Submit Assignment For Help

Go To Answer Directly
info@statisticsassignmenthelper.com

14.30 PROBLEM SET 5

TA: Tonja Bowen Bishop

Due: Tuesday, April 4, by 4:30 p.m.

PROBLEM SET 5
Note: The …rst three problems are required, and the remaining two are
practice problems. If you choose to do the practice problems now, you will
receive feedback from the grader. Alternatively, you may use them later in
the course to study for exams. Credit may not be awarded for solutions
that do not use methods discussed in class.

Problem 1
Suppose that X and Y represent coordinates on the Cartesian plane,
where X and Y both have a standard normal distribution and are indepen-
dent. Find the joint pdf of the polar coordinates r and . (Remember that
Y
r2 = X 2 + Y 2 and tan = X .) Are r and independent?

Problem 2
Suppose that n random variables Xi are independent and identically dis-
tributed as Normal: Xi N ( = 100; 2 = 225); i = 1; 2; :::; n:
a. Consider the case where n = 4. What is the probability that none
of the four random variables (X1 ; X2 ; X3 ; X4 ) is greater than 115?
P
b. De…ne the average of the n random variables as: X n = n1 ni=1 Xi .
What is the probability that X 4 is less than 115?

c. What is the smallest integer sample size n0 that one would need in
order to ensure that P X n0 5 0:95?

Problem 3
Assume the coin ‡ips below are independent ‡ips of a fair coin.
a. Calculate the exact probability that the number of heads in ten
‡ips is less than or equal to 4.

b. Use the normal approximation to the binomial distribution to ap-

proximate the probability you found in part a. and compare your answers
in terms of accuracy.

c. Now assume
Assume that you ‡ip the coin 100 times. The exact probability
(up to three digits after the decimal point) that the number of heads is less
than or equal to 40 is 0:028. What is the approximated probability (using
1

https://www.statisticsassignmenthelper.com/
2 14.30 PROBLEM SET 5

the normal distribution) that the number of heads is less than or equal to
40?

d. Suppose we now begin to use a di¤erent coin such that the ‡ips are
1
still independent but the probability of heads in each toss is now p = 20 .
Suppose you ‡ip this coin n = 100 times. Compute the exact probability
that H = 6 and compare it with an approximation based on the Poisson
distribution.

Problem 4
Let X1 and X2 be normally distributed random variables, so that
2
1 (x i )
fXi (x) = p e 2 2
2 i
for i = 1; 2, and suppose that X1 and X2 are independent.

a. Find the joint pdf of X1 and X2 .

b. De…ne Y = X1 + X2 . Find the pdf of Y using the 1-step method

(Hint: use Y2 = X1 X2 as an "ancilliary variable").

c. Find the mean and variance of Y .

Problem 5
Suppose that X is distributed chi-squared with p degrees of freedom,
where p is a positive integer.
a. Write down the pdf of X: Show that X can also be characterized
as having a gamma distribution. What are the parameters of the gamma
distribution?

b. Now suppose that Y is distributed normally with mean and vari-

2
Y
ance 2 , and that Y and X are independent. Let Z = + X. What
is the distribution of Z?

c. Suppose that p = 4: Use the appropriate table to …nd the value A

such that Pr (Z > A) = 0:05
 14.30 PROBLEM SET 5 - SUGGESTED ANSWERS

TA: Tonja Bowen Bishop

Problem 1
The joint pdf of X and Y will be equal to the product of the marginal
pdfs, since X and Y are independent.
fX,Y (x, y) = fX (x) fY (y)
1 1 2 1 1 2
= √ e− 2 x √ e− 2 y
2π 2π
1 − 1 (x2 +y2 )
= e 2
2π
The transformation into polar coordinates is
r2 = X 2 + Y 2
Y
tan θ =
X
with inverse transformations
X = r cos θ
Y = r sin θ
This yeilds the following matrix of partial derivatives.
∙ ¸
cos θ −r sin θ
sin θ r cos θ
The determinant of this matrix, the Jacobian, is
J = cos θ (r cos θ) − sin θ (−r sin θ)
= r cos2 θ + r sin2 θ
¡ ¢
= r cos2 θ + sin2 θ = r
The transformations are unique, so we can use the 1-step method without
modification.
1 1 2
frθ (r, θ) = r e− 2 r
2π
where r lies within [0, ∞] and θ lies within [0, 2π]. Because the ranges are
not dependent and the joint pdf is separable, r and θ are also independent.

