6
$\begingroup$

The question is: A graph G contains no 5-cliques and every two 3-cliques intersect in at least one vertex. Show that we can delete two vertices such that the resulting graph contains no 3-cliques.

I tried looking at ways of drawing 3-cliques such that every two intersect but more of them do not intersect in the same two points. The 5-clique is obviously a counter-example, but how can I show that if there is no five clique and all 3-cliques are non-disjoint then all of them contain (at least) one of two points {a,b}?

Maybe by looking at the 3-cliques as sets of vertices and then looking at the intersections between pairs of 3-cliques? But it only needs to contain one of the two points, so the intersection is too strong.

$\endgroup$
1
  • $\begingroup$ I would try to make a bipartite graph with one side being the triangles and the other side being the vertices of the graph used in the triangle, and do induction on the number of triangles in the graph to whack out all the cases. $\endgroup$
    – koifish
    Commented Jul 8 at 7:28

1 Answer 1

1
$\begingroup$

Assume, on the contrary, that there exists a graph $G$, $|V(G)|>1$ in which
(i) any two triangles have at least one vertex in common;
(ii) there are no $5$-cliques;
(iii) for any two vertices of $V(G)$, there exists a triangle without these two vertices.

In view (iii) and $|G|>1$, there exists triangles $abc$ and $axy$ in $G$ and the vertices $a,b,c,x,y$ are pairwise distinct (why?).

In view (ii) $abcxy$ is not a clique. Without loss of generality we can assume that $bx\notin E(G)$. Hence again in view of (iii), (i) for the pair of vertices $a,c$ there exists a triangle $buz$ in $G$. If $\{u,z\}\cap\{x,y\}=\varnothing$, then triangles $axy$ and $buz$ have no common vertices, which is impossible by (i). Let $u=y$. Then $z\notin\{x,y\}$.
Thus we have the subgraph of $G$ shown in Figure 1.

enter image description here

By (i) and (iii) for a pair $a,y$ there exists a triangle $bxw$ for some $w\in V(G)$.
But this is impossible, since according to our assumption $bx\notin E(G)$.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .