Wiki
Wiki

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

Updated


Statement

g2(n,m)g_2(n,m) is the largest size of a set of edges, every two on a 44-cycle of the whole graph, that every graph with nn vertices and mm edges contains for all sufficiently large nn (p. 269; see Theorem 3).

Theorem 5 (p. 271). For each function γ=γ(n)\gamma=\gamma(n) with γ(n)=o(n3/2)\gamma(n)=\mathrm o(n^{3/2}),

g2(n,(n2)−γ(n))≥(1−o(1))(n2).g_2\Bigl(n,\binom n2-\gamma(n)\Bigr)\ge(1-\mathrm o(1))\binom n2.

Source. Richard A. Duke, Paul Erdős and Vojtěch Rödl, Cycle-connected graphs, Discrete Math. 108 (1992), 261--278, doi:10.1016/0012-365X(92)90680-E; Lemma 4 on printed p. 270 and Theorem 5 on p. 271, read on the page images of the publisher's scan. The edition read is identified in the source digest.

Read depth. Claims checked: the statement and that of Lemma 4 were read clause by clause on the page images. No proof is printed beyond Lemma 4's, which was read for structure only.

Proof pointer

The paper calls the theorem a direct consequence of Lemma 4 (p. 271): with γ=o(n3/2)\gamma=\mathrm o(n^{3/2}) the deficit 32(2γ)2/3n\frac32(2\gamma)^{2/3}n there is o(n2)\mathrm o(n^2).

Dependencies

Lemma 4 (p. 270). For each choice of the function γ(n)\gamma(n),

g2(n,(n2)−γ(n))≥(n2)−32(2γ(n))2/3n(1+o(1)).g_2\Bigl(n,\binom n2-\gamma(n)\Bigr)\ge\binom n2-\tfrac32(2\gamma(n))^{2/3}n(1+\mathrm o(1)).

Its proof (pp. 270--271) shows the more general bound (10), g2(n,(n2)−γ)≥(n2)−γ−(2γ/ϵ+(ϵ2))ng_2(n,\binom n2-\gamma)\ge\binom n2-\gamma-(2\gamma/\epsilon+\binom{\epsilon}2)n for any ϵ(n)\epsilon(n), by discarding the edges at vertices meeting more than ϵ(n)\epsilon(n) deleted edges and the edges whose two ends are joined by deleted edges to one of the remaining vertices; the rest form a C4C_4-connected set, and ϵ=(2γ)1/3\epsilon=(2\gamma)^{1/3} gives the lemma.

Bears on

No problem page is reached by this theorem: it concerns 44-cycles in graphs missing o(n3/2)\mathrm o(n^{3/2}) edges, and no problem the corpus records asks about them.