Problem 2 ³ ´
a. For a single random variable: P (Xi ≤ 115) = P Xiσ−μ ≤ 115−μ
σ .
Xi −μ
Notice that Zi = σ is distributed standard normal (Zi ∼ N (0, 1)) so:
1
2 14.30 PROBLEM SET 5 - SUGGESTED ANSWERS
³ ´
P (Xi ≤ 115) = P Zi ≤ 115−100√
225
= P (Zi ≤ 1). Using the Table you can
find that this probability is approximately equal to: 0.8413. By indepen-
dence: P (X1 ≤ 115, X2 ≤ 115, X3 ≤ 115, X4 ≤ 115) =
P (X1 ≤ 115) P (X2 ≤ 115) P (X3 ≤ 115) P (X4 ≤ 115) = 0.84134 = 0.500 96.
P ³P Pn ¡ 1 ¢2 2 ´ ¡ ¢
b. X n = ni=1 n1 Xi ∼ N n 1
i=1 n μi , i=1 n σ i = N 100, 225 n =
µ ³ ´2 ¶ ³ ´
¡ ¢2
N 100, √15n , so: X 4 ∼ N 100, 15 . Thus: Z = X 4 −100 is a stan-
2 ( 15
2 )
µ ¶
¡ ¢
dard normal random variable: P X 4 < 115 = P X 4 −100 < 115−100
=
( 152 ) ( 15
2 )
P (Z < 2) = 0.9772.
ï ¯ !
¡¯ ¯ ¢ ¯ ¯ ³ √ √ ´
¯ ¯ 2
c. P ¯X n − μ¯ ≤ 5 = P ¯ ³ 15 ´ ¯ ≤ ³ 15 ´ = P − 3 n ≤ Z ≤ 3n =
X n −μ 5
¯ √n ¯ √
n

0.95. From the table we know that: P (Z ≤ 1.96) ' 0.975 and using the sym-
metry of √the normal distribution this implies that P (−1.96 ≤ Z ≤ 1.96) '
0.95, so 3n = 1.96 ⇒ n = (1.96 · 3)2 = 34. 574. We want the smallest
integer and it is n0 = 35.

Problem 3
a. The number ¡ of heads
¢ (H) in 10 independent flips of a fair coin is dis-
1 P4 ¡10¢ k 10−k
tributed Binomial 10, 2 . P (0 ≤ H ≤ 4) = k=0 k (0.5) (0.5) =
P4 ¡10¢ 10
k=0 k (0.5) =
10 £¡10¢ ¡10¢ ¡10¢ ¡10¢ ¡10¢¤ 386
= (0.5) 0 + 1 + 2 + 3 + 4 = 1024 = 0.376 95

