Math776: Graph Theory (I)

Fall, 2013
Homework 1 solution
1. [page 30, #2 ] Determine the average degree, number of edges, diameter, girth, and circumference of the hypercube graph Qd .
Solution by Shuliang Bai:
The average degree is d, according to its definition.

Number of edges is 2d1 d. According to it definition, the

Pvertex is coded by binary
strings of length d, so |V (Qd )| = 2d . Then |E(Qd )| = 12 vV d(v) = 21 2d d =
2d1 d.
Because the distance between any two vertices depends on the number of different
bits, so the diameter is d, i.e. diam G = d.

Girth(Q1 ) = , this is because there are no cycles on hypercube graph Q1 . Girth(Qd ) =

4, where d 2, this is because Qd = K2 Qd1 .

Circumference of Qd is 2d .

2. [page 30, #3 ] Let G be a graph containing a cycle C, and assume that G contains a path
of length
at least k between two vertices of C. Show that G contains a cycle of length at
least k.
Solution by Melissa Bechard: Let C = (Vc , Ec ) denote the cycle in G, and let
P = (Vp , Ep ) denote the path of length at least k in G. Notice, if |Vp Vc | = |Vp | then
we have a path of length k with distinct edges, and we can then connect the
end points
of this path by traversing C to result ina cycle of length greater
than
k

k. Assume

this is not the case. Let |V

k. Notice, if n k, then could take the path
p Vc | = n <
along C connecting these kdistinct vertices, and then traverse C and the result would
be a cycle of length at least k. Also, between each of the n vertices of P that lie on C
there exist some subpath of P . Let P0 P be a subpath of P with the property that
both of P0 s endpoints lie on C and P0 is the maximum length such subpath of P . Let
(P0 ) = m. Additionally, any other subpath of P has length less than or equal to m. We
then have the following inequality:
k (P )
nm

km

(as defined)
(since each path between the vertices in the intersection has length m)

This inequality gives us kk m, hence k m. Thus, we have a path, P0 , of length at

least k, whose end points lie on C. So,

we can traverse C to connect these end points
and result with a cycle of length at least k.
3. [page 30, #8 ] Show that every connected graph G contains a path of length at least
min{2(G), |G| 1}.

Solution by Nicholas Stiffler: Let P = x0 x1 . . . xm be a path in graph G of maximal

length.
We will denote the length of P by l.
If l |G| 1, we are done. Otherwise, the set O = V (G) \ V (P ) is nonempty, and since
the graph is connected, there exists a V (P ) O path, P = y0 y1 . . . yk that is non-trivial.
Proposition If l < 2(G) then there is a cycle spanning V (P ).
Proof Observe that N (x0 ) P and N (xm ) P , because if either endpoint of P is
adjacent to a vertex outside of P , then the path can be extended and is not maximal. If
x0 xi+1 and xm xi are both edges in P , then there is a cycle C = x0 . . . xi xm . . . xi+1 x0 .
The occurrence of these two edges can be shown by the Pigeonhole Principle.
A special case occurs where there is an edge x0 xm , since the other edge is given in
the path.
The vertex x0 has at least (G) 1 neighbors out of {x2 . . . xm1 } because it is adjacent to x1 and not adjacent to xm (would create an obvious cycle). For each neighbor
xi there is a corresponding vertex xi1 to which xm is not adjacent. So, xm must have at
least (G) 1 neighbors out of {x1 . . . xm2 }, and of those (G) 1 are forbidden. Since
m < 2(G), there are fewer than 2(G) 2 possible neighbors. So by the Pigeonhole
Principle one of the neighbors is forbidden, so there is a cycle.
By deleting an edge of the cycle spanning V (P ) incident with y0 you can extend the
remaining path with P , forming a path longer than P , which is a contradiction.

xi+1

x0

xm

xi+1

x0

xi

xi
subgraph

xi+1

x0
xi

C spanning V (P )

