35

Chapter 1: Fundamental Concepts

its neighbor on P . Note that P has at least three vertices. If G x v

is connected, let y = v . Otherwise, a component cut off from P x v in
G x v has at most one vertex; call it w . The vertex w must be adjacent
to v , since otherwise we could build a longer path. In this case, let y = w .

1.2.42. A connected simple graph having no 4-vertex induced subgraph that

is a path or a cycle has a vertex adjacent to every other vertex. Consider a
vertex x of maximum degree. If x has a nonneighbor y , let x, v, w be the
begining of a shortest path to y (w may equal y ). Since d(v) d(x), some
neighbor z of x is not adjacent to v . If z w , then {z, x, v, w} induce C 4 ;
otherwise, {z, x, v, w} induce P4 . Thus x must have no nonneighbor.

1.2.43. The edges of a connected simple graph with 2k edges can be partitioned into paths of length 2. The assumption of connectedness is necessary,
since the conclusion does not hold for a graph having components with an
odd number of edges.
We use induction on e(G); there is a single such path when e(G) = 2.
For e(G) > 2, let P = (x, y, z) be an arbitrary path of length two in G ,
and let G ! = G {x y, yz}. If we can partition E(G) into smaller connected
subgraphs of even size, then we can apply the induction hypothesis to each
piece and combine the resulting decompositions. One way to do this is to
partition E(G ! ) into connected subgraphs of even size and use P .
Hence we are finished unless G ! has two components of odd size (G !
cannot have more than three components, since an edge deletion increases
the number of components by at most one). Each odd component contains
at least one of {x, y, z}. Hence it is possible to add one of x y to one odd
component and yz to the other odd component to obtain a partition of G
into smaller connected subgraphs.

1.3. VERTEX DEGREES & COUNTING

1.3.1. A graph having exactly two vertices of odd degree must contain a
path from one to the other. The degree of a vertex in a component of G is
the same as its degree in G . If the vertices of odd degree are in different
components, then those components are graphs with odd degree sum.
1.3.2. In a class with nine students where each student sends valentine
cards to three others, it is not possible that each student sends to and receives
cards from the same people. The sending of a valentine can be represented
as a directed edge from the sender to the receiver. If each student sends to
and receives cards from the same people, then the graph has x y if and

Section 1.3: Vertex Degrees and Counting

36

only if y x . Modeling each opposed pair of edges by a single unoriented

edge yields a 3-regular graph with 9 vertices. This is impossible, since
every graph has an even number of vertices of odd degree.
1.3.3. If d(u) + d(v) = n + k for an edge uv in a simple graph on n vertices,
then uv belongs to at least k triangles. This is the same as showing that
u and v have at least k common neighbors. Let S be the neighbors of u
and T the neighbors of v , and suppose |S T | = j . Every vertex of G
appears in S or T or none or both. Common neighbors are counted twice,
so n |S| + |T | j = n + k j . Hence j k . (Almost every proof of this
using induction or contradiction does not need it, and is essentially just
this counting argument.)
1.3.4. The graph below is isomorphic to Q 4 . It suffices to label the vertices
with the names of the vertices in Q 4 so that vertices are adjacent if and
only if their labels differ in exactly one place.
0010

0110

0011

0111

1010 1110

1111 1011

1000 1100

1101 1001

0100

0000

0101

0001

!"
1.3.5. The k -dimensional cube Q k has 2k 2k copies of P3 .
Proof 1. To specify a particular subgraph isomorphic to P3 , the 3vertex path, we can specify !the
" middle vertex and its two neighbors. For
each vertex of Q k , there are 2k ways to choose two distinct neighbors, since
!"
Q k is a simple k -regular graph. Thus the total number of P3 s is 2k 2k .
Proof 2. We can alternatively choose the starting vertex and the next
two. There are 2k ways to pick the first vertex. For each vertex, there are
k ways to pick a neighbor. For each way to pick these vertices, there are
k 1 ways to pick a third vertex completing P3 , since Q k has no multiple
edges. The product of these factors counts each P3 twice, since we build it
from each end.
the total number of them is 2k k(k 1)/2.
!k " Thus
k2
Q k has 2 2
copies of C 4 .
Proof 1 (direct counting). The vertices two apart on a 4-cycle must
differ in two coordinates. Their two common neighbors each differ from
each in exactly one of these coordinates. Hence the vertices of a 4-cycle

37

Chapter 1: Fundamental Concepts

must use all 2-tuples in two coordinates while keeping the

! "remaining coordinates fixed. All such choices yield 4-cycles. There are 2k ways to choose
the two coordinates that vary and 2k2 ways to set a fixed value in the
remaining coordinates.
Proof 2 (prior result). Every 4-cycle contains four copies of P3 , and every P3 contains two vertices at distance 2 in the cube and hence extends to
exactly one 4-cycle. Hence the number of 4-cycles is one-fourth the number
of copies of P3 .
1.3.6. Counting components. If G has k components and H has l components, then G + H has k + l components. The maximum degree of G + H
is max{!(G), !(H )}.

1.3.7. Largest bipartite subgraphs. Pn is already bipartite. C n loses one

edge if n is odd, none #if n is$even. The largest bipartite subgraph of K n is
K "n/2#,$n/2% , which has n 2 /4 edges.
1.3.8. The lists (5,5,4,3,2,2,2,1), (5,5,4,4,2,2,1,1), and (5,5,5,3,2,2,1,1) are
graphic, but (5,5,5,4,2,1,1,1) is not. The answers can be obtained from
the Havel-Hakimi test; a list is graphic if and only if the list obtained by
deleting the largest element and deleting that many next-largest elements
is graphic. Below are graphs realizing the first three lists, found by the
Havel-Hakimi algorithm.

From the last list, we test (4, 4, 3, 1, 0, 1, 1), reordered to (4, 4, 3, 1, 1, 1, 0),
then (3, 2, 0, 0, 1, 0). This is not the degree list of a simple graph, since a
vertex of degree 3 requires three other vertices with nonzero degree.
1.3.9. In a league with two divisions of 13 teams each, no schedule has
each team playing exactly nine games against teams in its own division and
four games against teams in the other division. If this were possible, then
we could form a graph with the teams as vertices, making two vertices
adjacent if those teams play a game in the schedule. We are asking for
the subgraph induced by the 13 teams in a single division to be 9-regular.
However, there is no regular graph of odd degree with an odd number of
vertices, since for every graph the sum of the degrees is even.
1.3.10. If l, m, n are nonnegative integers with l + m = n 1, then there
exists a connected simple n -vertex graph with l vertices of even degree and m

Section 1.3: Vertex Degrees and Counting

38

vertices of odd degree if and only if m is even, except for (l, m, n) = (2, 0, 2).
Since every graph has an even number of vertices of odd degree, and the
only simple connected graph with two vertices has both degrees odd, the
condition is necessary.
To prove sufficiency, we construct such a graph G . If m = 0, let G = C l
(except G = K 1 if l = 1). For m > 0, we can begin with K 1,m1 , which has
m vertices of odd degree, and then add a path of length l beyond one of the
leaves. (Illustration shows l = 3, m = 4.)
Alternatively, start with a cycle of length l , and add m vertices of degree
one with a common neighbor on the cycle. That vertex of the cycle has even
degree because m is even. Many other constructions also work. It is also
possible to prove sufficiency by induction on n for n 3, but this approach
is longer and harder to get right than an explicit general construction.

1.3.11. If C is a closed walk in a simple graph G , then the subgraph consisting of the edges appearing an odd number of times in C is an even graph.
Consider an arbitrary vertex v V (G). Let S be the set of edges incident
to v , and let f (e) be the number of times an edge e is traversed
% by C . Each
time C passes through v it enters and leaves. Therefore, eS f (e) must
be even, since it equals twice the number of times that C visits v . Hence
there must an even number of odd contributions to the sum, which means
there are an even number of edges incident to v that appear an odd number of times in C . Since we can start a closed walk at any of its vertices,
this argument holds for every v V (G).
1.3.12. If every vertex of G has even degree, then G has no cut-edge.
Proof 1 (contradiction). If G has a cut-edge, deleting it leaves two
induced subgraphs whose degree sum is odd. This is impossible, since the
degree sum in every graph is even.
Proof 2 (construction/extremality). For an edge uv , a maximal trail in
G uv starting at u can only end at v , since whenever we reach a vertex
we have use an odd number of edges there. Hence a maximal such trail is
a (u, v)-trail. Every (u, v)-trail is a (u, v)-walk and contains a (u, v)-path.
Hence there is still a (u, v)-path after deletion of uv , so uv is not a cut-edge.
Proof 3 (prior results). Let G be an even graph. By Proposition 1.2.27,
G decomposes into cycles. By the meaning of decomposition, every edge

39

Chapter 1: Fundamental Concepts

of G is in a cycle. By Theorem 1.2.14, every edge in a cycle is not a cut-edge.

Hence every edge of G is not a cut-edge.
For k N, some (2k + 1)-regular simple graph has a cut-edge.
Construction 1. Let H, H ! be copies of K 2k,2k with partite sets X, Y
for H and X ! , Y ! for H ! . Add an isolated edge vv ! disjoint from these sets.
To H + H ! + vv ! , add edges from v to all of X and from v ! to all of X ! , and
add k disjoint edges within Y and k disjoint edges within Y ! . The resulting
graph G k is (2k + 1)-regular with 8k + 2 vertices and has vv ! as a cut-edge.
Below we sketch G 2 ; the graph G 1 is the graph in Example 1.3.26.

v!

X!

Y!