b. Since H is binomial we can calculate its mean and variance: E [H] =

10 · (0.5) = 5, V ar [H] = 10 · (0.5) (1 − 0.5) = 2.5. The approximation re-
lies on the assumption that H is distributed similar to a normal random
H−E[H]
variable, so: √ = H−5
√
2.5
' Z ∼ N (0, 1). Therefore: P (0 ≤ H ≤ 4) =
V ar[H]
µ ¶ ³ ´
0−E[H] H−E[H] 4−E[H]
P √ ≤√ ≤√ ' P √−5 2.5
≤ Z ≤ √−1
2.5
' P (−3.162 ≤ Z ≤ −0.632) =
V ar[H] V ar[H] V ar[H]
P (Z ≤ 3.162) − P (Z ≤ 0.632) ' 0.999 − 0.736 = 0.263 . Thus the approxi-
mation is not very accurate for n = 10.
µ ¶ ³ ´
0−E[H] H−E[H] 40−E[H] −50 −10
c. Now P (0 ≤ H ≤ 40) = P √ ≤√ ≤√ 'P √ 25
≤Z≤ √
25
=
V ar[H] V ar[H] V ar[H]
P (−10 ≤ Z ≤ −2) = P (Z ≤ 10) − P (Z ≤ 2) ' 1 − 0.977 = 0.023, which is
quite close to the exact probability.
¡ ¢ ¡ 1 ¢6 ¡ ¢
1 100−6
d. Exact calculation: P (H = 6) = 1006 20 1 − 20 = 0.15.
Approximation: as n → ∞, p → 0, (np) → λ the binomial distribution
converges to the Poisson distribution with parameter λ. Since here np =
 14.30 PROBLEM SET 5 - SUGGESTED ANSWERS 3

5 we can approximate the distribution with a Poisson distribution where

−λ 6 −5 6
λ = 5: P (H = 40) ' e 6!λ = e 6!5 = 0.146 . Clearly, this is a good
approximation.

Problem 4
(x−μi )2
− 2
First of all, to have valid pdfs, we must use fXi (x) = 2σ
i . As√1 e
2πσi
always, sorry about the typo (I omitted the negative sign).
a. Because X1 and X2 are independent, the joint pdf is again the
product of the marginal pdfs:

fX1 X2 (x1 , x2 ) = fX1 (x1 ) fX2 (x2 )

(x1 −μ1 )2 (x2 −μ2 )2
1 −
2σ 2
1 −
2σ 2
= √ e 1 √ e 2
2πσ 1 2πσ 2
µ³ ´2 ³ ´2 ¶
x1 −μ1 x −μ
1 − 12 σ1
+ 2σ 2
2
= e
2πσ 1 σ 2

b. We will use Y1 for Y . We start with the transformations

Y1 = X1 + X2
Y2 = X1 − X2

which will yeild the following inverse transformations:

Y1 + Y2
X1 =
2
Y1 − Y2
X2 =
2

Then the matrix of partial derivatives is

∙1 1
¸
2 2
1
2 − 12

¯¡ ¢ ¡ ¢ ¡ ¢ ¡ ¢¯
So the Jacobian is ¯ 12 − 12 − 12 12 ¯ = 1
2
4 14.30 PROBLEM SET 5 - SUGGESTED ANSWERS

The transformation is unique, so we can use the 1-step method without

modification.
⎛Ã !2 Ã !2 ⎞
y1 +y2 y1 −y2
− 12 ⎝ 2 −μ1
+ 2 −μ2
⎠ µ ¶
1 σ1 σ2
1
fY1 Y2 (y1 , y2 ) = e −
2πσ 1 σ 2 2
µ ³ ´2 ³ ´2 ¶
y1 +y2 y −y
1 − 21 2 σ 22 2
−μ1 +σ21 1 2 2 −μ2
= e 2σ1 σ2
4πσ 1 σ 2
µ ¶
σ2 σ2
1 − 21 2 42 (y12 +2y1 y2 +y22 −4μ1 y1 −4μ1 y2 +4μ21 )+ 4
1
(y12 −2y1 y2 +y22 −4μ2 y1 +4μ2 y2 +4μ22 )
= e 2σ1 σ2
4πσ 1 σ 2
(σ21 +σ22 )((y1 −(μ1 +μ2 ))2 +(y2 −(μ1 −μ2 ))2 )
1 −
8σ 2 2
= e 1 σ2 ×
4πσ 1 σ 2
2y1 y2 −2μ1 y2 −2μ1 y1 +2μ2 y1 −2μ2 y2 +2μ2 2 2 2
1 −2μ2 − −2y1 y2 +2μ1 y2 +2μ1 y1 −2μ2 y1 +2μ2 y2 −2μ1 −2μ2
−
8σ 2 8σ 2
e 1 2

To get the pdf of Y , we must integrate over Y2 .

