MathLabs

Worked solution: A rainbow-tree proof of Ringel's conjecture (2020)

Step 1 of 8: Ringel's 1963 conjecture: packing a complete graph with trees
In plain words

A tree with nn edges has n+1n+1 vertices and no cycles -- the simplest kind of connected shape. Ringel asked: can the complete graph K2n+1K_{2n+1}, where every pair among 2n+12n+1 points is joined, always be cut up perfectly into 2n+12n+1 copies of one chosen tree, using every edge exactly once? Walecki had solved the special case where the tree is just a long path back in 1882; Ringel's 1963 conjecture says the same works no matter how the tree branches.

K2n+1 decomposes into 2n+1 edge-disjoint copies of any tree T with n edgesK_{2n+1} \text{ decomposes into } 2n+1 \text{ edge-disjoint copies of any tree } T \text{ with } n \text{ edges}
Detailed analysis

Montgomery, Pokrovskiy and Sudakov (2021, Introduction) recall that Ringel posed Conjecture 1.1 in 1963: K2n+1K_{2n+1} can be decomposed into copies of any tree with nn edges. This is one of the oldest and best-known open problems on graph decompositions, verified previously only for special tree shapes (caterpillars, trees with at most 44 leaves, trees with at most 3535 vertices, and others) or with an approximate/bounded-degree restriction (Joos-Kim-Kuhn-Osthus proved it for bounded-degree trees; Ferber-Samotij and Adamaszek-Allen-Grosu-Hladky proved approximate versions for maximum degree O(n/log⁡n)O(n/\log n)). This paper's main theorem removes every restriction on the tree's shape or degree: for every sufficiently large nn, K2n+1K_{2n+1} decomposes into 2n+12n+1 copies of any tree TT with nn edges (Theorem 1.2).

Terms in this step
Graph decomposition
Partitioning the edges of a graph GG into edge-disjoint subgraphs, each isomorphic to a fixed graph HH, so every edge of GG belongs to exactly one copy of HH.
Tree
A connected graph with no cycles; a tree with nn edges automatically has n+1n+1 vertices.
Knowledge used in this step