ON A CLASS OF POLYNOMIALS ASSOCIATED WITH THE CLIQUES IN A GRAPH AND ITS APPLICATIONS

The clique polynomial of a graph is defined. An explicit formula is then derived for the clique polynomial of the complete graph. A fundamental theorem and a reduction process is then given for clique polynomials. Basic properties of the polynomial are also given. It is shown that the number theoretic functions defined by Menon are related to clique polynomials. This establishes a connection between the clique polynomial and decompositions of finite sets, symmetric groups and analysis.

spanning subgraph of G whose components are all cliques, i.e., a node disjoint set of cliques which cover all the nodes of G.With every clique in G, let us associate an indeterminate or weight w a, and with every cover C, a weight w(C) n w a where the product is taken over all the elements of C. Then the clique polynomial of G is w(c) where the summation is taken over all the covers of G.
Let us give each clique with n nodes a weight w Then the (general) clique n polynomial is denoted by K(G ), where (w I, w2, w 3 is a general weight vec- tor.This polynomial is a polynomial in the indeterminates w I, w2, etc.If we put Wn w, for all values of n (i.e., _w (w, w, w, ...)), then the resulting polynomial is called the simple clique polynomial of G.This polynomial is denoted by K(G w).
We derive explicit formulae for the general and simple clique polynomials of the complete graph and for their generating functions.A reduction algorithm is then given for clique polynomials.We show that a number-theoretic function defined by Menon [3] is really the clique polynomial of a complete graph, when weights are assigned in a particular manner.Finally, we establish some interesting connections between the clique polynomial and various combinatorial results.Central to this section of the article is Menon's paper [3].
In the material which follows, we denote the generating function for K(G (with indicator function t) by K(G H, t).Lower limits of summations are zero unless otherwise defined.When unspecified, upper limits are infinity, or the largest value of the variable which makes sense in the context of the summand.For brevity, we write k(G) for K(G ;w).

CLIQUE POLYNOMIALS OF COMPLETE GRAPHS.
If, instead of cliques, we take the components of a cover of G to be members of a general family F of connected graphs, then the resulting polynomial is called the F-poZnomaZ of G.We will not give an independent derivation of the generating function for the clique polynomial of the complete graph.Instead, we use a result given in Farrell [I] (Theorem I) on the generating function for the general F-poly- nomial of the complete graph.
The generating function for the F-polynomlal of the complete graph is where #i is the number of spanning subgraphs of K i which are members of the family F.
In the case of the clique polynomial, i is the number of spanning subgraphs of K i which are complete graphs.Thus #i I, for all i.Hence we get the following result.
THEOREM I.
wi ti The following corollary is obtained by extracting the coefficient of p! COROLLARY 1.1.P where the summation is taken over all solutions of Z i Ji p i We can obtain a recurrence relation for K(Kp ) from Theorem I, using standard techniques (or by direct substitution into Theorem 3 of [3]).This is given in the following theorem, in which the Blissard notation (see Riordan [4]) is used.THEOREM 2.
K p w(K + w) p-I r with K r K(K r ;_w) and w E Wr Results for the simple clique polynomial of K can be obtained from the above P results by putting (w, w, w From Theorem I, we get THEOREM 3. K(Kp;W, t) exp[w(exp t I)1 This generating function is also the generating function for Stirling numbers of the second kind (see [4], page 32).We denote the k-th Stirling number associated with n, by S(n,k).Hence we have the following interesting result.The number of clique covers with cardinality k in the labelled complete graph K P is S(p,k).
From Theorem 2 we obtain the following corollary.COROLLARY 2.1.K p w(K + I) p where K r E K(K r w)

