---
title: Hadwiger's Conjecture in Graph Theory
url: https://www.emergentmind.com/topics/hadwiger-s-conjecture
type: topic
---

# Hadwiger's Conjecture in Graph Theory

Hadwiger’s Conjecture is a foundational open problem in extremal and structural graph theory relating the chromatic number of a finite undirected graph to the containment of clique minors. Its numerous variants, asymptotic relaxations, and hereditary cases have deep connections with graph coloring, graph minors, perfect graphs, and topological methods. The conjecture also has a namesake geometric version (Hadwiger's covering conjecture) in convex geometry, but the focus here is purely on the graph-theoretic statement and its ramifications.

## 1. Formal Statement and Basic Definitions

Hadwiger’s Conjecture states that for every finite simple graph $G$, the chromatic number $\chi(G)$ is at most the Hadwiger number $\eta(G)$, defined as the largest $t$ such that $G$ contains $K_t$ (the complete graph on $t$ vertices) as a minor:

\[
\forall\, G, \quad \eta(G) \geq \chi(G)
\]

Equivalently, a graph with no $K_{t+1}$ minor is $t$-colorable. A graph minor is obtained via a sequence of vertex deletions, edge deletions, and edge contractions [1911.01491]. The variant for odd minors—where contractions must preserve odd cycle parities—is known as the odd Hadwiger’s Conjecture [2312.17130, 2109.02302].

## 2. Exact and Asymptotic Results: Chromatic Bounds vs. Clique Minors

Hadwiger’s Conjecture holds for $t \leq 6$: it is equivalent to the Four-Color Theorem for $t=5$; for $t=6$ the case was resolved combinatorially by Robertson, Seymour, and Thomas. For $t\geq 7$ the conjecture remains open.

Asymptotic bounds are a cornerstone of work on the conjecture. The following upper bounds on the chromatic number for $K_t$-minor-free graphs have been established:

| Bound | Authors | Statement |
|-------|---------|-----------|
| $O\left(t \sqrt{\log t}\right)$ | Kostochka, Thomason | For all $K_t$-minor-free $G$, $\chi(G)\leq O(t\sqrt{\log t})$ |
| $O\left(t (\log t)^{0.354}\right)$ | Norin, Song | Improved the logarithmic exponent [1911.01491] |
| $O\left(t (\log t)^{\beta}\right)$ for any $\beta > 1/4$ | Song (2020) | Further improved bound [2006.14945] |
| $t \cdot 2^{O((\log \log t)^{2/3})}$ | Song (2020) | Exponential-in-double-logarithm bound [2006.14945] |
| $O(t (\log \log t)^6)$ | Song (2020) | Polynomial-in-double-logarithm bound [2006.14945] |

The central qualitative strengthening due to Gerards and Seymour is the Odd Hadwiger’s Conjecture, which asserts that every graph with chromatic number $\geq t$ contains an *odd* $K_t$-minor [2312.17130, 2109.02302].

## 3. Finite, Infinite, and Weak Variants

For infinite graphs, Hadwiger’s Conjecture fails: there exist connected graphs $H$ with $\chi(H)=\omega$ but with no $K_{\omega}$ minor [1212.3093]. The “weak” form (for infinite cardinals $\kappa$):

> If $\chi(G)>\kappa$, then $K_\kappa$ is a minor of $G$.

holds for all infinite $\kappa$, but fails to capture the classical (finite) conjecture exactly [1312.2829].

A notable related form is the defective coloring or partition form: Given $G$ with no $K_{t+1}$ minor, partition $V(G)$ into $t$ classes such that the induced subgraph on each has bounded degree $f(t)$—a relaxation sharp up to maximum degree [1407.5236].

## 4. Structural and Hereditary Classes: Forbidden Subgraphs and Stability Number

The case $\alpha(G) = 2$ (i.e., independence number at most 2) is pivotal. Seymour conjectured that resolving Hadwiger for this class would likely solve the general case [2512.17114]. Substantial progress has been achieved for graphs with forbidden induced subgraphs and small independence number.

