Graph Theory Chapter 2
Graph Theory Chapter 2
VERTEX DEGREES
11
1.3
Vertex Degrees
dG (x2 ) = dG (x4 ) = 3.
a2 x1 a4 a3 x2
e1
x2
a1
Figure 1.9:
(b) a digraph D
A vertex of degree d is called a d-degree vertex. A 0-degree vertex is called an isolated vertex. A vertex is called to be odd or even if its degree is odd or even. A graph G is k -regular if dG (x) = k for each x V (G), and G is regular if it is k -regular for some k , and k is called the regularity of G. For instance, Kn is (n 1)-regular; Kn,n is n-regular; Petersen graph is 3-regular; the n-cube is n-regular. The maximum degree of G: (G) = max{dG (x) : x V (G)}. The minimum degree of G: (G) = min{dG (x) : x V (G)}. Clearly, (G) = k = (G) if G is k -regular.
For the digraph D shown in Figure 1.9 (b), for instance, d+ D (y1 ) = 2, d D (y1 ) = 2, d+ D (y2 ) = 1, d D (y2 ) = 2, d+ D (y3 ) = 1, d D (y3 ) = 3, d+ D (y4 ) = 3; d D (y4 ) = 0,
12
balanced if each of its vertices is balanced. The parameters + (D) = max{d+ D (y ) : y V (D )}, (D) = max{d D (y ) : y V (D )} and
are the maximum out-degree and maximum in-degree of D, respectively. The parameters + (D) = min{d+ D (y ) : y V (D )}, (D) = min{d D (y ) : y V (D )} and
are the minimum out-degree and minimum in-degree of D, respectively. The parameters (D) = max{+ (D), (D)}, (D) = min { + (D), (D)} and
are the maximum and the minimum degree of a digraph D, respectively. A digraph D is k -regular if (D) = (D) = k .
The First Theorem: Let G be a bipartite undirected graph with a bipartite {X, Y }. It is easy to see that the relationship between degree of vertices and the number of edges of G is as follows. dG (x) = (G) =
x X y Y
dG (y ).
(1.3)
dG (x).
(1.4)
Generally, for any a digraph D we have the following relationship between degree of vertices and the number of edges of G.
d+ D (x) =
x V
d D (x).
Proof: Let G be the associated bipartite graph with D of bipartition {X, Y }. Note that dG (x ) = d+ D (x), we have that d+ D (x) =
x V x X
dG (x ) = d D (x),
dG (x ) = (G) =
x Y
dG (x ) =
x V
d D (x).
dG (x)
1.3. VERTEX DEGREES and the number of vertices of odd degree is even.
13
Proof: Let D be the symmetric digraph of G. Then (D) = 2(G). Note that
dG (x) = d+ D (x) = dD (x),
x V.
d+ D (x) =
x V
d D (x) = (D ) = 2(G).
Let Vo and Ve be the sets of vertices of odd and even degree in G, respectively. Then dG (x) +
xVo xVe xVo
dG (x) =
x V
Since
xVe
Others Notations: The following notation and terminology are useful and
convenient to our discussions later on. Let D be a digraph, S and T are disjoint nonempty subset of V (D). The symbol ED (S, T ) denotes the set of edges of D whose tails are in S and heads are in T , and D (S, T ) = |ED (S, T )|. When just one graph is under discussion, we usually omit the letter D from these symbols and write (S, T ) and (S, T ) instead of ED (S, T ) and D (S, T ) for short. [S, T ] = (S, T ) (T, S ). If T = S = V (D) \ S , then we write
+ + ED (S ) (resp. ED (S )) instead of (S, S ) (resp. (S, S )), and d+ D (S ) = |ED (S )| (resp. d D (S ) = |ED (S )|). + The symbol ND (S ) (resp. ND (S )) denotes the set of heads (resp. tails) of edges
in ED [S ], which is called a set of out-neighbors (resp. in-neighbors) S in D. For instance, consider the digraph D shown in Figure 1.9. Let S = {y1 , y2 }, then
+ ED (S ) = {a3 }, ED (S ) = {a4 , a7 },
d+ D (S ) = 1, d D (S ) = 2,
+ ND (S ) = {y3 }, ND (S ) = {y3 , y4 }.
Similarly, for an undirected graph G and S V (G), the symbols EG (S ) and NG (S ) denote the set of edges incident with vertices in S in G and the set of neighbors of S in G, dG (S ) = |EG (S )|.
