Linial-Berge path partition duality

High ★★★ Graph Theory » Coloring

Status partial high confidence

The Linial-Berge path partition duality conjecture remains open for general directed graphs. After the problem was posted, the conjecture was proved for $k=2$ (Berger and Hartman, 2008) and for locally in/out-semicomplete digraphs (Sambinelli et al., 2019). A 2026 arXiv paper (de Paula Silva, da Silva, and Lee) establishes relaxations of both Linial conjectures by replacing stable sets with induced acyclic subdigraphs, but the original conjecture for arbitrary $k$ and general digraphs remains open.

Cited literature (3)

Reviewer notes. Multiple papers about a different 'Berge α-diperfect conjecture' (which concerns anti-directed odd cycles as forbidden induced subdigraphs) appeared in searches; these are distinct from the Linial-Berge path partition duality and were excluded. The Herskovics paper proving the conjecture for k ≥ λ-3 (published ca. 2015 in Discrete Applied Mathematics, pii S0166218X15003923) is mentioned in multiple reliable secondary sources (EGRES Open, Discrete Math survey) but could not be directly verified via WebFetch (ScienceDirect and ResearchGate both returned 403); it is therefore omitted from since_posted but is credible. The DOI 10.1007/s00373-019-02046-x for the Sambinelli et al. paper was found in search results and matches the Springer Graphs and Combinatorics 2019 link; the arXiv version was directly verified.

Auto-reviewed 2026-05-08 with claude-sonnet-4-6 (worker 02) (web search enabled).

Conjecture. The minimum $ k $ -norm of a path partition on a directed graph $ D $ is no more than the maximal size of an induced $ k $ -colorable subgraph.
Keywords: coloring · directed path · partition

Discussion

Definitions: Let $ D $ be a directed graph. A path partition of $ D $ is a set of vertex disjoint paths in it (some might be singletons), covering all vertices. Let $ k $ be a positive integer. The $ k $ norm of a path partition is the sum of $ \min\{|V(P)|,k\} $ for all paths $ P $ in it. This conjecture is known for acyclic graphs and for $ k =1,2 $ .