Construction 2a (inductive). Let G 1 be the graph at the end of Example 1.3.26 (or in Construction 1). This graph is 3-regular with 10 vertices
and cut-edge x y ; note that 10 = 4 1 + 6. From a (2k 1)-regular graph
G k1 with 4k + 2 vertices such that G k1 x y has two components of order
2k + 1, we form G k . Add two vertices for each component of G k1 x y , adjacent to all the vertices of that component. This adds degree two to each old
vertex, gives degree 2k + 1 to each new vertex, and leaves x y as a cut-edge.
The result is a (2k + 1)-regular graph G k of order 4k + 6 with cut-edge x y .
Construction 2b (explicit). Form Hk from K 2k+2 by removing k pairwise disjoint edges and adding one vertex that is adjacent to all vertices
that lost an incident edge. Now Hk has 2k + 2 vertices of degree 2k + 1
and one of degree 2k . Form G k by taking two disjoint copies of Hk and
adding an edge joining the vertices of degree 2k . The graphs produced in
Constructions 2a and 2b are identical.
1.3.13. Meeting on a mountain range. A mountain range is a polygonal
curve from (a, 0) to (b, 0) in the upper half-plane; we start A and B at
opposite endpoints. Let P be a highest peak; A and B will meet there. Let
the segments from P to (a, 0) be x 1 , . . . , xr , and let the segments from P
to (b, 0) be y1 , . . . , ys . We define a graph to describe the positions; when
A is on xi and B is on y j , the corresponding vertex is (i, j). We start at
the vertex (r, s) and must reach (1, 1). We introduce edges for the possible
transitions. We can move from (i, j) to (i, j + 1) if the common endpoint of
yj and y j+1 has height between the heights of the endpoints of x i . Similarly,
(i, j) is adjacent to (i + 1, j) if the common endpoint of x i and xi+1 has
height between the heights of the endpoints of x j . To avoid triviality, we
may assume that r + s > 2.
We prove that (r, s) and (1, 1) are the only vertices of odd degree in
G . This suffices, because every graph has an even number of vertices of

Section 1.3: Vertex Degrees and Counting

40

odd degree, which implies that (r, s) and (1, 1) are in the same component,
connected by a path.
The possible neighbors of (i, j) are the pairs obtained by changing i
or j by 1. Let X and Y be the intervals of heights attained by x i and y j ,
and let I = X Y . If the high end of I is the high end of exactly one of X
and Y , then exactly one neighboring vertex can be reached by moving past
the end of the corresponding segment. If it is the high end of both, then
usually one or three neighboring vertices can be reached, the latter when
both segments reach peaks at their high ends. However, if (i, j) = (1, 1),
then the high end of both segments is P and there is no neighbor of this
type. Similarly, the low end of I generates one or three neighbors, except
that when (i, j) = (r, s) there is no neighbor of this type.
No neighbor of (i, j) is generated from both the low end and the high
end of I . Since the contributions from the high and low end of I to the
degree of (i, j) are both odd, each degree is even, except for (r, s) and (1, 1),
where exactly one of the contributions is odd.
1.3.14. Every simple graph with at least two vertices has two vertices of
equal degree. The degree of a vertex in an n -vertex simple graph is in
{0, . . . , n 1}. These are n distinct values, so if no two are equal then all
appear. However, a graph cannot have both an isolated vertex and a vertex
adjacent to all others.
This does not hold for graphs allowing loops. In the 2-vertex graph
with one loop edge and one non-loop edge, the vertex degrees are 1 and 3.
This does not hold for loopless graphs. In the 3-vertex loopless graph
with pairs having multiplicity 0, 1, 2, the vertex degrees are 1, 3, 2.
1.3.15. Smallest k -regular graphs. A simple k -regular graph has at least
k + 1 vertices, so K k+1 is the smallest. This is the only isomorphism class of
k -regular graphs with k + 1 vertices. With k + 2 vertices, the complement of
a k -regular graph must be 1-regular. There is one such class when k is even
((k + 2)/2 isolated edges), none when k is odd. (Two graphs are isomorphic
if and only if their complements are isomorphic.)
With k + 3 vertices, the complement is 2-regular. For k 3, there are
distinct choices for such a graph: a (k + 3)-cycle or the disjoint union of a
3-cycle and a k -cycle. Since these two 2-regular graphs are nonisomorphic,
their complements are nonisomorphic k -regular graphs with k + 3 vertices.
1.3.16. For k 2 and g 2, there exists a k -regular graph with girth g .
We use strong induction on g . For g = 2, take the graph consisting of two
vertices and k edges joining them.
For the induction step, consider g > 2. Here we use induction on k .
For k = 2, a cycle of length g suffices. For k > 2, the induction hypothesis

41

Chapter 1: Fundamental Concepts

provides a (k 1)-regular graph H with girth g . Since "g/2# < g , the global
induction hypothesis also provides a graph G with girth "g/2# that is n(H )regular. Replace each vertex v in G with a copy of H ; each vertex in the
copy of H is made incident to one of the edges incident to v in G .
Each vertex in the resulting graph inherits k 1 incident edges from
H and one from G , so the graph is k -regular. It has cycles of length g in
copies of H . A cycle C in G is confined to a single copy of H or visits more
than one such copy. In the first case, its length is at least g , since H has
girth g . In the second case, the copies of H that C visits correspond to a
cycle in G , so C visits at least "g/2# such copies. For each copy, C must
enter on one edge and then move to another vertex before leaving, since
the copy is entered by only one edge at each vertex. Hence the length of
such a cycle is at least 2 "g/2#.
1.3.17. Deleting a vertex of maximum degree cannot increase the average
degree, but deleting a vertex of minimum degree can reduce the average degree. Deleting any vertex of a nontrivial regular graph reduces the average
degree, which proves the second claim. For the first claim, suppose that G
has n vertices and m edges, and let a and a ! be the average degrees of G
and G x , respectively. Since G x has m d(x) edges and degree sum
2m 2d(x), we have a ! = na2d(x)
(n2)a
< a if d(x) a > 0. Hence deletn1
n1
ing a vertex of maximum degree in nontrivial graph reduces the average
degree and cannot increase it.

1.3.18. If k 2, then a k -regular bipartite graph has no cut-edge. Since

components of k -regular graphs are k -regular, it suffices to consider a connected k -regular X, Y -bigraph. Let uv be a cut-edge, and let G and H be the
components formed by deleting uv . Let m = |V (G) X | and n = |V (G) Y |.
By symmetry, we may assume that u V (G) Y and v V (H ) X .
We count the edges of G . The degree of each vertex of G in X is k , so G
has mk edges. The degree of each vertex of G in Y is k except for d G (u) =
k 1, so G has nk 1 edges. Hence m k = nk 1, which is impossible because
one side is divisible by k and the other is not. The proof doesnt work if
k = 1, and the claim is false then.
If vertex degrees k and k + 1 are allowed, then a cut-edge may exist.
Consider the example of 2 K k,k plus one edge joining the two components.
1.3.19. A claw-free simple graph with maximum degree at least 5 has a 4cycle. Consider five edges incident to a vertex v of maximum degree in such
a graph G . Since G has no induced claw, the neighbors of v must induce at
least three edges. Since these three edges have six endpoints among the
five neighbors of v , two of them must be incident, say x y and yz . Adding
the edges xv and zv to these two completes a 4-cycle.
There are arbitrarily large 4-regular claw-free graphs with no 4-cycles.

Section 1.3: Vertex Degrees and Counting

42

Consider a vertex v in such a graph G . Since v has degree 4 and is not

the center of an induced claw and does not lie on a 4-cycle, the subgraph
induced by v and its neighbors consists of two edge-disjoint triangles sharing v (a bowtie). Since this happens at each vertex, G consists of pairwise
edge-disjoint triangles, with each vertex lying in two of them. Hence each
triangle has three neighboring triangles. Furthermore, two triangles that
neighbor a given triangle in this way cannot neighbor each other; that
would create a 4-cycle in the graph.
Define a graph H with one vertex for each triangle in G ; let vertices be
adjacent in H if the corresponding triangles share a vertex in G . Now H is
a 3-regular graph with no 3-cycles; a 3-cycle in H would yield a 4-cycle in
G using two edges from one of the corresponding triangles. Also H must
have no 4-cycles, because a 4-cycle in G could be built using one edge from
each of the four triangles corresponding to the vertices of a 4-cycle in H .
Note that e(G) = 2n(G) and n(H ) = e(G)/3 = 2n(G)/3.
On the other hand, given any 3-regular graph H with girth at least 5,
reversing the construction yields G with the desired properties and 3n(H )/2
vertices. Hence it suffices to show that there are arbitrarily large 3-regular
graphs with girth at least 5. Disconnected such examples can be formed
by taking many copies of the Petersen graph as components. The graph G
is connected if and only if H is connected. Connected instances of H can be
obtained from multiple copies of the Petersen graph by applying 2-switches
(Definition 1.3.32).
Alternatively, arbitrarily large connected examples can be constructed
by taking two odd cycles (say length 2m + 1) and joining the i th vertex
on the first cycle to the 2i th vertex (modulo 2m + 1) on the second cycle
(this generalizes the Petersen graph). We have constructed a connected
3-regular graph. Since we add disjoint edges between the cycles, there is
no triangle. A 4-cycle would have to alternate edges between the two odd
cycles with one edge of each, but the neighbors of adjacent vertices on the
first cycle are two apart on the second cycle.
1.3.20. K n has (n 1)!/2 cycles of length n , and K n,n has n !(n 1)!/2 cycles
of length 2n . Each cycle in K n is a listing of the vertices. These can be listed
in n ! orders, but we obtain the same subgraph no matter where we start
the cycle and no matter which direction we follow, so each cycle is listed
2n times. In K n,n , we can list the vertices in order on a cycle (alternating
between the partite sets), in 2(n !)2 ways, but by the same reasoning each
cycle appears (2n) 2 times.
! "! "
1.3.21. K m,n has 6 m3 3n 6-cycles. To extend an edge in K m,n to a 6-cycle, we
choose two more vertices from each side to be visited in order as we follow
the cycle. Hence each edge in K n,n appears in (m 1)(n 1)(m 2)(n 2)