1 4
Proof: Arbitrarily choose xy E (G). Since G is simple and contains no triangle, it follows that [dG (x) 1] + [dG (y ) 1] v 2,
14 that is, dG (x) + dG (y ) v. Then summing over all edges in G yields d2 G (x) v .
x V
d2 G (x)
1 v
dG (x)
x V
4 2 , v
1 4
v2 .
(v 1) in G is odd
(the self-complementary graph is dened in the exercise 1.2.6). Proof: Let Vo and Ve be the sets of vertices of odd and even degree in G, respectively. Then |Vo | is even by Corollary 1.1. Since v 1 (mod 4), v must be odd and, thus, |Ve | is odd and
1 2
V = {x V (G) : dG (x) =
To prove the conclusion, we only need to show that |V | is even. To the end, let x V with dGc (x) =
1 2
with dG (yx ) = dGc (x). Note that dG (yx ) = dGc (x) = (v 1) dGc (x) = 1 (v 1). 2 (1.5)
Thus, yx = x from (1.5) and yx V . Furthermore, yx = yz if x, z V and x = z . This fact implies that the vertices in V occur in pairs, which shows that |V | is even.
15
1.4
A subgraph is one of the most basic concepts in graph theory. In this section, we rst introduce various subgraphs induced by operations of graphs.
x4
x3
x4
x3
x4 G {x1 , x3 } x1
A spanning subgraph of G x1 e1 x2 x5 e5
e1 x2 e8
x4
x3
x4 G [ {x 1 , x 2 , x 4 }]
x4
x3
Figure 1.10:
16
and edge-set E (G1 ) E (G2 ). We write G1 + G2 for G1 G2 if G1 and G2 are disjoint, and G1 G2 for G1 G2 if G1 and G2 are edge-disjoint. If Gi = H for each i = 1, 2, , n, then write nH for G1 + G2 + + Gn . The intersection G1 G2 of G1 and G2 is dened similarly if V (G1 ) V (G2 ) = . These operations of graphs are depicted in Figure 1.11.
x1 x4 x1 x4
x2 x1 x3 x2 x3 x4
=
x2 x3
x2 x1 x3 x2 x1 x3
=
x2 x1 x3
x2
x5
x2
x5
x2
x5
x3
x4
x3
x4
x3
x4
Figure 1.11:
An edge e of G is said to be contracted if it is deleted and its end-vertices are identied; the resulting graph is denoted by G e. This is illustrated in Figure 1.12.
x1 x1
x5
x2 x3 = x5
x2
x4
x3 G
x4 Ge
Figure 1.12:
d+ H (x) =
x X
d H (x). Thus,
d H (x) = dG (X ) x X
d+ G (X ) =
x X
d+ G (x)
x X
d+ H (x) =
x X
d G (x)
as required.
17
Proof: Let H be a bipartite spanning subgraph of G with as many edges as possible, and let {X, Y } be a bipartition. Arbitrarily choose x V (G), without loss of generality, say x X . Let d = dG (x) dH (x). We claim that d dH (x). In fact, suppose to the contrary that d > dH (x). Let X = X \ {x} and Y = Y {x}. Consider a bipartite spanning subgraph H of G with the bipartition {X , Y }. Then (H ) (H ) = (H ) + d dH (x) > (H ), a contradiction. Thus, dG (x) = d + dH (x) 2 dH (x). Summing up all vertices in G yields that (G) 2 (H ) by Corollary 1.1.
of n identical complete graph K2 , see Figure 1.8 for Q1 , Q2 , Q3 and Q4 . The hypercube is an important class of topological structures of interconnection networks, some of whose properties will be further discussed in some sections in this book.
18
011
111 11
110
001
000
010
101
111
110
Figure 1.13:
Assume that L(G) is the line graph of a graph G. If L(G) is non-empty and has no isolated vertices, then its line graph L(L(G)) exists. For integers n 1, Ln (G) = L(Ln1 (G)), where L0 (G) and L1 (G) denote G and L(G), respectively, and Ln1 (G) is assumed to be non-empty and has no isolated vertex. The graph Ln (G) is called the nth iterated line graph of a graph G. Example 1.4.4 Two important classes of graphs, the well-known n-dimensional d-ary Kautz digraph K (d, n) and de Bruijn digraphs B (d, n) can be dened as K (d, n) = Ln1 (Kd+1 )
+ B (d, n) = Ln1 (Kd ),
pending one loop at each vertex. The digraphs in Figure 1.13 are B (2, 1), B (2, 2) and B (2, 3). The original denitions of K (d, n) and B (d, n) will be given in Section 1.8,
Exercises:
1.3.5, 1.3.6