THE FUNDAMENTAL THEOREM AND ALGORITHM
Let e be an edge in a graph G.We say that e is (clique) incorporated in G if e is required to belong to every clique cover of G as part of a component K for n

n
If a graph G contains an incorporated edge, G is called a restricted graph.We normally use an asterisk to denote such a graph.Let H be a subgraph of G. Th'em denotes the graph obtained from G by removing the nodes of H.In order to distinguish between the trivial complete graphs with one and two nodes (i.e. a node and an edge) and the non-trivial ones with more than two nodes, we call K a proper complete graph, n when n > 2. A proper (clique) cover is a cover in which all components are proper complete graphs.
Let e be an edge of G.We can put the covers in G into three classes (i) those not containing e, (ii) those in which e is a component and (ill) those in which e is part of a proper complete graph.These considerations lead to the followlng theorem, called the Fundamental Theorem for clique polynomials.Let G be a graph containing an (unincorporated) edge e.Let G" be the graph obtained from G by deleting e, G" the graph obtained from G by removing the nodes at the ends of e and G* the graph obtained from G by incorporating e.Then k(G)

An illustration of the Theorem
The diagram is self-explanatory.The internal arrows indicate the edge e.The asterisk indicates the incorporated edge e.

Figure
The following corollaries are easily deduced from the definitions and Theorem 5.If e is adjacent to an incorporated edge, then k(G") The following theorem gives a technique for finding clique polynomials of restricted graphs.THEOREM 6.Let G* be a restricted graph containing an incorporated edge e.Then k(G*) r. w n (G-Hn) where H is a clique of G with n nodes (n > 2) and containing the edge e, and the n summation is taken over all such clique subgraphs of G. PROOF.
Let e be part of a clique H n with n > 2 nodes Then the weight of Hn as a compo- The rest of the cover is a cover of the graph G-H The nent of clique cover is w n n result can therefore be deduced If we restrict the cliques to K and K 2, then the clique polynomial becomes the matching polynomial (see Farrell [2]).The matching polynomial of G is denoted by The following result is straightforward, kl(G) counts proper clique covers.
If G does not have triangles (i.e.G is triangle-free) then the only elements that a cover of G can have are K and K 2. It follows that every clique cover is a match- ing in G. Hence we have the following result.THEOREM 8.
If G is a triangle-free graph, then k(G) re(G) Theorem 5 yields an algorithm for finding clique polynomials of arbitrary graphs.
We simply apply the theorem recursively, until we obtain graphs H i, for which k(H i) is known.This algorithm is called the Fwulamental AlgoPthm for clique polynomials or for brevity, the reduction process.We illustrate this igorlthm using the graph G in Figure I.Notice that Corollaries 5. I, 5.2 and Theorem 6 can be very useful when applying the reduction process.Suppose that wrl Wr2 Wrn is a term in k(G) with non-zero coefficient.Then G has a clique cover consisting of the components Kri(i I, 2 n).Consider the component K m, where m r i, for some 0 < i n.This could be replaced by "smaller" clique covers.The weights of these covers will be the terms of k(Km).
Hence we have the following theorem.
If k(G) contains a term in w r w r w r with nonzero coefficient, then every 2 n term of the polynomial wrl Wr2 Wrj-i Wrj+l wrn k(Krj) also occurs in k(G) with nonzero coefficient, for J I, 2 n.
The following theorem characterizes complete graphs in terms of clique polynomials.

THEOREM II.
Let G be a graph with p nodes.Then G is a complete graph if and only if k(G) contains the term w P 5. CONNECTION WITHMENON'S FUNCTIONS Let S be a finite set with cardinality n.Let f be a function defined on all finite sets and depending only on the number of elements in the set; so that f(S) can be written as f(ISl) f(n).Let D be a partition of S with k elements.We extend the definition of f to all partitions of S by where S i is an element of D with n i elements, and the product is taken over all the k elements in D (IDI k) Next, we define the function F on finite sets S by

F(S) Z f(D) D
where the summation is taken over all the partitions D of S. We refer to the functions f and F as Menont8 notio8.These functions were defined in [3].
A correspondence between the structures of this section and that of Section can be established by letting f be the weight function f(S) Wls I, D be a cover C of some graph G, and Finally, F(S) corresponds to the clique polynomial of G, i.e.,