43

Chapter 1: Fundamental Concepts

6-cycles. Since each 6-cycle contains 6 edges, we conclude that K n,n has
mn(m 1)(n 1)(m 2)(n 2)/6 6-cycles.
Alternatively, each 6-cycle
! "! " uses three vertices from each partite set,
which we can choose in m3 3n ways. Each such choice of vertices induces
a copy of K 3,3 with 9 edges. There are 3! = 6 ways to pick three disjoint
edges to be omitted by a 6-cycle, so each K 3,3 contains 6 6-cycles.
1.3.22. Odd girth and minimum degree in nonbipartite triangle-free n vertex graphs. Let k = (G), and let l be the minimum length of an odd
cycle in G . Let C be a cycle of length l in G .
a) Every vertex not in V (C) has at most two neighbors in V (C). It
suffices to show that any two neighbors of such a vertex v on C must have
distance 2 on C , since having three neighbors would then require l = 6.
Since G is triangle-free, v does not have consecutive neighbors on C .
If v has neighbors x and y on C separated by distance more than 2 on C ,
then the detour through v can replace the x, y -path of even length on C to
form a shorter odd cycle.
b) n kl/2 (and thus l 2n/k ). Since C is a shortest odd cycle, it has
no chords (it is an induced cycle). Since (G) = k , each vertex of C thus has
at least k 2 edges to vertices outside C . However, each vertex outside C
has at most two neighbors on C . Letting m be the number of edges from
V (C) to V (G) V (C), we thus have l(k 2) m 2(n l). Simplifying
the inequality yields n kl/2.
c) The inequality of part (b) is sharp when k is even. Form G from the
cycle Cl by replacing each vertex of Cl with an independent set of size k/2
such that two vertices are adjacent if and only if the vertices they replaced
were adjacent. Each vertex is now adjacent to the vertices arising from the
two neighboring classes, so G is k -regular and has lk/2 vertices. Deleting
the copies of any one vertex of Cl leaves a bipartite graph, since the partite
sets can be labeled alternately around the classes arising from the rest of
Cl . Hence every odd cycle uses a copy of each vertex of C l and has length
at least l , and taking one vertex from each class forms such a cycle.
1.3.23. Equivalent definitions of the k -dimensional cube. In the direct definition of Q k , the vertices are the binary k -tuples, with edges consisting of
pairs differing in one place. The inductive definition gives the same graph.
For k = 0 both definitions specify K 1 . For the induction step, suppose k 1.
The inductive definition uses two copies of Q k1 , which by the induction
hypothesis is the 1-place difference graph of the binary (k 1)-tuples. If
we append 0 to the (k 1)-tuples in one copy of Q k1 and 1 to the (k 1)tuples in the other copy, then within each set we still have edges between
the labels differing in exactly one place. The inductive construction now
adds edges consisting of corresponding vertices in the two copies. This is

Section 1.3: Vertex Degrees and Counting

44

also what the direction definition does, since k -tuples chosen from the two
copies differ in the last position and therefore differ in exactly one position
if and only if they are the same in all other positions.
e(Q k ) = k 2k1 . By the inductive definition, e(Q k ) = 2e(Q k1 ) + 2k1 for
k 1, with e(Q 0 ) = 0. Thus the inductive step for a proof of the formula is
e(Q k ) = 2(k 1)2k2 + 2k1 = k2k1 .
1.3.24. K 2,3 is the smallest simple bipartite graph that is not a subgraph
of the k -dimensional cube for any k . Suppose the vectors x, y, a, b, c are the
vertices of a copy of K 2,3 in Q k . Any one of a, b, c differs from x in exactly one
coordinate and from y in another (it cant be the same coordinate, because
then x = y ). This implies that x and y differ in two coordinate i, j . Paths
from x to y in two steps can be formed by changing i and then j or changing
j and then i ; these are the only ways. In a cube two vertices have at
most two common neighbors. Hence K 2,3 is forbidden. Any bipartite graph
with fewer vertices or edges is contained in K 2,3 e or K 1,5 , but K 2,3 e
is a subgraph of Q 3 , and K 1,5 is a subgraph of Q 5 , so K 2,3 is the smallest
forbidden subgraph.

1.3.25. Every cycle of length 2r in a hypercube belongs to a subcube of

dimension at most r , uniquely if r 3. Let C be a cycle of length 2r in
Q k ; V (C) is a collection of binary vectors of length k . Let S be the set of
coordinates that change at some step while traversing the vectors in V (C).
In order to return to the first vector, each position must flip between 0 and
1 an even number of times. Thus traversing C changes each coordinate in
S at least twice, but only one coordinate changes with each edge. Hence
2 |S| 2r , or |S| r . Outside the coordinates of S , the vectors of V (C) all
agree. Hence V (C) is contained in a |S|-dimensional subcube.
As argued above, at most two coordinates vary among the vertices of
a 4-cycle; at least two coordinates vary, because otherwise there are not
enough vectors available to have four distinct vertices. By the same reasoning, exactly three three coordinates vary among the vertices of any
6-cycle; we cannot find six vertices in a 2-dimensional subcube. Thus the
r -dimensional subcube containing a particular cycle is unique when r 3.
Some 8-cycles are contained in 3-dimensional subcubes, such as 000x ,
001x , 011x , 010x , 110x , 111x , 101x , 100x , where x is a fixed vector of
length n 3. Such an 8-cycle is contained in n 3 4-dimensional subcubes,
obtained by letting some position in x vary.
1.3.26. A 3-dimensional
! " cube contains 16 6-cycles, and the k -dimensional
cube Q k contains 16 3k 2k3 6-cycles. If we show that every 6-cycle appears
in exactly one 3-dimensional subcube, then multiplying the number of 3dimensional subcubes by the number of 6-cycles in each subcube counts
each 6-cycle exactly once.

45

Chapter 1: Fundamental Concepts

For any set S of vertices not contained in a 3-dimensional subcube,

there must be four coordinates in the corresponding k -tuples that are not
constant within S . A cycle through S makes changes in four coordinates.
Completing the cycle requires returning to the original vertex, so any coordinate that changes must change back. Hence at least eight changes
are needed, and each edge changes exactly one coordinate. The cycle has
length at least 8; hence 6-cycles are contained in 3-dimensional subcubes.
Furthermore, there are only four vertices possible when k 2 coordinates are fixed, so every 6-cycle involves changes in three coordinates.
Hence the only 3-dimensional subcube containing the 6-cycle is the one
that varies in the same three coordinates
as the 6-cycle.
!"
By Example 1.3.8, there are 3k 2k3 3-dimensional subcubes, so it remains only to show that Q 3 has 16 cycles of length 6. We group them by
the two omitted vertices. The two omitted vertices may differ in 1, 2, or
3 coordinates. If they differ in one place (they are adjacent), then deleting them leaves a 6-cycle plus one edge joining a pair of opposite vertices.
Since Q 3 has 12 edges, there are 12 6-cycles of this type. Deleting two complementary vertices (differing in every coordinate) leaves only a 6-cycle.
Since Q 3 has four such pairs, there are four such 6-cycles. The remaining pairs differ in two positions. Deleting such a pair leaves a 4-cycle plus
two pendant edges, containing no 6-cycle. This considers all choices for the
omitted vertices, so the number of 6-cycles in Q 3 is 12 + 4.
1.3.27. Properties of the middle-levels graph. Let G be the subgraph of
Q 2k+1 induced by vertices in which the numbers of 1s and 0s differs by 1.
These are the (2k + 1)-tuples of weight k and weight k + 1, where weight
denotes the number of 1s.
Each vertex of weight k has k + 1 neighbors of weight k + 1, and!each"
vertex of weight k + 1 has k + 1 neighbors of weight k . There are 2k+1
k
vertices of each weight. Counting edges by the Degree-Sum Formula,
!
"
! "
e(G) = (k + 1) n(G)
= (k + 1) 2k+1
= (2k + 1) 2kk .
2
k+1

The graph is bipartite and has no odd cycle. The 1s in two vertices of
weight k must be covered by the 1s of any common neighbor of weight k + 1.
Since the union of distinct k -sets has size at least k + 1, there can only be
one common neighbor, and hence G has no 4-cycle. On the other hand, G
does have a 6-cycle. Given any arbitary fixed vector of weight k 1 for the
last 2k 2 positions, we can form a cycle of length six by using 110, 100,
101, 001, 011, 010 successively in the first three positions.

