You are on page 1of 5

# Math 151.

## Rumbos Solutions to Assignment #21

Spring 2008

1. Fortyeight measurements are recorded to several decimal places. Each of these 48 numbers is rounded o to the nearest integer. The sum of the original 48 numbers is approximated by the sum of those integers. Assume that the errors made in rounding o are independent, identically distributed random variables with a uniform distribution over the interval (0.5, 0.5). Compute approximately the probability that the sum of the integers is within two units of the true sum. Solution: Let X1 , X2 , . . . , Xn , where n = 48 denote the 48 measurements, and Y1 , Y2 , . . . , Yn be the corresponding nearest integers after rounding o. Then Xi = Yi + Ui for i = 1, 2, . . . , n,

where U1 , U2 , . . . , Un are independent identically distributed uniform random variables on the interval (0.5, 0.5). We then have that E(Ui ) = 0 for all i, and var(Ui ) = 1 , 12

n n

S=
i=1 n

Yi +
i=1

Ui ,

where
i=1

## Yi is the sum of the integer approximations. Put

n n

W =S
i=1

Yi =
i=1

Ui .

We would like to estimate Pr(|W | 2). We will do this by applying the Central Limit Theorem to U1 , U2 , U3 , . . . Observe that W = nU n , where U n is the sample mean of the rounding o values U1 , U2 , . . . , Un . By the Central Limit Theorem Pr Un / n z Pr(Z z),

## Math 151. Rumbos

Spring 2008

for all z R, where Z Normal(0, 1). We therefore get that Pr for z > 0. It then follows that Pr(|W | 2) = Pr(|U n | = Pr |U n | / n 2/n) 2 |U n | / n z Pr(|Z| z),

n 2

Pr |Z| 2

n 1

= 2FZ

## = 2FZ (1) 1 0.6826, or about 68.3%.

2. Let X denote a random variable with pdf 1 if 1 < x < , 2 x fX (x) = 0 otherwise. Consider a random sample of size 72 from this distribution. Compute approximately the probability that 50 or more observations of the random sample are less than 3. Solution: The probability that a random observation from the distribution is less than 3 is given by
3 3

p=

fX (x) dx =
1

1 2 dx = . 2 x 3

## Math 151. Rumbos

Spring 2008

Let Y denote the number of observation out of the 72 which are less than 3. Then, Y Binomial(p, n), where n = 72. It then follows that the mean of Y is = np = 48 and the standard deviation of Y is = np(1 p) = 4. By the Central Limit Theorem we then get that Y np z Pr(Z z), Pr np(1 p) for all z R, where Z Normal(0, 1), or Pr Y 48 4 z Pr(Z z). 49), where 49 48 4

Pr (Z 0.5987.

## Thus, Pr(Y > 49) 1 0.5987 = 0.4013 or about 40.13%.

3. [Exercise 5 on page 235 in the text] How large a random sample must be taken from a given distribution in order for the probability to be at least 0.99 that the sample mean will be within 2 standard deviations of the mean of the distribution? Solution: Applying Chebyshevs Inequality to the sample mean, X n , we get that var(X n ) 2 Pr(|X n | ) = 2, 2 n for any > 0. Thus, Pr(|X n | < ) Taking = 2 we get that Pr(|X n | < 2) 1 2 1 =1 . 2 n(2) 4n 1 2 . n2

Math 151. Rumbos We want n to be so that 1 which yields 4n be at least 25. 100, or n 1 4n 0.99,

Spring 2008

## 25. Thus, we want the sample size to

4. [Exercise 6 on page 235 in the text] Suppose that X1 , X2 , . . . , Xn is a random sample of size n from a distribution for which the mean is 6.5 and the variance is 4. Determine how large the value of n must be in order for the following relation to be satised: Pr(6 Solution: Observe that 6 0.5 or |X n 6.5| 0.5. Using the inequality Pr(|X n | < ) 1 2 , n2 Xn Xn 7) 0.8.

## 7 if and only if 0.5

X n 6.6

with = 6.5, 2 = 4, and = 0.5, we obtain that Pr(|X n 6.5| < 0.5) so that Pr(|X n 6.5| 0.5) Pr(|X n 6.5| < 0.5) 1 8 . n 1 4 , n(0.5)2

Thus, in order to make Pr(|X n 6.5| 0.8, we need to choose n so that 1 This yields n 40. 8 n 0.8.

## Math 151. Rumbos 5. [Exercise 8(a) on page 235 in the text]

Spring 2008

Suppose that 30% of the items in a large manufactured lot are of poor quality. Suppose also that a random sample of n items is to be taken from the lot, and let Qn denote the proportion of the items in the sample that are or poor quality. Use the Chebyshev inequality to nd the value of n such that Pr(0.2 Qn 0.4) 0.75.

## Solution: From the Chebyshev inequality we obtain that Pr(|X n | < ) 2 1 2. n

Applying this inequality to the case in which X n = Qn , = p = 0.3, 2 = p(1 p) = 0.21 and = 0.1, we obtain that Pr(|Qn 0.3| < 0.1) We then have that Pr(0.2 Qn 0.4) Qn 1 This yields n 84. Pr(|Qn 0.3| < 0.1) 0.4) 21 n 1 21 . n 1 0.21 . n(0.1)2