Scribd
Upload
33 views1 page

ECE 534: Exam I: Monday October 10, 2011 7:00 P.M. - 8:15 P.M. 269 Everitt Laboratory

This document provides the questions and solutions for an exam on ECE534. It includes 4 questions covering topics like: 1) Finding the mean squared error of an estimator and properties of conditional expectations. 2) Convergence of random sequences and their expectations and variances. 3) Determining if a series converges by comparing it to an integral test. 4) Estimating a random variable given observations from a joint Gaussian distribution.

Uploaded by

Thinh
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
33 views1 page

ECE 534: Exam I: Monday October 10, 2011 7:00 P.M. - 8:15 P.M. 269 Everitt Laboratory

This document provides the questions and solutions for an exam on ECE534. It includes 4 questions covering topics like: 1) Finding the mean squared error of an estimator and properties of conditional expectations. 2) Convergence of random sequences and their expectations and variances. 3) Determining if a series converges by comparing it to an integral test. 4) Estimating a random variable given observations from a joint Gaussian distribution.

Uploaded by

Thinh
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

University of Illinois Fall 2011

ECE534: Exam I
Monday October 10, 2011
7:00 p.m. 8:15 p.m.
269 Everitt Laboratory
1. (a) E[X|Y
1
, Y
2
] = (1, 2)

2 2
2 5

Y
1
Y
2

=
Y
1
6
+
Y
2
3
.
(b) MSE=E[X
2
] E[(
Y
1
6
+
Y
2
3
)
2
] = 1
2
36

5
9
=
14
36
=
7
18
.
(c) First,

Y
1
= Y
1
because Y
1
has mean zero. Secondly,

Y
2
= Y
2


E[Y
2
|Y
1
] = Y
1

2
5
Y
2
.
2. (a) No. For example, since P{X
n+1
= X
n
+ 1} = 0.5, E[(X
n+1
X
n
)
2
] 0.5 for all n, so
(X
n
: n 1) is not a Cauchy sequence in the m.s. sense, so it does not converge in the
m.s. sense.
(b) Taking expectations on each side of the recursion dening the random sequence yields
E[X
n+1
] = (0.5)E[X
n
]/2 + (0.5)(E[X
n
] + 1) = (0.75)E[X
n
] + 0.5. Solving the recursion
yields E[X
n
] = 2 (3/4)
n
; lim
n
E[X
n
] = 2.
(c) Assuming lim
n
Var(X
n
) exists, and knowing that lim
n
E[X
n
] exists, it follows that
m
2

= lim
n
E[X
2
n
] exists. Squaring both sides of the recursion dening the random
sequence yields E[X
2
n+1
] = (0.5)(E[X
n
]
2
/4) + (0.5)(E[X
2
n
] + 2E[X
n
] + 1). Taking the
limit on each side as n , and using the result of part (b) yields the following xed
point equation for the limit m
2
: m
2
=
5
8
m
2
+ 2.5 which has solution m
2
=
20
3
so
lim
n
Var(X
n
) =
8
3
.
3. Let s
n
denote the n
th
partial sum of the series. Then s
3n
= a
n
+b
n
where a
n
= 1 +
1
4
+
1
7
+
+
1
3n2
and b
n
= (
1
2

1
3
) + (
1
5

1
6
) + (
1
8

1
9
) + + (
1
3n1

1
3n
). Since b
n
> 0 for all
n, s
3n
a
n

1
3

3n
1
1
u
du =
ln n
3
, it follows that s
3n
as n . (It easily follows that
s
n
as well.) So s
n
does not converge to a nite limit.
4. (a) Note that Y = ln(Z) where Z = X + N. Since ln( ) is a bijection (one-to-one and
onto function), Y and Z hold the same information. Equivalently, any function of Y
can be expressed as a function of Z, and vice versa. Thus, E[X|Y ] = E[X|Z], putting
the problem within the realm of estimation involving joint Gaussians. So E[X|Y ] =
E[X|Z] =
Cov(X,Z)Z
Var(Z)
=
Z
1+
2
=
e
Y
1+
2
.
(b) By the formulas for linear MMSE, E[(X E[X|Y ])
2
] = E[(X E[X|Z])
2
] =

2
1+
2
.

You might also like