Search

Let G=(V,E) be a graph. For r≥1, let be the family of independent vertex r-sets of G. For vV(G), let denote the star . G is said to be r-EKR if there exists vV(G) such that for any non-star family of pair-wise intersecting sets in . If the inequality is strict, then G is strictly r-EKR. Let Γ be the family of graphs that are disjoint unions of complete graphs, paths, cycles, including at least one singleton. Holroyd, Spencer and Talbot proved that, if GΓ and 2r is no larger than the number of connected components of G, then G is r-EKR. However, Holroyd and Talbot conjectured that, if G is any graph and 2r is no larger than μ(G), the size of a smallest maximal independent vertex set of G, then G is r-EKR, and strictly so if 2r<μ(G). We show that in fact, if GΓ and 2r is no larger than the independence number of G, then G is r-EKR; we do this by proving the result for all graphs that are in a suitable larger set Γ′Γ. We also confirm the conjecture for graphs in an even larger set Γ″Γ′.

For a graph G, vertex v of G and integer r >= 1, we denote the family of independent r-sets of V(G) by I^(r)(G) and the subfamily {A in I^(r)(G): v in A} by I^(r)_v(G); such a family is called a star. Then, G is said to be r-EKR if no intersecting subfamily of I^(r)(G) is larger than the largest star in I^(r)(G). If every intersecting subfamily of I^(r)_v(G) of maximum size is a star, then G is said to be strictly r-EKR. We show that if a graph is r-EKR then its lexicographic product with any complete graph is r-EKR. For any graph G, we define mu(G) to be the minimum size of a maximal independent vertex set. We conjecture that, if 1 <= r <= 1/2 mu(G), then G is r-EKR, and if r < 1/2 mu(G), then G is strictly r-EKR. This is known to be true when G is an empty graph, a cycle, a path or the disjoint union of complete graphs. We show that it is also true when G is the disjoint union of a pair of complete multipartite graphs.

For a graph G and integer r >= 1 we denote the collection of independent r-setsof G by I^(r)(G). If v is in V(G) then I^(r)_v(G) is the collection of all independent r-sets containing v. A graph G is said to be r-EKR, for r >= 1, iff no intersecting family A of I^(r)(G) is larger than max_{v in V(G)} |I^(r)_v(G)|. There are various graphs that are known to have this property; the empty graph of order n >= 2r (this is the celebrated Erdos-Ko-Rado theorem), any disjoint union of atleast r copies of K_t for t >= 2, and any cycle. In this paper we show how these results can be extended to other classes of graphs via a compression proof technique. In particular we extend a theorem of Berge, showing that any disjoint union of at least r complete graphs, each of order at least two, is r-EKR. We also show that paths are r-EKR for all r >= 1.

Let S be a Steiner triple system and G a cubic graph. We say that G is S-colourable if its edges can be coloured so that at each vertex the incident colours form a triple of S. We show that if S is a projective system PG(n, 2), n >= 2, then G is S-colourable if and only if it is bridgeless, and that every bridgeless cubic graph has an S-colouring for every Steiner triple system of order greater than 3. We establish a condition on a cubic graph with a bridge which ensures that it fails to have an S-colouring if S is an affine system, and we conjecture that this is the only obstruction to colouring any cubic graph with any non-projective system of order greater than 3.

For a graph G, vertex v of G and integer r >= 1, we denote the family of independent r-sets of V(G) by I^(r)(G) and the subfamily {A in I^(r)(G): v in A} by I^(r)_v(G); such a family is called a star. Then, G is said to be r-EKR if no intersecting subfamily of I^(r)(G) is larger than the largest star in I^(r)(G). If every intersecting subfamily of I^(r)_v(G) of maximum size is a star, then G is said to be strictly r-EKR. We show that if a graph is r-EKR then its lexicographic product with any complete graph is r-EKR. For any graph G, we define mu(G) to be the minimum size of a maximal independent vertex set. We conjecture that, if 1 <= r <= 1/2 mu(G), then G is r-EKR, and if r < 1/2 mu(G), then G is strictly r-EKR. This is known to be true when G is an empty graph, a cycle, a path or the disjoint union of complete graphs. We show that it is also true when G is the disjoint union of a pair of complete multipartite graphs.

For a graph G and integer r >= 1 we denote the collection of independent r-setsof G by I^(r)(G). If v is in V(G) then I^(r)_v(G) is the collection of all independent r-sets containing v. A graph G is said to be r-EKR, for r >= 1, iff no intersecting family A of I^(r)(G) is larger than max_{v in V(G)} |I^(r)_v(G)|. There are various graphs that are known to have this property; the empty graph of order n >= 2r (this is the celebrated Erdos-Ko-Rado theorem), any disjoint union of atleast r copies of K_t for t >= 2, and any cycle. In this paper we show how these results can be extended to other classes of graphs via a compression proof technique. In particular we extend a theorem of Berge, showing that any disjoint union of at least r complete graphs, each of order at least two, is r-EKR. We also show that paths are r-EKR for all r >= 1.

Let S be a Steiner triple system and G a cubic graph. We say that G is S-colourable if its edges can be coloured so that at each vertex the incident colours form a triple of S. We show that if S is a projective system PG(n, 2), n >= 2, then G is S-colourable if and only if it is bridgeless, and that every bridgeless cubic graph has an S-colouring for every Steiner triple system of order greater than 3. We establish a condition on a cubic graph with a bridge which ensures that it fails to have an S-colouring if S is an affine system, and we conjecture that this is the only obstruction to colouring any cubic graph with any non-projective system of order greater than 3.