Z −
(σ21 +σ22 )((y1 −(μ1 +μ2 ))2 +(y2 −(μ1 −μ2 ))2 )
∞
1 8σ 2 2
1 σ2
fY (y1 ) = e ×
−∞ 4πσ 1 σ 2
2y1 y2 −2μ1 y2 −2μ1 y1 +2μ2 y1 −2μ2 y2 +2μ2 2 2 2
1 −2μ2 − −2y1 y2 +2μ1 y2 +2μ1 y1 −2μ2 y1 +2μ2 y2 −2μ1 −2μ2
−
8σ 2 8σ 2
e 1 2 dy2
(σ2 2
)
1 +σ 2 (y1 −(μ1 +μ2 ))
2 Z ∞ (σ2 2
)
1 +σ 2 (y2 −(μ1 −μ2 ))
2
1 −
8σ 2 σ 2 −
8σ2 σ 2
= e 1 2 e 1 2 ×
4πσ 1 σ 2 −∞
2y1 y2 −2μ1 y2 −2μ1 y1 +2μ2 y1 −2μ2 y2 +2μ2 2 2 2
1 −2μ2 − −2y1 y2 +2μ1 y2 +2μ1 y1 −2μ2 y1 +2μ2 y2 −2μ1 −2μ2
−
8σ 2 8σ 2
e 1 2 dy2

which has
¡ no closed form,¢in general. By other methods, it can be proved
that Y ˜N μ1 + μ2 , σ 21 + σ 22 . We can show this here if we let σ 1 = σ 2 = σ.
2 2 Z ∞
1 − (2σ )(y1 −(μ 1 +μ2 ))
−
(2σ2 )(y2 −(μ1 −μ2 ))2
fY (y1 ) = e 8σ 4 e 8σ 4 ×
4πσ 2 −∞
2y1 y2 −2μ1 y2 −2μ1 y1 +2μ2 y1 −2μ2 y2 +2μ2 2 2 2
1 −2μ2 − −2y1 y2 +2μ1 y2 +2μ1 y1 −2μ2 y1 +2μ2 y2 −2μ1 −2μ2
e− 8σ 2 8σ 2 dy2
(y −(μ1 +μ2 ))2
Z ∞ ( 2σ 2 (y2 −(μ1 −μ2 ))2
)
1 − 1 1
= √ e 4σ 2 √ e− 8σ 4 dy2
π2σ −∞ π2σ
1 (y1 −(μ1 +μ2 ))2
= √ e− 4σ 2
π2σ

R∞ (2σ2 )(y2 −(μ1 −μ2 ))2

because −∞ √ 1 e− 8σ 4 dy2 is a standard normal, and must
π2σ
integrate to 1.
 14.30 PROBLEM SET 5 - SUGGESTED ANSWERS 5

c. E (Y ) = E (X1 + X2 ) = μ1 + μ2 , since X1 and X2 are independent,

normally distributed randome variables. Similarly, V (Y ) = V (X1 + X2 ) =
σ 21 + σ 22 (since X1 and X2 are independent, their covariance is zero).

Problem 5
a. X is distributed χ2 with p degrees of freedom, so its pdf is
1 p x
f (x) = ¡ p ¢ p x 2 −1 e− 2
Γ 2 22
A gamma distribution for a random variable Y is of the form
1 y
α−1 − β
f (y) = αy e
Γ (α) β
p
You can see that if we let α = 2 and β = 2, X has a gamma distribution.

b. We learned in class that the

The square of a standard normal ran-
dom variable has a χ2 distribution with one degree of freedom. Thus
¡ y−μ ¢2 2
σ ˜χ(1) . In addition, we learned that the sum of two independent
χ variables will also have a χ2 distribution, with degrees of freedom equal
2

to the sum of the degrees of freedom of initial random variables. Because

Y and X are independent, Y 2 and X will also be independent, and we can
¡ ¢2
apply this property to conclude that y−μσ + X˜χ2(p+1) .

c. If p = 4, we use the fourth row of the table given in class, and look
for the column corresponding to α = 0.05. We can see that A = 9.488.