Fast Partial-Differential Synthesis of the Matching Polynomial of C
72-100
James M. Salvador,* Adrian Hernandez, Adriana Beltran, Richard Duran, and Anthony Mactutis
Department of Chemistry, University of Texas at El Paso, El Paso, Texas 79968
Received February 11, 1998
The Thesis algorithm uses partial-differential edge operators and a grammatical structure to generate and
avoid expanding the Matching Polynomial. To run the algorithm efficiently, the vertexes of fullerene graphs
C
60-100
were sorted into three-dimensional sectors.
INTRODUCTION
The Matching,
1
Reference,
2
or Acyclic
3
Polynomial is the
counting polynomial of nonadjacent edges of a graph as a
function of isolated vertexes. Its lowest order term is the
number of perfect matchings or Kekule ´ structures of a given
π-system. Hosoya’s Z index is the number of terms in a
Matching Polynomial. The Z index is elegantly calculated
through graph “retrosynthesis” and easily correlates the
hydrogen suppressed graph of acyclic alkanes to their boiling
point. The difference in the “occupied” roots of the
Matching and Characteristic Polynomials provides a method
of comparing the stability of different size π-systems and is
called a Topological Resonance Energy (TRE). Because of
recent interest in calculating the TRE of fullerenes up to
C
70
,
4-8
we report the development of the Thesis algorithm
to quickly calculate the Matching Polynomial of larger
fullerenes using partial-differential edge operators.
In matrix terms, for a graph G the Characteristic Polyno-
mial, CP(G) or the determinant of its adjacency matrix with
elements a
i,j
, is a listing of all walks across the matrix without
stepping twice on the same row or column. The Matching
Polynomial, MP(G), is a listing of all determinant walks that
contain only matching steps a
i,j
and a
j,i
. MP(G) equals CP-
(G) for acyclic graphs. MP(G) is a subset of CP(G) for
cyclic graphs because the latter includes walks that explicitly
list the Hamiltonian paths of the graph.
Matching off-diagonal walks are permutations of the
diagonal walk since the diagonal and off-diagonal positions
of an adjacency matrix can correspond to the vertexes (V
i
)
and edges (e
i,j
) of G, respectively, as shown in Figure 1.
Rosenfeld and Gutman showed that these permutations are
a product of partial-differential operators on the product of
diagonal elements.
9
We use partial-differential edge opera-
tors to develop Thesis as follows.
METHODS
MP(G) for the trivial graph of n vertexes is the product
of the diagonal elements of the adjacency matrix, eq 1. An
edge operator is a permutor of the diagonal walk to its
matching off-diagonal elements, eq 2. MP(G) for a nontrivial
graph is the product of edge operators on the trivial graph
monomial as shown in eq 3. Partial-differential operators
ensure that no row or column is repeated in a walk since the
Figure 1. An adjacency matrix.
Figure 2. A Mathematica program to generate MP(C
60
).
Figure 3. (((V
1
V
2
)(V
3
V
4
))((V
5
V
6
)(V
7
V
8
))).
1105 J. Chem. Inf. Comput. Sci. 1998, 38, 1105-1110
10.1021/ci9800155 CCC: $15.00 © 1998 American Chemical Society
Published on Web 09/18/1998