Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated


Source. Example 1, p. 22, of G.-T. Chen, P. Erdős and E. T. Ordman, Clique partitions of split graphs, in: Y. Alavi, D. R. Lick and J. Liu (eds.), Combinatorics, Graph Theory, Algorithms and Applications (Beijing, 1993), World Scientific, Singapore, 1994, pp. 21--30; the edition read is identified on the source card.

Statement

Setting (pp. 21--22). cp⁡(G)\operatorname{cp}(G) is the least number of cliques of GG containing each edge of GG exactly once. Gn=Kn−Kˉ2n/3G_n=K_n-\bar K_{2n/3} is the split graph on nn vertices with n/3n/3 vertices in the clique and 2n/32n/3 in the independent set, all 2n2/92n^2/9 connecting edges present; it is also a threshold graph (p. 22).

Example 1 (p. 22, quoted). "The clique partition number of Gn=Kn−Kˉ2n/3G_n=K_n-\bar K_{2n/3} is n2/6+n/6n^2/6+n/6, provided 6 divides nn."

The paper introduces the example (p. 22) to show that cp⁡\operatorname{cp} of a chordal graph on nn vertices can exceed n2/6n^2/6 by a term linear in nn, and calls the construction well known; the abstract says that a split graph on nn vertices "may require as many as n2/6+n/6n^2/6+n/6 cliques" (p. 21). When 6∤n6\nmid n the paper says the value grows by a term linear in nn and thereafter writes its bounds with O(n)O(n) (p. 23).

Read depth. Claims checked: the statement, the construction and the attribution of minimality were read on the page images (pp. 21--23). The minimality is not proved in the paper and was not checked here. Nothing here is independently reviewed.

Proof pointer

Upper bound, p. 22: the clique Kn/3K_{n/3} splits into n/3−1n/3-1 perfect matchings of n/6n/6 edges each (this uses 6∣n6\mid n); joining each matching to its own independent vertex turns each of the j=n3(n3−1)/2j=\tfrac n3(\tfrac n3-1)/2 clique edges into the base of a triangle, and the 2n2/9−2j2n^2/9-2j connecting edges left over are taken singly, for j+2n2/9−2j=n2/6+n/6j+2n^2/9-2j=n^2/6+n/6 cliques. Lower bound: the paper cites its references [7] (Erdős, Faudree and Ordman, Discrete Math. 72 (1988)) and [14] (Pullman and Donald, Utilitas Math. 19 (1981)) and gives an informal counting reason: connecting edges can be merged into larger cliques only by spending clique edges, and one triangle per clique edge is the best rate.

Dependencies

The minimality rests on P. Erdős, R. Faudree and E. Ordman, Clique coverings and clique partitions, Discrete Math. 72 (1988), 93--101, or N. J. Pullman and A. Donald, Clique coverings of graphs -- II: Complements of cliques, Utilitas Math. 19 (1981), 207--213, as the paper cites them.

Bears on

Problem 81 asks whether every chordal graph on nn vertices has a clique partition into n2/6+O(n)n^2/6+O(n) cliques. GnG_n is split, hence chordal (p. 22), and needs n2/6+n/6n^2/6+n/6 cliques, so the constant 16\tfrac16 in the question cannot be lowered; the example says nothing about the upper bound.