Wiki
Wiki

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

Updated


Statement

Proposition (p. 190, unnumbered, quoted). "If qq is even and GG a quadrilateral-free graph on q2+q+1q^2+q+1 points, then ∣E(G)∣⩽12q(q+1)2|E(G)|\leqslant\frac12q(q+1)^2."

Here a graph is finite and simple, and quadrilateral-free means it contains no cycle of length four as a subgraph, so the Proposition says ex⁡(q2+q+1,C4)≤q(q+1)2/2\operatorname{ex}(q^2+q+1,C_4)\leq q(q+1)^2/2 for every even qq. It is an upper bound only: no prime-power hypothesis is made, and no matching construction is claimed for even qq that is not a prime power.

The paper introduces the Proposition in Section 4 (Remarks) as what the proof of the Theorem in fact establishes.

Equality case (announced, not proved). In the same section the paper says it can show, with the proof omitted for brevity, that equality holds in the Proposition if and only if the projective plane on q2+q+1q^2+q+1 points has a polarity with q+1q+1 fixed points (for example q=2kq=2^k) and GG is the Erdős-Rényi polarity graph. This characterization is an announcement; the paper gives no proof of it.

Source. Z. Füredi, Graphs without Quadrilaterals, J. Combin. Theory Ser. B 34 (1983), 187-190: Section 4, the unnumbered Proposition on p. 190. The edition read is identified on the source card.

Read depth. Claims checked: the statement and the announcement were read clause by clause on the printed page. No independent proof review is recorded.

Proof pointer

The proof of the Theorem on p. 188, whose upper-bound part uses only that qq is even: the Lemma handles maximum degree at least q+2q+2, and when the maximum degree is at most q+1q+1, the evenness of qq shows that every vertex of degree q+1q+1 has a neighbour of degree at most qq, so at least q+1q+1 vertices have degree at most qq.

Dependencies

Lemma (p. 188) of the same paper.

Bears on

  • Problem 765: an upper bound on ex⁡(n;C4)\operatorname{ex}(n;C_4) at the orders n=q2+q+1n=q^2+q+1 with qq even, matching the polarity-graph lower bound when qq is also a prime power. It concerns those orders only and does not by itself give the asymptotic formula the problem asks for.