Spanning Trees

Spanning Trees
Let G be a graph. Recall from Section 1.4 that a subgraph H of G is called a spanning subgraph
of G if the vertex set of H is the same as the vertex set of G.
A spanning tree of a graph G is a spanning subgraph of G that is a tree.
Our first result of this section shows that the graphs which have spanning trees are easily
described.
Theorem 2.12 a graph G is connected if and only if it has a spanning tree.
Proof suppose that G is connected with n vertices and q edges. Then, by corollary 2.10, we have
q ≥ n – 1. If q = n – 1 then, by (iii) → (i) of theorem 2.11, G is a tree and so we can take T = G as
a spanning tree of G.
If q > n – 1 then, by theorem 2.4 (or by (i) → (iii) of theorem 2.11), G is not a tree and so G
must contain a cycle. Let e1 be an edge of such a cycle. Then the subgraph G – e1 is
connected (since e1 is not a bridge), has n vertices, and has q – 1 edges. If q – 1 = -1 then,
repeating the above argument gives T = G – e1 has a spanning tree of G.
If q – 1 > n – 1 then G – e1 is not a tree so, as before, there is a cycle in G – e1. Removing an
edge e2 from such a cycle gives a subgraph G – {e1, e2} = (G – e1) – e2 which is connected,
has n vertices and q – 2 edges. We keep on repeating this process, deleting q – n + 1 edges
altogether, to eventually produce a subgraph T which is connected, has n vertices and q –
(q – n + 1) = n – 1 edges. Thus by teorem 2.11, T is a tree and since it has the same vertex
set as G it is a spanning tree of G.
Conversely, if G has a spanning subtree T, then given any two vertices u and v of G then u
and v are also vertices of the connected subgraph T. thus u and v are connected by a path in
T and so by a path in G. this shows that G is connected.
Figure 2.12 and 2.13 ilustrate the theorem.

Figure 2.12: a connected graph and a spanning tree.
Figure 2.13: K4 and its 16 different spanning trees
Note that in the 16 different spanning trees of K4 shown in Figure 2.13 there are only two non-
isomorphic ones – the first 12 shown are isomorphic to each other, while the last four are also
isomorphic to each other. K6, the complete graph on 6 vertices, has 1296 different spanning
trees, but just 6 non-isomorphic ones. Another way of saying this is, given 6 vertices, then there
are 1296 different ways of joining these vertices to form a tree if we label the vertices 1, 2, 3, 4,
5, 6, but if we drop these labels then there are only 6 different ways.
The subject of counting how many spanning trees and non-isomorphic spanning trees there are
for a given graph was probably initiated by the English mathematician Arthur Cayley, who used
trees to try to count the number of saturated hydrocarbons CnH2n+2 containing a given number of
carbon atoms. Cayley was the first person to use the term “tree” (in 1857) and in 1889 [12] he
proved the following result which tells us that given n vertices, labeled 1, . . . , n, then there are
nn-2 different ways of joining them to form a tree.
Theorem 2.13 (Cayley, 1889) the complete graph Kn has nn-2 different spanning trees.
Proof we omit the proof but, for those interested, see pages 32-35 of bondy and murty [7] or
pages 50-52 of Wilson [65].
Bondy and Murty page 35
Theorem 2.9 τ(Kn) = nn – 2.
Proof let the vertex set of Kn be N = {1, 2, …, n}. we note that nn – 2 is the number of sequences
of length n- 2 that can be formed from N. thus, to prove the theorem, is suffices to establish a
one-one correspondence between the set of spanning trees of Kn and the set of such sequeces.
With each spanning tree T of Kn, we associate a unique sequence (t1, t2, …, tn-2) as follows.
Regarding N as an ordered set, let s1 be the first vertex of degree one in T; the vertex adjacent to
s1 is taken as t1. We now delete s1 from T, donate by s2 the first vertex of degree one in T – s1,
and take the vertex adjacent to s2 as t2. This operation is repeated until tn – 2 has been defined and
a tree with just two vertices remains; the tree in figure 2.9, for instance, gives rise to sequace (4,
3, 5, 3, 4, 5). It can be seen that different spanning trees of Kn determine difference sequences.
Figure
The reverse procedure is equally straightforward. Observe, first, that any vertex v of T occurs
dτ(v) – 1 times in (t1, t2, …, tn – 2). Thus the vertices of degree one in T are precisely those that do
not appear in this sequence. To reconstruct T from (t1, t2, …, tn – 2), we therefore proceed as
follows. Let s1 be the first vertex of N not in (t1, t2, …, tn – 2); join s1 to t1. Next, let s2 be the first
vertex of N\{s1} not in (t2, …, tn – 2), and join s2 to t2. Continue in this way until the n – 2 edges
s1t1, s2t2, …, sn – 2tn – 2 have been determined. T is now obtained by adding the edge joining the
two remaining vertices of N\{s1, s2, …, sn – 2}. It is easily verified that different sequence give
rise to different spanning trees of Kn. We have thus established the desired one-one
correspondence.
Note that nn – 2 is not the number of ninisomorphic spanning trees of Kn, but the number of
distinct spanning trees of Kn; there are just six nonisomorphic spanning trees of K6, where as
there are 64 = 1296 distinct spanning trees of K6

Spanning Trees

Uploaded by

Copyright:

Available Formats

Spanning Trees

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Spanning Trees

Uploaded by

Copyright:

Available Formats

Spanning Trees

A spanning tree of a graph G is a spanning subgraph of G that is a tree.

Theorem 2.12 a graph G is connected if and only if it has a spanning tree.

Figure 2.12 and 2.13 ilustrate the theorem.

Figure 2.13: K4 and its 16 different spanning trees

Bondy and Murty page 35

Theorem 2.9 τ(Kn) = nn – 2.

You might also like