Figure 1: Visualization of proof

xm

xm

4. [page 30, #9 ] Show that a connected graph of diameter k and minimum degree d has at
least about kd/3 vertices but need not have substantially more.
Solution by Robert Wilcox: We want to show that a connected graph of diameter
k and minimum degree d has at least about kd
3 vertices but need not have substantially
more. Let x0 and xk be vertices such that the shortest path, P, between the two has
length k, the diameter. Let v be a vertex not on P that is adjacent to a vertex on P.
Let i be the smallest integer such that xi is adjacent to v. If j > i + 2 then xj cannot
be adjacent to v or x0 P xi vxj P xk would be a path from x0 to xk of length less than k,
a contradiction. Thus any vertex off of P can only be adjacent to at most 3 vertices on
P. Since there are k vertices on P, each with at least d neighbors and only 3 vertices can
share a neighbor there must be at least kd/3 vertices in G. However constructing such
a path with each vertex off of the path adjacent to exactly 3 vertices of P will yield a
graph with not substantially more than kd
3 vertices.
5. [page 30, #12 ] Determine (G) and (G) for G = Pm , Cn , Kn , Km,n , and Qd ; d, m, n
3.
Proof by Garner Cochran:
(Pm ) = 1
If you remove an interior vertex of Pm , you disconnect graph.
(Pm ) = 1
If you delete any edge, the graph becomes disconnected.
(Cn ) = 2
If you delete any vertex of Cn you are left with Pn1 , therefore you must delete one more
vertex to disconnect it, so (Cn ) = 2.
(Cn ) = 2
If you delete any edge of Cn you are left with Pn , therefore you must delete one more
vertex to disconnect it, so (Cn ) = 2.
(Kn ) = n 1

You must delete all but one vertex to disconnect the graph because every vertex is has
an edge to every other vertex.
(Kn ) = n 1

Again since every vertex has an edge to every other vertex, the way to disconnect Kn by
deleting the least number of edges is deleting all of the edges incident to one vertex.
(Km,n ) = min{m, n}
The minimum number of vertices to disconnect the graph Km,n is by deleting whichever
side is smaller, if you delete any less than that, you will still have vertices on each side,
and since there are edges connecting every vertex on one side of the graph to the other
side, the graph will still be connected.
(Km,n ) = min{m, n}
Let m n, you must delete all the edges incident to a vertex on the side with n vertices.
This will allow you to disconnect a vertex from the graph. If you try to do it by deleting
less than m edges, you will be able to find some vertex on the n side which is still
3

completely connected. From the neighborhood of this vertex, we note that since none of
the vertices on the m side are disconnected, we may find a path from these vertices to
those on the n side.
(Qd ) = d
We must delete d vertices to disconnect the graph. If we delete all vertices with only one
1 in the string, then there is no way to get from the string of all zeroes to the string of
all ones.
If we try to delete only d 1 vertices, there will always be some bit it is possible to change
to continue exploring the graph.
(Qd ) = d
If we delete all edges incident to one vertex we disconnect the graph, and since d(v) = d
for all v G then it is possible to disconnect the graph by deleting d edges.

Again, by deleting only d 1 edges, we have not disconnected any individual vertex, and
since we are always able to explore the graph further, for the same reason as above, d 1
edges will not disconnect the graph.

6. [page 31, #18 ] Show that a tree without a vertex of degree 2 has more leaves than other
vertices. Can you find a very short proof that does not use induction?
Proof by Rade Musulin: Let G be a tree with no vertex of degree 2. Looking at the
average degree:
P
2|E|
2(|V | 1)
vV d(v)
=
=
< 2.
|V |
|V |
|V |

Let L be the set of leaves in G and let O be the set of vertices which are not leaves in
G. Note that the minimum degree of an element of O is 3 because no vertex has degree
2. So,
P
P
P
vO d(v)
vL d(v) +
vV d(v)
2>
=
|V |
|V |
P
P
1 + vO 3
|L| + 3|O|
|V | + 2|O|
2|O|
=
=
=1+
.
vL
|V |
|V |
|V |
|V |
So,

1>

2|O|
,
|V |

|V | > 2|O|,
|L| + |O| > 2|O|,
|L| > |O|.