Preliminary Exam: Probability and Mathematical Statistics May 2006

Name:
Instructions: Clearly mark the answers and show all the work. You may use calculators when appropriate.
Solve three problems from each part. Clearly marke the ones you are NOT solving
PART ONE: Probabillity
Solve three out of the next four problems
1. (50 points) This question consists of three parts.
(a) (15 points) Let X
i
be a sequence of random variables such that
lim
n
V arS
n
n
2
= 0, (1)
where S
n
=

n
i=1
X
i
. Show that
S
n
ES
n
n
P
0 as n . (2)
(b) (15 points) Now, suppose that we replace the condition (1) with the assumption that the X
i
are
pairwise uncorrelated and satisfy sup
i
EX
2
i
< . Show that the result (2) above also holds under
these alternative assumptions.
(c) (20 points) Assuming that the X
i
are independent random variables such that sup
i
EX
4
i
< ,
show that the convergence in probability in (2) can be strengthened to convergence almost surely.
Page 1 of 8 Please go on to the next page. . .
Preliminary Exam: Probability and Mathematical Statistics May 2006
2. (50 points points) Given a probability space (, F
0
, P) , an F
0
-measureable random variable X and
another -eld F F
0
, the conditional expectation of X given F is dened to be any random
variable Y which is F-measureable and satises
_
A
XdP =
_
A
Y dP
for all A F.
(a) (15 points) To warm up, consider how this denition relates to the one taught in undergraduate
probability. Specically, suppose that
1
,
2
, . . . is a nite or innite partition of into disjoint
sets each of which has positive probability (with respect to P), and let F = (
1
,
2
, . . . ) the
-eld generated by these sets. Then show that on each
i
,
E (X|F ) =
E (X;
i
)
P (
i
)
.
(b) (15 points) Let F
1
F
2
be two -elds on . Then show that
1. E (E (X|F
1
) |F
2
) = E (X|F
1
) , and
2. E (E (X|F
2
) |F
1
) = E (X|F
1
) .
(c) (20 points) Let = {a, b, c}. Give an example of P, F
1
, F
2
, and X in which
E (E (X|F
1
) |F
2
) = E (E (X|F
2
) |F
1
) .
Page 2 of 8 Please go on to the next page. . .
Preliminary Exam: Probability and Mathematical Statistics May 2006
3. (50 points) This problem consists of two parts.
(a) (25 points) Let X be N(0, 1) random variable. Let
M(s) = E{e
s X
} =
_

2
exp
_
s x
x
2
2
_
dx.
Show that M(s) = e
s
2
/2
.
(b) (25 points) Show that for any positive integer n we have E{X
2n+1
} = 0 and
E{X
2n
} =
(2n)!
2
n
n!
= (2n 1)(2n 3) . . . 3 1.
(Hint: Note that e
s
2
/2
=

k=0
s
2k
2
k
k!
Page 3 of 8 Please go on to the next page. . .
Preliminary Exam: Probability and Mathematical Statistics May 2006
4. (50 points) This question consists of two parts.
(a) (25 points) Show that for any c.d.f. F and any a 0
_
[F(x +a) F(x)]dx = a
(b) (25 points) Let X be a random variable with range {0,1,2,. . . }. Show that if EX < then
EX =

i=1
P(X i)
Page 4 of 8 Please go on to the next page. . .
Preliminary Exam: Probability and Mathematical Statistics May 2006
PART TWO: Mathematical Statistics
Solve three out of the next four problems
5. (50 points) Let U
1
, . . . , U
n
be i.i.d. random variables having uniform distribution on [0,1] and Y
n
=
(

n
i=1
U
i
)
1/n
. Show that

n(Y
n
e)
D
N(0, e
2
)
where e = exp(1).
Page 5 of 8 Please go on to the next page. . .
Preliminary Exam: Probability and Mathematical Statistics May 2006
6. (50 points) Let be a UMP test of level (0, 1) for testing simple hypothesis P
0
vs P
1
. If is the
power of the test, show that with equality if and only if P
0
= P
1
.
Page 6 of 8 Please go on to the next page. . .
Preliminary Exam: Probability and Mathematical Statistics May 2006
7. (50 points) Let X
1
, . . . , X
n
be independently and identically distributed with density
f(x, ) =
1

exp
_

_
, x ,
where = (,
2
), < < ,
2
> 0.
(a) Find maximum likelihood estimates of and
2
.
(b) Find the maximum likelihood estimate of P

(X
1
t) for t > .
Page 7 of 8 Please go on to the next page. . .
Preliminary Exam: Probability and Mathematical Statistics May 2006
8. (50 points) Let X
1
, . . . , X
n
be iid from Bernoulli distribution with unknown probability of success
P(X
1
= 1) = p (0, 1).
(a) (20 points) Show that S =
n

i=1
X
i
is a complete and sucient statistic.
(Hint:
n

k=0
_
n
k
_
g(k)p
k
(1 p)
nk
= (1 p)
n
_
n

k=0
_
n
k
_
g(k)
k
_
where =
p
1 p
.)
(b) (30 points) Find UMVUE for p
m
when m n is a positive integer.
Page 8 of 8 Done !