1.3.28. Alternative description of even-dimensional hypercubes. The simple graph Q "k has vertex set {0, 1}k , with u v if and only if u and v agree

Section 1.3: Vertex Degrees and Counting

46

in exactly one coordinate. Let the odd vertices be the vertices whose name
has an odd number of 1s; the rest are even vertices.
When k is even, Q "k
= Q k . To show this, rename all odd vertices by
changing 1s into 0s and 0s into 1s. Since k is even, the resulting labels
are still odd. Since k is even, every edge in Q "k joins an even vertex to
an odd vertex. Under the new naming, it joins the even vertex to an odd
vertex that differs from it in one coordinate. Hence the adjacency relation
becomes precisely the adjacency relation of Q k .
$ Q k , because Q "k contains an odd cycle and hence is
When k is odd, Q "k
=
not bipartite. Starting from one vertex, form a closed walk by successively
following k edges where each coordinate is the coordinate of agreement
along exactly one of these edges. Hence each coordinate changes exactly
k 1 times and therefore ends with the value it had at the start. Thus this
is a closed walk of odd length and contains an odd cycle.
1.3.29. Automorphisms of Q k .
a) A subgraph H of Q k is isomorphic to Q l if and only if it is the subgraph induced by a set of vertices agreeing in some set of k l coordinates.
Let f be an isomorphism from H to Q l , and let v be the vertex mapped to
the vertex 0 of Q l whose coordinates are all 0. Let u 1 , . . . , u l be the neighbors of v in H mapped to neighbors of 0 in Q l by f . Each u i differs from v
in one coordinate; let S be the set of l coordinates where these vertices differ from v . It suffices to show that vertices of H differ from v only on the
coordinates of S . This is immediate for l 1.
For l 2, we prove that each vertex mapped by f to a vertex of Q l
having weight j differs from v in j positions of S , by induction on j . Let
x be a vertex mapped to a vertex of weight j in Q l . For j 1, we have
already argued that x differs from v in j positions of S . For j 2, let y
and z be two neighbors of x whose images under f have weight j 1 in
Q l . By the induction hypothesis, y and z differ from v in j positions of S .
Since f (y) and f (z) differ in two places, they have two common neighbors
in Q l , which are x and another vertex w . Since w has weight j 2, the
induction hypothesis yields that w differs from v in j 1 positions of S .
Since the images of x, y, z, w induce a 4-cycle in Q l , also x, y, z, w induce
a 4-cycle in H . The only 4-cycle in Q k that contains all of y, z, w adds the
vertex that differs from v in the j 2 positions of S where w differs, plus
the two positions where y and z differ from w . This completes the proof
that x has the desired property.
b) The k -dimensional cube Q k has exactly 2k k ! automorphisms. (Part
(a) is unnecessary.) Form automorphisms of Q k by choosing a subset of
the k coordinates in which to complement 0 and 1 and, independently, a
permutation of the k coordinates. There are 2k k ! such automorphisms.

47

Chapter 1: Fundamental Concepts

We prove that every automorphism has this form. Let 0 be the all-0
vertex. Let f be the inverse of an automorphism, and let v be the vertex
mapped to 0 by f . The neighbors of v must be mapped to the neighbors of
0. If these choices completely determine f , then f complements the coordinates where v is nonzero, and the correspondence between the neighbors
of 0 and the neighbors of v determines the permutation of the coordinates
that expresses f as one of the maps listed above.
Suppose that x differs from v in coordinates r 1 , . . . , r j . Let u 1 , . . . , u j be
the neighbors of v differing from v in these coordinates. We prove that f (x)
is the k -tuple of weight j having 1 in the coordinates where f (u 1 ), . . . , f (u j )
have 1. We use induction on j .
For j 1, the claim follows by the definition of u 1 , . . . , u j . For j 2,
let y and z be two neighbors of x that differ from v in j 1 coordinates.
Let w be the common neighbor of y and z that differs from v in j 2 coordinates. By the induction hypothesis, f (y) and f (z) have weight j 1 (in
the appropriate positions), and f (w) has weight j 1. Since f (x) must be
the other common neighbor of f (y) and f (z), it has weight j , with 1s in the
desired positions.
1.3.30. The Petersen graph has twelve 5-cycles. Let G be the Petersen
graph. We show first that each edge of G appears in exactly four 5-cycles.
For each edge e = x y in G , there are two other edges incident to x and two
others incident to y . Since G has no 3-cycles, we can thus extend x y at both
ends to form a 4-vertex path in four ways. Since G has no 4-cycle, the endpoints of each such path are nonadjacent. By Proposition 1.1.38, there is
exactly one vertex to add to such a path to complete a 5-cycle. Thus e is in
exactly four 5-cycles.
When we sum this count over the 15 edges of G , we have counted 60
5-cycles. However, each 5-cycle has been counted five timesonce for each
of its edges. Thus the total number of 5-cycles in G is 60/5 = 12.

1.3.31. Combinatorial !proofs

graphs. ! "
"
!kwith
"
n
. Consider the complete
a) For 0 k n , 2 = 2 + k(n k) + nk
2
!"
graph K n , which has 2n edges. If we partition the vertices of K n into a k set and an (n k)-set, then we can count the edges as those within one

Section 1.3: Vertex Degrees and Counting

48

block of the partition and

a vertex from each. Hence the
! "those
! choosing
"
total number of edges is 2k + nk
+
k(n

k).
#
# !ni " 2 !n "
n i = n , then
2 . Again consider the edges of K n , and
b) If
2
partition the vertices into sets with n i being the size of the i th set. The left
side of the inequality counts the edges in K n having both ends in the same
Si , which is at most all of E(K n ).
n1
1.3.32. For n 1, there are 2( 2 ) simple even graphs with a fixed vertex set of size n . Let A be the set of simple even graphs with vertex set
n1
v1 , . . . , vn . Since 2( 2 ) is the size of the set B of simple graphs with vertex
set v1 , . . . , vn1 , we establish a bijection from A to B .
Given a graph in A, we obtain a graph in B by deleting vn . To show
that each graph in B arises exactly once, consider a graph G B . We form
a new graph G " by adding a vertex vn and making it adjacent to each vertex
with odd degree in G , as illustrated below.
The vertices with odd degree in G have even degree in G " . Also, vn
itself has even degree because the number of vertices of odd degree in G
is even. Thus G " A. Furthermore, G is the graph obtained from G " by
deleting vn , and every simple even graph in which deleting vn yields G must
have vn adjacent to the same vertices as in G " .
Since there is a bijection from A to B , the two sets have the same size.

vn

G"

1.3.33. Triangle-free graphs in which every two nonadjacent vertices have

exactly two common
! "neighbors.
n(G) = 1 + k+1
, where k is the degree of a vertex x in G . For every
2
pair of neighbors of x , there is exactly one nonneighbor of x that they have
as a common neighbor. Conversely, every nonneighbor of x has exactly one
pair of neighbors of x in its neighborhood, because these are its common
neighbors with x . This establishes a bijective correspondence between the
pairs in N (x) and the!nonneighbors
"
! " of x . Counting x , N (x), and N (x), we
have n(G) = 1 + k + 2k = 1 + k+1
. Since this argument holds for every
2
x V (G), we conclude that G is k -regular.
Comment: Such graphs exist only for isolated values of k . Unique
graphs !exist
" for k = 1, 2, 5. Viewing the vertices as x , N (x) = [k ], and
N (x) = [k]
, we have i adjacent to the pair { j, k} if and only if i { j, k}. The
2
! "
lack of triangles guarantees that only disjoint pairs in [k]
can be adjacent,
2

49

Chapter 1: Fundamental Concepts

! "
! "
but each pair in [k]
must have exactly k 2 neighbors in [k]
. For k =
2
2
! "
5, this implies that N (x) induces the 3-regular disjointness graph of [5]
,
2
which is the Petersen graph. Since the Petersen graph has girth 5 and
diameter 2, each intersecting pair has exactly one common neighbor in
N (x) in addition to its one common neighbor in N (x), so this graph has the
desired properties.
Numerical conditions eliminate k 3 (mod 4), because G would be
regular of odd degree with an odd number of vertices. There are stronger
necessary conditions. After k = 5, the next possibility is k = 10, then 26,
37, 82, etc. A realization for k = 10 is known to exist, but in general the
set of realizable values is not known.
1.3.34. If G is a kite-free simple n -vertex graph such that every pair of
nonadjacent vertices has exactly two common neighbors, then G is regular.
Since nonadjacent vertices have common neighbors, G is connected. Hence
it suffices to prove that adjacent vertices x and y have the same degree. To
prove this, we establish a bijection from A to B , where A = N (x) N (y)
and B = N (y) N (x).
Consider u A. Since u ! y , there exists v N (u) N (y) with v %= x .
Since G is kite-free, v ! x , so v B . Since x and v have common neighbors
y and u , the vertex v cannot be generated in this way from another vertex
of A. Hence we have defined an injection from A to B . Interchanging the
roles of y and x yields an injection from B to A. Since these sets are finite,
the injections are bijections, and d(x) = d(y).

1.3.35. If every induced k -vertex subgraph of a simple n -vertex graph G has

the same number of edges, where 1 < k < n 1, then G is a complete graph
or an empty graph.
a) If l k and G ! is a graph on l vertices
which
every induced k ! " in
! l2
"
vertex subgraph has m edges, then e(G ! ) = m kl / k2
. Counting the edges
!"
in all the k -vertex subgraphs of G ! yields m kl , but each edge appears in
! l2 "
of these subgraphs, once for each k -set of vertices containing it. (Both
k2
! l2 "
!"
sides of k2
e(G ! ) = m kl count the ways to pick an edge of G ! and a k -set
of vertices in G ! containing that edge. On the right, we pick the set first;
on the left, we pick the edge first.)
b) Under the stated conditions, G = K n or G = K n . Given vertices u
and v , let A and B be the sets of edges incident to u and v , respectively. The
set of edges with endpoints u and v is A B . We compute
#
#
#
#
# # # # #
#
|A B| = e(G) # A B # = e(G) # A B # = e(G) # A# # B # + # A B # .
In this formula, A and B are the edge sets of induced subgraphs of order
n 1, and A B is the edge set of an induced subgraph of order n 2. By
part (a), the sizes of these sets do not depend on the choice of u and v .

Section 1.3: Vertex Degrees and Counting

50

1.3.36. The unique reconstruction of the graph with vertex-deleted subgraphs below is the kite.
Proof 1. A vertex added to the first triangle may be joined to 0,1,2,
or 3 of its vertices. We eliminate 0 and 1 because no vertex-deleted subgraph has an isolated vertex. We eliminate 3 because every vertex-deleted
subgraph of K 4 is a triangle. Joining it to 2 yields the kite.

