(X( ),Y ( )). Thus, GX GY G.
34. Let GX := {Xn X} and GXY := {X = Y }, where P(GXc ) = P(GXY c ) = 0. Put G :=
Y
c
{Xn Y }. We must show that P(GY ) = 0. Our plan is to show that GX GXY GY .
It follows that GYc GXc GXY
c , and we can then write

P(GYc ) P(GXc ) + P(GXY


c
) = 0.

For GX GXY , Xn ( ) X( ) and X( ) = Y ( ). Hence, for such , Xn ( )


Y ( ); i.e., GY . Thus, GX GXY GY .
35. Let GX := {Xn X} and GY := {Xn Y }, where P(GXc ) = P(GYc ) = 0. Put GXY :=
{X = Y }. We must show that P(GXY c ) = 0. Our plan is to show that G G G .
X Y XY
c c c
It follows that GXY GX GY , and we can then write
c
P(GXY ) P(GXc ) + P(GYc ) = 0.

For GX GY , Xn ( ) X( ) and Xn ( ) Y ( ). Since limits of sequences of


numbers are unique, for such , X( ) = Y ( ); i.e., GXY . Thus, GX GY GXY .
T
36. Let GX := {Xn X}, GY := {Yn Y }, and GI := c
n=1 {Xn Yn }, where P(GX ) =
c c
P(GY ) = P(GI ) = 0. This last equality follows because

P(GIc ) P(Xn > Yn ) = 0.
n=1

Let G := {X Y }. We must show that P(G c ) = 0. Our plan is to show that GX


GY GI G. It follows that G c GXc GYc GIc , and we can then write

P(G c ) P(GXc ) + P(GYc ) + P(GIc ) = 0.

For GX GY GI , Xn ( ) X( ), Yn ( ) Y ( ), and for all n, Xn ( ) Yn ( ).


By properties of sequences of real numbers, for such , we must have X( ) Y ( ).
Thus, GX GY GI G.
37. Suppose Xn converges almost surely to X, and Xn converges in mean to Y . Then Xn
converges in probability to X and to Y . By Problem 5, X = Y a.s.
248 Chapter 14 Problem Solutions

38. If X( ) > 0, then cn X( ) . If X( ) < 0, then cn X( ) . If X( ) = 0, then


cn X( ) = 0 0. Hence,

+, if X( ) > 0,
Y ( ) = 0, if X( ) = 0,

, if X( ) < 0.

39. By a limit property of probability, we can write


   
\ [ [
P
{Xn = j} X0 = i = lim P
{Xn = j} X0 = i
M
N=1 n=N n=M

(n)
lim
M
P(Xn = j|X0 = i) = lim
M
pi j ,
n=M n=M

(n)
which must be zero since the pi j are summable.
40. Following the hint, write
 
\ E[S]
P(S = ) = P {S > n} = lim P(S > N) lim = 0.
N N N
n=1

41. Since Sn converges in mean to S, by the problems referred to in the hint,


n n
E[S] = lim E[Sn ] = lim lim |hk |E[|Xk |] B1/p |hk | < .
E[|hk ||Xk |] = n
n n
k=1 k=1 k=1

42. From the inequality


Z n+1 Z n+1
1 1 1
p
dt p
dt = ,
n t n (n + 1) (n + 1) p
we obtain
Z Z n+1
1 1 1 1
>
t p
dt = t p
dt (n + 1) p = np ,
1 n=1 n n=1 n=2

which implies
n=1 1/n
p < .
43. Write
 n  n  n  n 
1
E[Mn4 ] = E X i X j Xl Xm
n4 i=1 j=1 l=1 m=1
n n n n n
1
= 4 E[Xi4 ] + E[Xi Xi Xl Xl ] + E[Xi X j Xi X j ]
n i=1 i=1 l=1,l6=i i=1 j=1, j6=i
n n n n 
3
+ E[Xi X j X j Xi ] + 4 E[Xi X j ]
i=1 j=1, j6=i i=1 j=1, j6=i | {z }
=0
1
= 4 [n + 3n(n 1) 4 ].
n
Chapter 14 Problem Solutions 249

44. It suffices to show that


n=1 P(|Xn | ) < . To this end, write
Z

P(|Xn | ) = 2 n nx
2e dx = enx = en .