- **Forbidden holes:** If $G$ with $\alpha(G) \geq 3$ has no holes (induced cycles) of length between $4$ and $2\alpha(G)-1$, then $h(G)\ge\chi(G)$. For $\alpha(G) \geq 2$, forbidding all holes of length $4$ to $2\alpha(G)$ guarantees the odd Hadwiger bound $oh(G)\ge\chi(G)$ [1607.06718].
- **Specific hereditary classes:** Hadwiger’s Conjecture holds for
  - $\{ \text{co-claw}, \text{co-gem} \}$-free graphs, each admitting a $K_{\chi(G)}$-minor model with all branch sets of size at most 2.
  - $\{ \text{fork}, \text{antifork} \}$-free graphs, where all but possibly one branch set in the $K_{\chi(G)}$-minor model have size at most 2 [2605.28050].
- **Forbidden induced subgraphs and minimum degree:** If $G$ is $W_5$-free and $\alpha(G)\le 2$, or $G$ is $K_8$-free with $\alpha(G)\le 2$ and $\delta(G)\ne 19$, then $h(G)\geq\chi(G)$ [2305.05868, 1901.06985].

For further hereditary classes and inflations (blow-ups) of special graphs, a variety of positive results for the $\alpha(G)=2$ setting have been established [2512.17114].

## 5. Methodologies and Prototype Proofs

Classical and modern approaches employ a blend of extremal combinatorics, coloring, structural graph theory, and minor theory.

- **Chromatic coordinate models:** For $t=8$, decomposing graphs into stacked parallel planes of limited chromatic number (e.g., two $K_4$ minors on separate planes, interconnected to form $K_8$) proves Hadwiger for all 8-colorable graphs [2104.13519].
- **Packing and covering arguments:** Use of cliques, matchings, and partitioning strategies (e.g., “maximal inflations of $C_5$” in $\alpha(G)=2$ settings) facilitate reduction to perfect graphs or forceable minor formation [1901.06985, 2305.05868].
- **Extremal and density bounds:** Application of extremal function results (e.g., Mader’s bound for $K_t$-minors) and iterative coloring via greedy approaches, together with the use of generalized Kempe chains and repair via disjoint paths [2208.07338, 1304.6510].
- **Defective colorings and relaxations:** Partitioning into $t$ parts inducing subgraphs of bounded maximum degree as an approach toward the $t$-coloring goal, with explicit sharpness bounds [1407.5236].
- **Infinite graphs and tree-decompositions:** For infinite chromatic number, equivalence with well-founded tree-decomposition width and colorability [1312.2829].

## 6. Connections to Topological Bounds and Special Constructions

Topological lower bounds on chromatic number, such as the zig-zag number $zig(G)$ and the Dol’nikov–Kříž bound, admit translated Hadwiger-type statements: every graph with $zig(G)\ge t$ contains a $K_{\lfloor t/2 \rfloor + 1}$ odd minor, and if the Dol’nikov–Kříž bound attains $t$, then $G$ admits a $K_t$ as a minor [2312.17130]. For Kneser and Schrijver graphs, these bounds are tight.

Geometric analogues (Hadwiger's covering conjecture for convex bodies) are addressed for specific families like cap bodies in all dimensions $n\geq 3$ [2510.25968], but these geometric versions are distinct from the combinatorial statement.

## 7. Open Cases, Failures, and Future Directions

The original statement for $t\ge7$ remains open—specifically, whether every $K_7$-minor-free graph is 6-colorable or whether $h(G) \geq \chi(G)$ for arbitrarily large chromatic number.

- **Odd Hadwiger equivalence:** Asymptotic bounds of the form $\chi(G)=O(t)$ for $K_t$-minor-free graphs are transferred up to a factor of 2 to the odd-minor variant [2109.02302].
- **Infinite cases:** Hadwiger's Conjecture fails for infinite chromatic number; only a weaker form is true [1212.3093, 1312.2829].
- **Bounded defect partitions:** Approaching the gap via defective coloring and component size, seeking $t$ parts with bounded-size induced components remains open for $t\geq 7$ [1407.5236].
- **Further hereditary classes:** Characterizing all hereditary classes $C$ such that every $G\in C$ satisfies Hadwiger’s Conjecture or its strengthened branch-set models is a major ongoing effort [2512.17114, 2605.28050].
- **Search for counterexamples:** Random triangle-free graph constructions and dense families not precluded by current results are candidate ground for possible counterexamples or further reductions [2512.17114].

Significant advances have come from increasingly precise structure theorems, forbidden subgraphs, and extremal bounds. Nonetheless, neither the core conjecture nor its combinatorial or topological analogues for large $t$ have been fully settled. The problem remains central for graph theorists, with its resolution likely to have far-reaching implications across combinatorics and discrete mathematics.

Source: https://www.emergentmind.com/topics/hadwiger-s-conjecture