Proof 2. The graph G must have four vertices, and by Proposition

1.3.11 it has five edges. The only such simple graph is the kite.
1.3.37. Retrieving a regular graph. Suppose that H is a graph formed
by deleting a vertex from a regular graph G . We have H , so we know
n(G) = n(H ) + 1, but we dont know the vertex degrees in G . If G is d regular, then G has dn(G)/2 edges, and H has dn(G)/2 d edges. Thus
d = 2e(H )/(n(G) 2). Having determined d , we add one vertex w to H and
add d d H (v) edges from w to v for each v V (H ).
1.3.38. A graph with at least 3 vertices is connected if and only if at least
two of the subgraphs obtained by deleting one vertex are connected. The
endpoints of a maximal path are not cut-vertices. If G is connected, then
the subgraphs obtained by deleted such vertices are connected, and there
are at least of these.
Conversely, suppose that at least two vertex-deleted subgraphs are
connected. If G v is connected, then G is connected unless v is an isolated
vertex. If v is an isolated vertex, then all the other subgraphs obtained by
deleting one vertex are disconnected. Hence v cannot be isolated, and G is
connected.
1.3.39. Disconnected graphs are reconstructible. First we show that G is
connected if and only if it has at least two connected vertex-deleted subgraphs. Necessity holds, because the endpoints of a maximal path cannot
be cut-vertices. If G is disconnected, then G v is disconnected unless v is
an isolated vertex (degree 0) in G and G v is connected. This happens for
at most one vertex in G .
After determining that G is disconnected, we obtain which disconnected graph it is from its vertex-deleted subgraphs. We aim to identify
a connected graph M that is a component of G and a vds in the deck that
arises by deleting a specified vertex u of M . Replacing M u by M in that
subgraph will reconstruct G .

51

Chapter 1: Fundamental Concepts

Among all components of all graphs in the deck, let M be one with
maximum order. Since every component H of a potential reconstruction
G appears as a component of some G v , M cannot belong to any larger
component of G . Hence M is a component of G . Let L be a fixed connected
subgraph of M obtained by deleting a leaf u of some spanning tree of M .
Then L is a component of G u . We want to reconstruct G by substituting
M for L in G u ; we must identify G u . There may be several isomorphic
copies of G u .
As in the disconnected graph G shown above, M may appear as a component of every vds G v . However, since M cannot be created by a vertex
deletion, a vds with the fewest copies of M must arise by deleting a vertex
of M . Among these, we seek a subgraph with the most copies of L as components, because in addition to occurrences of L as a component of G , we
obtain an additional copy if and only if the deleted vertex of M can play
the role of u . This identifies G u , and we obtain G by replacing one of its
components isomorphic to L with a component isomorphic to M .

1.3.40. Largest graphs of specified types.

a) Largest n -vertex simple graph with an independent set of size a .
Proof 1. Since! there
set, such a
" ! "are no edges within the
! independent
"
graph has at most 2n 2a edges, which equals na
+
(n

a)a
.
This
bound
2
is achieved by the graph consisting of a copy H of K na , an independent set
S of size a , and edges joining each vertex of H to each vertex of S .
Proof 2. Each vertex of an independent set of size a#
has degree at most
n a . Each other vertex has degree at most n 1. Thus d(v) a(n a) +
(n a)(n 1). By the Degree-Sum Formula, e(G) (n a)(n 1 + a)/2.
This formula equals those above and is achieved by the same graph, since
this graph achieves the bound for each vertex degree.
b)" The maximum size of an n -vertex simple graph with k components is
!nk+1
. The graph consisting of K nk+1 plus k 1 isolated vertices has k
2
!
"
components and nk+1
edges. We prove that other n -vertex graphs with k
2
components dont have maximum size. Let G be such a graph.
If G has a component that is not complete, then adding edges to make
it complete does not change the number of components. Hence we may
assume that every component is complete.
If G has components with r and s vertices, where r s > 1, then we
move one vertex from the s -clique to the r -clique. This deletes s 1 edges

Section 1.3: Vertex Degrees and Counting

52

and creates r edges, all incident to the moved vertex. The other edges
remain the same, so we gain r s + 1 edges, which is positive.
Thus the number of edges is maximized only when every component is
a complete graph and only one component has more than one vertex.
c) The
number of edges in a disconnected simple n -vertex
! maximum
"
graph is n1
,
with
equality
only for K 1 + K n1 .
2
Proof
1
(using
part
(b)).
The maximum over graphs with k compo!
"
nents is nk+1
,
which
decreases
as k increases. For disconnected graphs,
2
! "
k 2. We maximize the number of edges when k = 2, obtaining n1
.
2
Proof 2 (direct argument). Given a disconnected simple graph G , let
S be the vertex set of! one
" component of G , and let t = |S|. Since no edges
join S and S , e(G) 2n t (n t). This bound is weakest when t (n t) is
smallest,
! "which for 1 t! n" 1 happens when t {1, n 1}. Thus always
e(G) 2n 1(n 1) = n1
, and equality holds when G = K 1 + K n1 .
2
Proof
3
(induction
on
n
). When n = 2, the only simple graph !with
!"
"
e(G) > 12 = 1 is K 2 , which is connected. For n > 2, suppose e(G) > n1
.
2
If !(G) = n 1, then G is connected.
Otherwise,
we
may
select
v
with
! "
! "
d(v) n 2. Then e(G v) > n1
n + 2 = n2
. By the induction
2
!n1" 2
hypothesis, G v is connected. Since e(G) > 2 and G is simple, we have
d(v) > 0, so there is an edge from v to G v , and G is also connected.
then G is connected,
Proof 4 (complementation).
! " If G is disconnected,
! "
so e(G) n 1 and e(G) 2n (n 1) = n1
.
In
fact,
G must contain
2
a spanning complete bipartite subgraph, which is as small as n 1 edges
only when G = K 1,n1 and G = K 1 + K n1 .

1.3.41. Every n -vertex simple graph with maximum degree %n/2& and minimum degree 'n/2( 1 is connected. Let x be a vertex of maximum degree. It
suffices to show that every vertex not adjacent to x has a common neighbor
with x . Choose y
/ N (x). We have |N (x)| = %n/2& and |N (y)| 'n/2( 1.
Since y ! x , we have N (x), N (y) V (G) {x, y}. Thus

|N (x) N (y)| = |N (x)|+|N (y)||N (x) N (y)| %n/2&+'n/2(1(n2) = 1.

1.3.42. Strongly independent sets. If S is an independent set with no common neighbors in a graph G , then the vertices of S have pairwise-disjoint
closed neighborhoods of size at least (G) + 1. Thus there are at most
'n(G)/((G) + 1)( of them. Equality is achievable for the 3-dimensional
cube using S = {000, 111}.
Equality is not achievable when G = Q 4 , since with 16 vertices and
minimum degree 4 it requires three parwise-disjoint closed neighborhoods
of size 5. If v S , then no vertex differing from v in at most two places is
in S . Also, at most one vertex differing from v in at least three places is in

53

Chapter 1: Fundamental Concepts

S , since such vertices differ from each other in at most two places. Thus
only two disjoint closed neighborhoods can be found in Q 4 .
1.3.43. Every simple graph has a vertex whose neighbors have average degree as large as the overall average
! degree. Let t (w)
! be the!average degree of
the neighbors of w . In the sum wV (G) t (w) = wV (G) yN (w) d(y)/d(w),
we have the terms d(u)/d(v) and d(v)/d(u) for each edge uv . Since
x/y + y/x 2 whenever x, y are positive real numbers (this is!equivalent
y)2 0), each such contribution is at least 2. Hence
t (w)
! to (x d(u)
d(v)
+

2
e(G)
.
Hence
the
average
of
the
neighborhood
avuvE(G) d(v)
d(u)
erage degrees is at least the average degree, and the pigeonhole principle
yields the desired vertex.
It is possible that every average neighborhood degree exceeds the average degree. Let G be the graph with 2n vertices formed by adding a
between a complete graph and an independent set. Since G has
"matching
#
n
+
n
edges
and 2n vertices, G has average degree (n + 1)/2. For each
2
vertex of the n -clique, the neighborhood average degree is n 1 + 1/n . For
each leaf, the neighborhood average degree is n .
1.3.44. Subgraphs with large minimum degree. Let G be a loopless graph
with average degree a .
a) If x V (G), then G " = G x has average degree at least a if and only
if d(x) a/2. Let a " be the average degree of G " , and let n be the order of
G . Deleting x reduces the degree sum by 2d(x), so (n 1)a " = na 2d(x).
Hence (n 1)(a " a) = a 2d(x). For n > 1, this implies that a " a if and
only if d(x) a/2.
Alternative presentation. The average degree of G is 2e(G)/n(G). Since
G " has e(G) d(x) edges, the average degree is at least a if and only if
2[e(G)d(x)]
a . Since e(G) = n(G)a/2, we can rewrite this as n(G)a
n(G)1
2d(x) = 2e(G) 2d(x) an(G) a . By canceling n(G)a , we find that the
original inequality is equivalent to d(x) a/2.
b) If a > 0, then G has a subgraph with minimum degree greater than
a/2. Iteratively delete vertices with degree at most half the current average degree, until no such vertex exists. By part (a), the average degree
never decreases. Since G is finite, the procedure must terminate. It ends
only by finding a subgraph where every vertex has degree greater than a/2.
c) The result of part (b) is best possible. To prove that no fraction of
a larger than 12 a can be guaranteed, let G n be an n -vertex tree. We have
a(G n ) = 2(n 1)/n = 2 2/n , but subgraphs of G n have minimum degree
at most 1. Given > 12 , we can choose n large enough so that 1 a(G n ).
1.3.45. Bipartite subgraphs of the Petersen graph.
a) Every edge of the Petersen graph is in four 5-cycles. In every 5-cycle
through an edge e, the edge e is the middle edge of a 4-vertex path. Such

Section 1.3: Vertex Degrees and Counting

54

a path can be obtained in four ways, since each edge extends two ways
at each endpoint. The neighbors at each endpoint of e are distinct and
nonadjacent, since the girth is 5.
Since the endpoints of each such P4 are nonadjacent, they have exactly one common neighbor. Thus each P4 yields one 5-cycle, and each
5-cycle through e arises from such a P4 , so there are exactly four 5-cycles
containing each edge.
b) The Petersen graph has twelve 5-cycles. Since there are 15 edges,
summing the number of 5-cycles through each edge yields 60. Since each
5-cycle is counted five times in this total, the number of 5-cycles is 12.
c) The largest bipartite subgraph has twelve edges.
Proof 1 (breaking odd cycles). Each edge is in four 5-cycles, so we
must delete at least 12/4 edges to break all 5-cycles. Hence we must delete
at least three edges to have a bipartite subgraph. The illustration shows
that deleting three is enough; the Petersen graph has a bipartite subgraph
with 12 edges (see also the cover of the text).