Then

e
P(|Xn | ) = (e )n = 1 e
< .
n=1 n=1

45. First use the Rayleigh cdf to write


2 /2
P(|Xn | ) = P(Xn ) = e(n ) .

Then

2 2 2 1
P(|Xn | ) = e n /2 (e /2 )n = 2
1 e /2
< .
n=1 n=1 n=1

Thus, Xn converges almost surely to zero.


46. For n > 1/ , write
Z
p 1 p 1
P(|Xn | ) = p1
x dx = .
n n p1 p1
Then

1 1
P(|Xn | ) = P(|Xn | ) +
p1 n p1
<
n>1 n<1/ n>1/

since p 1 > 1. Thus, Xn converges almost surely to zero.


47. To begin, first observe that

E[|Xn X|] = E[|n2 (1)nYn |] = n2 E[Yn ] = n2 pn .

Also observe that

P(|Xn X| ) = P(n2Yn ) = P(Yn /n2 ) = P(Yn = 1) = pn .

In order to have Xn converge almost surely to X, it is sufficient to consider pn such


that
n=1 pn < .

(a) If pn = 1/n3/2 , then 2


n=1 pn < , but n pn = n
1/2 6 0. For this choice of p ,
n
Xn converges almost surely but not in mean to X.
(b) If pn = 1/n3 , then 2
n=1 pn < and n pn = 1/n 0. For this choice of pn , Xn
converges almost surely and in mean to X.

48. To apply the weak law of large numbers of this section to (1/n) ni=1 Xi2 requires only
that the Xi2 be i.i.d. and have finite mean; there is no second-moment requirement on
Xi2 (which would be a requirement on Xi4 ).
250 Chapter 14 Problem Solutions

49. By writing
Nn 1 n
= Nk Nk1 ,
n n k=1
which is a sum of i.i.d. Poisson( ) random variables, we see that Nn /n converges
almost surely to by the strong law of large numbers. We next observe that Nbtc
Nt Ndte . Then
Nbtc Nt Ndte
,
t t t
and it follows that
Nbtc Nt Ndte
,
dte t btc
and then
btc Nbtc Nt Ndte dte
.
dte btc t dte btc
|{z} |{z} |{z} |{z}
1 1

Hence Nt /t converges almost surely to .


50. (a) By the strong law of large numbers, for not in a set of probability zero,

1 n
Xk ( ) .
n k=1

Hence, for > 0, for all sufficiently large n,


n
1
Xk ( ) < ,
n
k=1

which implies
1 n
Xk ( ) < ,
n k=1
from which it follows that
n
Xk ( ) < n( + ).
k=1

(b) Given M > 0, let > 0 and choose n in part (a) so that both n( + ) > M and
n M hold. Then
n
Tn ( ) = Xk ( ) < n( + ).
k=1

Now, for t > n( + ) Tn ( ),



Nt ( ) = I[0,t] (Tk ( )) n M.
k=1
Chapter 14 Problem Solutions 251

(c) As noted in the solution of part (a), the strong law of large numbers implies
Tn 1 n
= Xk a.s.
n n k=1
Hence, n/Tn 1/ a.s.
(d) First observe that
Nt Nt
YNt = ,
TNt t
and so
1 Nt
= lim YNt lim .
t t t
Next,
Nt + 1 Nt + 1 Nt 1
YNt +1 = = + ,
TNt +1 t t t
and so
1 Nt
= lim YNt +1 lim + 0.
t t t
Hence,
Nt 1
lim = .
t t
51. Let Xn := nI(0,1/n] (U), where U uniform(0, 1]. Then for every , U( ) (0, 1]. For
n > 1/U( ), or U( ) > 1/n, Xn ( ) = 0. Thus, Xn ( ) 0 for every . However,
E[Xn ] = nP(U 1/n) = 1.
Hence, Xn does not converge in mean to zero.
52. Suppose D contains n points, say a x1 < < xn b. Then
n
n < G(xk +) G(xk )
k=1
n1 n
= G(xn +) + G(xk +) G(xk ) G(x1 )
k=1 k=2
n1 n
G(b) + G(xk+1 ) G(xk1 ) G(a)
k=1 k=2
n1
= G(b) + [G(xk+1 ) G(xk )] G(a)
k=1
G(b) + G(xn ) G(x1 ) G(a) 2[G(b) G(a)].

