Lower bounds for the clique and the chromatic numbers of a graph

G is any simple graph with m edges and n vertices where these vertices have vertex degrees d(1)≥d(2)≥?≥d(n), respectively. General algorithms for the exact calculation of χ(G), the chromatic number of G, are almost always of ‘branch and bound’ type; this type of algorithm requires an easily constructed lower bound for χ(G). In this paper it is shown that, for a number of such bounds (many of which are described here for the first time), the bound does not exceed cl(G) where cl(G) is the clique number of G.In 1972 Myers and Liu showed that cl(G)≥n?(n?2m?n). Here we show that cl(G)≥ n?[n?(2m?n)(1+c²_v)^{$\text{1}\text{2}$}], where c_v is the vertex degree coefficient of variation in G, that cl(G)≥ Bondy's constructive lower bound for χ(G), and that cl(G)≥n?(n?W(G)), where W(G) is the largest positive integer such that W(G)≤d(W(G)+1) [W(G)+1 is the Welsh and Powell upper bound for χ(G)]. It is also shown that cl(G)+$\text{1}\text{3}$>n?(n?L(G))≥n?(n?λ₁) and that χ(G)≥ n?(n?λ₁); λ₁ is the largest eigenvalue of A, the adjacency matrix of G, and L(G) is a newly defined single-valued function of G similar to W(G) in its construction from the vertex degree sequence of G [L(G)≥W(G)].Finally, a ‘greedy’ lower bound for cl(G) is constructed from A and it is shown that this lower bound is never less than Bondy's bound; thereafter, for 150 random graphs with varying numbers of vertices and edge densities, the values of lower bounds given in this paper are listed, thereby illustrating that this last greedily obtained lower bound is almost always better than each of the other bounds.     

The extremal graph problem of the icosahedron

P. Turán has asked the following question:Let I¹² be the graph determined by the vertices and edges of an icosahedron. What is the maximum number of edges of a graph Gⁿ of n vertices if Gⁿ does not contain I¹² as a subgraph?We shall answer this question by proving that if n is sufficiently large, then there exists only one graph having maximum number of edges among the graphs of n vertices and not containing I¹². This graph Hⁿ can be defined in the following way:Let us divide n ? 2 vertices into 3 classes each of which contains [$\text{(n?2)}\text{3}$] or $\text{[}\text{(n?2)}\text{3}\text{] + 1}$ vertices. Join two vertices iff they are in different classes. Join two vertices outside of these classes to each other and to every vertex of these three classes.     

On complementary graphs with no isolated vertices

Let $\text{G}$ denote the complement of a graph G, and x(G), β₁(G), β₄(G), α₀(G), α₁(G) denote respectively the chromatic number, line-independence number, point-independence number, point-covering number, line-covering number of G, Nordhaus and Gaddum showed that for any graph G of order $\text{p}\text{, {2√p} ? x(G) + x(}\text{G}\text{) ? p + 1}\text{and p}\text{? x(G)·x(}\text{G}\text{) ? [(}\text{1}\text{2}\text{(p + 1))}{}^2\text{]}$. Recently Chartrand and Schuster have given the corresponding inequalities for the independence numbers of any graph G. However, combining their result with Gallai's well known formula β₁(G) + α₁(G) = ?, one is not deduce the analogous bounds for the line-covering numbers of $\text{G}\text{and}\text{G}$, since Gallai's formula holds only if G contains no isolated vertex. The purpose of this note is to improve the results of Chartland and Schuster for line-independence numbers for graphs where both $\text{G}\text{and}\text{G}$ contain no isolated vertices and thereby allowing us to use Gallai's formula to get the corresponding bounds for the line-covering numbers of G.     

Oscillatory properties of strongly superlinear differential equations with deviating arguments

We regard a graph G as a set {1,…, v} together with a nonempty set E of two-element subsets of {1,…, v}. Let p = (p₁,…, p_v) be an element of $\text{R}$^nv representing v points in $\text{R}$ⁿ and consider the realization G(p) of G in $\text{R}$ⁿ consisting of the line segments [p_i, p_j] in $\text{R}$ⁿ for {i, j} ?E. The figure G(p) is said to be rigid in $\text{R}$ⁿ if every continuous path in $\text{R}$^nv, beginning at p and preserving the edge lengths of G(p), terminates at a point q ? $\text{R}$^nv which is the image (Tp₁,…, Tp_v) of p under an isometry T of $\text{R}$ⁿ. We here study the rigidity and infinitesimal rigidity of graphs, surfaces, and more general structures. A graph theoretic method for determining the rigidity of graphs in $\text{R}$² is discussed, followed by an examination of the rigidity of convex polyhedral surfaces in $\text{R}$³.     

