Scribd
Upload
43 views

Recitation Guide - Week 11: N+K 1 K N 1

This document provides solutions to 3 problems about geometric distributions, matchings in graphs, and the relationship between matchings and vertex covers. Problem 1 calculates the expectation of a geometric random variable X with parameter p to be 1/p. Problem 2 proves that a perfect matching exists in a bipartite graph modeling a chessboard with pieces in each row and column, allowing 8 non-attacking pieces to be selected. Problem 3 establishes that the size of a maximum matching is less than or equal to the size of a minimum vertex cover.

Uploaded by

Chenyang Fang
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
43 views

Recitation Guide - Week 11: N+K 1 K N 1

This document provides solutions to 3 problems about geometric distributions, matchings in graphs, and the relationship between matchings and vertex covers. Problem 1 calculates the expectation of a geometric random variable X with parameter p to be 1/p. Problem 2 proves that a perfect matching exists in a bipartite graph modeling a chessboard with pieces in each row and column, allowing 8 non-attacking pieces to be selected. Problem 3 establishes that the size of a maximum matching is less than or equal to the size of a minimum vertex cover.

Uploaded by

Chenyang Fang
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/ 5

CIS 160

Recitation Guide - Week 11

Topics Covered: Geometric Distribution, Matchings, Planar Graphs

Problem 0:
Prove the Memoryless Property for geometric random variables: For a geometric random vari-
able X with parameter p and for n > 0,

Pr[X = n + k | X > k] = Pr[X = n]

Solution:

Pr[X = n + k ∩ X > k]
Pr[X = n + k | X > k] =
Pr[X > k]
Pr[X = n + k]
= (since n > 0)
Pr[X > k]
p(1 − p)n+k−1
=
(1 − p)k
= p(1 − p)n−1 = Pr[X = n]

1
Problem 1:
Calculate the expectation of a geometric random variable X with parameter p.
Solution:
First, we need some more mechanics:

Conditional Expectation. The following is the definition of conditional expectation.


X
E[Y | Z = z] = y · Pr[Y = y | Z = z],
y

where the summation is over all possible values y that the random variable Y can assume.

Law of Total Expectation For any random variables X and Y ,


X
E[X] = E[X|Y = y] Pr[Y = y]
y

Proof:
X
E[X] = x · Pr[X = x]
x
X X
= x· Pr[X = x|Y = y] Pr[Y = y]
x y
X X
= Pr[Y = y] · x · Pr[X = x|Y = y]
y x
X
= Pr[Y = y] · E[X|Y = y]
y

Now let us calculate the expectation of a geometric random variable X using the memoryless
property of the geometric random variable. Let Y be a random variable that is 0 if the first flip
results in tails, and 1 otherwise.
Using total expectation we have:

E[X] = E[X|Y = 0] Pr[Y = 0] + E[X|Y = 1] Pr[Y = 1]

Let us try to determine E[X|Y = 0]:


∞
X
E[X|Y = 0] = x · Pr[X = x|Y = 0]
x=1
= 1 · Pr[X = 1|Y = 0]
X∞
+ x · Pr[X = x|Y = 0]
x=2

2
Note that Pr[X = 1|Y = 0] = 0 (consider what these events mean):

∞
X
= x · Pr[X = x|Y = 0]
x=2

By the memoryless property, Pr[X = x|Y = 0] = Pr[X = x|X > 1] = Pr[X = x − 1]:

∞
X
= x · Pr[X = x − 1]
x=2
∞
X
= (x + 1) · Pr[X = x]
x=1
∞
X ∞
X
= x · Pr[X = x] + Pr[X = x]
x=1 x=1
∴ E[X|Y = 0] = E[X] + 1

Returning to E[X]:

E[X] = E[X|Y = 0] Pr[Y = 0] + E[X|Y = 1] Pr[Y = 1]


= (E[X] + 1)(1 − p) + 1 · p
∴ p · E[X] = 1
1
E[X] =
p

3
Problem 2:
Consider a normal chessboard (an 8x8 grid), in which in each row and in each column there are
exactly n pieces, where 0 < n ≤ 8. Prove that we can pick 8 pieces such that no two of them are
in the same row or column.
Solution:
We construct a bipartite graph G as follows. Let X be the set of rows modeled as vertices. Let Y
be the set of columns modeled as vertices. Let E be the set of edges such that if a piece exists in
row i and column j, then there is an edge between xi ∈ X and yj ∈ Y . Note that the graph must
be bipartite because no edges exist between two vertices in X or two vertices in Y .
The question asks us to find a matching: can we match each of the 8 rows to a unique column?
Note that this would mean that we could pick 8 edges (in our matching) that are not in the same
row or same column.
We must prove the existence of such a perfect matching. Note that the size of our two bipartite
sets X and Y are the same; in other words, |X| = |Y |. If we can find a matching that saturates X,
then it must also saturate Y . We show that Hall’s Condition is satisfied, that is that |NG (S)| ≥
|S|, ∀S ⊆ X, to prove the existence of this matching.
Consider an arbitrary but particular subset A ⊆ X (of the rows). Recall that there are n pieces in
each row and n pieces in each column. Thus, there must be n|A| edges from A to NG (A). We also
know that each column in NG (A) has at most n edges back to A, meaning that there are at most
n|NG (A)| edges from NG (A) to A. This means that n|A| ≤ n|NG (A)|, meaning that |A| ≤ |NG (A)|.
This satisfies Hall’s Condition, leading us to prove the existence of our matching.
This method can be applied to prove that any k-regular bipartite graph has a perfect matching.

4
Problem 3:
A vertex cover of graph G is a set of vertices, S, such that every edge in G is incident to at least
one vertex in S. Find a relationship between the size of a maximum matching and the size of a
minimum vertex cover of a graph G.
Solution:
Let M be a maximum matching of G, and let S be a minimum vertex cover. For every edge
e = {u, v} ∈ M , there exists an endpoint of e (u or v) in S, since for every edge in a graph,
one of its endpoints must belong to S. Thus, because edges of a matching share no endpoints,
|S| ≥ |M |.

You might also like