53. Write
   
[ [ 2
P U Kn = P {U Kn } P(U Kn ) 2n = 2 .
n=1 n=1 n=1 n=1

Since > 0 is arbitrary, the probability in question must be zero.


CHAPTER 15
Problem Solutions

1. Using the formula FWt (x) = FW1 (t H x), we see that



FW1 () = 1, x > 0,
lim FWt (x) =
t0 FW1 () = 0, x < 0,

which is the cdf of the zero random variable for x 6= 0. Hence, Wt converges in
distribution to the zero random variable.
Next,
, if W1 ( ) > 0,
X( ) := lim t H W1 ( ) = 0, if W1 ( ) = 0,
t
, if W1 ( ) < 0.
Thus,

P(X = ) = P(W1 > 0) = 1 FW1 (0), P(X = ) = P(W1 < 0) = FW1 (0),

and
P(X = 0) = P(W1 = 0) = FW1 (0) FW1 (0).

2. Since the joint characteristic function of a zero-mean Gaussian process is completely


determined by the covariance matrix, we simply observe that

E[W t1 W t2 ] = 2 min( t1 , t2 ) = 2 min(t1 ,t2 ),

and
E[( 1/2Wt1 )( 1/2Wt2 )] = 2 min(t1 ,t2 ).

3. Fix > 0, and consider the process Zt := Wt Wt . Since the Wiener process is
Gaussian with zero mean, so is the process Zt . Hence, it suffices to consider the
covariance
E[Zt1 Zt2 ] = E[(Wt1 Wt1 )(Wt2 Wt2 )].
The time intervals involved do not overlap if t1 < t2 or if t2 < t1 . Hence,

0, |t2 t1 | > ,
E[Zt1 Zt2 ] =
2 (|t2 t1 | + ), |t2 t1 | ,

which depends on t1 and t2 only through their difference.


4. For H = 1/2, qH ( ) = I(0,) ( ), and CH = 1. So,
Z
BH (t) BH (s) = [I(,t) ( ) I(,s) ( )] dW = Wt Ws .

252
Chapter 15 Problem Solutions 253

5. It suffices to show that


Z
[(1 + )H1/2 H1/2 ]2 d < .
1

Consider the function f (t) := t H1/2 . By the mean-value theorem of calculus,

(1 + )H1/2 H1/2 = f 0 (t), for some t ( , + 1).

Since f 0 (t) = (H 1/2)t H3/2 ,



(1 + )H1/2 H1/2 |H 1/2|/ 3/2H .

Then
Z Z
1
[(1 + )H1/2 H1/2 ]2 d (H 1/2)2 d < ,
1 1 32H
since 3 2H > 1.
6. The expression
 M 
n
P 1H y = 1
/n
says that
y
= Mn with probability 1 .
n1H
Hence, the width of the confidence interval is
2y 2y
1H
= nH1/2 .
n n

7. Suppose E[Yk2 ] = 2 k2H for k = 1, . . . , n. Substituting this into the required formula
yields  
2
E[Yn+1 ] 2 n2H 2 n2H 2 (n 1)2H = 2C(n),
which we are assuming is equal to
 
2 (n + 1)2H 2n2H + (n 1)2H .
2 ] = 2 (n + 1)2H .
It follows that E[Yn+1
8. (a) Clearly,
m
(m)
Xe := (Xk )
k=( 1)m+1

is zero mean. Also,

 (m) (m)  m nm
E Xe Xen = C(k l)
k=( 1)m+1 l=(n1)m+1
254 Chapter 15 Problem Solutions

m m
= C(k [i + (n 1)m])
k=( 1)m+1 i=1
m m
= C([ j + ( 1)m] [i + (n 1)m])
i=1 j=1
m m
= C( j i + m( n)).
i=1 j=1

(m)
Thus, Xe is WSS.
(b) From the solution of part (a), we see that
m m
Ce(m) (0) = C( j i)
i=1 j=1
m1
= mC(0) + 2 C(k)(m k)
k=1
= E[Ym2 ] = m 2 2H
, by Problem 7.

Thus, C(m) (0) = 2 m2H /m2 = 2 m2H2 .


