MathLabs

Worked solution: Kang–Kelly–Kühn–Methuku–Osthus asymptotic proof via iterative absorption (2021)

Step 8 of 8: Closing the loop: the Erdős–Faber–Lovász conjecture holds for large nn
In plain words

Every piece of the puzzle is now in place: whether the linear hypergraph H\mathcal{H} resembles a finite projective plane, a near-complete graph KnK_n, a near-pencil, or sits far away from all three extremal shapes, the stratified colouring pipeline packs all its hyperedges into at most nn colour classes.

What Erdős, Faber and Lovász posed at a party in 1972 as a seemingly modest test problem for hypergraph theory is thus resolved for all sufficiently large nn by uniting fifty years of probabilistic combinatorics -- the Rödl nibble, locally sparse graph colouring, vertex absorption, and 11-factorisation.

∀ n≥n0, ∀ linear H on n vertices:χ′(H)≤n\forall\, n \ge n_0,\ \forall\, \text{linear } \mathcal{H} \text{ on } n \text{ vertices}: \quad \chi'(\mathcal{H}) \le n
Detailed analysis

Kang, Kelly, Kühn, Methuku and Osthus (2023, Section 8) combine the case divisions of Sections 5, 6, and 7 to complete the proof of Theorem 1.1: there exists n0n_0 such that for every n≥n0n \ge n_0, every linear hypergraph H\mathcal{H} on nn vertices has chromatic index χ′(H)≤n\chi'(\mathcal{H}) \le n. By duality, this simultaneously proves all three equivalent formulations from Section 1.1: edge-colouring linear hypergraphs, vertex-colouring unions of nn cliques of size at most nn sharing at most one vertex pairwise, and rainbow-type set colourings of nn sets of size nn with pairwise intersection at most 11.

As the authors note in Section 1.2, every probabilistic step of the proof can also be derandomised into a randomized polynomial-time colouring algorithm for large nn.

Knowledge used in this step