Single Conflict Coloring in Graphs & Hypergraphs

• Single Conflict Coloring is a graph and hypergraph paradigm that defines colorings by avoiding locally assigned forbidden color pairs.
• It employs probabilistic techniques, degeneracy-based orientations, and the Lovász Local Lemma to establish bounds like O(√d log n) for d-degenerate graphs.
• The concept generalizes separation and adaptable choosability, linking structural combinatorics with conflict-free and distributed coloring strategies.

Single conflict coloring is a contemporary graph and hypergraph coloring paradigm synthesizing local orientation, list coloring, and forbidden pair frameworks. The core objective is to determine the chromatic threshold under which, for any local assignment of forbidden color pairs (or tuples in hypergraphs), there exists a feasible vertex coloring that avoids all induced edge conflicts. This concept interfaces with separation choosability, adapted choosability, and several conflict-free coloring models, providing a unifying lens for both structural combinatorics and algorithmic graph theory.

1. Fundamental Definitions and Notions

Let $G = (V, E)$ be a (multi)graph, potentially with parallel edges (no loops). A local $k$-partition assigns to each vertex $v$ a labeling function on incident edges: $c_v \colon E(v) \to \{1, 2, \dots, k\}$ where $E(v)$ is the set of edges incident with $v$. Each edge $e=uv$ thus possesses a pair of local colors $(c_u(e), c_v(e))$.

A coloring $\varphi\colon V \to \{1, \dots, k\}$ is a conflict $\{c_v\}$-coloring (or single-conflict coloring) if for every edge $k$0, it holds that

$k$1

The single-conflict chromatic number $k$2 (also denoted $k$3, $k$4, or $k$5 in the literature) is the minimum $k$6 such that for every local $k$7-partition, such a coloring exists. For $k$8-uniform hypergraphs $k$9, a local $v$0-partition gives each vertex $v$1 a map $v$2, so each edge $v$3 receives an $v$4-tuple $v$5; $v$6 avoids $v$7 if for some $v$8, $v$9, i.e., the color vector avoids matching the forbidden conflict tuple on $c_v \colon E(v) \to \{1, 2, \dots, k\}$0 (Dvořák et al., 2018, Casselgren et al., 11 Jan 2026, Casselgren et al., 17 Sep 2025).

This coloring type generalizes separation choosability and adaptable choosability:

• Adaptable choosability $c_v \colon E(v) \to \{1, 2, \dots, k\}$1: requires for every $c_v \colon E(v) \to \{1, 2, \dots, k\}$2-list assignment and $c_v \colon E(v) \to \{1, 2, \dots, k\}$3 edge-coloring, a vertex coloring avoiding $c_v \colon E(v) \to \{1, 2, \dots, k\}$4 on any edge $c_v \colon E(v) \to \{1, 2, \dots, k\}$5.
• Separation choosability $c_v \colon E(v) \to \{1, 2, \dots, k\}$6: for every $c_v \colon E(v) \to \{1, 2, \dots, k\}$7-list assignment with $c_v \colon E(v) \to \{1, 2, \dots, k\}$8 on $c_v \colon E(v) \to \{1, 2, \dots, k\}$9, $E(v)$0 is $E(v)$1-colorable.

Inequalities $E(v)$2 always hold (Dvořák et al., 2018, Casselgren et al., 17 Sep 2025).

2. Structural and Extremal Results

Degeneracy and General Upper Bounds

For $E(v)$3-degenerate graphs with $E(v)$4 vertices and edge multiplicity $E(v)$5, Bradshaw–Masařík proved: $E(v)$6 For general multiplicity $E(v)$7, the best bound is

$E(v)$8

where $E(v)$9 is the maximum degree (Bradshaw et al., 2021). The proof leverages degeneracy-based orientation, random inventory assignment, and the Lovász Local Lemma (and Moser–Tardos algorithmic LLL for constructivity).

A prior approach for topological surfaces established (for Euler genus $v$0 and simple graphs): $v$1 and more generally for multigraphs with $v$2 edges and maximum multiplicity $v$3: $v$4 These bounds are sharp up to logarithmic factors—complete graphs have $v$5, and for complete multigraphs the lower bound is $v$6 (Dvořák et al., 2018).

Classical Families

• Planar Graphs: If $v$7 is planar, $v$8 via orientation arguments (as planar graphs are 3-degenerate) (Dvořák et al., 2018). For single-conflict (open-neighborhood conflict-free) coloring, tight results are known: $v$9 for all planar graphs, with the extremal bounds established using structure-theoretic decompositions and contractions, and explicit inductive arguments for outerplanar graphs yielding $e=uv$0 (Bhyravarapu et al., 2019).
• Kneser Graphs: For $e=uv$1, $e=uv$2 for $e=uv$3, both upper and lower bounds shown via explicit coloring and hitting-set constructions (Bhyravarapu et al., 2019).
• String Graphs: For string graph $e=uv$4 with classical chromatic number $e=uv$5, $e=uv$6 (Keller et al., 2017).