A
B
B

A
A
B B
A
A
Proof 2 (study of bipartite subgraphs). The Petersen graph G has
an independent set of size 4, consisting of the vertices {ab, ac, ad, ae} in
the structural description. The 12 edges from these four vertices go to the
other six vertices, so this is a bipartite subgraph with 12 edges.
Let X and Y be the partite sets of a bipartite subgraph H . If |X |
4, then e(H ) 12, with equality only when X is an independent 4-set
in G . Hence we need only consider the case |X | = |Y | = 5. To obtain
e(G) > 10, some vertex x X must have three neighbors in Y . The two
nonneighbors of x in Y have common neighbors with x , and these must lie
in N (x), which is contained in Y . Hence e(G [Y ]) 2. Interchanging X and
Y in the argument shows that also e(G [ X ]) 2. Hence e(H ) 11.
1.3.46. When the algorithm of Theorem 1.4.2 is applied to a bipartite graph,
it need not find the bipartite subgraph with the most edges. For the bipartite graph below, the algorithm may reach the partition between the upper
vertices and lower vertices.

55

Chapter 1: Fundamental Concepts

This bipartite subgraph with eight edges has more than half of the
edges at each vertex, and no further changes are made. However, the
bipartite subgraph with the most edges is the full graph.
1.3.47. Every nontrivial loopless graph G has a bipartite subgraph containing more than half its edges. We use induction on n(G). If n(G) = 2,
then G consists of copies of a single edge and is bipartite. For n(G) > 2,
choose v V (G) that is not incident to all of E(G) (at most two vertices can
be incident to all of E(G)). Thus e(G v) > 0. By the induction hypothesis,
G v has a bipartite subgraph H containing more than e(G)/2 edges.
Let X, Y be a bipartition of H . If X contains at least half of N G (v),
then add v to Y ; otherwise add v to X . The augmented partition captures
a bipartite subgraph of G having more than half of E(G v) and at least
half of the remaining edges, so it has more than half of E(G).
Comment. The statement can also be proved without induction. By
Theorem 1.3.19, G has a bipartite subgraph H with at least e(G)/2 edges.
By the proof of Theorem 1.3.19, equality holds only if d H (v) = dG (v)/2 for
every v V (G). Given an edge uv , each of u and v has exactly half its neighbors in its own partite set. Switching both to the opposite set will capture
those edges while retaining the edge uv , so the new bipartite subgraph has
more edges.
1.3.48. No fraction of the edges larger than 1/2 can be guaranteed for the
bipartite subgraph. If G n is the complete graph K 2n , then e(G n ) =
!largest
"
2n
=
n(
2n 1), and the largest bipartite subgraph is K n,n , which has n 2
2
2
edges. Hence limn f (G n )/e(G n ) = limn 2nn2 n = 12 . For large enough
n , the fraction of the edges in the largest bipartite subgraph is arbitrarily
close to 1/2. (In fact, in every graph the largest bipartite subgraph has
more than half the edges.)
1.3.49. Every loopless graph G has a spanning k -partite subgraph H such
that e(H ) (1 1/k)e(G).
Proof 1 (local change). Begin with an arbitrary partition of V (G) into
k parts V1 , . . . , Vk , and consider the k -partite subgraph H containing all
edges of G consisting of two vertices from distinct parts. Given a partition
of V (G), let V (x) denote the part containing x . If in G some vertex x has
more neighbors in Vj than in some other part, then shifting x to the other
part increases the number of edges captured by the k -partite subgraph.
Since G has finitely many edges, this shifting process must terminate.
It terminates when for each #
x V (G) the number |N (x) Vi | is minimized
by Vi = V (x). Then dG (x) = i |NG (x) Vi | k|N G (x) V (x)|. We conclude
that |N G (x) V (x)| (1/k)dG (x), and hence d H (x) (1 1/k)dG (x) for all
x V (G). By the degree-sum formula, e(H ) (1 1/k)e(G).

Section 1.3: Vertex Degrees and Counting

56

Proof 2 (induction on n ). We prove that when G is nontrivial, some

such H has more than (1 1/k)e(G) edges. This is true when n = 2.
We procede by induction for n > 2. Choose a vertex v V (G). By the
induction hypothesis, G v has a spanning k -partite subgraph with more
than (1 1/k)e(G v) edges. This subgraph partitions V (G v) into k
partite sets. One of these sets contains at most 1/k neighbors of v . Add
v to that set to obtain the desired k -partite subgraph H . Now e(H ) >
(1 1/k)e(G v) + (1 1/k)dG (v) = (1 1/k)e(G).

1.3.50. For n 3, the minimum number of edges in a connected n -vertex

graph in which every edge belongs to a triangle is &3(n 1)/2'. To achieve
the minimum, we need only consider simple graphs. Say that connected
graphs with each edge in a triangle are good graphs. For n = 3, the only
such graph is K 3 , with three edges.
When n is odd, a construction with the claimed size consists of (n 1)/2
triangles sharing a common vertex. When n is even, add one vertex to the
construction for n 1 and make it adjacent to both endpoints of one edge.
For the lower bound, let G be a smallest n -vertex good graph. Since G
has fewer than 3n/2 edges (by the construction), G has a vertex v of degree
2. Let x and y be its neighbors. Since each edge belongs to a triangle, x
y . If n > 3, then we form G $ by deleting v and, if x y have no other neighbor,
contracting x y . Every edge of G $ belongs to a triangle that contained it
in G . The change reduces the number of vertices by 1 or 2 and reduces
the number of edges by at least 3/2 times the $reduction in the
% number of
vertices. By the induction hypothesis, e(G $ ) 3(n(G $ ) 1)/2 , and hence
the desired bound holds for G .

1.3.51. Let G be
# a simple n -vertex graph.
e(Gv)
(G)
. If we count up all the edges in all the subgraphs
a) e(G) = vV n2
obtained by deleting one vertex, then each edge of G is counted exactly n 2
times, because it shows up in the n 2 subgraphs obtained by deleting a
vertex other than its endpoints.
b) If n 4 and G has more than n 2 /4 edges, then G has a vertex whose
deletion leaves a graph with more than (n 1)2 /4 edges. Since G has more
than n 2 /4 edges and e(G) is an integer, we have e(G) (n 2 + 4)/4 when n
is even and e(G) (n 2 + 3)/4 when n is odd (since (2k + 1)2 = 4k 2 + 4k + 1,
every square of an odd number is one more than a multiple of 4). Thus
2
always we have e(G) (n#
+ 3)/4.
(n 2 + 3)/4. In the sum we have n
By part (a), we have vV (G) e(Gv)
n2
terms. Since the largest number in a set is at least the average, there is a
2
vertex v such that e(Gv)
1n n 4+3 . We rewrite this as
n2

e(G v)

(n 2 + 3)(n 2)
n 3 2n 2 + 3n 6
n 2 2n + 1 2n 6
+
=
=
4n
4n
4
4n

57

Chapter 1: Fundamental Concepts

When n 4, the last term is positive, and we obtain the strict inequality
e(G v) > (n 1)2 /4.
c) Inductive proof that G contains a triangle if e(G) > n 2 /4. We use
induction on n . When n 3, they only simple graph with more than n 2 /4
edges is when n = 3 and G = K 3 , which indeed contains a triangle. For
the induction step, consider n 4, and let G be an n -vertex simple graph
with more than n 2 /4 vertices. By part (b), G has a subgraph G v with
n 1 vertices and more than (n 1)2 /4 edges. By the induction hypothesis,
G v therefore contains a triangle. This triangle appears also in G .

1.3.52. K !n/2",#n/2$ is the only n -vertex triangle-free graph of maximum size.

As in the proof of Mantels result, let x be a vertex of maximum degree.
Since N (x) is an independent set, x and its non-neighbors capture all the
edges, and we have e(G) (n !(G))!(G). If equality holds, then summing the degrees in V (G) N (x) counts each edge exactly once. This
requires that V (G) N (x) also is an independent set, and hence G is bipartite. If G is bipartite and has (n !(G))!(G) edges, then G = K (n!(G)),!(G) .
Hence e(G) is maximized only by K !n/2",#n/2$ .

1.3.53. The bridge club with 14 members (no game can be played if two of
the four people table have previously been partners): If each member has
played with four others and then six additional games have been played,
then the arrival of a new member allows a game to be played. We show that
the new player yields a set of four people among which no two have been
partners. This is true if and only if the previous games must leave three
people (in the original 14) among which no two have been partners.
The graph of pairs who have NOT been partners initially is K 14 . For
each game played, two edges are lost from this graph. At the breakpoint in
the session, each vertex has lost four incident edges, so 28 edges have been
deleted. In the remaining six games, 12 more edges are deleted. Hence
40 edges have been deleted. Since e(K 14 ) = 91, there remain 51 edges for
pairs that have not yet been partners.
By Mantels Theorem (Theorem 1.3.23), the maximum
!
" number of edges
in a simple 14-vertex graph with no triangle is 142 /4 . Since 51 > 49,
the graph of remaining edges has a triangle. Thus, when the 15th person
arrives, there will be four people of whom none have partnered each other.