The asymptotic distribution of weighted empirical distribution functions

Let G_n denote the empirical distribution based on n independent uniform (0, 1) random variables. The asymptotic distribution of the supremum of weighted discrepancies between G_n(u) and u of the forms 6w_v(u)D_n(u)6 and 6w_v(G_n(u))D_n(u)6, where D_n(u) = G_n(u)?u, w_v(u) = (u(1?u))^?1+v and 0 ? v < $\text{1}\text{2}$ is obtained. Goodness-of-fit tests based on these statistics are shown to be asymptotically sensitive only in the extreme tails of a distribution, which is exactly where such statistics that use a weight function w_v with $\text{1}\text{2}$ ? v ? 1 are insensitive. For this reason weighted discrepancies which use the weight function w_v with 0 ? v < $\text{1}\text{2}$ are potentially applicable in the construction of confidence contours for the extreme tails of a distribution.     

Minimum number of edges in graphs that are both P2- and Pi-connected

A simple graph with n vertices is called P_i-connected if any two distinct vertices are connected by an elementary path of length i. In this paper, lower bounds of the number of edges in graphs that are both P₂- and P_i-connected are obtained. Namely if $\text{i?}\text{1}\text{2}\text{(n+1)}$, then |E(G)|?((4i?5)/(2i?2))(n?1), and if $\text{i >}\text{1}\text{2}\text{(n+ 1)}$, then |E(G)|?2(n?1) apart from one exeptional graph. Furthermore, extremal graphs are determined in the former.     

Extremal values of the interval number of a graph,II

It is shown that the interval number of a graph on n vertices is at most $\text{[}\text{1}\text{4}\text{(n+1)]}$, and this bound is best possible. This means that we can represent any graph on n vertices as an intersection graph in which the sets assigned to the vertices each consist of the union of at most $\text{[}\text{1}\text{4}\text{(n+1)]}$ finite closed intervals on the real line. This upper bound is attained by the complete bipartite graph K_[n/2],[n/2].     

Spheres tangent to all the faces of a simplex

There are at most 2ⁿ spheres tangent to all n + 1 faces of an n-simplex. It has been shown that the minimum number of such spheres is $\text{2}{}^{\mathrm{n}}\text{? c(n,}\text{1}\text{2}\text{(n + 1))}$ if n is odd and $\text{2}{}^{\mathrm{n}}\text{? c(n,}\text{1}\text{2}\text{(n + 1))}$ if n is even. The object of this note is to show that this result is a consequence of a theorem in graph theory.     

On subgraph number independence in trees

For finite graphs F and G, let N_F(G) denote the number of occurrences of F in G, i.e., the number of subgraphs of G which are isomorphic to F. If $\text{F}$ and $\text{G}$ are families of graphs, it is natural to ask then whether or not the quantities N_F(G), F∈$\text{F}$, are linearly independent when G is restricted to $\text{G}$. For example, if $\text{F}$ = {K₁, K₂} (where K_n denotes the complete graph on n vertices) and $\text{F}$ is the family of all (finite) trees, then of course N_K1(T) ? N_K2(T) = 1 for all T∈$\text{F}$. Slightly less trivially, if $\text{F}$ = {S_n: n = 1, 2, 3,…} (where S_n denotes the star on n edges) and $\text{G}$ again is the family of all trees, then Σ_n=1^∞(?1)ⁿ⁺¹N_Sn(T)=1 for all T∈$\text{G}$. It is proved that such a linear dependence can never occur if $\text{F}$ is finite, no F∈$\text{F}$ has an isolated point, and $\text{G}$ contains all trees. This result has important applications in recent work of L. Lovász and one of the authors (Graham and Lovász, to appear).     