Graph Class	$e=uv$7 or $e=uv$8 Upper Bound	Reference
$e=uv$9-degenerate	$(c_u(e), c_v(e))$0	(Bradshaw et al., 2021)
Planar	$(c_u(e), c_v(e))$1 ($(c_u(e), c_v(e))$2), $(c_u(e), c_v(e))$3 ($(c_u(e), c_v(e))$4)	(Dvořák et al., 2018, Bhyravarapu et al., 2019)
Outerplanar	$(c_u(e), c_v(e))$5 ($(c_u(e), c_v(e))$6)	(Bhyravarapu et al., 2019)
Kneser $(c_u(e), c_v(e))$7	$(c_u(e), c_v(e))$8 (if $(c_u(e), c_v(e))$9)	(Bhyravarapu et al., 2019)
Embeddable genus $\varphi\colon V \to \{1, \dots, k\}$0	$\varphi\colon V \to \{1, \dots, k\}$1	(Dvořák et al., 2018)

3. Relations to Adaptable and Separation Choosability

A local $\varphi\colon V \to \{1, \dots, k\}$2-partition gives rise to conflict tuples corresponding to list assignments and forbidden edge labels. The following chain holds for every (multi)graph $\varphi\colon V \to \{1, \dots, k\}$3: $\varphi\colon V \to \{1, \dots, k\}$4 There exist graphs (for every $\varphi\colon V \to \{1, \dots, k\}$5) with arbitrarily large gaps between $\varphi\colon V \to \{1, \dots, k\}$6, $\varphi\colon V \to \{1, \dots, k\}$7, $\varphi\colon V \to \{1, \dots, k\}$8, and $\varphi\colon V \to \{1, \dots, k\}$9 (Casselgren et al., 17 Sep 2025). For example, $\{c_v\}$0 has $\{c_v\}$1 substantially less than $\{c_v\}$2. Constructions exist where $\{c_v\}$3, and explicit planar multigraphs with $\{c_v\}$4 (Casselgren et al., 17 Sep 2025).

Within planar graphs, a no-gap theorem precludes $\{c_v\}$5 except as described above, and $\{c_v\}$6 remains open for simple planars.

4. Algorithmic and Probabilistic Techniques

Constructions for $\{c_v\}$7 colorings invoke randomized inventory assignment per vertex, deletion of colors violating forbidden pairs, and application of LLL to control the exponentially small probability of failure at each vertex (Bradshaw et al., 2021). Algorithmic versions use the Moser–Tardos resampling method, with local event dependencies and efficient expected runtime.

Distributed variants, as in the LOCAL model, frame single-conflict coloring as a two-round list/color-reduction process; for maximum outdegree $\{c_v\}$8, list sizes of $\{c_v\}$9 suffice for deterministic completion in two rounds, matching earlier Linial-style color reduction but for single-conflict constraints (Maus et al., 2020).

For $k$00-uniform hypergraphs, threshold-type results are governed by the balance between edge density and available palette size; for random conflict-tuple assignments, the sharp threshold for $k$01 is at $k$02. For general $k$03, palette sizes $k$04 suffice for $k$05-degenerate $k$06-uniforms (Casselgren et al., 11 Jan 2026).

5. Generalizations and Variants

Hypergraphs

Single-conflict coloring generalizes naturally. Each vertex $k$07 in $k$08-uniform $k$09 is assigned $k$10; a coloring $k$11 is proper if, for every $k$12, $k$13. The single-conflict chromatic number $k$14 denotes the global minimum $k$15 allowing all local $k$16-partitions to be avoided. Threshold phenomena for random conflict assignments parallel random $k$17-SAT scaling (Casselgren et al., 11 Jan 2026).

Variants

• Conflict-free coloring on open/closed neighborhoods: For open neighborhoods, the conflict-free chromatic number $k$18 is the minimum $k$19 so that every $k$20 has a neighbor whose color is unique in its open neighborhood. This is NP-complete and admits sharp planar, outerplanar, and Kneser-type bounds (Bhyravarapu et al., 2019).
• Proper conflict-free coloring: Requires a proper coloring with additional uniqueness-of-color-in-neighbor constraint, with explicit bounds for planar large-girth graphs achieved via greedy augmentations and discharge (Anderson et al., 2024).
• Strong conflict-free coloring: For $k$21, coincides with the perfect dominating set; algorithms parameterized by treewidth are fixed-parameter tractable (Agrawal et al., 2019).

6. Open Problems and Future Directions

Key open questions include:

• Is the $k$22 bound for $k$23-degenerate simple graphs optimal, or can the logarithmic factor be removed?
• For hypergraphs, does a similar sharp threshold for $k$24 hold for general (not only linear or sparse) $k$25-uniforms, and can palette sparsification be generalized?
• For specific planar structures, is there a simple planar $k$26 with $k$27
• How large can the gap between separation/adaptable and single-conflict chromatic numbers be, particularly in linear hypergraphs (Casselgren et al., 17 Sep 2025, Casselgren et al., 11 Jan 2026)?
• Are there deterministic polynomial-time algorithms matching the probabilistic upper bounds, especially for the surface genus case (Bradshaw et al., 2021, Dvořák et al., 2018)?

Further, the structural characterization of graphs (or hypergraphs) with $k$28 for small $k$29 remains incomplete, with only partial results for $k$30 (Casselgren et al., 17 Sep 2025).

7. Connections and Broader Context

Single conflict coloring is positioned at the intersection of list coloring, DP-coloring, and local avoidance constraints. The hierarchy

$k$31

provides a stratified perspective for investigating robustness and flexibility of vertex colorings under adversarial and combinatorial constraints.

Algorithmically, developments link probabilistic combinatorics (the Lovász Local Lemma, sharp thresholds) with distributed and fixed-parameter paradigms. The emergence of sharp transition phenomena in uniform hypergraphs, the impact of palette sparsification, and gaps between list-type parameters are central current research themes (Casselgren et al., 11 Jan 2026, Casselgren et al., 17 Sep 2025, Maus et al., 2020).

Open problems in tractability, chromatic gaps, and explicit characterizations suggest further research directions with implications for structural graph theory, distributed algorithms, and the theory of graph colorings under local and global forbidden configurations.