F(S) Z w(C)
Since all the elements are equivalent, all the nodes of the corresponding graph G must be equivalent.Therefore G must by K The following theorem formalizes our n discussion.
Let f and F be the two Menon functions defined above and let S be a set with This result can be formally proved by equating the expression for F(n) given in [3] (Equation 2.6) with the expression for K(K n ) given in Corollary 1.1.
The following corollary is immediate from Theorem 12, by using the generating function given in Theorem I, and denoting it by G(t).This result is also given in [3].COROLLARY 12.1. where From Theorem 2, we obtain the following recurrence for F(n), in which the Blissard notation is used.
COROLLARY 12.This corollary is equivalent to the recurrence given in [3] (Equation 4.1), which was obtained by a combinatorial argument.
The clique polynomial can be applied to enumerating problems connected with the decompositions of finite sets.For example, by giving each clique a weight of I, in the clique polynomial of K n, we can obtain the total number of partitions of a set with n elements.Also, the coefficient of w k in the simple clique polynomial of K n will be the number of partitions of the set into k elements.These results are stated formally in the following theorem.THEOREM Let Dk(n) be the number of partitions of a set of n elements into exactly k elements and D(n), the total number of partitions of S. Then (i) Dk(n) coefficient of w k in K(K n w) S(n,k) and (ii) D(n) K(K n ;w) where w (1, 1, 1, ...) Theorem 13 can be easily proved.Alternatively, the result follows from Theorem and Equations 5.4 and 5.5 of [3].
The clique polynomial can also be applied to certain problems in elementary analysis.This is clear from the following theorem, which is also given in [3], in terms of Menon's functions.
Let h be a function of the single variable x, on the complex numbers, and let the PROOF.
We will use the result given in Theorem I. Give each clique with i nodes a weight Put w i-h(1)(x) i=l i=l h(x + t) -h(x), from Taylor's Theorem.
Then from Corollary 12. I, we get The coefficient of T in G(t) (using the Taylor expansion of eh(x+t)) is e-h(x) d { x n eh(X)) But G(t) is the generating function for K(K n _w).Therefore K(;H) e-h(x) dxn eh(x Hence the result follows.
Menon has used the functions in order to obtain various enumeration theorems about symmetric groups.Thus the clique polynomial of K n can also be applied to these kinds of enumeration problem.This is not surprising, since Theorem 4 of [1] establishes a connection between the F-polomial of K and the cycle index of the symmetric group n on n elements.
The connection between the clique polynomial and Menon's functions is an important one.Menon's functions were abstract number theoretic functions.Now they have been given a tangible combinatorial foR.This we hope will contribute to the use of the functions in enumerative problems.

Call for Papers
Thinking about nonlinearity in engineering areas, up to the 70s, was focused on intentionally built nonlinear parts in order to improve the operational characteristics of a device or system.Keying, saturation, hysteretic phenomena, and dead zones were added to existing devices increasing their behavior diversity and precision.In this context, an intrinsic nonlinearity was treated just as a linear approximation, around equilibrium points.
Inspired on the rediscovering of the richness of nonlinear and chaotic phenomena, engineers started using analytical tools from "Qualitative Theory of Differential Equations," allowing more precise analysis and synthesis, in order to produce new vital products and services.Bifurcation theory, dynamical systems and chaos started to be part of the mandatory set of tools for design engineers.
This proposed special edition of the Mathematical Problems in Engineering aims to provide a picture of the importance of the bifurcation theory, relating it with nonlinear and chaotic dynamics for natural and engineered systems.Ideas of how this dynamics can be captured through precisely tailored real and numerical experiments and understanding by the combination of specific tools that associate dynamical system theory and geometric tools in a very clever, sophisticated, and at the same time simple and unique analytical environment are the subject of this issue, allowing new methods to design high-precision devices and equipment.
Authors should follow the Mathematical Problems in Engineering manuscript format described at http://www .hindawi.com/journals/mpe/.Prospective authors should submit an electronic copy of their complete manuscript through the journal Manuscript Tracking System at http:// mts.hindawi.com/according to the following timetable: COROLLLARY $.I.If G has no triangles, then k(G*) 0. Therefore k(G) k(G') + w2k(G").COROLLARY 5. 2.
can be easily proved.THEOREM 9.Let G be a graph consisting of r components G I,G 2, F(r) and fr _= f(r)

First
Round of Reviews March 1, 2009