Isomorphism of circulant graphs and digraphs

Let S? {1, …, n?1} satisfy ?S = S mod n. The circulant graph G(n, S) with vertex set {v₀, v₁,…, v_n?1} and edge set E satisfies v_iv_j?E if and only if j ? i ∈ S, where all arithmetic is done mod n. The circulant digraph G(n, S) is defined similarly without the restriction S = ? S. Ádám conjectured that $\text{G(n, S) ? G(n, S′)}$ if and only if S = uS′ for some unit u mod n. In this paper we prove the conjecture true if n = pq where p and q are distinct primes. We also show that it is not generally true when n = p², and determine exact conditions on S that it be true in this case. We then show as a simple consequence that the conjecture is false in most cases when n is divisible by p² where p is an odd prime, or n is divisible by 2⁴.     

Solution of the Hamiltonian problem for self-complementary graphs

A graph G is said to be highly constricted if there exists a nonempty subset S of vertices such that (i) G ? S has more than |S| components, (ii) S induces the complete graph, and (iii) for every u ∈ S and v ? S, we have d_G(u) > d_G(v), where d_G(u) denotes the degree of u in G. In this paper it is shown that a non-hamiltonian self-complementary graph G of order p is highly constricted, unless p = 4N and G is a particular graph $\text{G}{}^1\text{(4N)}$. It is also proved that if G is a self-complementary graph of order p(≥8) and π its degree sequence, then G is pancyclic if π has a realization with a hamiltonian cycle, and G has a 2-factor if π has a realization with a 2-factor, unless p = 4N and $\text{G = G}{}^1\text{(4N)}$.     

Comparability graphs and intersection graphs

A function diagram (f-diagram) D consists of the family of curves {$\text{1}\text{?}$ñ} obtained from n continuous functions $\text{f}{}_{\mathrm{i}}\text{:[0,1]→}\text{R}\text{(1?i?n)}$. We call the intersection graph of D a function graph (f-graph). It is shown that a graph G is an f-graph if and only if its complement ? is a comparability graph. An f-diagram generalizes the notion of a permulation diagram where the f_i are linear functions. It is also shown that G is the intersection graph of the concatenation of ?k permutation diagrams if and only if the partial order dimension of $\text{G}\text{?}\text{is ?k+1}$. Computational complexity results are obtained for recognizing such graphs.     

On total matching numbers and total covering numbers of complementary graphs

Best upper and lower bounds, as functions of n, are obtained for the quantities $\text{β}{}_2\text{(G)+β}{}_2\text{(}\text{G}\text{?}\text{)}$ and $\text{α}{}_2\text{(G)+α}{}_2\text{(}\text{G}\text{?}\text{)}$, where β₂(G) denotes the total matching number and α₂(G) the total covering number of any graph G with n vertices and with complementry graph ?.The best upper bound is obtained also for α₂(G)+β₂(G), when G is a connected graph.     

Ordered colourings of graphs

An ordered colouring of a graph with k colours is a vertex colouring with colours {1, 2,…,k} such that each vertex coloured j is joined to at least one vertex-of colour i for each i less than j. Examples of ordered colourings are those produced by the greedy colouring algorithm. Some properties are investigated of τ(G), the maximum k for which the graph G has an ordered k-colouring (in fact τ(G) is an upper bound for the number of colours used by the greedy algorithm). It is shown that if G has order n, then $\text{τ(G) + τ(}\text{G}\text{) ≤ [}\text{(5n + 2)}\text{4}\text{]}$.     

Connected matroids with the smallest whitney numbers

In “The Slimmest Geometric Lattices” (Trans. Amer. Math. Soc.). Dowling and Wilson showed that if G is a combinatorial geometry of rank r(G) = n, and if _X(G) = Σμ(0, x)λ^{r ? r(x)} = Σ (?1)^{r ? k}W_kλ^k is the characteristic polynomial of G, then

$\text{w}{}_{\mathrm{k}}\text{?}\text{r}\text{k}\text{+n}\text{r?1}\text{k}$

Thus γ(G) ? 2^{r ? 1} (n+2), where γ(G) = Σw_k. In this paper we sharpen these lower bounds for connected geometries: If G is connected, r(G) ? 3, and n(G) ? 2 ((r, n) ≠ (4,3)), then