1.3.54. The minimum number of triangles t (G) in an n -vertex graph G and

its complement.
#$
# $
%
. Let d1 , . . . , dn denote the vertex
a) t (G) = 3n (n 2)e + vV (G) d(v)
2
degrees. We prove that the right side of the formula assigns weight 1 to
the vertex triples that induce a triangle
in G or G and weight 0 to all
#$
other triples. Among these terms, 3n counts all triples, (n 2)e counts
% #di $
those determined by an edge of G and a vertex off that edge, and
2

Section 1.3: Vertex Degrees and Counting

58

counts 1 for each pair of incident edges. In the table below, we group these
contributions by how many edges the corresponding triple induces in G .
#n $
% #di $
t (G)
in G
(n 2)e
3
2
1
0
0
1

3 edges
2 edges
1 edge
0 edges

1
1
1
1

3
2
1
0

3
1
0
0

b) t (G) n(n 1)(n 5)/24. Begin with the formula for k 3 (G) + k3 (G)
from part (a). Using the convexity of quadratic functions, we get a lower
bound for the sum on the right by# replacing
the vertex
$
# $degrees by the average degree 2e/n . The bound is 3n (n 2)e + n 2e/n
, which reduces to
2
#n $
#n $
#$
2e( 2 e)/n . As a function of e, this is minimized when e = 12 2n . This
3
substitution and algebraic simplification produce t (G) n(n 1)(n 5)/24.
Comment. The proof of part (b) uses two minimizations. These imply
that equality
can hold only for a regular graph (di = 2e/n for all i ) with
#$
e = 12 2n . There is such a regular graph if and only if n is odd and (n 1)/2
is even. Thus we need n = 4k + 1 and G is 2k -regular.
1.3.55. Maximum size with no induced P4 . a) If G is a simple connected
graph and G is disconnected, then e(G) !(G)2 , with equality only for
K !(G),!(G) . Since G is disconnected,
!(G) n(G)/2, with equality only if
%
G = K !(G),!(G) . Thus e(G) =
di /2 n(G)!(G)/2 !(G)2 . As observed,
equality when G is disconnected requires G = K !(G),!(G) .
b) If G is a simple connected graph with maximum degree D and no
induced subgraph isomorphic to P4 , then e(G) D 2 . It suffices by part (a)
to prove that G is disconnected when G is connected and P4 -free. We use
induction on n(G) for n(G) 2; it is immediate when n(G) = 2. For the
induction step, let v be a non-cut-vertex of G . The graph G ' = G v is also
P4 -free, so its complement is disconnected, by the induction hypothesis.
Thus V (G) v has a vertex partition X, Y such that all of X is adjacent to
all of Y in G . Since G is connected, v has a neighbor z X Y ; we may
assume be symmetry that z Y . If G is connected, then G has a v, z -path.
Let y be the vertex before z on this path; note that y Y . Also G connected
requires x X such that vx E(G). Now {v, z, x, y} induces P4 in G .
%
1.3.56. Inductive proof that for
di even there is a multigraph with vertex
degrees d1 , . . . , dn .
%
%
Proof 1 (induction on
di ). If
di = 0, then all di are 0, and the
n -vertex %
graph with no edges has degree list d . For the induction step,
suppose
di > 0. If only one di is nonzero, then it must be even, and the

59

Chapter 1: Fundamental Concepts

graph consisting of n 1 isolated vertices plus di /2 loops at one vertex has

degree list d (multigraphs allow loops).
Otherwise, d has at least two nonzero entries, di and d j . Replacing
these with di 1 and d j 1 yeilds a list d ! with smaller even sum. By
the induction hypothesis, some graph G ! with degree list d ! . Form G by
adding an edge with endpoints u and v to G ! , where dG ! (u) = di 1 and
dG ! (v) = d j 1. Although u and v may already be adjacent in G ! , the
resulting multigraph G has degree list d .
Proof 2 (induction on n ). For n = 1, put d1 /2 loops at v1 . If dn is even,
put dn /2 loops at vn and apply the induction hypothesis. Otherwise, put an
edge from
! vn to some other vertex corresponding to positive di (which exists
since
di is even) and proceed as before.

1.3.57. An n -tuple of nonnegative integers with largest entry k is graphic if

the sum is even, k < n , and every entry is k or k 1. Let A(n) be the set of n tuples satisfying these conditions. Let B(n) be the set of graphic n -tuples.
We prove by induction on n that n -tuples in A(n) are also in B(n). When
n = 1, the only list in A(n) is (0), and it is graphic.
For the induction step, let d be an n -tuple in A(n), and let k be its
largest element. Form d ! from d by deleting a copy of k and subtracting 1
from k largest remaining elements. The operation is doable because k < n .
To apply the induction hypothesis, we need to prove that d ! A(n 1).
Since we delete an instance of k and subtract one from k other values, we
reduce the sum by 2k to obtain d ! from d , so d ! does have even sum.
Let q be the number of copies of k in d . If q > k + 1, then d ! has k s and
(k 1)s. If q = k + 1, then d ! has only (k 1)s. If q < k + 1, then d ! has
(k 1)s and (k 2)s. Also, if k = n 1, then the first possibility cannot
occur. Thus d ! has length n 1, its largest value is less than n 1, and its
largest and smallest values differ by at most 1. Thus d ! A(n 1), and we
can apply the induction hypothesis to d ! .
The induction hypothesis (d ! A(n 1)) (d ! B(n 1)) tells us that
!
d is graphic. Now the Havel-Hakimi Theorem implies that d is graphic.
(Actually, we use only the easy part of the HH Theorem, adding a vertex
joined to vertices with desired degrees.)
1.3.58. If d is a nonincreasing list of nonnegative integers, and d ! is obtained by deleting dk and subtracting 1 from the k largest other elements,
then d is graphic if and only if d ! is graphic. The proof is like that of the
HavelHakimi Theorem. Sufficiency is immediate. For necessity, let w be
a vertex of degree dk in a simple graph with degree sequence d . Alter G by
2-switches to obtain a graph in which w has the dk highest-degree other vertices as neighbors. The argument to find a 2-switch increasing the number
of desired neighbors of w is as in the proof of the HavelHakimi Theorem.

Section 1.3: Vertex Degrees and Counting

60

1.3.59. The list d = (d1 , . . . , d2k ) with d2i = d2i1 = i for 1 i k is graphic.
This is the degree list for the bipartite graph with vertices x 1 , . . . , x k and
y1 , . . . , yk defined by xr ys if and only if r + s > k . Since the neighborhood
of xr is {yk , yk1 , . . . , ykr+1 }, the degree of xr is r . Thus the graph has two
vertices of each degree from 1 to k .
1.3.60. Necessary and sufficient conditions for a list d to be graphic when d
consists of k copies of a and n k copies of b, with a b 0. Since the degree
sum must be even, the quantity ka + (n k)b must be even. In addition,
the inequality ka k(k 1) + (n k) min{k, b} must hold, since each vertex
with degree b has at most min{k, b} incident edges whose other endpoint
has degree a . We construct graphs with the desired degree sequence when
these conditions hold. Note that the inequality implies a n 1.
Case 1: b k and a n k . Begin with K k,nk , having partite sets
X of size k and Y of size n k . If k(a n + k) and (n k)(b k) are even,
then add an (a n + k)-regular graph on X and a (b k)-regular graph on
Y . To show that this is possible, note first that 0 a n + k k 1 and
0 b k a k n k 1. Also, when pq is even, a q -regular graph on
p vertices in a circle can be constructed by making each vertex adjacent to
the 'q/2( nearest vertices in each direction and also to the opposite vertex
if q is odd (since then p is even).
Note that k(a n +k) and (n k)(b k) have the same parity, since their
difference ak (n k)b differs from the given even number ka + (n k)b by
an even amount. If they are both odd, then we delete one edge from K k,nk ,
and now one vertex in the subgraph on X should have degree a n + k + 1
and one in the subgraph on Y should have degree b k + 1. When pq is odd,
such a graph on vertices v0 , . . . , v p1 in a circle (q -regular except for one
vertex of degree q + 1) can be constructed by making each vertex adjacent
to the (q 1)/2 nearest vertices in each direction and then adding the edges
{vi vi+( p1)/2 : 0 i ( p 1)/2. Note that all vertices are incident to one of
the added edges, except that v( p1)/2 is incident to two of them.
Case 2: k 1 a < n k . Begin by placing a complete graph on a set
S of k vertices. These vertices now have degree k 1 and will become the
vertices of degree a , which is okay since a b. Put a set T of n k additional
vertices in a circle. For each vertex in S , add a k + 1 consecutive neighbors
in T , starting the next set immediately after the previous set ends. Since
a n 1, each vertex in S is assigned a k + 1 distinct neighbors in T .
Since k(a k + 1) (n k)b and the edges are distributed nearly equally
to vertices of T , there is room to add these edges.
For the subgraph induced by T , we need a graph with n k vertices
and [(n k)b k(a k + 1)]/2 edges and degrees differing by at most 1.
The desired number of edges is integral, since ka + (n k)b is even, and it

