Scribd
Upload
112 views3 pages

Adjacency Matrix Notes

1) The adjacency matrix of a graph represents the connections between vertices, with 1s indicating edges and 0s indicating no edge. Powers of the adjacency matrix count the number of walks between vertices. 2) The eigenvalues of a graph's adjacency matrix provide information about the graph. For example, regular graphs have their degree as an eigenvalue. 3) Moore graphs are regular graphs with the maximum possible number of vertices given the degree and girth. There is a Moore graph if and only if the degree is 2, 3, 7 or possibly 57.

Uploaded by

Kavita M
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
112 views3 pages

Adjacency Matrix Notes

1) The adjacency matrix of a graph represents the connections between vertices, with 1s indicating edges and 0s indicating no edge. Powers of the adjacency matrix count the number of walks between vertices. 2) The eigenvalues of a graph's adjacency matrix provide information about the graph. For example, regular graphs have their degree as an eigenvalue. 3) Moore graphs are regular graphs with the maximum possible number of vertices given the degree and girth. There is a Moore graph if and only if the degree is 2, 3, 7 or possibly 57.

Uploaded by

Kavita M
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Graphs and Matrices

1 The Adjacency Matrix of a Graph

The adjacency matrix A of a graph is defined by numbering the vertices, say


from 1 up to n, and then putting aij = aji = 1 if there is an edge from i to j, and
aij = 0 otherwise. We can do the same for a digraph: putting aij = 1 if there
is an arc from i to j, and aij = 0 otherwise. For example, here is an adjacency
matrix of a directed cycle on 4 vertices:
 
0 1 0 0
 
 0 0 1 0 
 
 
 0 0 0 1 
 
1 0 0 0

2 Powers of the Adjacency Matrix

The powers of the adjacency matrix counts things. In particular,

entry i, j in As gives the number of walks from i to j of length s.

The proof is by induction argument. For example, the number of walks of length 2
is the number of vertices k such that there is an arc from i to k and an arc from
k to j. And this is exactly the i, j entry in A2 , by the definition of matrix
multiplication.

(If we have weights on the edges, then the result is still valid. We redefine
the adjacency matrix to have the weights as its entries, and define the weight of
a walk as the product of the weights of the arcs. Then if want to know the total
sum of weights of i, j paths of given length, that is the entry in the appropriate
power.)
3 Eigenvalues of Graphs

We consider next undirected graphs without loops. The eigenvalues of a graph


are the eigenvalues of its adjacency matrix. For example: the triangle K3 has
eigenvalues, 2, −1, −1.

Observation: An r-regular graph has r as an eigenvalue. Proof: It has the


all-1 eigenvector.

The complete bipartite graph Km,n has an adjacency matrix with rank 2.
So it has eigenvalue 0 with multiplicity m + n − 2. It can be checked that the
√
remaining two eigenvalues are ± mn.

4 Moore Graphs

Consider an r-regular simple graph of girth 5. That is, for every vertex v there
is no edge between the neighbors of v, and none of the neighbors have a common
neighbor apart from v. Then we must have at least 1 + r + r(r − 1) = r2 + 1
distinct vertices. A Moore graph is an r-regular graph with r2 + 1 vertices
with that property. For example, a 5-cycle is the Moore graph for r = 2 and the
Petersen graph for r = 3.

Theorem 1 There is a Moore graph if and only if r = 2, 3, 7 or possibly 57.

Proof: Assume a Moore graph exists and let A be its adjacency matrix. Then if
we consider any pair of vertices, they are either neighbors or they have a unique
common neighbor, but not both. Consider what this means for the matrix A2 .
Off the diagonal, it has a 1 wherever A has a 0 and vice versa. On the diagonal
it has r (just because the graph is r-regular). This means that

A2 + A − (r − 1)I = J,

where J is the all-1 matrix.

Now let λ be an eigenvalue of A with eigenvector v. There are two possibilities.


First, assume the entries of v sum to something nonzero. Then we can normalize v
such that its entries sum to 1. Then

[λ2 + λ − (r − 1)]v = J,

1
where 1 is the all-1 vector. It follows that v = n J, where n is the number of
vertices; and that λ = r (with multiplicity 1).

The second possibility is that the entries of v sum to zero. Then

[λ2 + λ − (r − 1)]v = 0.

And so we have λ2 + λ − (r − 1) = 0. Hence


p
−1 ± 1 + 4(r − 1)
λ= .
2
√
Now, suppose the λ+ = (−1 + 4r − 3)/2 is an eigenvalue b times. It follows
√
that λ− = (−1 − 4r − 3)/2 is an eigenvalue n − b − 1 = r2 − b times.

And the trace of A is clearly 0. So we have that

bλ+ + (r2 − b)λ− + r = 0.

And thus that


r(r − 2)
2b − r2 = √ .
4r − 3
The right-hand side must therefore be an integer. If the numerator is 0, which
happens when r = 2, then we are okay. Otherwise, we certainly need that
√
4r − 3 be rational, which only happens if it is an integer, say m.

Substitute this into the RHS, and we get

m4 − 2m2 − 15
16m
Thus, m must be a divisor of 15, which forces m ∈ {1, 3, 5, 15}. Thus r ∈
{1, 3, 7, 57}. It can be checked that in each case of r, b is a whole number.
[EndProof]

Moore graph have been constructed for r = 2, 3, 7, but the existence for
r = 57 is unsolved.

You might also like