9. Starting with

C(m) (n) 2  
lim = |n + 1|2H 2|n|2H + |n 1|2H ,
m m2H2 2
put n = 0 to get
C(m) (0)
lim = 2 .
m m2H2
Then observe that
 m 2 
1
(m) E (Xk )

C(m) (0) E[(X1 )2 ] m k=1
= = .
m2H2 m2H2 m2H2

10. Following the hint, observe that


2 2
 2 2

Ce(m) (n) 1 E[Y(n+1)m ] E[Ynm ] E[Ynm ] E[Y(n1)m ]
=
m2H 2 m2H
 2
E[Y(n+1)m ] 2 2
E[Y(n1)m ] 
1 2H E[Ynm ]
= (n + 1)2H 2n + (n 1)2H
2 [(n + 1)m]2H (nm)2H [(n 1)m]2H
2
[(n + 1)2H 2n2H + (n 1)2H ].
2

11. If the equation cited in the text holds, then in particular,

Ce(m) (0)
lim = 2 ,
m m2H2
Chapter 15 Problem Solutions 255

and
Ce(m) (n) Ce(m) (n)/m2H2
lim (m) (n) = lim = lim
m m C e(m) (0) m Ce(m) (0)/m2H2
1
= [|n + 1|2H 2|n|2H + |n 1|2H ].
2
Conversely, if both conditions hold,
Ce(m) (n) Ce(m) (0) (m) (n)
lim 2H2
= lim
m m m m2H2
Ce (0)
(m)
= lim 2H2 lim (m) (n)
m m m
2
= [|n + 1|2H 2|n|2H + |n 1|2H ].
2
12. To begin, write


j2 f 2d
j f j f
j f 2d
e j f e j f 2d

S( f ) = 1 e = e [e e ]
= 2 j
2j
2d   d
= 2 sin( f ) = 4 sin2 ( f ) .
Then
Z 1/2  d
C(n) = 4 sin2 ( f ) e j2 f n d f
1/2
Z 1/2  d
= 4 sin2 ( f ) cos(2 f n) d f
1/2
Z 1/2  d
= 2 4 sin2 ( f ) cos(2 f n) d f
0
Z  d d
= 2 4 sin2 ( /2) cos(n )
0 2
Z
1  2
d
= 4 sin ( /2) cos(n ) d . ()
0
Next, as suggested in the hint, apply the change of variable = 2 to
Z 2  d
1
4 sin2 ( /2) cos(n ) d

to get Z
1  d
4 sin2 ([2 ]/2) cos(n[2 ]) d ,
0
which, using a trigonometric identity, is equal to (). Thus,
Z
1 2  d
C(n) = 4 sin2 ( /2) cos(n ) d
2 0
Z
1  d
= 4 sin2 (t) cos(2nt) dt.
0
256 Chapter 15 Problem Solutions

By the formula provided in the hint,

cos(n )(2 2d)2 p1 21p


C(n) =
(1 2d)((2 2d + 2n)/2)((2 2d 2n)/2)
(1)n (1 2d)
= .
(1 d + n)(1 d n)

13. Following hint (i), let u = sin2 and dv = 3 d . Then du = 2 sin cos d and
v = 2 /( 2). Hence,
Z r Z r
2 sin2 r 1
3 sin2 d = 2 sin 2 d .
2 2

Next,
Z r 2r Z
1 2 1 dt
sin 2 d = (t/2) 2 sint
2 2 2 2
1 Z 2r
(1/2)
= t 2 sint dt
2 2
 2r Z 2r 
21 t 1 1
= sint + t 1 cost dt .
2 1 2 1 2

Now write
Z 2r Z 2r
t 1 cost dt = Re t 1 e jt dt Re e j /2 ( ) = cos( /2)( )
2 2

as 0 and r by hint (iii). To obtain the complete result, observe that


 sin 2 sint
2 sin2 = and t 1 sint = t
t
both tend to zero as their arguments tend to zero or to infinity.
14. (a) First observe that
1 1
1 1 1
Q( f ) = |i f |2H+1 = | l f |2H+1 = 2H+1
.
i=1 l= i= |i + f |

Hence,
h 1 i
S( f ) = Q( f ) + + Q( f ) sin2 ( f ),
| f |2H+1
and then

S( f ) 2H1 2 sin2 ( f )
= | f | [Q( f ) + Q( f )] sin ( f ) + 2.
| f |12H f2
Chapter 15 Problem Solutions 257