61

Chapter 1: Fundamental Concepts

is nonnegative, since k(a k + 1) (n k)b. The largest degree needed is

#(n k)b k(a k + 1)$ n k . This is at most b, which is less than n k
since b a < n k . The desired graph now exists by Exercise 1.3.57.
Case 3: b < k and a n k . Put the set S of size k in a circle. For
each vertex in the set T of size n k , assign b consecutive neighbors in
S ; these are distinct since b < k . Since a n k , no vertex of S receives
too many edges. On S we put an almost-regular graph with k vertices and
[ak b(n k)]/2 edges. Again, this number of edges is integral, and in
the case specified it is nonnegative. Existence of such a graph requires
a b(n k)/k k 1, which is equivalent to the given inequality k(a k +
1) (n k)b. Now again Exercise 1.3.57 provides the needed graph.
Case 4: b < k and a < min{k 1, n k}. Since a < n k , also b < n k .
Therefore, we can use the idea of Case 1 without the complete bipartite
graph. Again take disjoint vertex sets X of size k and Y of size n k . If ka
and (n k)b are even, then we use an a -regular graph on X and a b-regular
graph on Y . As observed before, these exist.
Note that ka and (n k)b have the same parity, since their sum is
given to be even. If they are both odd, then we put min{k, n k} disjoint
edges with endpoints in both X and Y . We now complete the graph with
a regular graph of even degree on one of these sets and an almost-regular
graph guaranteed by Exercise 1.3.57 on the other.
1.3.61. If G is a self-complementary n -vertex graph and n is odd, then G
has a vertex of degree (n 1)/2. Let d1 , . . . , dn be the degree list of G in
nonincreasing order. The degree list of G in nonincreasing order is n
1 dn , . . . , n 1 d1 . Since G
= G , the lists are the same. Since n is
odd, the central elements in the list yield d(n+1)/2 = n 1 d(n+1)/2 , so
d(n+1)/2 = (n 1)/2.
1.3.62. When n!is" congruent to 0 or 1 modulo 4, there is an n -vertex simple
graph G with 12 2n edges such that !(G) (G) 1. This is satisfied by the
construction given in the answer to Exercise 1.1.31.
More generally, let G be any 2k -regular simple graph with 4k + 1 vertices, where n = 4k + 1. Such a graph can be constructed by placing 4k + 1
vertices around a circle and joining each vertex to the k closest vertices! in
"
each direction. By the Degree-Sum Formula, e(G) = (4k + 1)(2k)/2 = 12 2n .
For the case where n = 4k , delete one vertex from the graph con!"
structed above to form G ! . Now e(G ! ) = e(G) 2k = (4k 1)(2k)/2 = 12 2n .
1.3.63. The non-negative integers
d1 dn are the vertex degrees of a
#
loopless graph if and only if #di is even and d1 d2 + + dn . Necessity.
If such a graph exists, then
di counts two endpoints of each edge and
must be even. Also, every edge incident to the vertex of largest degree

Section 1.3: Vertex Degrees and Counting

62

has its other end counted among the degrees of the other vertices, so the
inequality holds.
Sufficiency. Specify vertices v1 , . . . , vn and construct a graph so that
d(vi ) = di . Induction on n has problems: It is not enough to make
dn edges join v1 and vn degrees and apply the induction hypothesis to
(d1 dn ), d2 , . . . , dn1 . Although d1 dn d2 + + dn1 holds, d1 dn
may not be the largest of these numbers.
#
#
Proof 1 (induction on
di ). #
The basis step is
di = 0, realized by
an
independent
set.
Suppose
that
d
>
0;
we
consider
two cases. If d1 =
i
#n
d
,
then
the
desired
graph
consists
of
d
edges
from
v1 to v2 , . . . , vn . If
i
1
i=2 #
n
d1 < i=2
di , then the difference is at least 2, because the total degree sum
is even. Also, at least two of the values after d1 are nonzero, since d1 is the
largest. Thus we can subtract one from each of the last two nonzero values
to obtain a list d ! to which we can apply the induction hypothesis (it has
even sum, and the largest value is at most the sum of the others. To the
resulting G ! , we add one edge joining the two vertices whose
are
#degrees
n
the reduced values. (This can also be viewed as induction on ( i=2
di )d1 .)
#
#
Proof 2 (induction on
di ). Basis as above. Consider
di > 0. If
d1 > d2 , then we can subtract 1 from d1 and from d2 to obtain d ! with smaller
sum. Still d1 1 is a largest value in d ! and is bounded by the sum of the
other values. If d1 = d2 , then we subtract 1 from each of the two smallest
values to form #
d ! . If these are d1 and d2 , then d ! has the desired properties,
n
and otherwise i=2
di exceeds d1 by at least 2, and again d ! has the desired
properties. In each case, we can apply the induction hypothesis to d ! and
complete the proof as in Proof 1.
Proof 3 (local change). Every nonnegative integer sequence with even
sum is realizable when loops and multiple edges are allowed. Given such a
realization with a loop, we change it to reduce the number of loops without
changing vertex degrees. Eliminating them all produces the desired realization. If we have loops at distinct vertices u and v , then we replace two
loops with two copies of the edge uv . If we have loops only at v and have an
edge x y between two vertices other than v , then we replace one loop and
one copy of x y by edges vx and vy . Such an edge x y must exist because the
sum of the degrees of the other vertices is as large as the degree of v .
1.3.64. A simple graph with degree sequence d1 d2 dn is connected
if d j j for all j such that j n 1 dn . Let V (G) = {v1 , . . . , vn }, with
d(vi ) = di , and let H be the component of G containing vn ; note that H
has at least 1 + dn vertices. If G is not connected, then G has another
component H ! . Let j be the number of vertices in H ! . Since H has at least
1 + dn vertices, we have j n 1 dn . By the hypothesis, d j j . Since H !
has j vertices, its maximum degree is at least d j . Since d j j , there are at

63

Chapter 1: Fundamental Concepts

least j + 1 vertices in H ! , which contradicts the definition of j . Hence G is

in fact connected.
1.3.65. If D = {ai } is a set of distinct positive integers, with 0 < a1 < <
ak , then there is a simple graph on ak + 1 vertices whose set of vertex degrees
(repetition allowed) is D .
Proof 1 (inductive construction). We use induction on k . For k = 1,
use K a1 +1 . For k = 2, use the join K a1 K a2 a1 +1 . That is, G consists of a
clique Q with a1 vertices, an independent set S with a2 a1 + 1 vertices,
and all edges from Q to S . The vertices of S have degree a1 , and those of Q
have degree a2 .
For k 2, take a clique Q with a1 vertices and an independent set
S with ak ak1 vertices. Each vertex of S has neighborhood Q , and
each vertex of Q is adjacent to all other vertices. Other vertices have
a1 neighbors in Q and none in S , so the degree set of G Q S should
be {a2 a1 , . . . , ak1 a1 }. By the induction hypthesis, there is a simple
graph H with ak1 a1 + 1 vertices having this degree set (the degree set
is smaller by 2). Using H for G Q S completes G as desired.
Proof 2 (induction and complementation). Again use induction on k ,
using K a1 +1 when k = 1. For k > 1 and 0 < a1 < ak , the complement
of the desired graph with a1 + 1 vertices has degree set {ak a1 , . . . , ak
ak1 , 0}. By the induction hypothesis, there is a graph of order ak a1 + 1
with degree set {ak a1 , , ak ak1 }. Add a1 isolated vertices and take
the complement to obtain the desired graph G .
1.3.66. Construction of cubic graphs not obtainable by expansion alone. A
simple cubic graph G that cannot be obtained from a smaller cubic graph by
the expansion operation is the same as a cubic graph on which no erasure
can be performed, since any erasure yielding a smaller H from G could be
inverted by an expansion to obtain G from H . An edge cannot be erased by
this operation if and only if one of the subsequent contractions produces a
multiple edge. This happens if the other edges incident to the edge being
erased belong to a triangle, or in one other case, as indicated below.

and

Finally, we need only provide a simple cubic graph with 4k vertices

where every edge is non-erasable in one of these two ways. To do this place
copies of G 1 , . . . , G k of K 4 e (the unique 4-vertex graph with 5 edges)
around in a ring, and for each consecutive pair G i , G i+1 add an edge joining

Section 1.3: Vertex Degrees and Counting

64

a pair of vertices with degree two in the subgraphs, as indicated below,

where the wraparound edge has been cut.

1.3.67. Construction of 3-regular simple graphs

a) A 2-switch can be performed by performing a sequence of expansions
and erasures. We achieve a 2-switch using two expansions and then two
erasures as shown below. If the 2-switch deletes x y and zw and introduces
xw and yz , then the first expansion places new vertices u and v on x y and
zw , the second introduces s and t on the resulting edges ux and vz , the first
erasure deletes su and its vertices, and the second erasure deletes tv and its
vertices. The resulting vertices are the same as in the original graph, the
erasures were legal because they created only edges that were not present
originally, and we have deleted x y and zw and introduced xw and yz .

y
s
v u
w

y
z
t
v
w

y
z
t
v
w

z
w

b) Every 3-regular simple graph can be obtained from K 4 by a sequence

of expansions and erasures. Erasure is allowed only if no multiple edges
result. Suppose H is the desired 3-regular graph. Every 3-regular graph
has an even number of vertices, at least four. Any expansion of a 3-regular
graph is a 3-regular graph with two more vertices. Hence successive expansions from K 4 produce a 3-regular graph G with n(H ) vertices. Since G
and H have the same vertex degrees, there is a sequence of 2-switches from
G to H . Since every 2-switch can be produced by a sequence of expansions
and erasures, we can construct a sequence of expansions and erasures from
K 4 to H by going through G .
1.3.68. If G and H are X, Y -bigraphs, then d G (v) = d H (v) for all v X Y
if and only if there is a sequence of 2-switches that transforms G into H
without ever changing the bipartition. The condition is sufficient, since 2switches do not change vertex degrees. For necessity, assume that d G (v) =
d H (v) for all v . We build a sequence of 2-switches transforming G to H .
Proof 1 (induction on |X |). If |X | = 1, then already G = H , so we
may assume that |X | > 1. Choose x X and let k = d(x). Let S be a
selection of k vertices of highest degree in Y . If N (x) '= S , choose y S
and y ! Y S so that x ! y and x y ! . Since d(y) d(y ! ), there exists
x ! X so that y x ! and y ! ! x ! . Switching x y ! , x ! y for x y, x ! y ! increases