Fractional 3-coloring of triangle-free planar requests
Problem 1 · arXiv:1702.00588
Status disproved medium confidence
The source paper (1702.00588) proves in Theorem 1 that Problem 1 (RGEN: existence of \alpha>0 such that every planar triangle-free request graph admits a 3-coloring satisfying an \alpha-fraction of its requests) is equivalent to Thomassen's conjecture (EXP: every planar triangle-free graph on n vertices has at least 2^{\beta n} 3-colorings for some \beta>0). EXP has since been disproved: Dvo\v{r}\'{a}k (arXiv:2108.12669, 2021) constructs planar triangle-free graphs with at most 2^{6n^{0.731}} 3-colorings (sub-exponential), improving on Thomassen's own earlier counterexample to his conjecture. By the equivalence RGEN \Leftrightarrow EXP, no positive real \alpha of the type sought in Problem 1 can exist.
Cited literature (2)
-
For a related list-4-coloring setting, proves that every triangle-free planar graph with all lists of size at least 4 admits an L-coloring satisfying a constant fraction of vertex color preferences; does not directly resolve the 3-coloring request graph formulation of Problem 1.
-
Constructs planar triangle-free graphs with sub-exponentially many 3-colorings, disproving Thomassen's EXP conjecture; since RGEN \Leftrightarrow EXP by Theorem 1 of the source paper, this indirectly disproves Problem 1.
Reviewer notes. The disproof of Problem 1 is indirect: no paper provides an explicit counterexample to a request graph instance, but the logical chain is rigorous since RGEN \Leftrightarrow EXP is Theorem 1 of the source paper and EXP is disproved by 2108.12669 (and an earlier, unverified counterexample by Thomassen himself). The doi 10.1002/jgt.22634 for 1902.02971 is inferred from the Wiley Online Library URL in search results (direct fetch returned HTTP 402). The flexibility result (1902.02971) represents a positive partial result for a related but strictly different problem.
Context
This problem arises from the work of Asadi et al. [1] and is introduced as a reformulation vehicle for Thomassen's exponential-coloring conjecture. A request graph $(G, R_=, R_{\neq}, w)$ consists of a planar triangle-free graph $G$ together with degree-two vertices partitioned into equality requests $R_=$ and inequality requests $R_{\neq}$ with positive rational weights; a coloring satisfies a request if the two neighbors receive equal (resp. different) colors.
Source paper
Do triangle-free planar graphs have exponentially many 3-colorings?
Zdeněk Dvořák, Jean-Sébastien Sereni · 2017-09-19
https://arxiv.org/abs/1702.00588
PDF source