(b) We have
Z 1/2
4 cos([1 H] )(2 2H)
S( f ) d f = 2 = 2
1/2 (2 )22H (2H 1)2H
(2 )2H cos( H)(2 2H)
= .
2H(1 2H)

15. Write
(1)n (1 2d)
C(n) =
(n + 1 d)(1 d n)
(1)n (1 2d)
=
(n + 1 d)(1)n (d)(1 d)/(n + d)
(1 2d) (n + d)
= .
(1 d)(d) (n + 1 d)
Now, with = 1 d, observe that

(n + 1 ) (n + 1 )n+1 1/2 e(n+1 )



(n + ) (n + )n+ 1/2 e(n+ )
nn+1/2 [1 + (1 )/n)]n+1/2
= e12d
nn+ 1/2 (1 + /n)n+ 1/2
[1 + (1 )/n)]n+1/2
= e12d n2d1 .
(1 + /n)n+ 1/2
Thus,
1/2
12d (n + 1 ) 12d e1 12d 1/2
n e  1/2 = e e = e1/2d .
(n + ) e

Thus, = 1 2d and c = e1/2d (1 2d)/[(1 d)(d)].


16. Evaluating the integral, we have

In n1 k1 1 (k/n)1 1
= = .
n1 (1 )n1 1 1

Now, given a small > 0, for large n, In /n1 > 1/(1 ) , which implies
In /n > n(1/(1 ) ) . With Bn as in the hint, we have from Bn + n
k In Bn that
Bn 1 k In Bn
1
+ + 1
1 1
n n n n n
or
In Bn In 1 k
1 1 .
n
1 n1 n n n
Thus, Bn /n1 1/(1 ).
258 Chapter 15 Problem Solutions

17. We begin with the inequality


n1 Z n n1
1 k
t 1 dt ( + 1)1 .
=k | {z } =k
| {z } | {z }
=: Bn =: In
= Bn k1 +n1

Next, since In = (n2 k2 )/(2 ),

In 1 (k/n)2 1
= .
n2 2 2
Also,
In Bn k1 1
2 +
n2 n2 n n
and
Bn In
2
n2 n
imply Bn /n2 1/(2 ) as required.
18. Observe that

n


|hn | = k bnk
n=q n=q k=nq

|k ||bnk |I[nq,n] (k)
n=q k=0

= |k | |bnk |I[0,q] (n k)
k=0 n=q
 q
M i (1 + /2)k
|b |
i=0 k=0
  q
1
= M |bi | < .
i=0 1 1/(1 + /2)

19. Following the hint, we first compute


 
mnYn = mn ak Xnk = mn anl Xl
n n k n l
= Xl mn anl = Xl m( +l) a ,
l n l
| {z }
= (ml)

where the reduction to the impulse follows because the convolution of n and an
corresponds in the z-transform domain to the product [1/A(z)] A(z) = 1, and 1 is the
transform of the unit impulse. Thus, n mnYn = Xm .
Chapter 15 Problem Solutions 259

Next,
 
mnYn = mn bk Znk = mn bnl Zl
n n k n l
= Zl mn bnl = Zl m(l+k) bk
l n l k
| {z }
= hml

since this last convolution corresponds in the z-transform domain to the product
[1/A(z)] B(z) =: H(z).
20. Since Xn is WSS, E[|Xn |2 ] is a finite constant. Since
k=0 |hk | < , we have by an
example in Chapter 13 or by Problem 26 in Chapter 13 that m k=0 hk Xnk converges in
mean square as m . By another example in Chapter 13,
 
E[Yn ] = E hk Xnk = hk E[Xnk ].
k=0 k=0

If E[Xnk ] = , then

E[Yn ] = hk
k=0

is finite and does not depend on n. Next, by the continuity of the inner product
(Problem 24 in Chapter 13),
     m  m
E Xl hk Xnk = lim E Xl hk Xnk = lim hk E[Xl Xnk ]
m m
k=0 k=0 k=0
m
= lim
m
hk RX (l n + k) = hk RX ([l n] + k).
k=0 k=0

Similarly,
    m  
E[YnYl ] = E hk Xnk Yl = lim E hk Xnk Yl
m
k=0 k=0
m m
= lim
m
hk E[XnkYl ] = lim
m
hk RXY (n k l)
k=0 k=0

= hk RXY ([n l] k).
k=0

Thus, Xn and Yn are J-WSS.