$\text{w}{}_{\mathrm{i}}\text{?}\text{r}\text{i}\text{+ n}\text{r}\text{i+1}\text{for i>1; w}{}_1\text{?r+n}\text{r}\text{2}\text{? 1;}$

|μ| ? (r? 1)n; and γ ? (2^{r ? 1} ? 1)(2n + 2). These bounds are all achieved for the parallel connection of an r-point circuit and an (n + 1)point line. If G is any series-parallel network, $\text{r(G) = r(}\text{G}\text{?}\text{) = 4}$, and $\text{n(G) = n(}\text{G}\text{?}\text{) = 3}$ then $\text{(w}{}_1\text{(G))}{}^4{}_{\mathrm{t-G}}\text{? (w}{}_1\text{(}\text{G}\text{?}\text{)) = (8, 20, 18, 7, 1)}$. Further, if β is the Crapo invariant,

$\text{β(G)=}\text{d}{}_{\mathrm{X}}\text{(G)}\text{dλ}\text{(1)}\text{,}$

then β(G) ? max(1, n ? r + 2). This lower bound is achieved by the parallel connection of a line and a maximal size series-parallel network.     

A characterization of robert's inequality for boxicity

F.S. Roberts defined the boxicity of a graph G as the smallest positive integer n for which there exists a function F assigning to each vertex x?G a sequence F(x)(1),F(x)(2),…, F(x)(n) of closed intervals of R so that distinct vertices x and y are adjacent in G if and only if F(x)(i)∩F(y)(i)≠? fori = 1, 2, 3, …, n. Roberts then proved that if G is a graph having 2n + 1 vertices, thentheboxicityofGisatmostn. In this paper, we provide an explicit characterization of this inequality by determining for each n ? 1 the minimum collection $\text{C}$_n of graphs so that a graph G having 2n + 1 vertices has boxicity n if and only if it contains a graph from $\text{C}$_n as an induced subgraph. We also discuss combinatorial connections with analogous characterization problems for rectangle graphs, circular arc graphs, and partially ordered sets.     

On the minimum order of graphs with given semigroup

Denote by M(n) the smallest positive integer such that for every n-element monoid M there is a graph G with at most M(n) vertices such that End(G) is isomorphic to M. It is proved that $\text{2}\text{(1 + o(1))n}\text{log}{}_2\text{n}\text{≤M(n)≤n ·}\text{2n}\text{+ O(n)}$. Moreover, for almost all n-element monoids M there is a graph G with at most 12 · n · log₂n + n vertices such that End(G) is isomorphic to M.     

Omega theorems for the iterated additive convolution of a nonnegative arithmetic function

Let R_k(n) denote the number of ways of representing the integers not exceeding n as the sum of k members of a given sequence of nonnegative integers. Suppose that $\text{1}\text{2}\text{< β < k}$, $\text{δ =}\text{β}\text{2}\text{?}\text{β}\text{(4}\text{min}\text{(β,}\text{k}\text{2}\text{))}$ and

$\text{ξ=}\text{1/2β if β<k/2,}\text{β?1/2 if β=1/2,}\text{(k ? 2)(k + 1)/2k if k/2<β<k.}$

R. C. Vaughan has shown that the relation R_k(n) = G(n) + o(n^δ log^?ξn) as n → +∞ is impossible when G(n) is a linear combination of powers of n and the dominant term of G(n) is cn^β, c > 0. P. T. Bateman, for the case k = 2, has shown that similar results can be obtained when G(n) is a convex or concave function. In this paper, we combine the ideas of Vaughan and Bateman to extend the theorems stated above to functions whose fractional differences are of one sign for large n. Vaughan's theorem is included in ours, and in the case $\text{β <}\text{k}\text{2}$ we show that a better choice of parameter improves Vaughan's result by enabling us to drop the power of log n from the estimate of the error term.     

Characterization of quadratic and cubic σ-polynomials

Here quadratic and cubic σ-polynomials are characterized, or, equivalently, chromatic polynomials of the graphs of order p, whose chromatic number is p ? 2 or p ? 3, are characterized. Also Robert Korfhage's conjecture that if σ² + bσ + a is a σ-polynomial then $\text{a ≤}\text{1}\text{2}\text{(b}{}^2\text{? 5b + 12)}$ is verified. In general, if σ(G) = Σ₀ⁿa_iσⁱ is a σ-polynomial of a graph G, then a_n?2 is determined.