Scribd
Upload
39 views

Recitation Guide - Week 10: 1 2 3 M 1 M I J I J I J I J I J

1) The document covers topics on Eulerian graphs, Markov's inequality, and variance. It includes 3 problems and their solutions. 2) Using Markov's inequality, the probability that a Penn student named Jake sleeps less than 8 hours per night can be lower bounded by 3/4, given that students average 6 hours of sleep. 3) If a 10-digit number with no zeros is randomly generated, the expected number of missing digits is 9*(8/9)^10 and the variance is 9*(8/9)^10 + 72*(7/9)^10 - 81*(8/9)^20.

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
39 views

Recitation Guide - Week 10: 1 2 3 M 1 M I J I J I J I J I J

1) The document covers topics on Eulerian graphs, Markov's inequality, and variance. It includes 3 problems and their solutions. 2) Using Markov's inequality, the probability that a Penn student named Jake sleeps less than 8 hours per night can be lower bounded by 3/4, given that students average 6 hours of sleep. 3) If a 10-digit number with no zeros is randomly generated, the expected number of missing digits is 9*(8/9)^10 and the variance is 9*(8/9)^10 + 72*(7/9)^10 - 81*(8/9)^20.

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/ 3

CIS 160

Recitation Guide - Week 10

Topics Covered: Eulerian Graphs, Markov’s, Variance

Problem 1:
Prove that if a graph has an Eulerian circuit, its edges can be partitioned into a set of disjoint,
simple cycles.
Solution:
We will prove this by induction on the number of edges m.
BC: m = 3. For a graph with 3 edges to have an Eulerian circuit, it must be the cycle graph on
3 vertices. In this case, our partition of the edges would just have one set- the entire set of edges,
which gives one simple cycle. Therefore, we can partition our edge set into a set of simple cycles.
X
(Why can’t the base case be a number of edges less than 3?)
IH: Assume the claim holds for all 3 ≤ j ≤ k, where j, k ∈ N. That is, assume that for a graph
with j edges, if that graph has an Eulerian circuit, then its edges can be partitioned into a set of
disjoint, simple cycles.
IS: We now consider an arbitrary graph G with k + 1 edges that contains an Eulerian circuit. Let
us consider the Eulerian circuit in G. It will be of the form v1 v2 v3 ...vm . We know that since this
is an Eulerian circuit, we must eventually visit a vertex we have seen earlier (at the least, v1 must
equal vm for this to even be a circuit).
Following the above reasoning, we consider a pair of vertices on this circuit vi and vj , i < j, such
that vi = vj . More specifically, let us consider the pair of vi and vj where j − i is minimized. We
know that since vi = vj and we can follow the circuit from vi back to itself (now enumerated vj ), we
have found a circuit. Moreover, since j − i is minimized, we know that this circuit cannot contain
a smaller circuit; therefore, it must be a simple cycle.
Let us remove all the edges in this simple cycle that we have found, and add them to a set S for
now. We now have a graph G0 , where E(G0 ) = E(G) \ S. G0 may be disconnected, and we cannot
say for sure that it has an Eulerian circuit. However, we know one thing. We have only removed
the edges in our cycle that we found earlier. For each of the vertices in that cycle, we took away
2 from each of their degrees. For all other vertices in the graph, we did not change their degree.
Since we originally had an Eulerian circuit, all vertices in G must have had even degree, and in G0 ,
they must still have even degree.
For each connected component of G0 , then, it must either be a lone vertex or contain an Eulerian
circuit (becuase all vertices have even degree within the CC). Furthermore, each of these connected
components cannot possibly have more than k edges. Therefore, by our IH, each connected com-
ponent’s (not the lone vertex CCs) edges can be partitioned into a set of disjoint, simple cycles.
Between partitions for each CC, there cannot be any shared edges (that would contradict that they
are separate CCs). Therefore, we can combine each CC’s set of disjoint cycles with S to construct a
partition on G’s k +1 edges that forms a set of disjoint, simple cycles. This completes the IS.

1
Problem 2:
Suppose it is known that students at Penn sleep an average of 6 hours per night (i.e. the expected
number of hours that each student sleeps every night is 6 hours). Jake, a disorganized Penn
student, is known to sleep a small number of hours every night. Using Markov’s inequality, provide
a lower-bound on the probability that he sleeps less than 8 hours per night.
Solution:
Let X be a random variable denoting the amount of sleep Jake gets every night. We are looking
to lower-bound Pr[X < 8]. Note that Markov’s Inequality gives information about upper bounds
on the probability that X is large. However, we also know that Pr[X < 8] = 1 − Pr[X ≥ 8].
Also, keep in mind we can only apply Markov’s Inequality because X is a duration, which must be
non-negative. We begin from the information that Markov’s Inequality guarantees us:

E[X]
Pr[X ≥ a] ≤ (Markov’s Inequality)
a
E[X]
Pr[X ≥ 8] ≤ (a = 8)
8
6
Pr[X ≥ 8] ≤ (from problem)
8
3
1 + Pr[X ≥ 8] ≤ 1 + (adding one to both sides)
4
3
1 − ≤ 1 − Pr[X ≥ 8] (rearranging)
4
1
≤ Pr[X < 8]
4

2
Problem 3:
A 10 digit number with no zeroes is chosen by independently and randomly selecting each digit (1
- 9). Let N be the number of digits missing from the 10 digit number. For example, if the number
is 1231452832, then we are missing the digits 6, 7, 9 so N = 3. Find E[N ] and Var[N ].
Solution:
Assign an indicator, Ik , for each digit which is 1 if digit k is missing and 0 otherwise. We note
8 10
that E[Ik ] = Pr[Ik = 1] = ( 89 )10 . By linearity of expectation, E[N ] = 9 · ( ) . We need E[N 2 ] =
9
E[(I1 +I2 +...+I9 )2 ] still. This expansion will yield 92 = 81 terms, 72 of which are of the form Ik ·Ij
7 10
where k 6= j, and 9 of which are of the form Ik 2 . Note that E[Ij · Ik ] = Pr[Ij = 1 ∩ Ik = 1] = ( )
9
2 8 10
and E[Ik ] = ( ) . Hence, we get our variance:
9

Var[N ] = E[N 2 ] − E[N ]2

 10  10  20


8 7 8
Var[N ] = 9 + 72 − 81
9 9 9

You might also like