UC Berkeley

Department of Electrical Engineering and Computer Sciences

EE126: Probability and Random Processes

Homework V
Spring 2020

Problem 1. Midterm Solve all of the problems on the midterm again (including
the ones you got correct).

Problem 2. Coupon Collector Convergence

Recall the coupon collector problem: for a positive integer n, there are n different
coupons, and you are trying to collect them all. Each time you purchase an item,
you receive one of the n coupons uniformly at random. Let Tn denote the amount
of time it takes to collect all n coupons.
Prove that Tn /(n ln n) → 1 in probability as n → ∞.

Problem 3. Two-Population Sampling

We are conducting a public opinion poll to determine the fraction p of people who
will vote for Mr. Whatshisname as the next president. We ask N1 college-educated
and N2 non-college-educated people, where N1 and N2 are positive integers. We
assume that the votes in each of the two groups are i.i.d. Ber(p1 ) and Ber(p2 ),
respectively in favor of Whatshisname. In the general population, the percentage of
college-educated people is known to be q.
(a) What is a 95% confidence interval for p, using an upper bound for the variance?

(b) How do we choose N1 and N2 subject to N1 + N2 = N to minimize the width

of that interval? (You may ignore the constraint that N1 and N2 must be
integers.)

Problem 4. Confidence Intervals: Chebyshev vs. Chernoff vs. CLT

Let X1 , . . . , Xn be i.i.d. Ber(q) random variables, with common mean µ = E[X1 ] = q
and variance σ 2 = var(X1 ) = q(1−q). We want to estimate the mean µ, and towards
this goal we use the sample mean estimator
∆ X1 + · · · + Xn
X̄n = .
n
Given some confidence level a ∈ (0, 1) we want to construct a confidence interval
around X̄n such that µ lies in this interval with probability at least 1 − a.

1. Use Chebyshev’s inequality in order to show that µ lies in the interval

 σ 1 σ 1 
X̄n − √ √ , X̄n + √ √
n a n a
with probability at least 1 − a.

1
 2. Recall that in Homework 4 you showed that
2
Pr(|X̄n − q| ≥ ) ≤ 2e−2n , for any  > 0.

Use this inequality in order to show that µ lies in the interval

1 1
 r r 
2 2 2 2
X̄n − √ 2 ln , X̄n + √ 2 ln
n a n a

with probability at least 1 − a.

3. Show that if Z ∼ N (0, 1), then

2
Pr(|Z| ≥ ) ≤ 2e− 2 , for any  > 0.

4. Use the Central Limit Theorem, and Part (c) in order to heuristically argue
that µ lies in the interval
 r r 
σ 2 σ 2
X̄n − √ 2 ln , X̄n + √ 2 ln
n a n a

with probability at least 1 − a.

5. Compare the three confidence intervals.

Problem 5. [Bonus] The CLT Implies the WLLN

1. Let {Xn }n∈N be a sequence of random variables. Show that if Xn converges

P
to c in distribution, where c is a constant, then Xn → c.

2. Let {Xn }n∈N be a sequence of i.i.d. random variables, with mean µ and finite
variance σ 2 . Show that the CLT implies the WLLN.